marker-assisted selection in wheat - ictsd

iiiContentsAcknowledgementsForewordAbbreviations and acronymsContributorsviviiixxvSection I – Introduction to marker-assisted selection 1Chapter 1Marker-assisted selection as a tool for genetic improvement ofcrops, livestock, forestry and fish in developing countries:an overview of the issues 3John Ruane and Andrea SonninoChapter 2An assessment of the use of molecular markers in developingcountries 15Andrea Sonnino, Marcelo J. Carena, Elcio P. Guimarães,Roswitha Baumung, Dafydd Pilling and Barbara RischkowskySection II – marker-assisted selection in crops – case studies 27Chapter 3Molecular markers for use in plant molecular breeding andgermplasm evaluation 29Jeremy D. Edwards and Susan R. McCouchChapter 4Marker-assisted selection in wheat: evolution, not revolution 51Robert Koebner and Richard SummersChapter 5Marker-assisted selection for improving quantitative traits of foragecrops 59Oene Dolstra, Christel Denneboom, Ab L.F. de Vos and E.N. van LooChapter 6Targeted introgression of cotton fibre quality quantitative trait lociusing molecular markers 67Jean-Marc Lacape, Trung-Bieu Nguyen, Bernad Hau and Marc Giband

ivChapter 7Marker-assisted selection in common beans and cassava 81Mathew W. Blair, Martin A. Fregene, Steve E. Beebe and Hernán CeballosChapter 8Marker-assisted selection in maize: current status, potential,limitations and perspectives from the private and public sectors 117Michel Ragot and Michael LeeChapter 9Molecular marker-assisted selection for resistance to pathogens intomato 151Amalia Barone and Luigi FruscianteSection Iii – marker-assisted selection in livestock – case studies 165Chapter 10Strategies, limitations and opportunities for marker-assistedselection in livestock 167Jack C.M. Dekkers and Julius H.J. van der WerfChapter 11Marker-assisted selection in poultry 185Dirk-Jan de Koning and Paul M. HockingChapter 12Marker-assisted selection in dairy cattle 199Joel Ira WellerChapter 13Marker-assisted selection in sheep and goats 229Julius H.J. van der WerfSection Iv – marker-assisted selection in forestry – case studies 249Chapter 14Marker-assisted selection in Eucalyptus 251Dario GrattapagliaChapter 15Marker-assisted selection in forestry species 283Penny Butcher and Simon Southerton

Section v – marker-assisted selection in fish – case studies 307Chapter 16Possibilities for marker-assisted selection in aquaculture breedingschemes 309Anna K. SonessonChapter 17Marker-assisted selection in fish and shellfish breeding schemes 329Victor MartinezSection Vi – selected issues relevant to applications ofmarker-assisted selection in developing countries 363Chapter 18Marker-assisted selection in crop and livestock improvement:how to strengthen national research capacity and internationalpartnerships 365Maurício Antônio LopesChapter 19Technical, economic and policy considerations on marker-assistedselection in crops: lessons from the experience at an internationalagricultural research centre 381H. Manilal William, Michael Morris, Marilyn Warburton andDavid A. HoisingtonChapter 20Impacts of intellectual property rights on marker-assisted selectionresearch and application for agriculture in developing countries 405Victoria Henson-ApollonioChapter 21Marker-assisted selection as a potential tool for geneticimprovement in developing countries: debating the issues 427Jonathan Robinson and John RuaneChapter 22Marker-assisted selection: policy considerations and options fordeveloping countries 441James D. Dargie

viAcknowledgementsPreparation of this major publication on marker-assisted selection would not havebeen possible without the committed support of a large number of individualsfrom the time that the idea of this publication was initially raised in early 2004.First and foremost among these are the members of the FAO Working Group onBiotechnology, in particular its former Chairperson, Jim Dargie, and his successor,Shivaji Pandey. Three of the members, Devin Bartley (FAO Aquaculture Managementand Conservation Service), Pierre Sigaud (formerly of FAO Forest ResourcesDevelopment Service) and Nuria Urquia (FAO Seed and Plant Genetic ResourcesService), as well as Paul Boettcher (Joint FAO/IAEA Division of Nuclear Techniquesin Food and Agriculture) and Eric Hallerman (Virginia Polytechnic Institute and StateUniversity, United States of America) are particularly thanked for commenting onindividual chapters in the book.Thanks are also extended to Adrianna Gabrielli, Michela Paganini and ChrissiRedfern for their assistance with editorial and layout aspects of this publication.Finally, special thanks go to all the authors for dedicating their time and effortsto the preparation of these chapters in an exemplary and timely fashion and for theirmeticulous attention to detail in the editing process.

viiForewordSince almost the beginning of human civilization, exploiting variation in thecharacteristics of the plant and animal genetic resources that are used for producingfood and other agricultural products through breeding has been at the heart of effortsto increase and diversify agricultural production and productivity, enhance foodsecurity and incomes, and adapt farming to changing environmental conditions andsocial needs. Initially, this was achieved simply by selecting and reproducing preferredindividuals or spontaneous variants, and indeed this practice remains important todayas the basis for producing new generations of cultivated landraces and indigenousbreeds. However, the crops, trees, livestock and fish that are farmed today have arisenlargely from the introduction of scientific breeding at the beginning of the twentiethcentury, with the inclusion of crosses into breeding schemes prior to artificialselection and application of Mendel’s laws of inheritance to improve both simple andquantitative traits providing the foundations for modern genetics.Today, thanks to continuing investments made in research and technologydevelopment, the process of producing improved varieties, clones, breeds and strains ofagriculturally important species has become progressively more accurate, reliable andefficient. Nevertheless, one of the continuing technical constraints to more effectivebreeding is that selecting material with one or a combination of the characteristicsrequired by farmers, foresters, industry and consumers still relies mainly on physicaland agronomic attributes (phenotype). Some of these characteristics are influencedby the environment and are therefore not necessarily a good guide to the actualheritable genetic composition (genotype) of the material in question. Others may notbe visible or may only be detected in mature plants and animals. Others again may bedifficult or very costly to screen, and many characters such as drought tolerance andmilk composition are controlled by a large number of genes whose mode of action aswell as their interaction with each other and with various environmental triggers ismainly unknown. Improving the identification, selection and monitoring of specificcharacters in plants and animals through breeding schemes is therefore a critical needto secure future improvements in genetic resources for food and agriculture.Since the first description of DNA structure over 50 years ago, scientists have madetremendous strides in identifying genes and gene functions, making it increasinglypossible to detect genetic differences (DNA polymorphisms) for traits amongindividual plants and animals in a much more direct way, thereby assisting in theselection of desired traits. The central technology involved is molecular markerassistedselection (MAS), using sequences and/or banding patterns of DNA that havebeen shown through linkage mapping to be located in or near genes that affect thephenotype. These molecular markers can then be used to assist breeders track whether

viiithe specific gene or chromosome segment(s) known to affect the phenotype of interestis present in the individuals or populations of interest.Although the ultimate goal of identifying the location, function and most favourablealleles of each gene through genome sequence and post-genomics research, and thenusing markers to select for economically important genes in breeding programmes,is still decades away, in recent years the use of MAS in agriculture has movedprogressively from theory to practical application. In the process, it has generatedboth high expectations for increasing genetic progress through breeding, and raiseda number of unresolved challenges. These include: selection of the most appropriatemethods and tools for MAS among the many now available for the task at hand,analysing and managing the data produced given the increasing trend towards highthroughputtechniques and the constraints imposed by suboptimal levels of resourcescurrently attached to breeding and science and technology including biotechnology,and dealing with intellectual property rights, especially in developing countries.Since its foundation, FAO has recognized that the biological basis for sustainableagricultural production, fighting hunger and world food security lies in the geneticresources used for food and agriculture. It has also recognized the enormouscontributions that have been made to the improvement of these resources throughboth traditional and more advanced breeding, as well as the ever-increasing role playedby biotechnology in improving breeding processes and products. As a knowledgeorganization, one of FAO’s major roles is to provide its Members and their institutionswith factual, comprehensive and current information relevant to sound stewardshipof crops, livestock, forestry and fisheries, thereby ensuring its availability as a globalpublic good. This book, by providing a comprehensive description and assessment ofthe use of MAS for increasing the rate of genetic gain in crops, livestock, forestry andfarmed fish, including the related policy, organizational and resource considerations,continues the Organization’s tradition of dealing with issues of importance toagricultural and economic development in a multidisciplinary and cross-sectoralmanner. As such it is hoped that the information and options presented and thesuggestions made will provide valuable guidance to scientists and breeders in boththe public and private sectors, as well as to government and institutional policyanddecision-makers.Shivaji PandeyChairpersonFAO Working Group on Biotechnology

ixAbbreviations and acronymsAATFAB-QTLACMVAFLPAIAMBIONETAMMANETAnGRASARECABACBCMNVBCMVBecABGYMVBIO-EARNBLUPbpBSAPsBtBTABYDVCAADPCAPSCBBCBSCBSDCCNcDNACGIARCGMCIAfrican Agricultural Technology FoundationAdvanced backcross QTLAfrican cassava mosaic virusAmplified fragment length polymorphismArtificial inseminationAsian Maize Biotechnology NetworkAfrican Molecular Marker Applications NetworkAnimal genetic resourcesAssociation for Strengthening Agricultural Research inEastern and Central AfricaBacterial artificial chromosomeBean common mosaic necrotic virusBean common mosaic virusBiosciences eastern and central AfricaBean golden yellow mosaic virusEast African Regional Programme and ResearchNetwork for Biotechnology, Biosafety andBiotechnology Policy DevelopmentBest linear unbiased predictionBase pairsBiodiversity Strategies and Action PlansBacillus thuriengensisBos taurus chromosomeBarley yellow dwarf virusComprehensive Africa Agriculture DevelopmentProgrammeCleaved amplified polymorphic sequencesCassava bacterial blightCassava brown streakCassava brown streak diseaseCereal cyst nematodeComplementary DNAConsultative Group on International AgriculturalResearchCassava green miteConfidence interval

CIATCIMMYTCIPCIRADcMCMDCMVCORPOICACRCTDArTDFIDDHDHPLCDMCDNADYDEACMVEBVECECOSOCeQTLESTESTPEUEUCAGENF 1F 2FAOFAO-BioDeCFHBInternational Center for Tropical Agriculture(Centro Internacional de Agricultura Tropical)International Maize and Wheat Improvement Center(Centro Internacional de Mejoramiento de Maíz yTrigo)International Potato Center Centro (Internacional dela Papa)French Agricultural Research Centre for InternationalDevelopment (Centre de coopération internationale enrecherche agronomique pour le développement)Centi-MorganCassava mosaic diseaseCassava mosaic virusColombian Agricultural Research Corporation(Corporación Colombiana de InvestigaciónAgropecuaria)Country reportComputer tomographyDiversity array technologyUnited Kingdom’s Department for InternationalDevelopmentDouble-haploidDenaturing high pressure liquid chromatographyDry matter contentDeoxyribonucleic acidDaughter yield deviationEast Africa cassava mosaic virusEstimated breeding valueEuropean CommissionEconomic and Social Council of the United NationsExpressed gene QTLExpressed sequence tagExpressed sequence tagged polymorphismEuropean UnionEucalyptus Genome NetworkFirst filial generationSecond filial generationFood and Agriculture Organization of the UnitedNationsFAO Biotechnology in Developing CountriesFusarium head blight

xiFIVIMSFNPFSCFSILGABIGASGCAGCPGDPGEGHGISGMOsGRDCGRFAGRMh 2HIPCHWEIACIAPIARCsIBDICARICMVICRISATICSUIFPRIIHNIITAILRIINIFAPIPIPGRIIPRIRRIRRIISAGFood Insecurity and Vulnerability Information andMapping SystemsFunctional nucleotide polymorphismForest Stewardship CouncilFull-sib intercross lineGenome analysis of the plant biological systemGene-assisted selectionGeneral combining abilityGeneration Challenge ProgrammeGross domestic productGenetic engineeringGrowth hormoneGeographical information systemsGenetically modified organismsGrains Research and Development CorporationGenetic resources for food and agricultureGametic relationship matrixHeritabilityHeavily indebted poor countriesHardy-Weinberg equilibriumInterAcademy CouncilInterAcademy PanelInternational agricultural research centresIdentity by descentIndian Council for Agricultural ResearchIndian cassava mosaic virusInternational Crops Research Institute for the Semi-Arid TropicsInternational Council for ScienceInternational Food Policy Research InstituteInfectious haematopoietic necrosisInternational Institute of Tropical AgricultureInternational Livestock Research InstituteNational Institute for Forestry, Agricultureand Livestock Research (Instituto Nacional deInvestigaciones Forestales y Agropecuarias)Intellectual propertyInternational Plant Genetic Resource InstituteIntellectual property rightInternal rate of returnInternational Rice Research InstituteInternational Society for Animal Genetics

xiiISNARISSRITPGRFAJGIKARILDLDLLD-MASLELE-MASLIMSLODMABCMA-BLUPMAIMALDI-TOFMARSMASMBLMCMDMDGMFAMHCmiRNAMLMoDADmRNAMSVMTANARESNARSNDANEPADNGONIRSNPVNUEInternational Service for National AgriculturalResearchInter-simple sequence repeatsInternational Treaty on Plant Genetic Resources forFood and AgricultureJoint Genome InstituteKenya Agricultural Research InstituteLinkage disequilibriumLinkage disequilibrium and linkageLinkage disequilibrium MASLinkage equilibriumLinkage equilibrium MASLaboratory information management systemLogarithm of the odds ratioMarker-assisted back-crossingMarker-assisted best linear unbiased predictionMarker-assisted introgressionMatrix-assisted laser desorption/ionization-time offlightMarker-assisted recurrent selectionMarker-assisted selectionMedical biotechnology laboratoriesMolecular characterizationMarek’s diseaseMillennium Development GoalsMicrofibril angleMajor histocompatibility complexMicroRNAMaximum likelihoodMeasurement of domestic animal diversityMessenger RNAMaize streak virusMaterial Transfer AgreementNational agricultural research and extension systemsNational agricultural research systemsNon-disclosure agreementNew Partnership for Africa’s DevelopmentNon-governmental organizationNear infrared reflectance spectroscopyNet present valueNitrogen use efficiency

xiiiOBMOECDOIEOPVPAGEPBRsPCRPGRFAPICPPBPPDPRSPsPTPVPQPMQTLQTL-NILsQTNR&DRAPDrDNARFLPRGARNARRAS&TSACMVSAGESBMVSCASCARSCNSCSSDS-PAGEsiRNASLS-MASSLUSMASNPOrange blossom midgeOrganisation for Economic Co-operation andDevelopmentWorld Organisation for Animal HealthOpen-pollinated varietyPolyacrylamide gel electrophoresisPlant breeders’ rightsPolymerase chain reactionPlant genetic resources for food and agriculturePolymorphic information contentParticipatory plant breedingPost-harvest physiological deteriorationPoverty reduction strategy papersProgeny testPlant variety protectionQuality protein maizeQuantitative trait loci (or locus)Near isogenic lines for QTLQuantitative trait nucleotideResearch and developmentRandom amplified polymorphic DNARibosomal DNARestriction fragment length polymorphismResistance gene analoguesRibonucleic acidRapid rural appraisalScience and technologySouth African cassava mosaic virusSerial analysis of gene expressionSoil-borne mosaic virusSpecific combining abilitySequence characterized amplified regionSoybean cyst nematodeSomatic cell scoreSodium dodecyl sulphate polyacrylamide gelelectrophoresisShort interfering RNASingle large-scale MASSwedish University of Agricultural SciencesSimple marker analysisSingle nucleotide polymorphism

xivSoW-AnGRSPS AgreementSSCPSSLPSSRSTBSTSSWSWapsTBT AgreementTCTEsTMVToMVTRIPS AgreementTSWVTUATYLCVUNUPOVUSAIDUSDAWECWFSWIPOWRIWSCWTOYMVState of the World’s Animal Genetic ResourcesWTO Agreement on the Application of Sanitary andPhytosanitary MeasuresSingle strand conformation polymorphismSimple sequence length polymorphismSimple sequence repeat (syn. microsatelllite)Septoria tritici blotchSequence-tagged sitesSeed weightSector-wide approachesWTO Agreement on Technical Barriers to TradeTissue cultureTransposable elementsTobacco mosaic virusTomato mottle virusWTO Agreement on Trade-Related Aspects ofIntellectual Property RightsTomato spotted wilt virusTechnology Use AgreementTomato yellow leaf curl virusUnited NationsInternational Union for the Protection of NewVarieties of PlantsUnited States Agency for International DevelopmentUnited States Department of AgricultureWorm egg countWorld Food SummitWorld Intellectual Property OrganizationWorld Resources InstituteWood specific consumptionWorld Trade OrganizationYellow mosaic virus

xvContributorsAmalia BaroneProfessor in Plant GeneticsUniversity of Naples “Federico II”Department of Soil, Plant, Environmentaland Animal Production SciencesVia Universita’ 10080055 Portici, Italye-mail: ambarone@unina.itRoswitha BaumungResearcher, Division Livestock SciencesUniversity of Natural Resources andApplied Life Sciences (BOKU),Gregor Mendelstr. 331180 Vienna, Austriae-mail: roswitha.baumung@boku.ac.atSteve E. BeebeBean BreederBean Project ManagerCentro Internacional de AgriculturaTropical (CIAT)Km 17, Recta Cali-PalmiraA.A. 6713, Cali, Colombiae-mail: s.beebe@cgiar.orgMathew W. BlairGermplasm Specialist/Bean BreederBiotechnology UnitCentro Internacional de AgriculturaTropical (CIAT)Km 17, Recta Cali-PalmiraA.A. 6713, Cali, Colombiae-mail: m.blair@cgiar.orgPenny ButcherConservation Research ScientistKings Park and Botanic GardenBotanic Gardens and Parks AuthorityFraser Avenue, West Perth 6005Western Australia, Australiae-mail: pbutcher@bgpa.wa.gov.auMarcelo J. CarenaAssociate Professor/DirectorCorn Breeding and GeneticsNorth Dakota State UniversityDepartment of Plant SciencesLoftsgard Hall 166Fargo, ND 58105-5051, USAe-mail: marcelo.carena@ndsu.eduHernán CeballosCassava BreedingCassava Breeding ProjectCentro Internacional de AgriculturaTropical (CIAT)Km 17, Recta Cali-PalmiraA.A. 6713, Cali, Colombiaand Universidad Nacional de Colombiae-mail: h.ceballos@cgiar.orgJames D. DargieFormer Director, Joint FAO/IAEADivision of Nuclear Techniques in Foodand AgricultureBrunnstubengasse 432102 Bisamberg, Austriae-mail: j.dargie@aon.at

xviJack C. M. DekkersProfessor and Section LeaderAnimal Breeding and GeneticsDepartment of Animal Science239D Kildee HallIowa State UniversityAmes, IA 50011-3150, USAe-mail: jdekkers@iastate.eduDirk-Jan de KoningPrincipal Investigator – Genetics andGenomicsThe Roslin InstituteRoslin BiocentreRoslin, MidlothianEH25 9PS, UKe-mail: DJ.deKoning@bbsrc.ac.ukChristel DenneboomPlant Research InternationalPO Box 166700 AA WageningenNetherlandse-mail: christel.denneboom@wur.nlAb L.F. de VosPlant Research InternationalPO Box 166700 AA WageningenNetherlandse-mail: ab.devos@wur.nlOene DolstraSenior ScientistPlant Research InternationalPO Box 166700 AA WageningenNetherlandse-mail: oene.dolstra@wur.nlJeremy D. EdwardsResearch AssociateUniversity of Arizona, Department ofPlant Sciences821B Marley Building, 1145 E 4 ST.Tucson, AZ 85721, USAe-mail: jdedw@cals.arizona.eduMartin FregeneSenior Scientist and Cassava GeneticistCentro Internacional de AgriculturaTropical (CIAT)Km 17, Recta Cali-PalmiraA.A. 6713, CaliColombiae-mail: m.fregene@cgiar.orgLuigi FruscianteProfessor in Plant GeneticsUniversity of Naples “Federico II”Department of Soil, Plant, Environmentaland Animal Production SciencesVia Universita’ 10080055 PorticiItalye-mail: fruscian@unina.itMarc GibandResearcher - Cotton Molecular GeneticsCIRAD - UPR Systemes Cotonniers /EMBRAPA AlgodãoRua Osvaldo Cruz 1143Centenario58.107-720 Campina Grande, PBBrazile-mail: marc.giband@cirad.frDario GrattapagliaSenior Scientist and Professor - PlantGenetics/GenomicsBrazilian Agricultural ResearchCorporation – EmbrapaGenetic Resources and BiotechnologyCenterAv. W5 Norte - Final - Plano Piloto70770-900 - Brasília, DF, BrazilandGenomic Science ProgramUniversidade Católica de BrasíliaSGAN 916 Modulo B70790-160 Brasília, DF, Brazile-mail: dario@cenargen.embrapa.br

xviiElcio Perpétuo GuimarãesSenior Officer - Cereals/Crops BreedingFood and Agriculture Organization ofthe United Nations (FAO)00153 Rome, Italye-mail: elcio.guimaraes@fao.orgBernard HauDoctor in Plant breedingCIRADUPR Systèmes CotonniersTA B-10/02Avenue Agropolis34980 Montpellier Cedex 5Francee-mail: bernard.hau@cirad.frVictoria W.K. Henson-ApollonioSenior Scientist, Project ManagerThe CGIAR Central Advisory Serviceon Intellectual Property (CAS-IP)Bioversity InternationalVia dei Tre Denari, 472/a00057 MaccareseItalye-mail: v.henson-apollonio@cgiar.orgPaul M. HockingPrincipal Investigator - Genetics andGenomicsThe Roslin InstituteRoslin Biocentre, RoslinMidlothian EH25 9PS, UKe-mail: Paul.Hocking@bbsrc.ac.ukDavid A. HoisingtonGlobal Theme Leader - BiotechnologyICRISATPatancheru, 502 324 Andhra Pradesh,Indiae-mail: d.hoisington@cgiar.orgRobert KoebnerCrop Breeding and BiotechnologyConsultantCropGen InternationalMockbeggars, Townhouse RoadOld CostesseyNorwich NR8 5BX, UKe-mail: mockbeggars@onetel.comJean-Marc LacapeCotton Molecular BreedingCIRADUMR Développement et Améliorationdes PlantesTA A-96/03Avenue Agropolis34980 Montpellier Cedex 5, Francee-mail: marc.lacape@cirad.frMichael LeeProfessor of Plant Breeding and GeneticsDepartment of AgronomyIowa State UniversityRoom 1553 Agronomy Hall100 Osborn DriveAmes, IA 50011-1010, USAe-mail: mlee@iastate.eduMaurício Antônio LopesResearch Scientist - Plant GeneticsBrazilian Agricultural ResearchCorporation – EmbrapaGenetic Resources and BiotecnologyCenterAv. W5 Norte - Final - Plano Piloto70770-900 - Brasília, DF, Brazile-mail: mlopes@cenargen.embrapa.brVictor MartinezAssistant ProfessorFacultad de Ciencias Veterinarias yPecuarias, Universidad de ChileAvda. Santa Rosa 11735La Pintana, Santiago, Chilee-mail: vmartine@uchile.cl

xviiiSusan R. McCouchProfessorPlant Breeding & Genetics162 Emerson HallCornell UniversityIthaca, NY 14853-1901USAe-mail: srm4@cornell.eduMichael MorrisLead Agricultural EconomistThe World Bank - MSN 6-6021818 Street, NW.Washington, DC 20433USAe-mail: mmorris3@worldbank.orgTrung-Bieu NguyenCotton BreedingCIRADUPR Systèmes CotonniersTA B-10/02Avenue Agropolis34980 Montpellier Cedex 5FranceDafydd PillingEditor State of the World AnGRFood and Agriculture Organization ofthe United Nations (FAO)00153 RomeItalye-mail: dafydd.pilling@fao.orgMichel RagotHead - Genetic Information ManagementSyngenta Seeds12 chemin de l’Hobit31790 Saint-SauveurFrancee-mail: michel.ragot@syngenta.comBarbara RischkowskyAt that time:Coordinator State of the World AnGRFood and Agriculture Organization ofthe United Nations (FAO)00153 Rome, ItalyCurrently:Senior Livestock ScientistInternational Center for AgriculturalResearch in the Dry Areas (ICARDA)PO Box 5466Aleppo, Syrian Arab Republice-mail: b.rischkowsky@cgiar.orgJonathan RobinsonConsultantTick-aho, Joroisniemenkehätie79600 Joroinen, Finlande-mail: jrobinson@tiscali.itJohn RuaneAgricultural Officer (Biotechnology)FAO Working Group on BiotechnologyFood and Agriculture Organization ofthe United Nations (FAO)00153 Rome, Italye-mail: john.ruane@fao.orgBeate D. ScherfAnimal Production OfficerAnimal Genetic Resources GroupFood and Agriculture Organization ofthe United Nations (FAO)00153 Rome, Italye-mail: beate.scherf@fao.orgAnna K. SonessonSenior Scientist, Genetics and BreedingAKVAFORSK (Institute of AquacultureResearch AS)PO Box 50101432 Ås, Norwaye-mail: anna.sonesson@akvaforsk.no

xixAndrea SonninoSenior Agricultural Research OfficerResearch and TechnologicalDevelopment ServiceFood and Agriculture Organization ofthe United Nations (FAO)00153 Rome, Italye-mail: andrea.sonnino@fao.orgSimon SouthertonSenior Research ScientistEnsis GeneticsThe joint forces of CSIRO and SCIONPO Box E4008 Kingston ACT 2604,Australiae-mail: simon.southerton@ensisjv.comRichard SummersCereal Breeding LeadRAGT Seeds LtdThe Maris Centre45 Hauxton RoadTrumpingtonCambridge CB2 2LQ, UKe-mail: RSummers@ragt.frJulius H. J. van der WerfProfessor in Animal Breeding andGeneticsSchool of Rural Science and AgricultureUniversity of New EnglandArmidale, NSW 2351, Australiae-mail: julius.vanderwerf@une.edu.auE.N. (Robert) van LooPlant Research InternationalPO Box 166700 AA Wageningen, Netherlandse-mail: robert.vanloo@wur.nlMarilyn WarburtonSenior Scientist, Molecular GeneticistGenetic Resources Enhancement UnitInternational Maize and WheatImprovement Center (CIMMYT)Apdo. Postal 6-64106600 Mexico, DF, Mexicoe-mail: m.warburton@cgiar.orgJoel Ira WellerResearch ScientistDepartment of Cattle and GeneticsInstitute of Animal SciencesAgricultural Research OrganizationThe Volcani CenterPO Box 6Bet Dagan 50250, Israele-mail: weller@agri.huji.ac.ilH. Manilal WilliamSenior Scientist, Global Wheat ProgramGenetic Resources Enhancement UnitInternational Maize and WheatImprovement Center (CIMMYT)Apdo. Postal 6-64106600 México, DF, Mexicoe-mail: m.william@cgiar.org

Section IIntroduction tomarker-assisted selection

Chapter 1Marker-assisted selection as atool for genetic improvement ofcrops, livestock, forestry and fish indeveloping countries:an overview of the issuesJohn Ruane and Andrea Sonnino

Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishSummaryThis chapter provides an overview of the techniques, current status and issues involved inusing marker-assisted selection (MAS) for genetic improvement in developing countries.Molecular marker maps, the necessary framework for any MAS programme, have beenconstructed for the majority of agriculturally important species, although the densityof the maps varies considerably among species. Despite the considerable resources thathave been invested in this field and despite the enormous potential it still represents, withfew exceptions, MAS has not yet delivered its expected benefits in commercial breedingprogrammes for crops, livestock, forest trees or farmed fish in the developed world. Whenevaluating the potential merits of applying MAS as a tool for genetic improvement indeveloping countries, some of the issues that should be considered are its economic costsand benefits, its potential benefits compared with conventional breeding or with applicationof other biotechnologies, and the potential impact of intellectual property rights (IPRs) onthe development and application of MAS.

Chapter 1 – An overview of the issuesIntroductionThe potential benefits of using markerslinked to genes of interest in breedingprogrammes, thus moving from phenotypebasedtowards genotype-based selection,have been obvious for many decades.However, realization of this potential hasbeen limited by the lack of markers. Withthe advent of DNA-based genetic markersin the late 1970s, the situation changedand researchers could, for the first time,begin to identify large numbers of markersdispersed throughout the genetic material ofany species of interest and use the markersto detect associations with traits of interest,thus allowing MAS finally to become areality. This led to a whole new field ofacademic research, including the milestonepaper by Paterson et al. (1988). This showedthat with the availability of large numbersof genetic markers for their species ofinterest (tomato), the effects and locationof marker-linked genes having an impact ona number of quantitative traits (fruit traitsin their case) could be estimated using anapproach that could be applied to dissectthe genetic make-up of any physiological,morphological and behavioural trait inplants and animals.Most of the traits considered in animaland plant genetic improvement programmesare quantitative, i.e. they are controlled bymany genes together with environmentalfactors, and the underlying genes have smalleffects on the phenotype observed. Milkyield and growth rate in animals or yieldand seed size in plants are typical examplesof quantitative traits. In classical geneticimprovement programmes, selection is carriedout based on observable phenotypesof the candidates for selection and/or theirrelatives but without knowing which genesare actually being selected. The developmentof molecular markers was thereforegreeted with great enthusiasm as it was seenas a major breakthrough promising to overcomethis key limitation. As Young (1999)wrote: “Before the advent of DNA markertechnology, the idea of rapidly uncoveringthe loci controlling complex, multigenictraits seemed like a dream. Suddenly, it wasdifficult to open a plant genetics journalwithout finding dozens of papers seeking topinpoint many, if not most, agriculturallyrelevant genes.” However, despite the considerableresources that have been investedin this field and despite the enormous potentialit still represents, with few exceptions,MAS has not yet delivered its expected benefitsin commercial breeding programmesfor crops, livestock, forest trees or farmedfish in the developed world. In developingcountries, where investments in molecularmarkers have been far smaller, delivery ofbenefits has lagged even further behind.The focus of this chapter is on the use ofmolecular markers for genetic improvementof populations through MAS, includingmarker-assisted introgression. Its aim is toprovide an easily understandable overviewof the techniques, applications and issuesinvolved in the use of DNA markers inMAS for genetic improvement of domesticplant and animal populations in developingcountries. In the next section of the chapter,a brief description of the technical aspectsof molecular markers and MAS is provided.The current status of the application ofMAS in crops, forestry, livestock and fishis then summarized, while the final sectionNote: This chapter is based on the Background Document to Conference 10 (on molecular marker-assistedselection as a potential tool for genetic improvement of crops, forest trees, livestock and fish in developingcountries) of the FAO Biotechnology Forum, 17 November–14 December 2003 (available at www.fao.org/biotech/C10doc.htm).

Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishhighlights issues that might be importantto applications of MAS in developingcountries. Although molecular markers maybe used for a wide range of different tasks,such as to quantify the genetic diversityand relationships within and betweenagricultural populations (e.g. livestockbreeds), to investigate biological processes(such as mating systems, pollen movementor seed dispersal in plants) or to identifyspecific genotypes (e.g. cloned forest trees),these applications are not considered here.Background to MASMolecular markersAll living organisms are made up of cellsthat are programmed by genetic materialcalled DNA. This molecule is made up ofa long chain of nitrogen-containing bases(there are four different bases – adenine [A],cytosine [C], guanine [G] and thymine [T]).Only a small fraction of the DNA sequencetypically makes up genes, i.e. that code forproteins, while the remaining and majorshare of the DNA represents non-codingsequences, the role of which is not yetclearly understood. The genetic material isorganized into sets of chromosomes (e.g.five pairs in Arabidopsis thaliana; 30 pairs inBos taurus [cow]), and the entire set is calledthe genome. In a diploid individual (i.e.where chromosomes are organized in pairs),there are two alleles of every gene – onefrom each parent.Molecular markers should not be consideredas normal genes as they usually donot have any biological effect. Instead, theycan be thought of as constant landmarksin the genome. They are identifiable DNAsequences, found at specific locations of thegenome, and transmitted by the standardlaws of inheritance from one generationto the next. They rely on a DNA assay,in contrast to morphological markers thatare based on visible traits, and biochemicalmarkers that are based on proteins producedby genes.Different kinds of molecular markersexist, such as restriction fragment lengthpolymorphisms (RFLPs), random amplifiedpolymorphic DNA (RAPDs) markers,amplified fragment length polymorphisms(AFLPs), microsatellites and single nucleotidepolymorphisms (SNPs). They maydiffer in a variety of ways – such as theirtechnical requirements (e.g. whether theycan be automated or require use of radioactivity);the amount of time, moneyand labour needed; the number of geneticmarkers that can be detected throughoutthe genome; and the amount of geneticvariation found at each marker in a givenpopulation. The information provided tothe breeder by the markers varies dependingon the type of marker system used. Each hasits advantages and disadvantages and, in thefuture, other systems are likely to be developed.More details on the individual markersystems are provided in Chapter 3.From markers to MASThe molecular marker systems describedabove allow high-density DNA markermaps (i.e. with many markers of known location,interspersed at relatively short intervalsthroughout the genome) to be constructedfor a range of economically importantagricultural species, thus providing theframework needed for eventual applicationsof MAS.Using the marker map, putative genesaffecting traits of interest can then be detectedby testing for statistical associationsbetween marker variants and any trait ofinterest. These traits might be geneticallysimple – for example, many traits for diseaseresistance in plants are controlled by one ora few genes (Young, 1999). Alternatively,

Chapter 1 – An overview of the issuesthey could be genetically complex quantitativetraits, involving many genes (i.e.so-called quantitative trait loci [QTL])and environmental effects. Most economicallyimportant agronomic traits tend tofall into this latter category. For example,using 280 molecular markers (comprising134 RFLPs, 131 AFLPs and 15 microsatellites)and recording populations of ricelines for various plant water stress indicators,phenology, plant biomass, yield andyield components under irrigated and waterstress conditions, Babu et al. (2003) detecteda number of putative QTL for droughtresistance traits.Having identified markers physicallylocated beside or even within genes ofinterest, in the next step it is now possibleto carry out MAS, i.e. to select identifiablemarker variants (alleles) in order to selectfor non-identifiable favourable variants ofthe genes of interest. For example, considera hypothetical situation where a molecularmarker M (with two alleles M1 and M2),identified using a DNA assay, is knownto be located on a chromosome close toa gene of interest Q (with a variant Q1that increases yield and a variant Q2 thatdecreases yield), that is, as yet, unknown.If a given individual in the population hasthe alleles M1 and Q1 on one chromosomeand M2 and Q2 on the other chromosome,then any of its progeny receiving the M1allele will have a high probability (how highdepends on how close M and Q are to eachother on the chromosome) of also carryingthe favourable Q1 allele, and thus wouldbe preferred for selection purposes. On theother hand, those that inherit the M2 allelewill tend to have inherited the unfavourableQ2 allele, and so would not be preferredfor selection. With conventional selectionwhich relies on phenotypic values, it is notpossible to use this kind of information.The success of MAS is influenced bythe relationship between the markersand the genes of interest. Dekkers (2004)distinguished three kinds of relationship:• The molecular marker is located withinthe gene of interest (i.e. within the geneQ, using the example above). In thissituation, one can refer to gene-assistedselection (GAS). This is the mostfavourable situation for MAS since, byfollowing inheritance of the M alleles,inheritance of the Q alleles is followeddirectly. On the other hand, these kindsof markers are the most uncommon andare thus the most difficult to find.• The marker is in linkage disequilibrium(LD) with Q throughout the wholepopulation. LD is the tendency of certaincombinations of alleles (e.g. M1 and Q1)to be inherited together. PopulationwideLD can be found when markersand genes of interest are physicallyvery close to each other and/or whenlines or breeds have been crossed inrecent generations. Selection using thesemarkers can be called LD-MAS.• The marker is not in linkage disequilibrium(i.e. it is in linkage equilibrium[LE]) with Q throughout the whole population.Selection using these markerscan be called LE-MAS. This is the mostdifficult situation for applying MAS.The universal nature of DNA, molecularmarkers and genes means that MAS can,in theory, be applied to any agriculturallyimportant species. Indeed, active researchprogrammes have been devoted to buildingmolecular marker maps and detecting QTLsfor potential use in MAS programmes in awhole range of crop, livestock, forest treeand fish species. In addition, MAS can beapplied to support existing conventionalbreeding programmes. These programmesuse strategies such as: recurrent selection (i.e.

Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishusing within-breed or within-line selection,important in livestock); development ofcrossbreds or hybrids (by crossing severalimproved lines or breeds) and introgression(where a target gene is introduced from,for example, a low-productive lineor breed (donor) into a productive line(recipient) that lacks the target gene (astrategy especially important in plants).See Dekkers and Hospital (2002) for moredetails. MAS can be incorporated into anyone of these strategies (e.g. for markerassistedintrogression by using markers toaccelerate introduction of the target gene).Alternatively, novel breeding strategies canbe developed to harness the new possibilitiesthat MAS raises.Current Status of Applications ofMAS in AgricultureBelow is a brief summary of the currentstatus regarding application of MAS inthe different agricultural sectors. For moredetails, a number of case studies for cropsare presented in Section II of the book andfor livestock, forestry and fish in SectionsIII, IV and V, respectively.CropsThe promise of MAS has possibly beengreeted with the most enthusiasm and expectationin this particular agricultural sector,stimulating tremendous investments in thedevelopment of molecular marker mapsand research to detect associations betweenphenotypes and markers. Molecular markermaps have been constructed for a widerange of crop species. Information on majorplant projects (such as the sequencing of theentire rice genome) can be found at www.ncbi.nlm.nih.gov/genomes/PLANTS/PlantList.html.In a recent review, however, Dekkers andHospital (2002) noted that “as theoreticaland experimental results of QTL detectionhave accumulated, the initial enthusiasmfor the potential genetic gains allowed bymolecular genetics has been tempered byevidence for limits to the precision of theestimates of QTL effects”, and that “overall,there are still few reports of successfulMAS experiments or applications.” Theyreported that marker-assisted introgressionof known genes was widely used in plants,particularly by private breeding companies,whereas marker-assisted introgressionof unknown genes had often proved to beless useful in practice than expected. AsYoung (1999) wrote: “even though markerassistedselection now plays a prominentrole in the field of plant breeding, examplesof successful, practical outcomes arerare. It is clear that DNA markers holdgreat promise, but realizing that promiseremains elusive.”There is also considerable divergencewith respect to the applications of MASamong different crop species. For example,Koebner (2003) highlighted the relativelyfast uptake of MAS in maize compared withwheat and barley, arguing that this largelyreflected the breeding structure. Thus,whereas maize breeding is dominated inindustrialized countries by a small numberof large private companies that produceF 1 hybrids, a system allowing protectionfrom farm-saved seed and competitor use,breeding for the other major cereal speciesis primarily by public sector organizationsand most varieties are inbred pure breedinglines, a system allowing less protection overthe released varieties. Progress in arablecrops is nevertheless quite advanced comparedwith horticultural crop species suchas apples and pears, where development ofmolecular marker maps has been slow andonly few QTL have been detected (Tartarini,2003), even if MAS can potentially be very

Chapter 1 – An overview of the issuesuseful for genetic improvement of suchlong-cycle plants.LivestockAgain, much effort has been put into thedevelopment of molecular marker mapsin this sector. The first reported map inlivestock was for chicken in 1992, whichwas quickly followed by the publication ofmaps for cattle, pigs and sheep. Since then,the search for useful markers has continuedand further species have been targeted,including goat, horse, rabbit and turkey(see www.thearkdb.org/ for the currentstatus regarding some major livestock species).Microsatellite markers have been ofmajor importance.Dekkers (2004) recently reviewed commercialapplications of MAS in livestockand noted that several gene or markertests are available on a commercial basisin different species and for different traits,and that the majority of uses involve GAS,where an important gene (e.g. responsiblefor a congenital defect) has been identifiedor, to a lesser degree, LD-MAS. He pointedout that documentation is poor since,although several genetic tests are available,the extent to which they are used in commercialapplications is unclear, as is themanner in which they are used and whethertheir use leads to greater responses to selection.He concluded that “opportunities forthe application of MAS exist, in particularfor GAS and LD-MAS and, to a lesserdegree, for LE-MAS because of greaterimplementation requirements. Regardlessof the strategy, successful application ofMAS requires a comprehensive integratedapproach with continued emphasis on phenotypicrecording programmes to enableQTL detection, estimation and confirmationof effects, and use of estimates inselection. Although initial expectations forthe use of MAS were high, the current attitudeis one of cautious optimism.”ForestryAs for crops, extensive efforts have beendevoted to construction of molecular markermaps for the major commercial genera,such as eucalypts, pines and acacia. RFLPs,RAPDs, microsatellites and AFLPs havebeen extensively used. The Web site http://dendrome.ucdavis.edu/index.php providesupdated information on the status regardingmolecular marker maps in forestry.Molecular maps have been used to locatemarkers associated with variation in forestrytraits of commercial interest, such asgrowth, frost tolerance, wood properties,vegetative propagation, leaf oil compositionand disease resistance. Since MAS allowsearly selection before traits of interest(e.g. wood quality) are expressed, a majorincentive for using molecular techniquesin tree breeding is to improve the rate ofgenetic gain by reducing the long generationinterval However, Butcher (2003)noted that “MAS has yet to be incorporatedin operational breeding programmes forplantation species” and she referred to thehigh costs of genotyping, the large familysizes required to detect QTL and the lackof knowledge of QTL interactions withgenetic background, tree age and environmentas explanatory factors.In a recent review of biotechnology inforestry, Yanchuk (2002) also highlighted thepotential advantage of early selection usingMAS, but again pointed out that MAS is notyet being applied routinely in tree breedingprogrammes, largely “because of economicconstraints (i.e. the additional genetic gainsare generally not large enough to offset thecosts of applying the technology). Thus it islikely that MAS will only be applied for ahandful of species and situations, e.g. a few

10Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishof the major commercially used pine andEucalyptus species. Molecular markers aretherefore primarily an information tool andare used to locate DNA/genes that can be ofinterest for genetic transformation, or informationon population structure, matingsystems and pedigree confirmation.”FishMolecular marker maps have been constructedfor a number of aquaculture species,e.g. tilapia, catfish, giant tiger prawn,kuruma prawn, Japanese flounder andAtlantic salmon, although their density isgenerally low. Density is high for the rainbowtrout, where the map published in 2003has over 1 300 markers spread throughoutthe genome – the vast majority are AFLPsbut it also includes over 200 microsatellitemarkers (Nichols et al., 2003). Some QTLsof interest have been detected (e.g. for coldand salinity tolerance in tilapia and for specificdiseases in rainbow trout and salmon).In a recent review of MAS in fish breedingschemes, Sonesson (2003) suggested thatMAS would be especially valuable for traitsthat are impossible to record on the candidatesfor selection such as disease resistance,fillet quality, feed efficiency and sexualmaturation, and concluded that MAS is notused in fish breeding schemes today andthat the lack of dense molecular maps is thelimiting factor.ConclusionsMolecular marker maps, the necessaryframework for any MAS programme,have been constructed for the majorityof agriculturally important species but thedensity of the maps varies considerablyamong species. Currently, MAS does notplay a major role in genetic improvementprogrammes in any of the agriculturalsectors. Enthusiasm and optimism remainconcerning the potential contributionsthat MAS offers for genetic improvement.However, this seems to be tempered bythe realization that it may be more difficultand therefore take longer than originallythought before genetic improvement ofquantitative traits using MAS is realized. Theconclusions from the review by Dekkers andHospital (2002) are a good reflection of this:“Further advances in molecular technologyand genome programmes will soon create awealth of information that can be exploitedfor the genetic improvement of plants andanimals. High-throughput genotyping,for example, will allow direct selection onmarker information based on populationwideLD. Methods to effectively analyseand use this information in selection are stillto be developed. The eventual applicationof these technologies in practical breedingprogrammes will be on the basis of economicgrounds, which, along with cost-effectivetechnology, will require further evidence ofpredictable and sustainable genetic advancesusing MAS. Until complex traits can befully dissected, the application of MASwill be limited to genes of moderate-tolargeeffect and to applications that donot endanger the response to conventionalselection. Until then, observable phenotypewill remain an important component ofgenetic improvement programmes, becauseit takes account of the collective effect of allgenes.”Some Factors Relevant toApplying MAS in DevelopingCountriesIn the debate on the role or value of MASas a potential tool for genetic improvementin developing countries, some of thepotential factors that should be consideredare described briefly below, as they mayinfluence applications of the technology.

Chapter 1 – An overview of the issues 11Economic factorsAs with any new technology promisingincreased benefits, the costs of applicationmust also be considered. According toDekkers and Hospital (2002), “economicsis the key determinant for the application ofmolecular genetics in genetic improvementprogrammes. The use of markers in selectionincurs the costs that are inherent to moleculartechniques. Apart from the cost of QTLdetection, which can be substantial, costsfor MAS include the costs of DNA collection,genotyping and analysis.” For example,Koebner (2003) suggested that the currentcosts of MAS would need to fall considerablybefore it would be used widely in wheatand barley breeding. In practice, therefore,although MAS may lead to increased geneticresponses, decision-makers need to considerwhether it may be cost-effective or whetherthe money and resources spent on developingand applying MAS might instead bemore efficiently used on improving existingconventional breeding programmes oradopting other new technologies.Little consideration has been givento this issue. Some results have, however,been published recently from studiesat the International Maize and WheatImprovement Center (CIMMYT) in Mexicoon the relative cost-effectiveness of conventionalselection and MAS for differentmaize breeding applications. One applicationconsidered by Morris et al. (2003) wasthe transfer of an elite allele at a single dominantgene from a donor line to a recipientline. Here, conventional breeding is lessexpensive but MAS is quicker. For situationslike this, where the choice betweenconventional breeding and MAS involvesa trade-off between time and money, theysuggested that the cost-effectiveness ofusing MAS depends on four parameters: therelative cost of phenotypic versus markerscreening; the time saved by MAS; the sizeand temporal distribution of benefits associatedwith accelerated release of improvedgermplasm and, finally, the availability tothe breeding programme of operating capital.They conclude that “all four of theseparameters can vary significantly betweenbreeding projects, suggesting that detailedeconomic analysis may be needed to predictin advance which selection technology willbe optimal for a given breeding project.”In the applications considered byCIMMYT, the costs of developing molecularmarkers associated with the trait ofinterest were not considered, as it wasassumed that they were already available.There is a distinction between developmentcosts (e.g. identifying molecular markers onthe genome, detecting associations betweenmarkers and the traits of interest) andrunning costs (typing individuals for theappropriate markers in the selection programme)of MAS. Development costs canbe considerable, so developing countriesneed to consider whether to develop theirown technology or, alternatively, to importthe technology developed elsewhere, ifavailable.Another aspect to be considered is how toevaluate the economic benefits of MAS. Fora publicly-funded breeding programme, itshould include economic benefits to farmersfrom genetic improvement of their plants oranimals. For private companies on the otherhand, the impacts of using MAS on theirmarket share, and not on rates of geneticimprovement, would be of greatest interest.The economics of MAS are consideredin more detail later, in particular inChapter 19.MAS versus conventional methodsAlthough conventional breeding programmesthat rely on phenotypic records

12Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishhave their limitations, they have shownover time that they can be highly successful.Application of MAS will not occur in avacuum and the potential benefits (genetic,economic, etc.) of using MAS need to becompared with those achieved or expectedfrom any existing conventional breedingprogrammes.In the different agricultural sectors, thisquestion has received much attention fromresearchers. There seems to be general consensusthat the relative success of MAScompared with conventional breeding maydepend on the kind of trait (or traits) to begenetically improved. If the trait is difficultto record or is not routinely recordedin conventional programmes, MAS willoffer more advantages than if it is routinelyrecorded. Similarly, if the trait is sex-limitedor can only be measured late in life thenMAS is favoured, as marker informationcan be used in both sexes and at any age.In considering the merits of MAS versusconventional breeding, it is also importantto keep in mind that the existence ofa strong breeding programme is a prerequisitefor the application of advancedmolecular technologies such as MAS. Insituations where the infrastructure andcapacity are insufficient to support a successfulconventional breeding programme,MAS will not provide a shortcut to geneticimprovement.MAS versus other biotechnologies forgenetic improvementThe relative costs and benefits of applyingMAS should be compared not onlywith conventional breeding but also withthe use of other new technologies that canpotentially improve agricultural populationsgenetically. These include tissue culturein crops and forest trees, reproductivetechnologies (e.g. embryo transfer or cloning)in livestock and triploidization or sexreversalin farmed fish. They also includegenetic modification, a technology that canbe applied to all sectors. Compared withgenetic modification, regulation of MAS,be it at the level of research and development,field testing, commercial release orimport/export of developed products, ismore relaxed; in addition, public acceptanceof the technology is not an issue.Intellectual property rights issuesAs discussed in Conference 6 of the FAOBiotechnology Forum (FAO, 2001), theissue of intellectual property rights (IPRs)is playing an ever greater role in foodand agriculture in developing countries.Participants in that conference, inter alia,suggested that this issue was having a generallynegative influence on the quality ofagricultural research carried out and on thenature of research collaborations betweenthe public and private sector and betweendeveloping and developed countries.It is therefore obvious that IPRs mayalso have an impact on the development andapplication of MAS in developing countries.For example, the AFLP molecular markermapping technique is patented. Molecularmarkers can be patented, although this canoften be overcome by using other markersnear the gene of interest. Individual genescan also be patented. With IPRs, however,there is nevertheless public disclosure ofthe invention or information. Non-disclosureof information, where patents are notsought but the information on markers ordetected QTL is nevertheless kept secret,can also have negative impacts, by denyingdeveloping countries access to potentiallyuseful information.More details on IPRs and MAS can befound in Chapter 20.

Chapter 1 – An overview of the issues 13REFERENCESBabu, R.C., Nguyen, B.D., Chamarerk, V., Shanmugasundaram, P., Chezhian, P., Jeyaprakash, P.,Ganesh, S.K., Palchamy, A., Sadasivam, S., Sarkarung, S., Wade, L.J. & Nguyen, H.T. 2003.Genetic analysis of drought resistance in rice by molecular markers: association between secondarytraits and field performance. Crop Sci. 43: 1457–1469.Butcher, P.A. 2003. Molecular breeding of tropical trees. In A. Rimbawanto & M. Susanta, eds. Proc.Int. Conf. Adv. in Genetic Improvement of Tropical Tree Species, 1–3 October 2002. Centre forForest Biotechnology and Tree Improvement, Yogyakarta, Indonesia.Dekkers, J.C.M. 2004. Commercial application of marker- and gene-assisted selection inlivestock: strategies and lessons. J. Anim. Sci. 82: E313–E328 (available at http://jas.fass.org/cgi/reprint/82/13_suppl/E313)Dekkers, J.C.M. & Hospital, F. 2002. The use of molecular genetics in the improvement of agriculturalpopulations. Nature Revs. Genet. 3: 22–32.FAO. 2001. Agricultural biotechnology for developing countries – results of an electronic forum, byJ. Ruane & M. Zimmermann. FAO Research and Technology Paper No. 8. Rome (also available atwww.fao.org/docrep/004/Y2729E/Y2729E00.htm).Koebner, R. 2003. MAS in cereals: green for maize, amber for rice, still red for wheat and barley. InProc. Int. Workshop on Marker-Assisted Selection: A Fast Track to Increase Genetic Gain in Plantand Animal Breeding? (available at www.fao.org/biotech/docs/Koebner.pdf).Morris, M., Dreher, K., Ribaut, J-M. & Khairallah, M. 2003. Money matters (II): costs ofmaize inbred line conversion schemes at CIMMYT using conventional and marker-assistedselection. Mol. Breed.11: 235–247.Nichols, K.M., Young, W.P., Danzmann, R.G., Robison, B.D., Rexroad, C., Noakes, M., Phillips, B.,Bentzen, P., Spies, I., Knudsen, K., Allendorf, F.W., Cunningham, B.M., Brunelli, J., Zhang, H.,Ristow, S., Drew, R., Brown, K.H., Wheeler, P.A. & Thorgaard, G.H. 2003. A consolidatedlinkage map for rainbow trout (Oncorhynchus mykiss). Anim. Genet. 34: 102–115.Paterson A.H., Lander, E.S., Hewitt, J.D., Peterson, S., Lincoln, S.E. & Tanksley, S.D. 1988.Resolution of quantitative traits into Mendelian factors by using a complete linkage map of restrictionfragment length polymorphisms. Nature 335: 721–726.Sonesson, A.K. 2003. Possibilities for marker-assisted selection in fish breeding schemes. In Proc. Int.Workshop on Marker-Assisted Selection: A Fast Track to Increase Genetic Gain in Plant and AnimalBreeding? (available at www.fao.org/biotech/docs/Sonesson.pdf).Tartarini, S. 2003. Marker-assisted selection in pome fruit breeding. In Proc. Int. Workshop onMarker-Assisted Selection: A Fast Track to Increase Genetic Gain in Plant and Animal Breeding?(available at www.fao.org/biotech/docs/Tartarini.pdf).Yanchuk, A. 2002. The role and implications of biotechnology in forestry. Forest Genetic Resources30: 18–22 (available at www.fao.org/docrep/005/Y4341E/Y4341E06.htm).Young, N.D. 1999. A cautiously optimistic vision for marker-assisted breeding. Mol. Breed. 5:505–510.

Chapter 2An assessment of the use ofmolecular markers indeveloping countriesAndrea Sonnino, Marcelo J. Carena, Elcio P. Guimarães,Roswitha Baumung, Dafydd Pilling and Barbara Rischkowsky

16Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishSummaryFour different sources of information were analysed to assess the current uses of molecularmarkers in crops, forest trees and livestock in developing countries: the FAO Biotechnologyin Developing Countries (FAO-BioDeC) database of biotechnology in developing countries;country reports evaluating the current status of applied plant breeding and relatedbiotechnologies; country reports on animal genetic resources management for preparingthe First Report on the State of the World’s Animal Genetic Resources (SoW-AnGR);and the results of a questionnaire survey on animal genetic diversity studies. Even if stilllargely incomplete, the current data show that molecular markers are widely used for plantbreeding in the developing world and most probably their use will increase in the future.In the animal sector the use of molecular markers seems less developed and limited orabsent in most developing countries. Major differences exist among and within regionsregarding the application of molecular marker techniques in plant and animal breeding andgenetics. These can be explained by the relatively high investments in infrastructure andhuman resources necessary to undertake research in these fields. The spectrum of applicationof molecular markers in crop plants is quite wide, covering many plants relevant tothe enhancement of food security, but other important plant species are still neglected. Thepractical results of marker-assisted selection (MAS) in the field are disappointingly modest,possibly due to: low levels of investment; limited coordination between biotechnologistsand practical breeders; instable, non-focused or ill-addressed research projects; and the lackof linkages between research and farmers. Partnerships between developed and developingcountries may be a means of better realizing the potential of molecular marker techniquesfor improving both animal and crop production.

Chapter 2 – An assessment of the use of molecular markers in developing countries 17IntroductionAssessments relating to the use of molecularmarkers in crop plants are based on twosources of information: (i) FAO-BioDeC,a searchable database of biotechnologyproducts and techniques in use and in thepipeline in developing and transition countries(available at www.fao.org/biotech/inventory_admin/dep/default.asp); and(ii) FAO country reports produced bynational agricultural research systems(NARS) as part of a survey of country informationand trends in resources allocated forapplied plant breeding and related biotechnology,with the aim of raising awareness,evaluating opportunities for investmentand designing national, regional and/orglobal strategies to strengthen the capacityof national plant breeding programmes(Guimarães, Kueneman and Carena, 2006).As the FAO-BioDeC database containslittle information on the use of molecularmarkers in relation to animals, it is evenmore difficult to give a comprehensiveoverview of the situation with respect tolivestock in developing countries than it isfor crops. However, information on the useof molecular markers was drawn from thecountry reports on animal genetic resources(AnGR) management submitted to FAOas part of the preparation of the FirstReport on the State of the World’s AnimalGenetic Resources (SoW-AnGR) and froma questionnaire survey on genetic diversitystudies. The country reports covered a widevariety of aspects of AnGR managementand contain only quite general informationabout the role of molecular techniques. Thequestionnaire survey looked specifically atthe use of molecular markers in livestockgenetic diversity studies and was directedto researchers involved in such studies. Assuch, it gives an indication of where geneticdiversity studies are being undertaken andwhich markers are primarily used, but itdoes not provide a complete picture.While this book focuses on the useof markers to assist in genetic selection(MAS), it is often difficult to obtain specificinformation on the extent to which markersare used for this purpose in developingcountries. For this reason, some of the datapresented in this chapter cover the overalluse of molecular markers in developingcountries and do not allow discriminationbetween molecular markers used forselection from uses for other purposes,such as the descriptive studies of geneticdiversity within populations or geneticdistance between populations. Other datapresented here describe the use of molecularmarkers for measuring genetic diversityonly. In this case, the information can beconsidered as an indicator of the humancapacity and infrastructure available foruse of markers in MAS. For these reasons,and due to the incomplete nature of someof the information available, this overviewshould be considered preliminary, but stillmeaningful.FAO-BioDeCAt the time of writing (September 2006),FAO-BioDeC includes 2 336 entries relatedto crops and 829 entries related toforest trees. The database currently covers74 developing countries, including countrieswith economies in transition.No quantitative information is availableconcerning the human capacity or fundinginvolved in any research initiative. Activitiescarried out in developed countries or atinternational research centres, such as thosethat are part of the Consultative Groupon International Agricultural Research(CGIAR), are not considered.To compile the data in FAO-BioDeC,several sources of information were

18Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishTable 1Number of research initiatives utilizing genetic markers in the crop and forestry sectors sorted bytype of markersMarkers Crop Forestry TotalRFLP 61 9 70RAPD 158 15 173SSRs/Microsatellites 68 19 87AFLP 65 3 68Isozymes 2 50 52Chloroplast DNA markers 0 11 11rDNA (ribosomal DNA sequences) 0 4 4Other or not specified 135 77 212Total 489 188 677Table 2Number of research initiatives utilizing genetic markers in the crop and forestry sectors according tothe development stage of the technique or productPhase Crop Forestry TotalExperimental phase 344 179 523Field tests 107 8 115Commercial phase 4 1 5Unspecified 34 0 34Total 489 188 677consulted (for a complete description seeFAO, 2005). In particular, information onplant biotechnology products and techniqueswas gathered from a survey undertakenin Latin America by the InternationalService for National Agricultural Research(ISNAR) and from country biotechnologystatus assessment reports prepared forFAO in South and Southeast Asia, Africaand transition countries in Eastern Europe.Other information was obtained fromcountry reports and published literature.The initial biotechnology applicationdata obtained was classified on a country/regional/continental basis, by species, traitanalysed or technique used, and by whetherthe application was in the research or fieldtesting phases or was already commerciallyreleased.FAO-BioDeC currently contains677 entries related to the use of molecularmarker techniques, 489 of which areassociated with crop plants and 188 withforest trees. Table 1 suggests that earlygeneration DNA-based molecular markerssuch as randomly amplified polymorphicDNAs (RAPDs) are more widely used thanthe more recently developed markers, e.g.amplified fragment length polymorphisms(AFLPs), while isozymes are still largelyused in the forestry sector.Only in five cases have the research initiativesreported reached the final stage ofdevelopment, giving rise to commercializedproducts (Table 2). These are one variety ofan unspecified ornamental plant released inBrazil; one variety of rice commercializedin Indonesia; one strain of Rhizobium etli,the soil bacterium inducing the formationof nitrogen-fixing nodules on the roots ofa common bean obtained in Mexico; onerice variety containing pyramided genes forbacterial leaf blight resistance obtained inthe Netherlands Antilles; and one varietyof an unspecified forest tree in Burundi. In115 cases (107 in the crop sector and eight

Chapter 2 – An assessment of the use of molecular markers in developing countries 19Table 3Number of research initiatives utilizing genetic markers in the crop and forestry sectors according togeographical originRegion Crop Forestry TotalAfrica 52 17 69Asia and Pacific 98 103 201Europe (transition countries) 42 13 55Latin America and Caribbean 249 55 304Near East and North Africa 48 0 48Total 489 188 677for forest trees), the research initiativeshave reached the field test stage, while in523 cases (344 of which are related to thecrop sector), they are at earlier stages.The use of molecular markers is widespreadin Latin America and the Caribbeanwith molecular research being reportedfrom ten countries. Special emphasis ison the crop sector and includes applicationson Andean local roots and tubers,sugar cane, rice, cocoa, banana, bean andmaize (Table 3). In the Asia and Pacificregion, research activities with molecularmarkers focus on forest trees, sugar cane,rice, jute, banana, coconut and wheat. TheFAO-BioDeC database shows that, whileresearch involving molecular markers inAfrica is under way in only a few countriesincluding Ethiopia, Nigeria, SouthAfrica and Zimbabwe, the crops understudy range from traditional commoditiesto tropical fruits. Molecular research in theNear East and North Africa is reportedfor only six countries and focuses on datepalm, durum and bread wheat, rice, barleyand olive trees. In transition countries ofEastern Europe, molecular markers targetseveral crop plants including wheat, maize,pulses, vegetables and tobacco across sevencountries.Table 4 shows that most attentionfocuses on cereals, especially durumand bread wheat, barley, maize and rice.Other important cereal or pseudo-cerealspecies such as sorghum, amaranthus andTable 4Number of research initiatives utilizing geneticmarkers according to the crop of applicationCrop groupNumber of projectsCereals and pseudo-cereals 134Pulses 54Root and tubers 51Fruit trees 53Vegetables 29Industrial crops 74Fodder crops 16Aromatics 5Other or not specified 73Total 489buckwheat receive less attention and noresearch initiatives are reported for teffor millets. Among the pulses, molecularresearch projects are reported for beans(18), chickpea (5), cowpea (9) and soybean(7) and little or no attention is dedicatedto lentil, pigeon pea, faba bean and otherlocally important leguminous plants such asbambara groundnut. Among root and tubercrops, potato, sweet potato and cassavaattract the most research effort involvingmolecular markers, but some research is alsoundertaken on Andean roots and tubers.Few or no records are available for root andtuber species important for food securityin many developing countries such as yam,taro (or dasheen), cocoyam and other aroids.Research on fruit trees involving molecularmarkers includes tropical fruit trees such asbanana, cocoa, coconut and papaya, as wellas plants more typical of temperate climatessuch as strawberry and apple, while less

20Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishresearch was reported for citrus, mango,pineapple and many other fruit trees largelycultivated in developing countries. Severalresearch initiatives are applying molecularmarkers to industrial crop species, such assugar cane, cotton, rubber, jute, coffee, flaxand oil palm.FAO plant breeding and relatedbiotechnology capacityassessmentIn 2002, a draft questionnaire was designedto gather country information on resourceallocation trends in plant breeding and biotechnologyrelated activities. Later in thesame year, a group of experts includingrepresentatives from CGIAR centres, thepublic and private sectors and non-governmentalorganizations (NGOs), met atFAO headquarters to discuss the natureof the information to be collected and theprocedure for its collection. This resultedin a questionnaire being developed andsent to all public and private applied plantbreeding programmes as well as to biotechnologylaboratories in developing countriesand countries in transition. Among otherissues, the survey gathered informationon the number of full-time equivalentplant breeders and biotechnologists availableduring each five-year period beginningfrom 1985. The questionnaire also requestedinformation concerning trends of resourcesallocated to biotechnology as well as togermplasm improvement (pre-breeding),line development and line evaluation.One of the objectives of the survey wasto assess the gap between biotechnologytools and their successful deployment inapplied breeding programmes (Guimarães,Kueneman and Carena, 2006). The surveytherefore also concentrated on priorities forbreeding, potential international support tostrengthen national breeding programmes,the number of varieties released and thefactors that are most likely to limit thesuccess of applied plant breeding programmes,including the current status ofbiotechnology. The work of gatheringthe information and preparing a technicalreport on the current status of nationalplant breeding and related biotechnologywas assigned to a well-known and respectednational plant breeding scientist. This hasbeen the key to identifying gaps in order todevelop strategies for strengthening effortsdirected at the sustainable use of plantgenetic resources for food and agriculture(PGRFA) in national programmes.For the purposes of this chapter, biotechnologydata were gathered from25 countries to complement the preliminaryassessments based on FAO-BioDeCon the use of molecular markers in developingcountries (Table 5). The data gatheredindicate that tissue culture is the mostcommon biotechnology technique as it wasused in 88 percent of all cases, followedby MAS (44 percent), the double-haploidtechnique (32 percent), interspecificcrosses (28 percent), molecular characterization(24 percent) and genetic engineering(12 percent).Applications of molecular markersinclude a number of categories withinbiotechnology such as MAS, molecularcharacterization, facilitating genetic engineeringand tracking desirable chromosomesegments when making wide crosses (e.g.interspecific crosses). The results in Table 5suggest that molecular markers might bean integral part of developing countryagricultural efforts. MAS seems to be thesecond most utilized biotechnology toolapplied after tissue culture, implying thatemphasis should be given to the developmentof molecular markers to make selectionmore efficient. However, rapid and efficient

Chapter 2 – An assessment of the use of molecular markers in developing countries 21Table 5Biotechnology applications in plant genetic resources for food and agriculture in use in25 developing countriesCountry TC MAS IC DH MC GEAlgeria X 1 X X X N 2 NAngola X N N N N NArmenia X N X X X NCameroon X X N N N NCosta Rica X N X N X XDominican Republic N N N N N NEthiopia X X N X N NGeorgia X X X N X NGhana X X N N N NMali X N N N N NKenya X X X X N XMalawi X N N N N NMoldova X N N N N NMozambique N N N N N NNicaragua X X N N X NNiger X X N N N NNigeria X X X X X NSenegal X N X X N NSierra Leone X N N N N NSri Lanka N N N N N NSudan X N N N N NTunisia X X N X N NUzbekistan X N N X X NZambia X N N N N NZimbabwe X X N N N X1One or more institutions in the country are using the tool. However, this does not measure its impact.2Not in use.TC = tissue culture; MAS = marker-assisted selection; IC = interspecific crosses; DH = double-haploid technology;MC = molecular characterization; GE = genetic engineeringadvancement of plant breeding efforts mightnot be achieved through MAS because ofthe complexity encountered in multitraitand multistage selection for economicallyimportant traits. Consequently, today inthe developed world, molecular markersdo not have a prominent role in breedingprogrammes (Hallauer, 1999).Use of molecular techniques inAnGR managementFAO invited 188 countries to participatein the preparation of the First Reporton the SoW-AnGR. One hundred andsixty-nine country reports (CR) on AnGRwere submitted (available at www.fao.org/dad-is/).The countries were offered guidelinesfor the preparation of the country reports,one section of which was to be devoted toreviewing the state of national capacitiesand assessing future capacity buildingrequirements (FAO, 2001). Countries wereassigned to seven regions on the basisof the regional classification establishedby FAO for the purpose of preparingthe SoW-AnGR. This analysis considered148 country reports available by July 2005,of which 42 were from Africa, 25 fromAsia, 39 from Europe and the Caucasus,22 from Latin America and the Caribbean,7 from the Near and Middle East, 2 fromNorth America and 11 from the SouthwestPacific (Pilling et al., 2007).

22Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishTable 6Use of molecular markers reported in country reports on AnGR managementRegionNumberprovidinginformationReporting useof molecularmarkers%Number withinformation onspeciesReporting use of molecular markersIn cattle%In other species%Europe 29 83 18 89 100Africa 29 14 3 100 33Asia 16 50 7 86 100Latin America and the Caribbean 15 73 9 78 89Southwest Pacific 9 11 0 - -North America 2 100 1 100 100Near and Middle East 5 40 2 0 100Not surprisingly, the information providedby the country reports indicates thatthere is a large gap between developed anddeveloping countries in terms of capacity toutilize molecular markers for the study andmanagement of AnGR (Table 6). Comparedwith other developing regions, a higherpercentage of countries from Asia andLatin America and the Caribbean reportedtheir use. In Africa, the Southwest Pacific(excluding Australia), the Near and MiddleEast, and Eastern Europe and the Caucasus,very few countries report the use of thesetechnologies, the prominent exception inthe last case being Ukraine which hascarried out molecular characterization andgenetic distance studies on a number oflivestock species (CR Ukraine, 2004).In Africa, only four countries describethe existence of characterization or geneticdistance studies based on the use of molecularmarkers and in all cases the studies relate tolocal breeds. One country report indicatesthat local breeds of goat, pig and chickenare the subject of molecular characterizationcarried out abroad. In no case is the use ofMAS reported from this region.Excluding Japan, seven Asian countries(out of 15 providing information onwhether or not the technologies are used)report molecular marker studies, of whichfive specify genetic distance studies and onementions research into MAS (CR Malaysia,2003). A range of species are the subjectof molecular characterization, the mostcommon being cattle, chickens, sheep, goatsand pigs; however, some studies involvingbuffaloes, ducks, horses, camels or deerare also reported. Systematic studies ofAsian breeds are being conducted by theSociety for Research on Native Livestockin Japan, including analysis of genetic relationshipsbased on mitochondrial DNApolymorphisms and other DNA markers(CR Japan, 2003).In Latin America and the Caribbean,11 countries out of the 15 that providedinformation indicate some use of molecularmarkers. Among nine countries providinginformation on the species involved inmolecular characterization studies, sevenmentioned cattle while smaller numbersmention sheep, pigs, chickens, horses,goats, buffaloes, llamas, alpacas, vicuñasor guanacos. Several countries indicate theinclusion of locally adapted breeds in suchstudies, but there was little indication thatmolecular markers have been incorporatedwithin breeding programmes. However,the report from Colombia (2003) notedthe potential significance of MAS programmesfor utilizing the genes of theBlanco Orejinegro cattle breed, which isreported to show resistance to brucellosisand which has been the subject of molecularcharacterization.

Chapter 2 – An assessment of the use of molecular markers in developing countries 23Apart from Australia, no countries inthe Southwest Pacific region report the useof molecular markers.In the Near and Middle East one report(CR Jordan, 2003) refers to molecularcharacterization and genetic distancestudies in indigenous goats, while another(CR Egypt, 2003) notes that molecularstudies of buffalo, sheep and goats hadrecently been initiated with the aid ofregional and international organizations.Survey on the use of molecularmarkers in genetic distancestudies in livestockMore specific and detailed information onthe use of molecular markers in AnGRresearch was obtained from a questionnairestudy launched in 2003. One hundredand thirty-two questionnaires were sentout via e-mail to research teams that hadbeen involved in genetic distance studiesduring the past ten years. The researcherswere identified through a literature searchand enquiry via several Internet discussiongroups. The points covered in thesurvey were: number of breeds and samplesizes; number and type of markers used;additional breed information such as phenotypictraits or geographic spread; andthe mathematical and statistical methodschosen for measuring genetic distance. Thestudy also aimed to verify the degree offamiliarity and acceptance of measurementof domestic animal diversity (MoDAD) recommendations,which had been proposedas standards for genetic diversity studiesby the International Society for AnimalGenetics (ISAG) and FAO about ten yearsearlier (FAO, 1998a; b). Compliance withthe recommendations was seen as importantas it would enable the compilationof results from different genetic distancestudies.Table 7Number of countries where samples werecollected for AnGR genetic distance studiesFAO regionNumber of countriesAfrica 13Asia and the Pacific 19Europe 37Latin America and the Caribbean 10Near East 9North America 2Total 93Information on 87 genetic distance studieswas obtained from 57 researchers. Thestudies covered breeds from 13 mammalianand avian species and investigated samplesfrom 93 countries; the largest number ofcountries was in Europe, followed by thosein Asia and the Pacific (Table 7). Most ofthe studies focused on ruminants. The sizeof the projects varied between one and120 breeds originating from up to 33 countries.However, a large number of nationalprojects focused on breeds within a specificcountry or region. There were also a fewlarge international projects involving cattleand goats (Table 8). A smaller number ofpig and chicken projects were implemented.No feedback was received regarding breedsof llamas, ducks, turkeys or geese.With regard to compliance with the recommendationsof the FAO/ISAG advisorygroup, 95 percent of all projects aimed tofulfil the minimum requirement of sampling25 animals per breed. Although microsatellitemarkers were used in 90 percent of thestudies, in only 23 percent were all markerstaken from the recommended marker list.In about 57 percent of studies some recommendedmicrosatellites were used. Thedegree of acceptance of the recommendationswas highest in pigs and lowest inchickens. More detailed information on theresults is given by Baumung, Simianer andHoffmann (2004) and FAO (2004).

24Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishTable 8Number of projects and countries in which samples were collected according to animal species andFAO regionsSpeciesNumber ofprojectsNumber ofcountriesFAO regionBuffalo 3 9 Africa, Asia and the Pacific, Europe, Latin America and the CaribbeanCattle 24 40 Africa, Asia and the Pacific, Europe, Latin America and the CaribbeanGoat 11 28 Africa, Asia and the Pacific, Europe, Latin America and the CaribbeanSheep 19 56 Africa, Asia and the Pacific, Europe, Latin America and the CaribbeanPig 6 19 Africa, Asia and the Pacific, EuropeAss 1 1 EuropeHorse 5 25 Africa, Asia and the Pacific, Europe, Latin America and the Caribbean,North AmericaBactrian camel 1 2 Asia and the PacificDromedary 2 7 Africa, Near EastAlpaca 3 2 Near East, Latin America and the CaribbeanRabbit 1 19 Africa, Asia and the Pacific, EuropeChicken 8 34 Africa, Asia and the Pacific, Europe, Latin America and the Caribbean,Near EastYak 2 8 Asia and the Pacific, Europe, Near EastConclusionsEven if still largely incomplete, the currentdata allow some general conclusions tobe drawn regarding the use of molecularmarkers in agricultural research anddevelopment in developing countries.Molecular markers are widely utilizedin the plant production sector of the developingworld even if the present uptake ofmolecular marker technologies does notreflect their actual potential. It might thereforebe speculated that a significant increasein their utilization might be expected in thenear future. However, it is recommendedthat each technique is carefully assessed forits actual potential for improving the efficiencyof plant breeding and germplasmcharacterization. Until this is demonstrated,the use of molecular markers would bea costly investment with limited returns.Publishing all marker research that has notbeen successful is also strongly encouragedin order to avoid potential failures and/orimporting inappropriate technologies fromdeveloped countries.Major differences exist between regions(and within regions) regarding the applicationof molecular marker techniques in plantbreeding and genetics. While some countrieshave developed quite extensive researchprogrammes, vast geographical areas, particularlyin Africa, remain excluded fromthese technological advancements or cancount only on minimal activities. This can beexplained by the relatively high investmentsin infrastructure and human resources necessaryto undertake research in this field.High costs can also be indicated as a causeof the low technological level of geneticmarker research in many countries, whichfocus on isozymes or on restriction fragmentlength polymorphisms (RFLPs) andhave not yet adopted the more advancedpolymerase chain reaction (PCR)-basedmarkers. However, the life span of PCRbasedmarkers is very short and it might bebetter to wait until improved markers suchas single nucleotide polymorphisms (SNPs)become available. The spectrum of applicationof molecular markers to crop plantsin developing countries is quite wide andcovers many plant species that are relevantfor the enhancement of food security orfor the improvement of farmers’ incomes

Chapter 2 – An assessment of the use of molecular markers in developing countries 25in tropical areas. However, other importantplant species are still neglected by theongoing research initiatives.According to the data reported in FAO-BioDeC, only five products obtainedthrough the use of molecular markers havebeen commercially released to date in developingcountries. Even if more commercialproducts have been released but are missingfrom the database, such as those reported byToenniessen, O’Toole and DeVries (2003)or others obtained by the internationalagricultural research centres or the privatesector, the totality of practical resultsobtained from using molecular markersis disappointingly modest compared withthe declared potential of the approach.The reasons for the poor results to dateare multiple and include: the low level ofinvestments in both biotechnology researchand applied plant breeding; the limitedcoordination between biotechnology laboratoriesand plant breeding programmes;managerial and political frailties leadingto instable, unfocused or ill-addressedresearch projects; legal, infrastructural ortechnical weaknesses of the seed productionand commercialization systems; andthe lack of linkages between research andpractical application of research productsby farmers.Applied plant breeding should continueto be the foundation for the application ofmolecular markers. Focusing useful moleculartechniques on the right traits will builda strong linkage between genomics andplant breeding in order to produce new andbetter cultivars. Therefore, more than ever,there is the need for better communicationand cooperation among scientists in plantbreeding and biotechnology. Public plantbreeding and biotechnology programmesin developing countries are being seriouslyeroded through lack of funding.This loss of public support affects breedingcontinuity and objectivity and, equallyimportantly, the training of future plantbreeders and biotechnologists and the utilizationand improvement of plant geneticresources currently available. The fact thatpoor farmers rely on public and privatebreeding institutions for solving long-termchallenges should influence policy-makersto reverse the trend of reduced funding.Cooperation between industry and publicinstitutions is a promising approach tofollow. Ensuring strong applied breedingprogrammes incorporating the applicationof molecular markers will be essential inensuring the sustainable use and enhancementof plant genetic resources.AnGR management shows a similarpattern to the use of MAS in plant breedingmanagement in terms of the differences thatexist among regions in the use of molecularmarker techniques. Within several regionsthere are also differences between more andless developed countries. The reasons aresimilar to those mentioned above, namelya lack of financial, human and technicalresources. In particular, human capacitiesin animal genetics and breeding are muchsmaller than those existing in the cropsector. Consequently, the use of moleculartechniques to evaluate genetic resources, toplan conservation efforts, or to facilitate theachievement of desired breeding objectivesis limited or absent in most developingcountries.Nevertheless, country reports expresseda strong desire to develop greater capacityto carry out molecular studies of nationalAnGR, and the responses to the FAOquestionnaire also indicated a high level ofinterest in doing so. For the near future,microsatellite loci will remain the mostuseful type of genetic marker for genetic distancestudies and for genetic improvement

26Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishprogrammes but SNPs were singled outas promising markers for the future. Withpartnerships between developed and developingcountries within or across regions,genetic diversity studies may be a means ofrealizing the potential of molecular markertechniques to improve decision-making onbreed development and the prioritizationof breeds for conservation programmes.The successful application of MAS inanimal breeding necessitates a high levelof expenditure in terms of establishmentand maintenance costs and requires skilledhuman resources, equipment, laboratoriesand supportive infrastructure. As such, thecost-effectiveness of these strategies has tobe carefully evaluated before promotingthem in resource-poor environments.ReferencesBaumung, R., Simianer, H. & Hoffmann, I. 2004. Genetic diversity studies in farm animals – asurvey. J. Anim. Breed. Genet. 121: 361-373.FAO. Country reports on the state of animal genetic resources (available at www.fao.org/dad-is/).FAO. 1998a. Secondary guidelines for development of National Farm Animal Genetic ResourcesManagement Plans. Measurement of domestic animal diversity (MoDAD): original Working Groupreport. Rome.FAO. 1998b. Secondary guidelines for development of National Farm Animal Genetic ResourcesManagement Plans. Measurement of domestic animal diversity (MoDAD): recommendedmicrosatellite markers. Rome.FAO. 2001. Preparation of the first report on the state of the world’s animal genetic resources.Guidelines for the preparation of CRs. Rome.FAO. 2004. Measurement of domestic animal diversity – a review of recent diversity studies. Documentprepared for the third session of the Intergovernmental Working Group on Animal GeneticResources. Rome (available at www.fao.org/ag/againfo/programmes/en/genetics/documents/ITWG3_Inf3.pdf).FAO. 2005. Status of research and application of crop biotechnologies in developing countries: preliminaryassessment, by Z. Dhlamini, C. Spillane, J.P. Moss, J. Ruane, N. Urquia & A. Sonnino. Rome(available at www.fao.org/docrep/008/y5800e/y5800e00.htm).Guimarães, E.P., Kueneman, E. & Carena, M.J. 2006. Assessment of national plant breeding andbiotechnology capacity in Africa and recommendations for future capacity building. Hort. Sci. 41:50–52.Hallauer, A.R. 1999. Temperate maize and heterosis. In J.G. Coors & S. Pandey, eds. The genetics andexploitation of heterosis in crops, pp. 353-361. Proc. Internat. Symp. of Heterosis in Crops. MexicoCity, 18–22 August 1997. ASA, CSSA and SSSA, Madison, WI, USA.Pilling, D., Cardellino, R., Zjalic, M., Rischkowsky, B., Tempelman, K.A. & Hoffmann, I. 2007.The use of reproductive and molecular biotechnology in animal genetic resources management – aglobal overview. Anim. Genet. Resources Information. FAO, Rome (in press).Toenniessen G.P., O’Toole, J.C. & DeVries, J. 2003. Advances in plant biotechnology and its adoptionin developing countries. Curr. Opin. in Plant Biol. 6: 191–198.

Section IIMarker-assisted selectionin crops – case studies

Chapter 3Molecular markers for use inplant molecular breeding andgermplasm evaluationJeremy D. Edwards and Susan R. McCouch

30Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishSummaryA number of molecular marker technologies exist, each with different advantages and disadvantages.When available, genome sequence allows for the development of greater numbersand higher quality molecular markers. When genome sequence is limited in the organism ofinterest, related species may serve as sources of molecular markers. Some molecular markertechnologies combine the discovery and assay of DNA sequence variations, and thereforecan be used in species without the need for prior sequence information and up-frontinvestment in marker development. As a prerequisite for marker-assisted selection (MAS),there must be a known association between genetic markers and genes affecting the phenotypeto be modified. Comparative databases can facilitate the transfer of knowledge ofgenetic marker-phenotype association across species so that discoveries in one species maybe applied to many others. Further genomics research and reductions in the costs associatedwith molecular markers will continue to provide new opportunities to employ MAS.

Chapter 3 – Molecular markers for use in plant molecular breeding and germplasm evaluation 31IntroductionMolecular markers are valuable tools forthe classification of germplasm and inMAS. The purpose of this chapter is toprovide guidance in selecting appropriatemolecular marker systems based on theavailability of technological resources invarious species and to provide some examplesof MAS applications. One of the manybenefits of the increasing amount of DNAsequence information in many organisms isthe expanding opportunity for the developmentof new molecular markers. As the fullgenome sequence will not be available formost species of interest in the near future,it is important to find strategies for developingand using molecular markers whensequence resources are limited. This chapterdescribes several technologies that existfor developing molecular markers withoutDNA sequence information. It also drawson some examples from rice (Oryza sativaL.) to illustrate how molecular markerdevelopment was influenced by the additionof each layer of sequence information,culminating in the present status of rice asthe first crop with nearly complete genomesequence information.Molecular marker technologiesRestriction fragment lengthpolymorphismsRestriction fragment length polymorphisms(RFLPs) were the first DNA-based molecularmarkers. An application of Southernanalysis (Southern, 1975), RFLPs exploitthe ability of single stranded DNA to bind(hybridize) to DNA with a complementarysequence. RFLP markers detect variationin DNA sequences at the same loci indifferent individuals or accessions. Technically,RFLP technology involves thehybridization of cloned DNA to restrictionfragments of differing molecular weightsfrom restriction enzyme-digested genomicDNA. The digested DNA fragments aresize-separated on agarose gels by electrophoresisand transferred as denatured(single stranded) arrays of fragments tofilters through capillary action. The filtersare then incubated with specific labelledprobes (genes or anonymous fragmentsof single stranded DNA), washedand exposed to x-ray film. To identifypolymorphisms between individuals oraccessions, the genomic DNA extractedfrom each individual is digested with a seriesof restriction enzymes to find enzymesthat produce fragments (bands) that differin molecular weight between accessionsand can be distinguished by hybridizationwith a given probe. To ensure that probeshybridize to single fragments on a gel, theDNA used as a probe should be from asingle or low copy (non-repetitive) regionof the genome. Probes may represent genes(i.e. derived from complementary DNA[cDNA]) or they may represent anonymoussequences derived from genomic DNA.Genomic probes are generated by shearingor digesting DNA and cloning the fragmentsinto a plasmid vector that allows foramplification of the cloned fragment in asuitable host. To increase the frequency oflow copy clones in a genomic library, theDNA may be digested with a methylationsensitiveenzyme, such as PstI. Therepetitive regions of a genome are typicallyheavily methylated and thus producefragments >25 kb when digested with amethylation-sensitive enzyme. As a result,these fragments do not clone efficientlyinto plasmid vectors and consequentlyare effectively filtered out of the analysis.Thus, use of methylation-sensitive enzymesincreases the representation of unmethylatedand typically low copy gene sequences inRFLP analysis. Sharing of anonymous,

32Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishunsequenced RFLP markers amongresearchers requires an infrastructure forthe maintenance and distribution of clonedprobes for use by multiple researchers.However, if end-sequence or full-clonesequence information is available, the probescan be amplified readily from genomicDNA via the polymerase chain reaction(PCR), and the cumbersome aspects ofclone maintenance and distribution areavoided. The polymorphisms detected byRFLPs may result from single base changescausing a loss of restriction sites or a gain ofnew restriction sites, or from insertions anddeletions (indels) between restriction sites(McCouch et al., 1988; Edwards, Lee andMcCouch, 2004).PCR-based markersMany advances in molecular marker technologyhave come through applications ofthe PCR method (Mullis et al., 1986). InPCR, a thermo-stable DNA polymeraseenzyme makes copies of a target sequencebeginning from two small pieces of syntheticallyproduced DNA (primers) thatare complementary to sequences bracketingthe target. Through iterations of theprocess with heating to separate the doublestranded DNA molecules and cooling toallow the primers to re-anneal, the targetsequence is exponentially amplified.Polymerase chain reaction-based markersrequire much less DNA per assay thanRFLPs and are more compatible with automatedhigh-throughput genotyping (i.e. theability to process large numbers of samplesquickly and efficiently).Randomly amplified polymorphic DNAmarkersRandomly amplified polymorphic DNAmarkers (RAPDs) use PCR to amplifystretches of DNA between single primersof arbitrary sequence (Williams et al., 1990;Welsh and McClelland, 1990). Amplificationoccurs only where sequences complementaryto the primers are in close enoughproximity for successful PCR. The typicaloligonucleotide used for RAPDs isten bases long and will amplify many locisimultaneously, allowing multiple markersto be assayed in a single PCR reactionand a single lane on an agarose gel. As theprimers are arbitrary, RAPD technologycan be applied directly to any species withno prior sequence knowledge. This technologyis particularly useful when thereis a need to assay loci across the entiregenome. The polymorphisms are detectedonly as the presence or absence of a bandof a particular molecular weight, and itis not possible to differentiate betweenhomozygous and heterozygous markers.RAPDs are notoriously unreliable because,aside from sequence differences, the amplificationor failure of amplification of anyband may be sensitive to any number offactors, including DNA template quality,PCR conditions, reagents and equipment.Amplified fragment lengthpolymorphismsAmplified fragment length polymorphisms(AFLPs) are molecular markers derivedfrom the selective amplification of restrictionfragments (Vos et al., 1995). GenomicDNA is digested with a pair of restrictionenzymes and oligonucleotide adaptors areligated to the ends of each restriction fragment.The fragments are amplified usingprimers that anneal to the adaptor sequenceand extend into the restriction fragment.Only a portion of restriction fragmentswill be within the range of sizes than can beamplified by PCR and visualized on polyacylamidegels (between 50 and 350 bp). Forlarge genomes, additional selective bases

Chapter 3 – Molecular markers for use in plant molecular breeding and germplasm evaluation 33can be added to the primers to reduce thenumber of co-amplified bands. AFLPs havemany of the advantages of RAPDs, but havemuch better reproducibility. AFLP technologyrequires greater technical skill thanRAPDs and, because AFLPs run on polyacrylamidegels instead of agarose, they alsorequire a larger investment in equipmentthan RAPDs. Using manual gels, AFLPbands are detectable using silver stain, orby labelling of the primers with a radioactiveisotope. Alternatively, for higherthroughput, AFLPs can be detected with anautomated DNA sequencer by using fluorescentlylabelled primers.Diversity array technology (DArT) isa modification of the AFLP procedureusing a microarray platform (Jaccoud et al.,2001) that greatly increases throughput. InDArT, DNA fragments from one sampleare arrayed and used to detect polymorphismsfor the fragments in other samplesby differential hybridization (Wenzl etal., 2004).Developing molecular markerswith DNA sequence informationWhen the DNA sequence is available, it ispossible to design primers to amplify acrossa specific locus. However not all loci willbe polymorphic. Targeting highly variablesequence features increases the likelihoodof detecting polymorphism. These highlyvariable features include tandem repeatssuch as microsatellites, and dispersed complexrepeats such as transposable elements.MicrosatellitesSimple sequence length polymorphisms(SSLPs), also known as simple sequencerepeats (SSRs), or microsatellites, consist oftandemly repeated di-, tri- or tetra-nucleotidemotifs and are a common feature ofmost eukaryotic genomes. The number ofrepeats is highly variable because slippedstrand mis-pairing causes frequent gain orloss of repeat units. With their high levelof allelic diversity, microsatellites are valuableas molecular markers, particularly forstudies of closely related individuals.PCR-based markers are designed toamplify fragments that contain a microsatelliteusing primers complementary tounique sequences surrounding the repeatmotif (Weber and May, 1989). Differencesin the number of tandem repeats are readilyassayed by measuring the molecular weightof the resulting PCR fragments. As the differencesmay be as small as two base pairs,the fragments are separated by electrophoresison polyacrylamide gels or usingcapillary DNA sequencers that providesufficient resolution.Without prior sequence knowledge,microsatellites can be discovered by screeninglibraries of clones. Clones containing therepeat motif must be sequenced to findunique sites for primer design flanking therepeats. Microsatellite marker developmentfrom pre-existing sequence is far more direct.Good reviews of microsatellite markerdevelopment include those of McCouchet al. (1997) and Zane, Bargelloni andAtarnello (2002). Microsatellites discoveredin non-coding sequence often have a higherrate of polymorphism than microsatellitesdiscovered in genes. However, in somespecies such as spruce (Picea spp.) withhighly repetitive genomes, SSR markersdeveloped from gene sequences have fewerinstances of null alleles, i.e. failure of PCRamplification (Rungus et al., 2004).Microsatellite markers have severaladvantages. They are co-dominant; the heterozygousstate can be discerned from thehomozygous state. The markers are easilyautomated using florescent primers on anautomated sequencer and it is possible

34Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishto multiplex (combine) several markerswith non-overlapping size ranges on asingle electrophoresis run. The results arehighly reproducible, and the markers areeasily shared among researchers simply bydistributing primer sequences. AlthoughSSRs are abundant in most eukaryoticgenomes, their genomic distribution mayvary. Uneven distributions of microsatelliteslimit their usefulness in some species.Inter-SSRs (ISSRs) are another type ofmolecular marker that makes use of microsatellitesequences. ISSRs use PCR primersanchored in the termini of the repeatsextending into the flanking sequence by severalnucleotides (Zietkietkiewicz, Rafalskiand Labuda, 1994). PCR products are producedfor each pair of microsatellites thatare in sufficient proximity for PCR tooccur, or may be generated by anchoringone primer in the SSR motif and using asecond “universal” primer correspondingto a sequence that has been ligated ontothe ends of restriction fragments (as in theAFLP technique described above, wheregenomic DNA is first digested with arestriction enzyme and oligonucleotideadaptors are ligated to the ends of eachrestriction fragment, except that one primerresides in an SSR motif that is bracketedby the restriction sites) (Gupta et al.,1994; Goodwin, Aitken and Smith, 1997).Markers at multiple loci are assayed as thepresence or absence of bands of particularsizes. ISSRs can be visualized on agarosegels, on silver stained polyacrylamide gelsor fluorescently labelled for detection withan automated DNA sequencer.Transposable element-based markersTransposable elements (TEs) are anotherrapidly changing feature of the genomethat can be exploited as a source of variabilityfor molecular markers. Discovery ofTE sequences is a prerequisite for their useas markers. While TEs may be discoveredas mutations in alleles of genes conferringmutant phenotypes, they have also beendiscovered directly in genomic sequence(reviewed by Feschotte, Jiang and Wessler,2002). Transposon display is a modifiedAFLP procedure that differs only in thatone of the two primers is designed withinthe consensus sequence of a TE family sothat amplification depends on the presenceof a TE insertion within a restrictionfragment (Casa et al., 2000). Using thisapproach, the presence or absence of a TEcan be assayed simultaneously at manyloci throughout the genome. To assay fora TE insertion at a specific locus, singlecopy “anchor markers” can be designedwith primers located in unique sequencesflanking the region of interest. A size polymorphismindicates the presence or absenceof the TE in that particular location. Anchormarkers are advantageous because they areco-dominant, can be run on a simple agarosegel system and are biologically informativein that they provide evidence of bothcomplete, or incomplete, insertion or excisionevents. This methodology can also beapplied to any known indel feature regardlessof whether or not it is derived from aTE.Single nucleotide polymorphismsSingle nucleotide polymorphisms (SNPs)are an abundant source of sequence variantsthat can be targeted for molecularmarker development. Of all the molecularmarker technologies available today,SNPs provide the greatest marker density.SNPs are often the only option forfinding markers very near or within agene of interest, and can even be used todetect a known functional nucleotide polymorphism(FNP). Discovery of SNPs

Chapter 3 – Molecular markers for use in plant molecular breeding and germplasm evaluation 35Table 1SNP technologiesAllele discrimination• Hybridization• Primer extension• Ligation• Invasive cleavageDetection methods• Gel separation• Arrays• Mass spectrometry• Plate readersrequires obtaining an initial DNA sequencein a reference individual followed by someform of re-sequencing in other varietiesto find variable base pairs. In addition todirect sequencing, SNPs can be discoveredthrough ecotilling with the CEL I enzyme(Comai et al., 2004) or by denaturing highpressure liquid chromatography (DHPLC)to measure small conformational differenceswhen PCR amplified sequences arehybridized to a reference sequence (Kwok,2001). In addition to SNP discovery, bothDHPLC and ecotilling are viable technologiesfor SNP detection. There is a myriad ofother SNP assay technologies in developmentand to date no single method standsout as superior to the others. Table 1 listssome examples of SNP allele discriminationmethods and detection systems that can becombined in various ways (see reviews byKwok, 2001 and Gut, 2001). The benefits ofSNP assays include increased speed of genotyping,lower cost and the parallel assay ofmultiple SNP.Single feature polymorphisms andmicroarray-based genotypingIndel polymorphisms, also known as singlefeature polymorphisms (SFPs), are particularlyamenable to microarray-basedgenotyping. These assays are done by labellinggenomic DNA (target) and hybridizingto arrayed oligonucleotide probes that arecomplementary to indel loci. Each SFPis scored by the presence or absence of ahybridization signal with its correspondingoligonucleotide probe on the array. Bothspotted oligonucleotides (Barrett et al.,2004) and Affymetrix-type arrays (Borevitzet al., 2003) have been used in these assays.The SFPs can be discovered throughsequence alignments or by hybridizationof genomic DNA with whole genomemicroarrays. The advantage of microarrayplatforms for genotyping is that they arehighly parallel, and they are well suited forapplications such as quantitative trait loci(QTL) analysis, where whole genome coveragewith many markers is desirable.Special considerations fordiversity studies and germplasmevaluationThe interpretation of molecular marker datafor germplasm classification and diversitycan be confounded by uncertainty about theunderlying sources of the polymorphismsand by homoplasy (false homology). ForRFLPs in rice, indels can account for asmuch or more of the polymorphism aschanges in the restriction sites themselves(Edwards, Lee and McCouch, 2004). AFLPsand RAPDs can also be sensitive to bothindels and base changes. The ratio of indelsto base changes is important for diversitystudies because, when molecular markersare used to estimate nucleotide divergence,the divergence will be overestimated ifindel-derived polymorphisms are common(Upholt, 1977; Nei and Miller, 1990; Innanet al., 1999). The greatest certainty of theunderlying polymorphism comes fromSNP technologies that directly assay forsingle base changes.For SSR markers among closely relatedindividuals, most polymorphism shouldbe caused by expansion or contractionof the number of repeat units. However,as genetic distance between the varietiesincreases, there is an increasing chancethat indel events will cause additional size

36Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishpolymorphism (Chen, Cho and McCouch,2002). Thus, the use of stepwise SSRmutation models would be inappropriate forhighly diverged populations. Homoplasy isalso a problem in SSR markers because thehyper-variability leads to some shared allelesizes through parallelism, convergence andreversion (Doyle et al., 1998). Homoplasyfrom reversions can affect transposonbasedmarkers or any markers withpolymorphisms potentially derived fromClass II DNA transposable elements. Thisclass of TEs has a cut and paste mechanismof transposition, so a TE may insert onto alocus and later excise.In RAPDs, ISSRs and AFLPs, homoplasycan occur when two or more lociproduce PCR fragments of similar molecularweight. Although it is desirable to havehigh numbers of bands to maximize theamount of information per lane, this mustbe balanced against the increasing risk ofhomoplasy as more loci are represented.Special considerations formarker-assisted selectionQuality markers for use in MAS shouldbe reliable and easily shared amongresearchers. Co-dominant markers are preferredto avoid the need for progeny testing.Sometimes less desirable markers for MASsuch as RAPDs, ISSRs and AFLPs areuseful for finding markers linked to thedesired allele. Once such a marker is found,it is possible to extract and sequence thecorresponding band. This sequence canbe used to develop co-dominant markerssuch as cleaved amplified polymorphicsequences (CAPS) (Konieczny andAusubel, 1993) or to sequence characterizedpolymorphic regions (SCARs) (Paranand Michelmore, 1993). SCAR and CAPSmarkers are co-dominant and simplify thescreening of large numbers of individuals.When a genetic map exists, markers canbe positioned on the map and other linkedmarkers can be substituted. The additionalmarkers are useful for higher resolutionmapping to find markers more closelylinked to the desired allele or ultimately forpositional cloning of the underlying gene.Reproducibility of molecular markerdataFor orphan species, clearly there is a hugevalue to the anonymous primer approaches(AFLP, DArTs, ISSRs and RAPDs) thatdo not require sequence information ormuch up-front investment. However,the data can be difficult to score, andreproducibility requires a lot of technicalskill. Technologies that depend on thepresence or absence of PCR amplifiedbands are susceptible to changes in PCRconditions and the quality of sample DNA,and the data from separate experimentsmay differ. Further, in any method thatdepends on accurate measurement ofmolecular weight differences between bands(e.g. SSRs), the exact mole-cular weightsassigned to each allele may be differentin each analysis because of differencesin labelling of PCR products, roundingof allele molecular weight estimates andbinning of alleles. Without controls foreach allele encountered, it is difficult orimpossible to merge separate sets of data.Despite discrepancies in the exact dataderived from molecular markers, the resultsand conclusions should be consistent withinindependent experiments. For reliabilityin making inferences across independentdata-sets, SNP markers are preferred. SNPdata-sets can be easily integrated basedon sequence, and SNPs have properties(such as a low mutation rate) that areparticularly valuable for evolutionaryinference (Nielsen, 2000).

Chapter 3 – Molecular markers for use in plant molecular breeding and germplasm evaluation 37Table 2Key features of common molecular marker technologiesMarkertypeUsesrestrictionenzymesPCRbasedPolymorphismAbundanceCodominantAutomationLoci per assaySpecialized equipmentRFLP no yes moderate moderate yes no 1 to few Radioactive isotopeRAPD yes no moderate moderate no yes many Agarose gelsAFLP yes no moderate moderate no yes many Polyacrylamide gels/capillaryISSR yes no moderate moderate no yes many Agarose/polyacrylamide gelsDArT yes yes moderate moderate no yes many MicroarrayCAPS yes yes variable moderate yes yes single Agarose gelsSCAR yes no low moderate yes yes single Agarose gelsSSR yes no low moderate yes yes 1 to about 20 Polyacrylamide gels/capillaryTE-Anchor yes no variable variable yes yes single Agarose gelsSNP yes no variable highest yes yes 1 to thousands VariableChoosing a molecular markertechnologyClearly there is no single best choice ofmolecular marker for all situations. Factorsinfluencing the decision may include theobjectives of the study, availability oforganism specific sequences, equipmentand technical resources, and biologicalfeatures of the species. Several importantadvantages/disadvantages for each type ofmolecular marker discussed are summarizedin Table 2 (see review by Powell etal., 1996).If available, microsatellite or SNPmarkers are often the best choice. The rateof adoption of SSR markers can be facilitated,and the costs reduced, by preparing“kits” of selected SSR markers for certainspecies to provide a reliable set of markerswith good amplification, reasonable polymorphismand good genome coverage.This was done in the early days of the riceSSR effort and SSR kits were distributed atvery low cost through Research Genetics(called Rice-Pairs; McCouch et al., 1997).Similarly, for SNPs, there is a need todevelop useful sets of markers that arewidely available and can be mass-produced(at reduced cost) for distribution to theinternational community. SNP kits wouldalso have a clear benefit for databasingand analysing datasets obtained from multiplelaboratories. In addition to kits ofmarkers, there is a need to distribute setsof “control genotypes” as samples, particularlyto address the problem surroundingthe difficulties in integrating SSR datasets.When SNPs or SSRs are not available, it issometimes possible to transfer molecularmarkers from closely related species (Guptaet al., 2003; La Rota et al., 2005; Zhang etal., 2005). When financial resources arerestricted, RAPDs, AFLPs and ISSRs canprovide large numbers of markers with alimited investment. AFLPs, SSRs and ISSRscan provide high throughput using an automatedsequencer, while RAPDs and ISSRscan be run on agarose gels with minimalinvestment in equipment. The effectivenessof each method may vary by species and byapplication. Therefore, it is reasonable totry to use more than one method, particularlyat the early stages of research.impact of the rice genomesequence: A case historyDNA sequence information greatlyaccelerates the development of molecularmarkers. This is evident in the historyof rice microsatellite marker proliferationcoinciding with the release of data fromrice genome sequencing projects. Figure 1

38Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishFigure 1Progress of microsatellite marker development in riceMcCouch et al., 2002Temnykh et al., 2001Temnykh et al., 2000Chen et al., 1997Akagi et al., 1996Panaud, Chen andMcCouch, 1996Wu and Tanksley, 1993Zhao and Kochert, 19930 500 1000 1500 2000Number of microsatellite markersScreened libraries Genes/ESTs BAC-ends Shotgun sequences Finished BACstracks the publication of rice microsatellitemarkers derived from screening libraries ofclones and from the various categories ofsequences deposited in public databases. Theearliest method of developing microsatellitemarkers in rice was by using microsatellitesequences as probes to isolate clones fromgenomic libraries (Zhao and Kochert, 1993;Wu and Tanksley, 1993; Panaud, Chen andMcCouch, 1996; Akagi et al., 1996; Chenet al., 1997; Temnykh et al., 2000). In 1996,Akagi et al. used microsatellite repeatsfound in rice sequences from databasesearches to develop 35 new markers andin 2000, Temnykh et al. published 91 newmicrosatellite markers developed fromexpressed sequence tag (EST) sequences.Temnykh et al. (2001) developed 200 newmarkers, mostly from end sequences of ricebacterial artificial chromosomes (BACs).However, the most dramatic increase inmicrosatellite markers (2 240 new markersin 2002 and 25 000 in 2004) was madepossible primarily though the use of wholegenome shotgun sequences (McCouch et al.,2002; G. Wilson, personal communication).Complete genome sequence provides anadditional advantage in electronicallydetermining the position of new markerson genetic and physical maps. However,full genomic sequence is not a requirementfor microsatellite marker development, andthere are a number of microsatellite markersthat have been developed for a wide arrayof crop species (Table 3) without the benefitof full genomic sequence.Marker-assisted selectionstrategies and examplesMAS in a breeding context involves scoringindirectly for the presence or absenceof a desired phenotype or phenotypiccomponent based on the sequences orbanding patterns of molecular markerslocated in or near the genes controlling thephenotype. The sequence polymorphismor banding pattern of the molecular markeris indicative of the presence or absence of aspecific gene or chromosomal segment thatis known to carry a desired allele.DNA markers can increase screeningefficiency in breeding programmes in a

Chapter 3 – Molecular markers for use in plant molecular breeding and germplasm evaluation 39Table 3Examples of SSR markers available across different plant speciesCommon name Species Number ofSSRsReferenceRice Oryza sativa 2240 McCouch et al., 2002Maize Zea mays 1669 MapPairs (mp.invitrogen.com)Soybean Glycine max 597 MapPairs (mp.invitrogen.com)Cassava Manihot esculenta 318 MapPairs (mp.invitrogen.com)Arabidopsis Arabidopsis thaliana 290 MapPairs (mp.invitrogen.com)Cotton Gossypium spp. 217 MapPairs (mp.invitrogen.com)Sugar cane Saccharum spp. 200 www.intl-pag.org/pag/9/abstracts/W30_04.htmlWheat Triticum aestivum 193 MapPairs (mp.invitrogen.com)Grape Vitis vinifera 152 noGroundnut Arachis hypogaea 110 Ferguson et al., 2004Cucumber Cucumis sativus 110 Fazio, Staub and Chung, 2002Peach Prunus persica 109 Aranzana et al., 2004Kiwifruit Actinidia spp. 105 Testolin et al., 2001Barley Hordeum vulgare 44 MapPairs (mp.invitrogen.com)Potato Solanum tuberosum 31 Ghislain et al., 2004Pine trees Pinus spp. 28 MapPairs (mp.invitrogen.com)Banana Musa spp. 28 MapPairs (mp.invitrogen.com)Sweet potato Ipomoea batatas 26 MapPairs (mp.invitrogen.com)Sugar beet Beta vulgaris 25 www.intl-pag.org/pag/10/abstracts/PAGX_W306.htmlEggplant Solanum melongena 23 www.intl-pag.org/pag/11/abstracts/P3b_P181_XI.htmlFrom: Thomson, Septiningsih and Sutrisno, 2003 (reprinted with permission of author)number of ways. For example, theyprovide:• the ability to screen in the juvenile stagefor traits that are expressed late in the lifeof the organism (i.e. grain or fruit quality,male sterility, photoperiod sensitivity);• the ability to screen for traits that areextremely difficult, expensive or timeconsuming to score phenotypically (i.e.quantitatively inherited or environmentallysensitive traits such as rootmorphology, resistance to quarantinedpests or to specific races or biotypes ofdiseases or insects, tolerance to certainabiotic stresses such as drought, salt andmineral deficiencies or toxicities);• the ability to distinguish the homozygousfrom the heterozygous conditionof many loci in a single generationwithout the need for progeny testing (asmolecular markers are co-dominant);• the ability to perform simultaneousMAS for several characters at one time(or to combine MAS with phenotypic orbiochemical evaluation).This section provides examples ofhow molecular markers are being used inbreeding and germplasm evaluation. Whilethese examples are drawn mostly from rice,they illustrate applications of MAS techniquesthat are used in other species.Before molecular markers can be usedfor selection purposes, their associationwith genes or traits of interest must befirmly established. While the number ofeconomically important genetic loci thathave been cloned or tagged via linkage tomolecular markers is still limited in mostspecies, work towards this end is acceleratingrapidly. This is particularly true inrice, due to the availability of completegenome sequence information.Nonetheless, a great deal of time andeffort is required to identify the geneticloci and specific allelic variants that areresponsible for the tremendous array of

40Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishcharacters that breeders are concernedabout in population or variety improvementprogrammes. Given the complexity ofquantitative traits, many different lines orcrosses must be carefully analysed overdifferent years and environments to unravelimportant components of gene interaction.In a breeding context, understanding thegenetic basis of genotype by genotypeinteraction (G x G) and genotype byenvironment interaction (G x E) is criticalas the basis for predicting how QTL arelikely to behave. Information from a largenumber of studies addressing each of thesepoints must then be assembled into adatabase that offers easy access to users andallows many different kinds of data to beintegrated with a simple query.The Gramene database represents abeginning in the quest to serve this usercommunity. Gramene is a comparativegenome database for grasses and currentlyoffers a complete inventory of all publishedQTL that have been identified in rice (www.gramene.org/qtl/index.html), allowing usersto find information about where along thechromosome a QTL is located, what phenotypeis associated with the QTL, how itwas measured, what germplasm was used,what molecular markers reside nearby, whatthe corresponding position is on a comparativemap of another grass species and withwhat statistical significance the QTL wasdetected. The database also provides a linkto the published article so that users canreadily find more information on the subject.Similar inventories and databases arebeing assembled for other families of plantsand are critical to the implementation ofeffective molecular breeding strategies.Comparative genome methods takeadvantage of the fact that some specieshave more developed genetic systems thanothers. Examples of well studied “model”organisms with available genomic sequenceinclude species such as Arabidopsis andrice for plants, Populus (Taylor, 2002) andEucalyptus (Poke et al., 2005) specificallyfor forestry, and Fugu (Aparicio et al.,2002) and zebrafish (Guryev et al., 2006)for fisheries. Relying heavily on the useof comparative maps and comparativesequence analysis, genome databases allowresearchers to make predictions about thelocation and phenotypic consequences ofhomologous genes in related species. Thus,understanding how a gene or QTL behavesin one species can potentially shortcutthe process of identifying a related geneor QTL in the genetic system of anotherspecies. This approach underscores thesearch for QTL associated with abioticstress tolerance in cereals. A global effortto identify loci associated with droughttolerance has recently been initiated underthe umbrella of the Generation ChallengeProgramme (www.generationcp.org).Markers associated with tolerance fora variety of environmental stresses rank asimportant targets for molecular MAS incereal breeding because these complex traitsare often prohibitively difficult to screenusing classical selection techniques. Effortsto identify QTL associated with tolerance todrought, salt and mineral deficiencies or toxicities(Champoux et al., 1995; Flowers et al.,2000; Nguyen et al., 2002; Kamoshita et al.,2002; Price et al., 2002; Gregorio, 2002) ina number of genetic backgrounds representan important first step towards achievingthis goal. Additional studies have specificallyaddressed the problems associated withG x G and G x E (Zheng et al., 2000; Li etal., 2003; Hittalmani et al., 2003).In the area of biotic stress, several geneshave been cloned and characterized forresistance to major diseases such as bacterialblight and blast (Song et al., 1995;

Chapter 3 – Molecular markers for use in plant molecular breeding and germplasm evaluation 41Yoshimura et al., 1998; Wang et al., 1999;Bryan et al., 2000; Sun et al., 2004) andmany other genes for disease resistancehave been tagged with linked markers. Thisopens the door for targeted approaches toMAS (Valent et al., 2001). While the diseaseresistance literature is too vast to summarizehere, it is important to note that advances inthis area are having an impact on varietalimprovement programmes (www.syix.com/rrb/98rpt/MarkerAssist.htm). Pyramidingof resistance genes into a single varietyand the construction of multiline varieties,each with one or more R genes (resistancegenes) that can be used in various combinations,are under way to develop moredurable forms of disease and insect resistance(Yoshimura et al., 1992; Yoshimura etal., 1995; Hittalmani et al., 1995; Blair andMcCouch, 1997; Ndjiondjop et al., 1999;Davierwala et al., 2001; Su et al., 2002;Conaway-Bormans et al., 2003; Lorieux etal., 2003; Hayashi et al., 2004).Marker-based selection is also helpfulin attempts to transfer genes from exoticgermplasm into cultivated lines. In rice,several workers have used RFLP andSSR markers to monitor introgression ofbrown planthopper resistance from O.officinalis (Kochert, Jena and Zhao, 1990),bacterial blight resistance from O. longistaminata(Ronald et al., 1992), aluminumtolerance or yield and quality-related traitsfrom O. rufipogon (Nguyen et al., 2002;Thomson et al., 2003; Septiningsih et al.,2003a, b) or from other wild species such asO. glumaepatula (Brondani et al., 2002) orO. glaberrima (Jones et al., 1997; Lorieuxet al., 2003) into cultivated O. sativa backgrounds.Marker-assisted introgressionstrategies have also been used in a numberof livestock breeding programmes but,because of longer generation intervals andlower reproductive rates, this is generallyfeasible for genes of large effect (Dekkers,2004; Chapter 10). Identifying the recombinantswith the least amount of donorDNA flanking the genes of interest isenhanced by the use of molecular markers(Monna et al., 2002; Takeuchi et al., 2003;Blair, Panaud and McCouch, 2003). Inthese examples, MAS offers a powerfulstrategy for making efficient use of thewealth of useful genetic variation that existsin the early landraces and wild speciesof cultivated food crops (Tanksley andMcCouch, 1997).As this kind of information accumulates,MAS permits rapid identification of individualsthat may contain only one geneticcomponent of a complex trait. Once identified,such an individual can be crossedwith another individual in a breeding programmeso that multiple, complementarygenes are combined to optimize a quantitativelyinherited trait. Individuals containingonly one gene of interest often defy accuratephenotypic identification wherepolygenic traits are concerned because varioustypes of epistasis, or gene interaction,may be required to generate the phenotypeof interest (Yamamoto et al., 2000; Zhenget al., 2000).Linkage disequilibrium (LD) mappingis another marker-assisted approachthat provides important informationthat is immediately relevant to breedingprogrammes (Remington, Ungererand Purugganan, 2001; Flint-Garcia,Thornsberry and Buckler, 2003). Usingcollections of distantly related germplasmaccessions rather than populations derivedfrom bi-parental crosses allows researchersto explore the relationship betweenphenotype and genotype in materials thathave been amply tested over years andenvironments, often as part of an appliedbreeding programme. This provides

42Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishcritical information about how specificcombinations of genes and alleles interactin relevant varietal backgrounds and allowsbreeders to compare the phenotypic effectof genes or chromosomal segments that havebeen inherited from a common ancestor andselected in multiple-cross combinations.In addition to the use of MAS in traditionalcrossing and selection programmes,breeders also have opportunities to adjustparticular traits or phenotypes via theintroduction of genes using a transgenicapproach (Ye et al., 2000; James, 2003;Nuffield Council on Bioethics, 2004). Onceintroduced into the gene pool, a transgenecan be tracked with the aid of molecularmarkers (designed to tag the transgenesequence itself) through subsequent crosses,just as would be done for any other gene ofinterest in a breeding programme.Another use of molecular markers invariety improvement involves markerassistedgermplasm evaluation (Xu, Ishiiand McCouch, 2003). Population structureanalysis offers insight about how diversityis partitioned within a species and canhelp define clusters, or subpopulations, ofgermplasm that are likely to contain highfrequencies of particular alleles (Garris,McCouch and Kresovich, 2003). This typeof analysis can also guide allele miningefforts aimed at identifying valuableaccessions in a germplasm collection foruse as parents in a breeding programme.Such approaches have the potential to makeparental selection more efficient, to expandthe gene pool of modern cultivars andultimately to speed up the development ofproductive new varieties. As information isgenerated about which genes and alleles areassociated with phenotypic characters ofagronomic importance, and as the complexinteractions among genes are enumerated inthe context of specific gene pools and theenvironments to which they are adapted,breeders are increasingly empowered tomake predictions about how to combinediverse alleles productively.To exploit molecular breeding strategiesfully, information resources must be developedso that the overwhelming amount ofinformation about genes, alleles and naturalgenetic variation can be funnelled into a usefultool for breeding applications. This willinvolve a very different approach to informationresources than currently employedby the large genome databases, which areoriented towards genomics researchersand molecular biologists rather than thebreeding community. Nonetheless, a fewexamples offer beacons of inspiration in thisarea, including the emerging InternationalRice Information System (IRIS) database(Bruskiewich et al., 2003; www.icis.cgiar.org/), the GeneFlow database (www.geneflow.com), the marker-assisted selectionwheat (MASwheat) database (http://maswheat.ucdavis.edu/)and software such asReal Time QTL (http://zamir.sgn.cornell.edu/Qtl/Html/home.htm).In conclusion, genomics research is generatinginformation about the location andphenotypic consequences of specific genesand alleles in a wide range of species. Thisinformation can be translated into tools forbreeders. Molecular marker technology canbenefit breeding objectives by increasing theefficiency and reliability of selection and byproviding essential insights into how genesbehave in different environments and indifferent genetic backgrounds. Once genesand QTL are identified, markers allowinteresting alleles to be traced throughthe pedigrees of breeding programmes ormined out of germplasm collections to serveas the basis for future varietal improvement.Using markers in combination withboth QTL and association approaches, the

Chapter 3 – Molecular markers for use in plant molecular breeding and germplasm evaluation 43effect of specific alleles on a phenotypecan be monitored with relative precision.As all this information is assembled andorganized in databases that provide easyquerying capabilities for plant breeders,breeders will take advantage of the powerthat comes from the application of genomebased strategies for plant improvement.ReferencesAkagi, H., Yokozeki, Y., Inagaki, A. & Fujimura, T.Y. 1996. Microsatellite DNA markers for ricechromosomes. Theor. Appl. Genet. 93: 1071–1077.Aparicio, S., Chapman, J., Stupka, E., Putman, N., Chia, J.M., Dehal, P., Christoffels, A., Rash,S., Hoon, S., Smit, A., Gelpke, M.D.S., Roach, J., Oh, T., Ho, I.Y., Wong, M., Detter, C.,Verhoef, F., Predki, P., Tay, A., Lucas, S., Richardson, P., Smith, S.F., Clark, M.S., Edwards,Y.J.K., Doggett, N., Zharkikh, A., Tavtigian, S.V., Pruss, D., Barnstead, M., Evans, C., Baden,H., Powell, J., Glusman, G., Rowen, L., Hood, L., Tan, Y.H., Elgar, G., Hawkins, T., Venkatesh,B., Rokhsar, D. & Brenner, S. 2002. Whole-genome shotgun assembly and analysis of the genomeof Fugu rubripes. Science 297: 1301–1310.Aranzana, M.J., Pineda, A., Cosson, P., Dirlewanger, E., Ascasibar, J., Cipriani, G., Ryder, C.D.,Testolin, R., Abbott, A., King, G.J., Iezzoni, A.F. & Arus, P. 2003. A set of simple-sequence repeat(SSR) markers covering the Prunus genome. Theor. Appl. Genet. 106: 819–825.Barrett, M.T., Scheffer, A., Ben-Dor, A., Sampas, N., Lipson, D., Kincaid, R., Tsang, P., Curry,B., Baird, K., Meltzer, P.S., Yakhini, Z., Bruhn, L. & Laderman, S. 2004. Comparative genomichybridization using oligonucleotide microarrays and total genomic DNA. Proc. Nat. Acad. Sci.USA 101: 17765–17770.Blair, M.W. & McCouch, S.R. 1997. Microsatellite and sequence-tagged site markers diagnostic forthe rice bacterial leaf blight resistance gene xa-5. Theor. Appl. Genet. 95: 174-184.Blair, M.W., Panaud O. & McCouch, S.R. 1999. Inter-simple sequence repeat (ISSR) amplification foranalysis of microsatellite motif frequency and fingerprinting in rice (Oryza sativa L.). Theor. Appl.Genet. 98: 780–792.Borevitz, J. O., Liang, D., Plouffe, D., Chang, H.S., Zhu, T., Weigel, D., Berry, C.C., Winzeler, E.& Chory, J. 2003. Large-scale identification of single-feature polymorphisms in complex genomes.Genome Res. 13: 513–523.Brondani, C., Rangel, N., Brondani V. & Ferreira, E. 2002. QTL mapping and introgression ofyield-related traits from Oryza glumaepatula to cultivated rice (Oryza sativa) using microsatellitemarkers. Theor. Appl. Genet. 104: 1192–1203.Bruskiewich, R.M., Cosico, A.B., Eusebio, W., Portugal, A.M., Ramos, L.M., Reyes, M.T., Sallan,M.A., Ulat, V.J., Wang, X., McNally, K.L., Sackville Hamilton, R. & McLaren, C.G. 2003.Linking genotype to phenotype: the International Rice Information System (IRIS). Bioinformatics19 Suppl 1: i63–i65.Bryan, G.T., Wu, K.S., Farrall, L., Jia, Y., Hershey, H.P., McAdams, S.A., Faulk, K.N., Donaldson,G.K., Tarchini, R. & Valent, B. 2000. A single amino acid difference distinguishes resistant andsusceptible alleles of the rice blast resistance gene Pi-ta. The Plant Cell 12: 2033–2046.Casa, A.M., Brouwer, C., Nagel, A., Wang, L., Zhang Q., Kresovich, S. & Wessler, S.R. 2000.Inaugural article: the MITE family heartbreaker (Hbr): molecular markers in maize. Proc. Nat.Acad. Sci. USA 97: 10083–10089.

44Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishChampoux, M.C., Sarkarung, G.W.S., Mackill, D.J., O’Toole, J.C., Huang, N. & McCouch, S.R.1995. Locating genes associated with root morphology and drought avoidance via linkage tomolecular markers. Theor. Appl. Genet. 90: 969–981.Chen, X., Cho Y.G. & McCouch, S.R. 2002. Sequence divergence of rice microsatellites in Oryza andother plant species. Mol. Genet. Genomics 268: 331–343.Chen, X., Temnykh, S., Xu, Y., Cho, X.G. & McCouch, S.R. 1997. Development of a microsatelliteframework map providing genome-wide coverage in rice (Oryza sativa L.). Theor. Appl. Genet.95: 553–567.Comai, L., Young, K., Till, B.J., Reynolds, S.H., Greene, E.A., Codomo, C.A., Enns, L.C., Johnson,J.E., Burtner, C., Odden, A.R. & Henikoff, S. 2004. Efficient discovery of DNA polymorphismsin natural populations by Ecotilling. Plant J. 37: 778–786.Conaway-Bormans, C.A., Marchetti, M.A., Johnson, C.W., McClung, A.M. &Park, W.D. 2003. Molecular markers linked to the blast resistance gene Pi-z in rice for use inmarker-assisted selection. Theor. Appl. Genet. 107: 1014–1020.Davierwala, A.P., Reddy, A.P., Lagu, M.D., Ranjekar, P.K. & Gupta, V.S. 2001. Marker assistedselection of bacterial blight resistance genes in rice. Biochem. Genet. 39: 261–278.Dekkers, J.C.M. 2004. Commercial application of marker- and gene-assisted selection in livestock:strategies and lessons. J. Anim. Sci. 82 E-Suppl: E313–E328.Doyle, J.J., Morgante, M., Tingey S.V. & Powell, W. 1998. Size homoplasy in chloroplast microsatellitesof wild perennial relatives of soybean (Glycine subgenus Glycine). Mol. Biol. Evol. 15:215–218.Edwards, J.D., Lee, V.M. & McCouch, S.R. 2004. Sources and predictors of resolvable indelpolymorphism assessed using rice as a model. Mol. Genet. Genomics 271: 298–307.Fazio, G., Staub, J.E. & Chung, S.M. 2002. Development and characterization of PCR markers incucumber. J. Am. Soc. Hortic. Sci. 127: 545–557.Ferguson, M.E., Burow, M.D., Schulze, S.R., Bramel, P.J., Paterson, A.H., Kresovich, S. & Mitchell,S. 2004. Microsatellite identification and characterization in peanut (A. hypogaea L.). Theor. Appl.Genet. 108: 1064–1070.Feschotte, C., Jiang N. & Wessler, S.R. 2002. Plant transposable elements: where genetics meetsgenomics. Nature Revs. Genet. 3: 329–341.Flint-Garcia, S.A., Thornsberry, J.M. & Buckler, E.S.T. 2003. Structure of linkage disequilibrium inplants. Ann. Rev. Plant Biol. 54: 357–374.Flowers, T.J., Koyama, M.L., Flowers, S.A., Sudhakar, C., Singh, K.P. & Yeo, A.R. 2000. QTL: theirplace in engineering tolerance of rice to salinity. J. Exp. Bot. 51: 99–106.Garris A., McCouch, S. & Kresovich, S. 2003. Population structure and its effect on haplotype diversityand linkage disequilibrium surrounding the xa5 locus of rice (Oryza sativa L.). Genetics 165:759–769.Ghislain, M., Spooner, D.M., Rodriguez, F., Villamon, F., Nunez, J., Vasquez, C., Waugh, R. &Bonierbale, M. 2004. Selection of highly informative and user friendly microsatellites (SSRs) forgenotyping of cultivated potato. Theor. Appl. Genet. 108: 881–890.Goodwin, I.D., Aitken, E.A.B. & Smith, L.W. 1997. Application of inter-simple sequence repeat(ISSR) markers to plant genetics. Electrophoresis 18: 1524–1528.Gregorio, G.B. 2002. Progress in breeding for trace minerals in staple crops. J. Nutr. 132:500S–502S.

Chapter 3 – Molecular markers for use in plant molecular breeding and germplasm evaluation 45Gupta, M., Chyi, Y.S., Romero-Severson, J. & Owen, J.L. 1994. Amplification of DNA markersfrom evolutionarily diverse genomes using single primers of simple-sequence repeats. Theor. Appl.Genet. 89: 998–1006.Gupta, P.K., Rustgi, S., Sharma, S., Singh, R., Kumar, N. & Balyan, H.S. 2003. Transferable EST-SSR markers for the study of polymorphism and genetic diversity in bread wheat. Mol. Genet.Genomics 270: 315–323.Guryev, V., Koudijs, M.J., Berezikov, E., Johnson, S.L., Plasterk, R.H., van Eeden,F.J. & Cuppen, E. 2006. Genetic variation in the zebrafish. Genome Res. 16: 491–497.Gut, I.G. 2001. Automation in genotyping of single nucleotide polymorphisms. Hum. Mutat. 17:475–492.Hayashi, K., Hashimoto, N., Daigen, M. & Ashikawa, I. 2004. Development of PCR-based SNPmarkers for rice blast resistance genes at the Piz locus. Theor. Appl. Genet. 108: 1212–1220.Hittalmani, S., Foolad, M.R., Mew, T., Rodriguez, R.L. & Huang, N. 1995. Development of aPCR-based marker to identify rice blast resistance gene, Pi-2(T), in a segregating population.Theor. Appl. Genet. 91: 9–14.Hittalmani, S., Huang, N., Courtois, B., Venuprasad, R., Shashidhar, H.E., Zhuang, J.Y., Zheng,K.L., Liu, G.F., Wang, G.C., Sidhu, J.S., Srivantaneeyakul, S., Singh, V.P., Bagali, P.G., Prasanna,H.C., McLaren, G. & Khush, G.S. 2003. Identification of QTL for growth- and grain yield-relatedtraits in rice across nine locations of Asia. Theor. Appl. Genet. 107: 679–690.Innan, H., Terauchi, R., Kahl, G. & Tajima, F. 1999. A method for estimating nucleotide diversityfrom AFLP data. Genetics 151: 1157-1164.Jaccoud, D., Peng, K., Feinstein, D. & Kilian, A. 2001. Diversity arrays: a solid state technology forsequence information independent genotyping. Nucl. Acids Res. 29: E25.James, C. 2003. Global review of commercialized transgenic crops: 2002 Feature: Bt Maize. ISAAABriefs.Jones, M.P., Dingkuhn, M. Aluko, G.K. & Semon, M. 1997. Interspecific O. sativa L.x O. glaberrimaSteud. Progenies in upland rice improvement. Euphytica 92: 237–246.Kamoshita, A., Wade, J., Ali, L., Pathan, S., Zhang, J., Sarkarung, S. & Nguyen, T. 2002. MappingQTLs for root morphology of a rice population adapted to rain-fed lowland conditions. Theor.Appl. Genet. 104: 880–893.Kochert, G., Jena, K. & Zhao, X.P. 1990. Genome characterization and introgression. In 4 th Ann.Mtg. Rockefeller Foundation’s Internat. Program on Rice Biotech. Los Baños, Philippines, IRRI.Konieczny, A. & Ausubel, F.M. 1993. A procedure for mapping Arabidopsis mutations using codominantecotype-specific PCR-based markers. Plant J. 4: 403–410.Kwok, P.Y. 2001. Methods for genotyping single nucleotide polymorphisms. Ann. Rev. GenomicsHum. Genet. 2: 235–258.La Rota, M., Kantety, R.V., Yu, J.K. & Sorrells, M.E. 2005. Non-random distribution and frequenciesof genomic and EST-derived microsatellite markers in rice, wheat, and barley. BMC Genomics6: 23.Li, Z.K., Yu, S.B., Lafitte, H.R., Huang, N., Courtois, B., Hittalmani, S., Vijayakumar, C.H.,Liu, G.F., Wang, G.C., Shashidhar, H.E., Zhuang, J.Y., Zheng, K.L., Singh, V.P., Sidhu, J.S.,Srivantaneeyakul, S. & Khush, G.S. 2003. QTL x environment interactions in rice. I. Heading dateand plant height. Theor. Appl. Genet. 108: 141–153.

46Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishLorieux, M., Reversat, G., Garcia Diaz, S.X., Denance, C., Jouvenet, N., Orieux, Y., Bourger, N.,Pando-Bahuon, A. & Ghesquiere, A. 2003. Linkage mapping of Hsa-1(Og), a resistance gene ofAfrican rice to the cyst nematode, Heterodera sacchari. Theor. Appl. Genet. 107: 691–696.McCouch, S.R., Kochert, G., Yu, Z.H., Wang, Z.Y., Khush, G.S., Coffman, W.R. & Tanksley, S.D.1988. Molecular mapping of rice chromosomes. Theor. Appl. Genet. 76: 815–829.McCouch, S.R., Chen, X., Panaud, O., Temnykh, S., Xu, Y., Cho, Y.G., Huang, N., Ishii, T. &Blair, M. 1997. Microsatellite marker development, mapping and applications in rice genetics andbreeding. Plant Mol. Biol. 35: 89–99.McCouch, S.R., Teytelman, L., Xu, Y., Lobos, K.B., Clare, K., Walton, M., Fu, B., Maghirang, R.,Li, Z., Xing, Y., Zhang, Q., Kono, I., Yano, M., Fjellstrom, R., DeClerck, G., Schneider, D.,Cartinhour, S., Ware, D. & Stein, L. 2002. Development and mapping of 2 240 new SSR markersfor rice (Oryza sativa L.) (supplement). DNA Res. 9: 257–279.Monna, L., Lin, X., Kojima, S., Sasaki, T. & Yano, M. 2002. Genetic dissection of a genomic regionfor a quantitative trait locus, Hd3, into two loci, Hd3a and Hd3b, controlling heading date in rice.Theor. Appl. Genet. 104: 772–778.Mullis, K., Faloona, F., Scharf, S., Saiki, R., Horn, G. & Erlich, H. 1986. Specific enzymatic amplificationof DNA in vitro: the polymerase chain reaction. Cold Spring Harb. Symp. Quant. Biol. 51Pt 1: 263–273.Ndjiondjop, M.N., Albar, L., Fargette, D., Fauquet, C. & Ghesquiere, A. 1999. The genetic basis ofhigh resistance to rice yellow mottle virus (RYMV) in cultivars of two cultivated rice species. PlantDisease 83: 931–935.Nei, M. & Miller, J.C. 1990. A simple method for estimating average number of nucleotide substitutionswithin and between populations from restriction data. Genetics 125: 873–879.Nguyen, B.D., Brar, D.S., Bui, B.C., Nguyen, T.V., Pham, L.N. & Nguyen, H.T. 2002. Identificationand mapping of the QTL for aluminum tolerance introgressed from the new source, Oryzarufipogon Griff., into indica rice (Oryza sativa L.). Theor. Appl. Genet. 106: 583–593.Nielsen, R. 2000. Estimation of population parameters and recombination rates from single nucleotidepolymorphisms. Genetics 154: 931–942.Nuffield Council on Bioethics. 2004. The use of genetically modified crops in developing countries.London, UK.Panaud, O., Chen, X. & McCouch, S.R. 1996. Development of microsatellite markers and characterizationof simple sequence length polymorphism (SSLP) in rice (Oryza sativa L.). Mol. Gen.Genet. 252: 597–607.Paran, I. & Michelmore, R.W. 1993. Development of reliable PCR-based markers linked to downymildew resistance genes in lettuce. Genetics 85: 985–993.Poke, F.S., Vaillancourt, R.E., Potts, B.M. & Reid, J.B. 2005. Genomic research in Eucalyptus.Genetica 125: 79–101.Powell, W., Morgante, M., Andre, C., Hanafey, M., Vogel, J., Tingey, S. & Rafalski, A. 1996. Thecomparison of RFLP, RAPD, AFLP and SSR (microsatellite) markers for germplasm analysis. Mol.Breed. 2: 225–238.Price, A.H., Townend, J., Jones, M.P., Audebert, A. & Courtois, B. 2002. Mapping QTLs associatedwith drought avoidance in upland rice grown in the Philippines and West Africa. Plant Mol.Biol. 48: 683–695.Remington, D.L., Ungerer, M.C. & Purugganan, M.D. 2001. Map-based cloning of quantitativetrait loci: progress and prospects. Genet. Res. 78: 213–218.

Chapter 3 – Molecular markers for use in plant molecular breeding and germplasm evaluation 47Ronald, P.C., Salmeron, J.M., Carland, F.M. & Staskawicz, B.J. 1992. The cloned avirulence geneavrPto induces disease resistance in tomato cultivars containing the Pto resistance gene. J. Bacteriol.174: 1604–1611.Rungus, D., Berube, Y., Zhang, J., Ralph, S., Ritland, C.E., Ellis, B.E., Douglas, C., Bohlmann, J.& Ritland, K. 2004. Robust simple sequence repeat markers for spruce (Picea spp.) from expressedsequence tags. Theor. Appl. Genet. 109: 1283–1294.Septiningsih, E.M., Prasetiyono, J., Lubis, E., Tai, T.H., Tjubaryat, T., Moeljopawiro, S. &McCouch, S.R. 2003a. Identification of quantitative trait loci for yield and yield components inan advanced backcross population derived from the Oryza sativa variety IR64 and the wild relativeO. rufipogon. Theor. Appl. Genet. 107: 1419–1432.Septiningsih, E.M., Trijatmiko, K.R., Moeljopawiro, S. & McCouch, S.R. 2003b. Identification ofquantitative trait loci for grain quality in an advanced backcross population derived from the Oryzasativa variety IR64 and the wild relative O. rufipogon. Theor. Appl. Genet. 107: 1433–1441.Song, W.Y., Wang, G.L., Chen, L.L., Kim, H.S., Pi, L.Y., Holsten, T., Gardner, J., Wang, B., Zhai,W.X., Zhu, L.H., Fauquet, C. & Ronald, P. 1995. A receptor kinase-like protein encoded by therice disease resistance gene, Xa21. Science 270: 1804–1806.Southern, E.M. 1975. Detection of specific sequences among DNA fragments separated by gel electrophoresis.J. Mol. Biol. 98: 503–510.Su, C.C., Cheng, X.N., Zhai, H.Q. & Wan, J.M. 2002. Detection and analysis of QTL for resistanceto the brown planthopper, Nilaparvata lugens (Stal), in rice (Oryza sativa L.), using backcrossinbred lines]. Yi Chuan Xue Bao 29: 332–338.Sun, X., Cao, Y., Yang, Z., Xu, C., Li, X., Wang, S. & Zhang, Q. 2004. Xa26, a gene conferringresistance to Xanthomonas oryzae pv. oryzae in rice, encodes an LRR receptor kinase-like protein.Plant J. 37: 517–527Takeuchi, Y., Lin, S.Y., Sasaki, T. & Yano, M. 2003. Fine linkage mapping enables dissection ofclosely linked quantitative trait loci for seed dormancy and heading in rice. Theor. Appl. Genet.107: 1174–1180.Tanksley, S.D. & McCouch, S.R. 1997. Seed banks and molecular maps: unlocking genetic potentialfrom the wild. Science 277: 1063–1066.Taylor, G., 2002. Populus: arabidopsis for forestry. Do we need a model tree? Ann. Bot. (London) 90:681–689.Temnykh, S.P., Park, W.D., Ayres, N., Cartinhour, S., Hauck, N., Lipovich, L., Cho,Y-G., Ishii,T. & McCouch, S.R. 2000. Mapping and genome organization of microsatellite sequences in rice(Oryza sativa L.). Theor. Appl. Genet. 100: 697–712.Temnykh, S., DeClerck, G., Lukashova, A., Lipovich, L., Cartinhour, S. & McCouch, S. 2001.Computational and experimental analysis of microsatellites in rice (Oryza sativa L.): frequency,length variation, transposon associations, and genetic marker potential. Genome Res. 11:1441–1452.Testolin, R., Huang, W.G., Lain, O., Messina, R., Vecchione, A. & Cipriani, G. 2001. A kiwifruit(Actinidia spp.) linkage map based on microsatellites and integrated with AFLP markers. Theor.Appl. Genet. 103: 30–36.Thomson M., Septiningsih E. and Sutrisno. 2003. Workshop on IPR in relation to Plant Breeding,Cipanas, Indonesia, 10–18 November.Thomson, M.J., Tai, T.H., McClung, A.M., Lai, X.H., Hinga, M.E., Lobos, K.B., Xu, Y., Martinez,C.P. & McCouch, S.R. 2003. Mapping quantitative trait loci for yield, yield components and

48Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishmorphological traits in an advanced backcross population between Oryza rufipogon and the Oryzasativa cultivar Jefferson. Theor. Appl. Genet. 107: 479–493.Upholt, W.B. 1977. Estimation of DNA sequence divergence from comparison of restriction endonucleasedigests. Nucl. Acids Res. 4: 1257–1265.Valent, B., Bryan, G.T., Jia, Y., Farrall, L., McAdams, S.A., Faulk, K.N. & Levy, M. 2001. Enhancingdeployment of rice blast resistance genes: opportunities from cloning a resistance gene/avirulencegene pair. In G.J. Khush, D.S. Brar & B. Hardy, eds. Rice genetics IV, pp. 309-322. Enfield, NH,USA, Science Publishers.Vos, P., Hogers, R., Bleeker, M., Reijans, M., van de Lee, T., Hornes, M., Frijters, A., Pot,J., Peleman, J. & Kuiper, M. 1995. AFLP: a new technique for DNA fingerprinting. Nucl.Acids Res. 23: 4407–4414.Wang Z.X., Yano, M., Yamanouchi, U., Iwamoto, M.I., Monna, L., Hayasaka, H., Katayose, Y. &Sasaki, T. 1999. The Pib gene for rice blast resistance belongs to the nucleotide binding and leucinerichrepeat class of plant disease resistance genes. Plant J. 19: 55–64.Weber, J.L. & May, P.E. 1989. Abundant class of human DNA polymorphisms which can be typedusing the polymerase chain reaction. Am. J. Hum. Genet. 44: 388–396.Welsh, J. & McClelland, M. 1990. Finger-printing genomes using PCR with arbitrary primers. Nucl.Acids Res. 18: 7213–7218.Wenzl, P., Carling, J., Kudrna, D., Jaccoud, D., Huttner, E., Kleinhofs, A. & Kilian, A. 2004.Diversity arrays technology (DArT) for whole-genome profiling of barley. Proc Natl Acad Sci USA101: 9915–9920.Williams, J.G., Kubelik, A.R., Livak, K.J., Rafalski J.A. & Tingey, S.V. 1990. DNA polymorphismsamplified by arbitrary primers are useful as genetic markers. Nucl. Acids Res. 18: 6531–6535.Wu, K.S. & Tanksley, S.D. 1993. Abundance, polymorphism and genetic mapping of microsatellitesin rice. Mol. Gen. Genet. 241: 225–235.Xu, Y., Ishii, T. & McCouch, S.R. 2003. Marker-assisted evaluation of germplasm resources for plantbreeding, In T.W. Mew, D.S. Brar, S. Peng, D. Dawe & B. Hardy, eds. Rice science: innovations andimpact for livelihoods. Proc. Internat. Rice Res. Conf., pp. 213-230. Beijing, China, IRRI, ChineseAcademy of Engineering and Chinese Academy of Agricultural Sciences.Yamamoto, T., Lin, H., Sasaki, T. & Yano, M. 2000. Identification of heading date quantitative traitlocus Hd6 and characterization of its epistatic interactions with Hd2 in rice using advanced backcrossprogeny. Genetics 154: 885–891.Ye, X., Al-Babili, S., Kloti, A., Zhang, J., Lucca, P., Beyer, P. & Potrykus, I. 2000. Engineering theprovitamin A (beta-carotene) biosynthetic pathway into (carotenoid-free) rice endosperm. Science287: 303–305.Yoshimura, S., Yoshimura, A., Saito, A., Kishimoto, N., Kawase, M., Yano, M., Nakagahara, M.,Ogawa, T. & Iwata, N. 1992. RFLP analysis of introgressed chromosomal segments in three nearisogeniclines of rice for bacterial blight resistance genes, Xa-1, Xa-3 and Xa-4. Jpn. J. Genet. 67:29–37.Yoshimura. S., Iwata, N.A.Y., McCouch, S.R., Abenes, M.L., Baraoidan, M.R., Mew, T.W. &Nelson, R.J. 1995. Tagging and combining bacterial blight resistance genes using RAPD and RFLPmarkers. Mol. Breed. 1: 375–387.

Chapter 3 – Molecular markers for use in plant molecular breeding and germplasm evaluation 49Yoshimura, S., Yamanouchi, U., Katayose, Y., Toki, S., Wang, Z.X., Kono, I., Kurata, N., Yano, M.,Iwata, N. & Sasaki, T. 1998. Expression of Xa1, a bacterial blight-resistance in rice, is induced by bacterialinoculation. Proc. Nat. Acad. Sci. USA 95: 1663–1668.Zane, L., Bargelloni, L. & Atarnello, T.P. 2002. Strategies for microsatellite isolation: a review. Mol.Ecol. 11: 1–16.Zhao, X. & Kochert, G. 1993. Phylogenetic distribution and genetic mapping of a (GGC)nmicrosatellite from rice (Oryza sativa L.). Plant Mol. Biol. 21: 607–614.Zhang, L.Y., Bernard, M., Leroy, P., Feuillet, C. & Sourdille, P. 2005. High transferability of breadwheat EST-derived SSRs to other cereals. Theor. Appl. Genet. 111: 677–687.Zheng, H.G., Babu, R.C., Pathan, M.S., Ali, L., Huang, N., Courtois, B. & Nguyen, H.T. 2000.Quantitative trait loci for root-penetration ability and root thickness in rice: comparison of geneticbackgrounds. Genome 43: 53–61.Zietkietkiewicz, E.A.R., Rafalski, A. & Labuda, D. 1994. Genome fingerprinting by simple sequencerepeat (SSR)-anchored polymerase chain reaction amplification. Genomics 20: 176–183.

Chapter 4Marker-assisted selection in wheat:evolution, not revolutionRobert Koebner and Richard Summers

52Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishSummaryThis chapter reviews the uptake of marker-assisted selection (MAS) in wheat in a Europeancontext. Although less intense than the scale of its application in maize, reflecting the factthat maize varieties are predominantly F 1 hybrids, the use of MAS in wheat has grown overthe last few years. This growth has been encouraged by an increase in the number of amenabletarget traits, but more significantly by a combination of technological improvements,particularly in the areas of DNA acquisition, laboratory management systems and integrationinto the breeding cycle, which together have served to reduce the per unit cost of eachdata point. Microsatellites (simple sequence repeats [SSRs]) are, and will likely remain forsome time, the marker of choice because of their flexibility and the knowledge base associatedwith them. Some current examples are provided of the use of MAS in a major UnitedKingdom commercial breeding programme.

Chapter 4 – Marker-assisted selection in wheat: evolution, not revolution 53IntroductionWheat is a very important world staplecrop. The 2005 United States Departmentof Agriculture (USDA) estimates for theglobal production of wheat (both breadand durum) and maize are, respectively,627 million tonnes and 708 million tonnes.In Europe, bread wheat is without doubtthe most important broad-acre crop, witha production in the extended EuropeanUnion of 25 states of 115 million tonnes(maize 48 million tonnes). The largest productionand highest productivity of breadwheat are achieved in northwest Europe.Historically, wheat has been bred largelyby government-sponsored national andregional programmes, but the introductionof plant variety rights into Europe inthe 1960s encouraged participation by theprivate sector. Currently, wheat breedingin northwest Europe is almost exclusivelycarried out by private companies, withsome research underpinning by the publicsector. Breeders continue to be successfulin the production of high-yielding, diseaseresistant,high-quality varieties and, in theUnited Kingdom at least, genetic advancesfor yield have been running at between 0.5to 1 percent per annum for many years.Wheat is a naturally inbreeding species,and although a level of heterosis canbe demonstrated, difficulties in enforcingcross-pollination in a reliable and cost-effectiveway have hindered the development ofany significant contribution of F 1 hybridsto the variety pool. Most varietal developmentprogrammes are therefore based onversions of the long-established pedigreebreeding system, where large F 2 populationsare generated and conventional phenotypicselection is carried out in early generationsfor highly heritable, qualitative traits(such as disease resistance) and in later onesfor quantitative traits (primarily yield andquality). Thus, most varieties are bred andgrown as inbred, pure breeding lines. As aresult, the unit value of seed and economicmargins for breeders are low. By contrast,maize is a naturally out-crossing species thatshows highly significant levels of heterosis.This has resulted in the majority of maizebreeding being geared to the productionof F 1 hybrids. In industrialized countries,maize hybrid breeding has for some timebeen dominated by a small number of largeprivate sector companies that are able to sustainprofitability through their control overthe genotype of their varieties. No revenue islost as a result of the use of farm-saved seed,and the inbred components of a successfulhybrid are not available to competitors touse as parental material for their own varietalimprovement programmes. This hasfar-reaching implications on the feasibilityof MAS in maize, and largely explains thelead that maize enjoys over wheat in thedeployment of MAS technology.The continuing development of molecularmarker technology over the last decadehas been a happy by-product of “bigbiology” genomics research. As recentlyas 1996, the definition of 5 000 SSR loci inthe human genome merited a major publicationin Nature (Dib et al., 1996), but thenumber of known human single nucleotidepolymorphisms (SNPs) now runs into millions.Thus, although marker availability,potentially at least, is no longer limitingin crops, and the clear potential benefitsof marker deployment to plant breedingare undisputed, only relatively recentlyhas it begun to make more than a marginalimpact on breeding methodology. Evenin maize, where the level of DNA markerpolymorphism is high, large-scale deploymentof MAS did not gather any significantmomentum until more than 15 years afterthe publication of the first restriction frag-

54Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishment length polymorphism (RFLP)-basedmaize genetic map. In the less geneticallyvariable cereals, prominently wheat,the level of polymorphism is not now inpractice likely to represent the major constraintto MAS uptake, although in the pastit was argued that this was the case. Whathas changed in recent times is that currentmarker technology, and systems ofDNA acquisition, laboratory managementand integration into the breeding cycle,have all developed to the extent where thebenefits of MAS can be increasingly realizedin actual practice. As many of theseimprovements are incremental rather thansudden, we argue that the trends in MASapplication in wheat are characteristicallyevolutionary rather than revolutionary.Target traits for MAS innorthwest European winterwheat breedingThe use of MAS to date has a history ofabout 20 years, and until recently involvedthe exploitation of just two non-DNAbasedassays. The first, which has beenretained with only slight modificationssince its inception, exploits a correlationbetween bread-making quality and allelicstatus at the Glu-1 (endosperm storageprotein subunit) loci. It uses electrophoreticprofiles obtained by the straightforward,robust and cheap procedure sodium dodecylsulphate polyacrylamide gel electrophoresis(SDS-PAGE) from crude seed proteinextracts, which have been shown to bepartially predictive of end-use quality. Thesecond is predictive for the presence of thegene Pch1, which confers a high level ofresistance to eyespot, a stem base diseasethat is difficult to screen using conventionalpathology methods. Both these targets havein the meanwhile become assayable bypolymerase chain reaction (PCR)-basedassays, although SDS-PAGE remains inroutine use thanks to its flexibility and costeffectiveness. In recent years, the numberof loci for which DNA-based assays havebeen generated has increased dramatically,the majority using PCR as a technologyplatform. Over 50 of these are described(specifically in a United States of Americacontext) at http://maswheat.ucdavis.edu/,which reports the output of an ongoingUnited States Department of Agriculture(USDA)-funded programme. The focusis heavily on disease and pest resistance,reflecting the generally simple inheritanceof genes conferring these traits.Some of the above traits are of sufficientrelevance to the United Kingdom contextthat identical or equivalent assays havebeen incorporated in a number of breedingprogrammes, where they are used as guidesto parental selection and/or in early generationselection. Prominent among these aremarkers for the genes Rht-1 (responsible forthe “Green Revolution” semi-dwarfism),Pinb (grain texture), Pch1, Lr37/Yr17 (agene complex conferring resistance to twoof the most important leaf fungal pathogens)and the wheat/rye translocation1B/1R (which is associated with high levelsof yield). Emerging MAS targets are necessarilyprogramme-dependent, but the broadfocus is on quantitative trait locus (QTL)targets that could have a major impact onbreeding efficiency. In the United Kingdom,as elsewhere worldwide, current focus is onresistance to the diseases Fusarium headblight (FHB), Septoria tritici blotch (STB)and barley yellow dwarf virus (BYDV), andon durable resistance to yellow rust. Othercurrent targets, more specific to the UnitedKingdom and northwest European context,but in routine use, are resistance to theinsect pest orange blossom midge (OBM),and soil-borne mosaic virus (SBMV).

Chapter 4 – Marker-assisted selection in wheat: evolution, not revolution 55FHBThe importance of FHB is less in its effecton yield reduction, but rather on the potentiallydamaging reduction in grain qualityassociated with infected grain, which canbe heavily contaminated by the fungal tricothecintoxins. An important source ofFHB resistance originates from the Chinesevariety Sumai 3, and a major component ofthis resistance (up to 50 percent) has beenassociated with a single QTL (Waldron etal., 1999; Anderson et al., 2001; Buerstmayret al., 2002). While this QTL is largelyeffective in preventing the spread of thepathogen following infection, a further QTLthat gives a significant degree of protectionagainst initial infection has been mapped toa different chromosome (Buerstmayr et al.,2003). Selection for FHB resistance by conventionalmeans is complicated both by thequantitative nature of the Sumai 3 resistanceand by difficulties in ensuring evenand reliable artificial infections in breedingnurseries. However, SSR-based MAS protocolshave been developed for both QTL(see http://maswheat.ucdavis.edu/ andBuerstmayr et al., 2003), and the urgencyof breeding for resistance has ensured thatincreasing use is being made of such assays.Both these QTL in concert do not explainall the genetic resistance of Sumai 3 to FHB,but the remainder appears to be determinedby QTL of minor effects and/or pleiotropiceffects associated with an ear morphology,which is inconsistent with a northwestEuropean winter wheat ideotype.STBSTB of wheat is caused by the fungusMycosphaerella graminicola (syn. Septoriatritici), and in recent years has become themajor leaf disease of wheat in many regionsof the world. In past years, good levels ofcontrol were achieved by the application ofstrobilurin fungicides, but their heavy usehas led to the emergence of pathogen strainsthat cannot be so easily controlled by chemicalmeans. A number both of major genesgiving near-complete resistance to specificraces of the pathogen and of quantitativerace non-specific resistances with polygenicinheritance have been defined, andone of the former, Stb6, which maps closeto the SSR locus Xgwm369 on chromosome3A (Chartrain, Brading and Brown,2004), is common in many gene pools. Thisensures that the gene has been retained inelite materials, and its known map positionhas made it relatively straightforward touse a marker assay to track its presence inbreeding populations.BYDVSignificant grain yield losses are attributableto natural infections of BYDV, and no majorsource of resistance has been identified todate in wheat. Control is achieved in theabsence of genetic resistance by insecticidalspray, which is associated with both aneconomic and an environmental cost.However, a potent resistance is present in therelated species Thinopyrum intermedium.It is possible to generate sexual hybridsbetween wheat and this grass, but the F 1plants are self-sterile and either have tobe rescued by chromosome doubling orback-crossed to wheat. By this route, adistal segment of the grass chromosomethat carries the BYDV resistance geneBdv2 has been introduced into wheat. Asthis introgression comprises a significantlength of non-wheat chromosome, it hasbeen relatively straightforward to generatemarkers suitable for MAS use (Ayala et al.,2001a; Zhang et al., 2004). A MAS approachfor screening is attractive because artificialinoculation involves the propagation ofvirus-bearing aphids, while natural infections

56Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishare unreliable. Interestingly, unlike theexperience with many alien introgressionsegments, no obvious negative effects of itspresence on agronomic performance haveyet been detected either in InternationalMaize and Wheat Improvement Center(CIMMYT) trials (Ayala et al., 2001b) or atRAGT Seeds (Cambridge, UK).Durable resistance to yellow rustYellow rust is historically the most damagingof the leaf fungal pathogens in temperateEurope. Control has been achieved in thepast largely by a combination of fungicideapplication and of combinations ofmajor seedling resistance genes, of whicha significant number have been describedin the literature. However, like most racespecificresistances, most of these majorgenes have lost their effectiveness, and thishas led to a renewed effort in the definitionof partial or adult plant resistances tothis disease. The French variety Cappelle-Desprez dominated the wheat crop acrossFrance and the United Kingdom during the1960s and 1970s, and maintained its levelof adult resistance to yellow rust over thewhole of this period. A major part of thegenetic basis for this durable resistance waslocated to a translocated wheat chromosome(Law and Worland, 1997), and this hasbeen confirmed by a rigorous QTL analysis(Mallard et al., 2005), which has provided anumber of informative SSR markers for thiseffect. Other independent sources of adultresistance have been identified in Frenchand Eastern European germplasm at RAGTSeeds, and the major QTL responsible havebeen defined and marked.OBMOBM larvae feed on developing grain andheavy infestations result in a significantreduction in grain quality and some loss inyield. As for many sporadic pests, phenotypicscreening is unreliable and an indirectmeans of selection would be valuable.The gene Sm1 confers resistance to OBM(Sitodiplosis mosellana) by the expression ofan antibiotic that kills or slows the developmentof larvae. Thomas et al. (2005) definedthe map position of Sm1 and proposed aclose linkage with an SSR locus Xbarc35.This linkage remains to be validated inUnited Kingdom breeding populations, asit remains unclear whether the antibioticeffect shown by a few United Kingdomwheat varieties is conferred by Sm1.SBMVSBMV is one of two known viral pathogenstransmitted by the soil fungus Polymyxagraminis (another one being yellow mosaicvirus [YMV]), and can be an important agentof yield loss in some areas. Chemical controlis not feasible, and once soil is infected bythe virus-bearing host, the only solutionspossible are to abandon wheat culture orto use resistant varieties. Phenotyping isparticularly difficult as plant infection isenvironmentally sensitive, and the detectionof infection is laborious and prone to error.A proprietary assay for resistance to SBMVoriginating from European germplasm hasbeen in routine use at RAGT Seeds since2000 with a very high level of marker/phenotype association. More recently, abulk segregant analysis along with a QTLapproach has allowed the definition of aresistance locus to YMV from Chinesegermplasm, and a number of linked SSRmarkers have been identified (Liu et al.,2005).High-throughputinfrastructuresTechnical considerations of DNA acquisition,laboratory information management

Chapter 4 – Marker-assisted selection in wheat: evolution, not revolution 57system (LIMS), laboratory automation anddata capture and analysis are generic forany MAS set-up, and these are well coveredelsewhere in this volume. The limitationsaffecting MAS deployment in wheat flowfrom the restricted revenue generated bybreeding a self-pollinated, homozygous,non-hybrid product. As a result, the volumeof capital investment affordable in maize isnot available to a wheat MAS programme.Financial constraints also affect the developmentof marker platforms. It is wellknown that the predictive ability of a linkedmarker will be disrupted by recombination,and therefore that “perfect” markers aremore desirable than linked ones. However,the development of genome-wide genebasedmarkers, pre-eminently SNPs, whichare particularly suited to high-throughputgenotyping on automated platforms, is stillsome way off. At present, an insufficientnumber of such assays has been established(grain hardness, semi-dwarfness and graintexture) to consider adjusting the presentmajor genotyping methodology, which isfounded on SSRs. Doubts have been raisedthat SNP frequency in exon sequence willbe high enough to generate informativeassays for many critical genes, but earlyexperience suggests that sequence polymorphismis more than adequate in introns andother untranslated regions of wheat genes.At present, the consensus is that there isplenty of mileage left in SSR technology,and wheat maps continue to be refined bythe addition of new SSR loci.ConclusionIn 1999, Young set out his “cautiouslyoptimistic vision” for MAS. Seven years on,the situation continues to crystallize. Thetechnology itself is no longer limiting. Withrespect to marker availability, SSRs remainuseful and SSR-based genetic maps arebecoming increasingly densely populated,while SNPs may eventually represent asource of plentiful perfect markers for genesof defined function. The “big biology”spawned by the genomics revolution hasbrought miniaturization and automation tobiological assays so that levels of throughputrelevant to the wheat breeding processare becoming attainable. The issue thatremains unresolved is the affordability oflarge-scale MAS. As wheat is a broad-acrecommodity product, its value is low, andthis impedes the ability of the industry toinvest in MAS infrastructure to the extentthat is possible for crops such as maizewhere the generation of F 1 hybrid seed is aviable proposition. However, as economiesof scale and improvements in technologycontinue to drive down assay price, thepenetration of MAS into commercial wheatbreeding will surely grow. This growthshould progressively allow a widening in therange of possible MAS targets, in particularextending to critical ones such as QTL foryield and its components (mean kernelsize, kernel number per ear and numberof fertile tillers per unit area). These arealready widely exploited in maize breedingand their definition and validation in wheatrepresent a significant research theme inboth the public and private sectors. In themeantime, much MAS use will be directedtowards specific purposes such as acceleratedselection of a few traits that are difficultto manage by conventional phenotyping,for the maintenance of recessive allelesin backcrossing programmes, for thepyramiding of disease resistance genes andfor guiding the choice of parents to be usedin crossing programmes.

58Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishReferencesAnderson, J.A., Stack, RW., Liu, S., Waldron B.L., Fjeld, A.D., Coyne, C., Moreno-Sevilla, B.,Mitchell Fetch, J., Song, Q.J., Cregan, P.B. & Frohberg, R.C. 2001. DNA markers for Fusariumhead blight resistance QTLs in two wheat populations. Theor. Appl. Genet. 102: 1164–1168.Ayala, L., Henry, M., Gonzalez-de-Leon, D., van Ginkel, M., Mujeeb-Kazi, A., Keller, B. &Khairallah, M.A. 2001a. Diagnostic molecular marker allowing the study of Th. intermediumderivedresistance to BYDV in bread wheat segregating populations. Theor. Appl. Genet. 102:942–949.Ayala, L., van Ginkel, M., Khairallah, M., Keller, B. & Henry, M. 2001b. Expression of Thinopyrumintermedium-derived barley yellow dwarf virus resistance in elite bread wheat backgrounds.Phytopath. 91: 55–62.Buerstmayr, H., Lemmens, M., Hartl, L., Doldi, L., Steiner, B., Stierschneider, M. & Ruckenbauer,P. 2002. Molecular mapping of QTLs for Fusarium head blight resistance in spring wheat. I.Resistance to fungal spread (Type II resistance). Theor. Appl. Genet. 104: 84–91.Buerstmayr, H., Steiner, B., Hartl, L., Griesser, M., Angerer, N., Lengauer, D., Miedaner, T.,Schneider, B. & Lemmens, M. 2003. Molecular mapping of QTLs for Fusarium head blight resistancein spring wheat. II. Resistance to fungal penetration and spread. Theor. Appl. Genet. 107:503–508.Chartrain, L., Brading, P.A. & Brown, J.K.M. 2004. Presence of the Stb6 gene for resistance toSeptoria tritici blotch (Mycosphaerella graminicola) in cultivars used in wheat-breeding programmesworldwide. Plant Pathol. 54: 134–143.Dib, C., Faure, S., Fizames, C., Samson, D., Drouot, N., Vignal, A., Millasseau, P., Marc, S.,Hazan, J., Seboun, E., Lathrop, M., Gyapay, G., Morissette, J. & Weissenbach, J. 1996. A comprehensivegenetic map of the human genome based on 5 264 microsatellites. Nature 380: 152–154.Law, C.N. & Worland, A.J. 1997. The control of adult plant resistance to yellow rust by the translocatedchromosome 5BS-7BS of bread wheat. Plant Breed. 116: 59–63.Liu, W.H., Nie, H., Wang, S.B., Li, X., He, Z.T., Han, C.G., Wang, J.R., Chen, X.L., Li, L.H. & Yu,J.L. 2005. Mapping a resistance gene in wheat cultivar Yangfu 9311 to yellow mosaic virus, usingmicrosatellite markers. Theor. Appl. Genet. 111: 651–657.Mallard, S., Gaudet, D., Aldeia, A., Abelard, C., Besnard, A.L., Sourdille, P. & Dedryver, F. 2005.Genetic analysis of durable resistance to yellow rust in bread wheat. Theor. Appl. Genet. 110:1401–1409.Thomas, J., Fineberg, N., Penner, G., McCartney, C., Aung, T., Wise, I. & McCallum, B. 2005.Chromosome location and markers of Sm1: a gene of wheat that conditions antibiotic resistance toorange wheat blossom midge. Mol. Breed. 15: 183–192.Waldron, B.L., Moreno-Sevilla, B., Anderson, J.A., Stack, R.W. & Frohberg, R.C. 1999. RFLPmapping of QTLs for Fusarium head blight resistance in wheat. Crop Sci. 39: 805–811.Young, N.D. 1999. A cautiously optimistic vision for marker-assisted selection. Mol. Breed. 5:505–510.Zhang, Z.Y., Xu, J.S., Xu, Q.J., Larkin, P. & Xin, Z.Y. 2004. Development of novel PCR markerslinked to the BYDV resistance gene Bdv2 useful in wheat for marker-assisted selection. Theor.Appl. Genet. 109: 433–439.

Chapter 5Marker-assisted selection forimproving quantitative traitsof forage cropsOene Dolstra, Christel Denneboom,Ab L.F. de Vos and E.N. van Loo

60Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishSummaryThis chapter provides an example of using marker-assisted selection (MAS) for breedingperennial ryegrass (Lolium perenne), a pasture species. A mapping study had shownthe presence of quantitative trait loci (QTL) for seven component traits of nitrogen useefficiency (NUE). The NUE-related QTL clustered in five chromosomal regions. TheseQTL were validated through divergent marker selection in an F 2 population. The criterionused for plant selection was a summation index based on the number of positive QTLalleles. The evaluation studies showed a strong indirect response of marker selectionon NUE. Marker selection using a summation index such as applied here proved to bevery effective for difficult and complex quantitative traits such as NUE. The strategy iseasily applicable in outbreeding crops to raise the frequency of several desirable allelessimultaneously.

Chapter 5 – Marker-assisted selection for improving quantitative traits of forage crops 61IntroductionMost agronomical characteristics of foragecrops have a quantitative, polygenic andmostly complex nature. For these reasons,genetic improvement of such traits islaborious and time consuming. Improvingnitrogen use efficiency (NUE) in perennialryegrass (Lolium perenne, 2n = 14), themajor grass species in northern Europe, isin this respect a good example. The highinput of nitrogen needed to attain highforage yields for animal husbandry hascaused severe water pollution (van Loo etal., 2003), and therefore lowering nitrogeninputs through improving nitrogen use bybreeding is of utmost importance.Selection for NUE, however, is not easilyimplemented in conventional grass breedingbased on field evaluations. Adequatetesting requires separate and long-term trialswith good control of the N stress, andsuch experiments tend to be rather inaccurate.To circumvent the disadvantagesof field testing, a hydroponics system wasused in this study in which the crop situationis simulated with growth-dependentN application (van Loo et al., 1992), theaim being to grow plants having an equalsuboptimal N content. The set-up has acapacity to test about 1 600 plants in paralleland enables all plants to experience moreor less the same N strain. Criteria used tomeasure NUE are several plant growthcharacteristics, such as tillering, and shootand root growth. Each test usually requiresfour to five cuts. The trait is vigour-relatedand complex, and is extremely important inrelation to regrowth after cutting. Together,all these aspects make NUE a very attractivetrait for MAS.Analysis of genetic variationThe genetic variation for NUE presentin an F 1 plant originating from a crossbetween two contrasting genotypes forNUE was first analysed by crossing theF 1 with a doubled haploid. The resultingtest cross progeny was then used to producea molecular marker map and analysethe variation. This approach was chosento avoid inbreeding effects and to be ableto use dominant molecular markers. Theperformance of the mapping population forNUE-related traits was studied on hydroponicswith the system set at a moderatelylow nitrogen deficiency (3.6 percent N ofleaf dry weight). The outcome of the mappingstudy was a genetic map with sevenlinkage groups.Putative genes (quantitative trait loci[QTL]) for the components of NUE werefound on four linkage groups. The locationof the selection markers for QTL is depictedin Figure 1. The map shows five genomicsites with 1-5 QTL. In total, 13 QTL forseven NUE related traits were found. Threesites contain more than one QTL.The findings of the current study aretypical for genetic analyses of quantitativetraits in forage crops and also indicativeof the problems associated with exploitationof QTL information through markerassistedbreeding. These included uncertaintieswith respect to effect and locationof QTL, the fairly large number of QTLoften found in genetic analyses, the cosegregationof QTL and the weighing ofthe different component traits of NUE andNUE-QTL. Below is a description of howthese breeding problems were solved orcircumvented in a divergent marker selectionstudy to validate the QTL found in themapping study.Divergent marker selectionThe plant materials used in the validationstudy were an F 2 generation obtained byselfing of the heterozygous F 1 genotype

62Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishFigure 1Map position of the selection markers (in blue) for QTL for seven components ofnitrogen use efficiencyLG1LG3LG4LG6E32M49-164E37M49-4030 K10F08-1415 K10F08-14410 E32M52-02R15 E32M49-21920 E32M54-05R25 E38M48-04R30 E38M48-13035E32M52-352E32M48-3034045 E41M48-23950 E32M52-01R55E36M48-220E41M47-42560E32M52-04R65E41M61-362707580859095100105110115120125130135140145150155160165170E36M48-180E36M50-094Uni-001-147K07H08-165E40M47-380E38M51-285E38M49-223LW1 E36M49-213LAE3 E37M49-343E37M49-118E35M48-252E38M49-309E41M48-479E38M51-429K14F12-113K14F12-123E32M54-03RE38M49-305E35M48-339E38M51-01RE33M49-233E32M51-313E37M49-201E37M48-265E41M48-205E38M49-242TN1DWS1LAE1DWT1LWR1Rye35-117K04D01-157K04D01-239E35M48-328E41M48-268E37M48-347E36M50-233E38M48-11RADH-1000Rye12-148Rye26-147E32M51-303E36M49-133E38M48-416E32M51-328E41M48-296E38M51-247E33M49-277E36M49-217E38M48-08RE37M48-331E38M51-14RE41M48-299E32M51-212E38M47-334E38M47-255E38M47-348E35M47-135E41M48-208E38M48-093E38M51-291E38M51-04RE32M51-304E33M51-03RE38M52-262E38M52-260E38M51-12RE38M51-267Meth300-250E36M48-108E37M47-441KXX104-108TN2DWS2LAE2DWT2DWR1Rye14-221Rye14-223E38M52-289E36M49-297E36M50-173E32M52-169E38M51-13RE38M48-10RE38M48-05RE38M48-06RE32M52-08RE38M48-110E38M51-264E32M52-126E32M52-06RE36M48-163E35M48-141E38M47-116E32M54-06RE37M47-327E32M52-137E32M52-07RE35M47-278LWR2TN: Tiller numberLAE: Leaf area expansionLW: Leaf widthLWR: Leaf weight ratioDWR: Dry weight rootsDWS: Dry weight shootsDWT: Total dry weightused to generate the mapping populationmentioned above (van Loo et al., 2003). Intotal, about 200 genotypes were genotypedfor five amplified fragment length polymorphism(AFLP) selection markers usingthe fluorescent AFLP technique developedby Applied BioSystems (Figure 1). Themarkers were co-dominantly scored usingthe heights of the fluorescence peaks relativeto those of homozygous fragments asa criterion.The genotyping data were used subsequentlyas a basis for a divergent massselection programme. The selection strategyis outlined in Figure 2. The selectioncriterion was a genotype-specific selectionindex, being the summation of allpositive QTL alleles (or chromosome segments)over the five QTL sites considered(Figures 1 and 2).Application of marker selectionThe AFLP technique is usually not themarker technology of choice for selectionpurposes because of its dominant nature andhigh costs per selection marker. However,co-dominant scoring of the five selectionmarkers was quite adequate. The trimodalfrequency distributions allowed properclassification of plants, although some misclassificationcannot be fully excluded. Theadvantages of co-dominant AFLP scoringfrom a selection point of view are so largethat a small number of genotyping errorsare acceptable.The decision to use a summation indexas the criterion for selection was made primarilybecause of the difficulty of weightingthe individual NUE related traits and theco-localization of QTL. The designationof the positive QTL alleles (chromosome

Chapter 5 – Marker-assisted selection for improving quantitative traits of forage crops 63Figure 2Strategy applied for marker selectionCo-dominant scoring ofmarkers associated withQTLsGenotypeNumber of plus allelesQTL1QTL2QLT3QTL4QTL5SumSelectioncriterionF 21 2 1 0 0 2 52 1 1 0 0 2 43 0 2 2 1 1 64 2 2 2 1 1 85 2 2 1 1 1 7...195 1 1 2 2 2 8196 1 1 2 0 2 6197 0 2 0 1 2 5198 0 2 0 1 1 4199 2 2 1 1 1 7200 2 1 2 2 1 8fragments) turned out to be straightforward.Figure 3 shows the F 2 frequency distributionfor the number of “plus alleles”.The population mean is somewhat belowthe expected number of five owing to thefact that the AFLP marker on LG1 showeda skewed segregation. This is likely dueto gametophytic selection in favour of thenegative QTL allele, perhaps due to linkagewith an incompatibility locus.The intensities of selection were setat about 25 percent, representing about50 genotypes per selection (Figure 3). Theselection pressure was kept fairly lowbecause of the need to have sufficient seedsfor measuring selection responses. In thisway, the influence of genetic drift accompanyingmarker selection was minimized. Thecut-off point for the top selection was sixpositive alleles and three for the oppositeselection (Figure 3). The frequency of theplus alleles was on average 0.66 and 0.27,respectively. Selection showed a positiveresponse for all NUE loci. However, thebetween-selection difference in allele frequencyof the loci ranged from 0.18 to 0.77,showing that index selection did not affectall NUE loci to the same degree. The differenceswere probably mainly due to chance.Indirect response to markerselectionThe selections were then multiplied usinga polycross scheme (after vegetative propagation)to obtain sufficient seeds forevaluation on hydroponics and under variousfield conditions. The marker selectionswere evaluated for NUE in a replicated trialwith two cuts on hydroponics at two Nlevels, being 2.5 and 5 percent N in leaves(van Loo et al., 2003). The same set of plantcharacteristics as in the original mappingstudies was monitored after each cut. Leafarea expansion rate, leaf length and width,as well as tiller number, were determinedone week after cutting. The determinationof shoot and root dry weight followedthree weeks later. The indirect responsesto marker selection are summarized inFigure 4. At low N supply, the NUEplus

64Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishFigure 3Frequency distribution for number of plus alleles found for five QTLs706050Number of plants4030201000 1 2 3 4 5 6 7 8 9 10Number of plus allelesThe two contrasting marker selections, i.e. NUEplus (right) and NUEmin (left), are marked in blue.Figure 4Performance of the marker selections, NUEmin (white) and NUEplus (blue) on hydroponicsat low N-supply (2.5%N in leaves) and high N-supply (5%N in leaves). The performance isexpressed as a percentage of the mean of the two selections60 80 100 12060 80 100 120Tiller numberDry weight rootsDry weight shootsLeaf lengthRoot weight ratioLow N - supply High N - supplyselection showed a remarkable 40 percenthigher tillering rate and dry matter productionthan the NUEmin selection. The40 percent higher tillering rate is associatedwith a 40 percent higher leaf area increaseafter defoliation (data not shown). Relativeroot growth (expressed as the ratio of rootto total growth) and leaf length were hardly

Chapter 5 – Marker-assisted selection for improving quantitative traits of forage crops 65changed through marker selection. At highN supply, the performances of NUEplusand NUEmin were fairly similar (Figure 4).The selections also showed striking differencesin field trials in Germany, Englandand the Netherlands. At suboptimal N,the NUEplus selection significantly outperformedits counterpart in yields of drymatter and water soluble carbohydrates,while total N uptake was slightly lower.ConclusionsDivergent mass selection has shown thatmarker selection using a summation indexcan be very effective for difficult andcomplex quantitative traits such as NUE. Acollateral advantage of such an approach isthat it offers a true validation of the putativegenes (QTL) for the traits of interest. Theassociated response to marker selectiondistinctively indicates the presence of truegenes affecting NUE, particularly in thevicinity of markers, which were stronglyaffected by the selection imposed. Theresults also indicate that recurrent massselection to increase the number of positivealleles is worthwhile. The strategy is easilyapplicable in outbreeding crops.AcknowledgementsThe research work was carried out withinthe framework of the EU-FAIR projectNIMGRASS (CT98-4063).Referencesvan Loo, E.N., Schapendonk, A.H.C.M. & de Vos, A.L.F. 1992. Effects of nitrogen supply ontillering dynamics and regrowth of perennial ryegrass populations. Netherlands J. Agric. Sci. 40:401–419.van Loo, E.N., Dolstra, O., Humphreys, M.O., Wolters, L., Luessink, W., de Riek, W. & BarkN. 2003. Lower nitrogen losses through marker assisted selection for nitrogen use efficiency andfeeding value (NIMGRASS). Vorträge Pflanzenzüchtung 59: 270–279.

Chapter 6Targeted introgression of cottonfibre quality quantitative trait lociusing molecular markersJean-Marc Lacape, Trung-Bieu Nguyen, Bernad Hau and Marc Giband

68Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishSummaryWithin the framework of a cotton breeding programme, molecular markers are used toimprove the efficiency of the introgression of fibre quality traits of Gossypium barbadenseinto G. hirsutum. A saturated genetic map was developed based on genotyping dataobtained from the BC 1 (75 plants) and BC 2 (200 plants) generations. Phenotypic measurementsconducted over three generations (BC 1 , BC 2 and BC 2 S 1 ) allowed 80 quantitative traitloci (QTL) to be detected for fibre length, uniformity, strength, elongation, fineness andcolour. Positive QTL, i.e. those for which favourable alleles came from the G. barbadenseparent, were harboured by 19 QTL-rich regions on 15 “carrier” chromosomes. In subsequentgenerations (BC 3 and BC 4 ), markers framing the QTL-rich regions were used toselect about 10 percent of over 400 plants analysed in each generation. Although BC plantsselected through the marker-assisted selection (MAS) process show promising fibre quality,only their full field evaluation will allow validation of the procedure.

Chapter 6 – Targeted introgression of cotton fibre quality quantitative trait loci using molecular markers 69INtroductionAmong the four species of Gossypium thatproduce seeds with spinnable fibres calledcotton, Gossypium hirsutum dominates theworld’s cotton fibre production, accountingfor approximately 90 percent of total worldproduction. The second most cultivated species,G. barbadense, includes superior extralong, strong and fine cottons. However,compared with G. hirsutum, the marketingadvantage of “high quality” G. barbadensecottons is offset by their lower productivityand a narrower adaptability to harshenvironments. Breeding approaches withinthese two species have essentially relied onhybridization and selection methods (subsequentto simple or complex crosses, apedigree system, sometimes combined withrecurrent selection, is applied). AlthoughG. hirsutum and G. barbadense displaycomplementary characteristics, attempts toutilize deliberate interspecific G. hirsutum/G. barbadense recombinations throughconventional breeding have had limitedimpact on cultivar development.In the past 10–15 years, DNA markersfor analyses of QTL and MAS have receivedconsiderable attention by plant and animalbreeders (Dekkers and Hospital, 2002).However, following an initial keen interestand promises for molecular-based breedingapproaches, the successful application ofthis technology has been shown to dependon the reliability and accuracy of the QTLanalyses, which in turn are strongly affectedby both population size and environmentalfactors (Schön et al., 2004). Examples ofapplied MAS in breeding programmes arestill scarce, particularly when complextraits (yield components, product quality)are under consideration.In the case of cotton, it is only recentlythat the results of efforts to gain a betterunderstanding of the genome and themolecular basis of fibre quality have beenpublished. Most of the earlier efforts incotton molecular breeding concentratedon interspecific hybridization, due to thefact that, intraspecifically, the major speciesG. hirsutum displayed a very lowlevel of molecular variability (Brubakerand Wendel, 2001). Based on studies ofinterspecific G. hirsutum x G. barbadensepopulations, published reports relate (i) tothe construction of high-resolution geneticmaps (Lacape et al., 2003; Rong et al.,2004); and (ii) to the identification of fibrequality-related QTL (Jiang et al., 1998;Kohel et al., 2001; Paterson et al., 2003;Lacape et al., 2005). In parallel, data haveaccumulated describing the cotton fibretranscriptome (reviewed by Wilkins andArpat, 2005). These studies confirmed thatkey fibre quality properties, such as length,fineness and strength, are controlled quantitatively,thus complicating conventionalbreeding for fibre improvement.Within the framework of a marker-assistedbackcross introgression scheme aimed attransferring fibre quality traits from a lowproductivityline of G. barbadense (donor)into a productive line of G. hirsutum (recipient),a saturated genetic map of tetraploidcotton was first developed (Lacape et al.,2003). This chapter describes how molecularmarkers were used in the early BC 1 and BC 2generations to identify QTL-rich regionsinvolved in determining fibre quality, asrecently reported by Lacape et al. (2005),and how MAS was actually implemented inthe later BC 3 and BC 4 generations.MethodologyThe major milestones (Figure 1) in themarker-assisted backcross selection processincluded the construction of two geneticmaps from the BC 1 and BC 2 populations,the detection of fibre quality QTL from

70Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishFigure 1Schematic representation of the marker-assisted backcross selection schemeG. hirsutum x G. barbadense(var. Guazuncho 2) (var. VH8 -4602)BC 1 generation:1160 loci (RFLP + SSR + AFLP)phenotypic data and QTL mappingNo selectionF 1 xBC 1 xG.hG.hBC 2 generation: 200 plants514 loci (SSR + AFLP)200 BC 2 + 200 BC 2 S 1 phenotypic data and QTL mappingNo selectionBC 2BC 2 S 1BC 3 x G.hxG.hBC 3 generation:411 plants with 35 SSR loci and43 plants with 340 AFLP lociMarker selectionBC 4 generation:450 plants with 50 SSR loci and 37 plants with 340 AFLP lociMarker and phenotypic selectionBC 4 x G.hBC 4 S 1BC 5PyramidingQTL - NILsthree phenotyping data sets (BC 1 , BC 2 andBC 2 S 1 ) and the actual marker-based selectionin the BC 3 and BC 4 generations, followedby the analysis of marker-trait associationsin the BC 3 and BC 4 generations.Plant materialThe initial interspecific cross involved theG. hirsutum variety Guazuncho 2 andthe G. barbadense variety VH8-4602.Guazuncho 2 is a modern pure lineG. hirsutum variety created in Argentina andwas chosen as a recipient in the backcrossgenerations for its good overall agronomicperformance. VH8-4602, a G. barbadensevariety of the Sea Island type, was thedonor parent for superior fibre quality, inparticular for length (+9 to +12 mm as comparedwith Guazuncho 2), strength (+12 to+16 g/tex) and fineness (-30 to -50 millitex);conversely its fibre colour indices (reflectanceand yellowness) are of lower value. 1The plant material used in the multigenerationQTL analyses included threepopulations: BC 1 , BC 2 and BC 2 S 1 (Lacapeet al., 2005). The first backcross generation(BC 1 ), consisted of 75 plants grownin a greenhouse in Montpellier (France)during the summer of 1999; these servedas female parents for the second backcrossto Guazuncho 2. Two hundred individualfield-grown BC 2 plants that had shown asatisfactory production of BC 3 seeds andoriginating from 53 different BC 1 plantswere used in 2000. Open pollinated seedsharvested from BC 2 plants were grown as200 BC 2 S 1 progenies in 2001 under fieldconditions in Brazil. Each BC 2 S 1 line was11 tex = 1 gram/kilometre

Chapter 6 – Targeted introgression of cotton fibre quality quantitative trait loci using molecular markers 71planted in two replications, each plot (onerow) measuring 5 m. The next BC 3 andBC 4 generations were grown under field(411 BC 3 in 2002) or greenhouse (450 BC 4in 2003) conditions in Montpellier. Everyplant in each BC 1-4 generation was used forDNA extraction from young fresh leavesusing different methods described elsewhere(Lacape et al., 2003; Nguyen et al.,2004). In each BC 3 and BC 4 generation, anearly genotyping was conducted (beforeflowering of BC 3 plants and at the seedlingstage for BC 4 plants), to reduce the numberof plants to be manipulated and raised toflowering for selfing and backcrossing.From each generation (75 BC 1 , 200 BC 2 ,400 BC 2 S 1 , 43 selected BC 3 and 37 selectedBC 4 ), the cotton seed harvest was ginned(separation of the fibre from the seed) on alaboratory roller gin and the fibre was sampledfor analyses at the Fibre TechnologyLaboratory of the French AgriculturalResearch Centre for International Development(CIRAD).Fibre analysesAll fibre quality measurements (11 traits)were conducted at CIRAD, Montpellier,on a high volume instrument line (ZellwegerUster 900, Uster Technologies, Switzerland).These included length, uniformity, strength,elongation and colour. A FMT3 maturimeter(Shirley Dev Ltd., UK) was used to determinemicronaire value, maturity and fineness.Molecular analysesThe different types of markers displayingpolymorphism between G. hirsutum andG. barbadense included restriction fragmentlength polymorphisms (RFLPs) (used onlyin the BC 1 generation), simple sequencerepeats (SSRs) and amplified fragment lengthpolymorphisms (AFLPs). Details of themarkers and protocols used are providedin Lacape et al. (2003) and Nguyen et al.(2004). The AFLP markers were all derivedfrom combinations of EcoRI/MseI primerpairs (64 pairs in the BC 1 , 45 in the BC 2 and30 in the BC 3 and BC 4 generations). Thecotton microsatellites were derived essentiallyfrom two public libraries, BrookhavenNational Laboratory (BNL) and CIRAD(CIR). In the BC 1 generation, the microsatellitesused included 188 polymorphic BNLmarkers out of the 216 available (Lacapeet al., 2003) and 204 CIR markers out of392 developed (Nguyen et al., 2004). Fromthe results of the combined QTL analysesof the BC 1 /BC 2 /BC 2 S 1 generations (Lacapeet al., 2005), QTL-rich regions were identifiedon “carrier” chromosomes, and SSRloci present within or in the vicinity ofthese regions were assembled for constitutinggroups of three SSRs (one group perregion) to be tested as multiplexes, takinginto account both annealing temperatureand compatibility of sizes of amplified fragments.A subset of 60 SSR (20 region-specifictriplexes) was used for early genotyping ofall 411 BC 3 and 450 BC 4 plants (see examplesin Figure 2). The individual plants selectedfrom BC 3 and BC 4 (43 and 37 plants respectively)were further analysed using knownAFLPs to provide broad genome coverage.In the context of our marker-assisted introgressionprogramme, the SSR markerstarget the QTL-rich regions, i.e. those lociof the “foreground genome” expected tohave been introgressed, while the AFLPmarkers essentially serve to cover the rest ofthe genome, i.e. the “background genome”,aimed at returning to the recipient genomecomposition.Construction of genetic mapThe BC 1 (75 individuals) and BC 2 (200individuals) maps were constructed separatelyusing the MapMaker 3.0 software

72Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishFigure 2Examples of autoradiograms showing the segregation of 2 SSR triplexes observed amongBC 4 plants and used for the targeted genotyping of specific regions along chromosomes c5and c25 (a subset of around 15 BC 4 plants from the 450 plants analysed is shown)c5BNL3995CIR034CIR301CIR034CIR301BNL3995CIR280ac25CIR109CIR109BNL1047aBNL1047aCIR280a(Lander et al., 1987). The MapMaker“group” (using a logarithm of the oddsratio [LOD] of 5.0 and 30 as a maximalrecombination frequency), “order” and“sequence” commands were used in eachcase. After aligning the BC 1 and BC 2 mapsusing common loci, a consensus frameworkBC 1 /BC 2 map was constructed bysimple extrapolation of the positions of theadditional BC 2 loci on the BC 1 map usedas a backbone map. The allelic constitutionthroughout the 26 chromosomes of allBC 1-4 individuals was displayed graphicallyusing Graphical GenoTyping software (R.van Berloo, Laboratory of Plant Breeding,Wageningen, Netherlands) and representedalong the consensus BC 1 map data.QTL analysesThe combined marker and phenotypic datathen served for three (BC 1 , BC 2 and BC 2 S 1 )separate QTL analyses of fibre qualitycomponents. The association betweenphenotype and marker genotype was investigatedthrough simple marker analysis(SMA), interval mapping (IM) and compositeinterval mapping (CIM) using thecomputer software QTL Cartographer 1.13(Basten, Weir and Beng, 1999) as describedin Lacape et al. (2005). In each data set(trait, generation), permutation-basedthresholds were considered at a 5 percentrisk at the genome level. Interval methodsrelied on the positions of the loci on theconsensus BC 1 map. Molecular data offurther generations (BC 3 and BC 4 ) werealso combined with phenotypic measurementsfor conducting the SMA option ofQTL Cartographer. Cotton fibre propertieswere considered from a product transformationperspective, meaning thatdecreases the in fibre fineness and yellow-

Chapter 6 – Targeted introgression of cotton fibre quality quantitative trait loci using molecular markers 73Table 1Range of parental (G. hirsutum [Gh] and G. barbadense [Gb]) values over the five sets of dataGh Gb BC 1N=75BC 2N=200BC 2S 1N=200BC 3N=43BC 4N=37Length (mm)* 27.5–31.8 39.2–43.7 33.8(27.8–38.2)Length uniformity 81.3–85.5 83.9–87.1 85.0(82.0–88.2)Strength (g/tex) 26.5–32.5 41.4–46.7 35.7(29.7–41.6)Elongation 5.1–6.4 5.5–6.0 6.3(5.7–7.4)Fineness (mtex)** 207–243 178–191 218(177–308)Colour reflectance 71.2–77.7 74.6–75.6 74.3(65.9–81.1)28.6(22.9–35.3)81.3(73.3–86.7)28.3(17.8–43.7)5.5(3.9–7.6)224(165–379)72.2(56.8–81.3)30.7(26.9–36.1)83.3(80.6–85.1)29.0(23.7–34.5)6.3(5.4–7.4)225(176–285)74.1(69.8–77.5)30.0(25.0–33.9)81.9(77.6–86.4)24.5(16.8–32.8)5.7(4.5–7.0)128***(117–148)71.5(64.7–76.6)31.7(28.1–37.1)86.1(82.5–89.5)33.8(29.7–39.1)6.3(4.9–7.4)243(192–283)75.1(67.1–82.0)* Length is upper half mean length (UHML), ** standard fineness, *** low fibre fineness values in BC 3 generation because ofpoor maturitiesNote: Mean values and range (in brackets) observed in each BC 1–4 generation (number of plants, N, indicated) of fibretechnological parameters.ness index, for example, were positivelyconsidered.Details of the plant material used andthe types of analyses undertaken duringthe different steps of the MAS process aregiven in Figure 1.ResultsPhenotypic variationThe two parents were characterized by theircontrasting fibre properties (Table 1) withsignificant advantages for the G. barbadenseparent in terms of length (+9.7 mm onaverage over all data sets), strength (+15.9 g/tex) and fineness (-38 mtex). By contrast,the G. hirsutum parent displayed better yellownessindex/colour reflectance. For eachBC population, it was observed that thedata fitted normal distributions, that transgressivesegregants were regularly in thelower range of phenotypic values and that,although progeny values rarely reachedthose of G. barbadense, high phenotypicvalues were observed, including within themost advanced BC 4 generation (Table 1).Genetic mappingThe first step in the programme involvedthe construction of two genetic maps oftetraploid cotton by combining RFLP, SSRand AFLP markers generated separatelyfrom the first two backcross generations(BC 1 and BC 2 ). The initial BC 1 map comprising888 loci grouped in 37 linkagegroups and spanning 4 400 cM (Lacape etal., 2003), benefited from the developmentand integration of new additional microsatellitemarkers (Nguyen et al., 2004). Thisupdated saturated BC1 map spans 5 500 cMand comprises a total of 1 160 loci orderedalong 26 chromosomes or linkage groups(Nguyen et al., 2004). On the other hand,the BC 2 map constructed using AFLP andSSR markers had 514 loci in total. The twomaps agreed perfectly for loci order. Theyhad 373 loci in common (between sevenand 26 per chromosome throughout the 26chromosomes), thus allowing their mergerinto a combined consensus map. The consensusframework map comprises 1 306loci and spans 5 597 cM, with an averagemarker interval of 4.3 cM.QTL detectionThe QTL analyses, conducted throughcomposite interval mapping, used twomolecular data sets (BC 1 and BC 2 ) andthree sets of fibre measurements (per plant

74Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishTable 2Number of QTL for each trait and range of observed phenotypic effects conferred by theG. barbadense alleles (either positive, “Gb +”, or negative, “Gb –”) detected over the threepopulations (BC 1 , BC 2 and BC 2 S 1 )QTLGb +Range phenotypic effectsQTLGb –Range phenotypic effectsLength (mm)* 12 +0.7 to +2.1 3 –1.6 to –1.8Length uniformity 3 +0.5 to +1.5 3 –1.1 to –3.3Strength (g/tex) 8 +0.8 to +2.8 4 –0.9 to –3.4Elongation 6 +0.2 to +0.5 4 –0.3 to –0.6Fineness (mtex)** 13 –10 to –20 8 +9 to +40Colour reflectance 3 +1.8 to +2.5 13 –0.9 to –3.5Total 45 35* Length is upper half mean length (UHML), ** standard fineness.basis for BC 1 and BC 2 and per-line basiswith two replicates for BC 2 S 1 ). The generationsBC 1 and BC 2 were conducted withno selection, except for choosing thoseplants that produced backcrossed seeds.The fibre measurements, which initiallyincluded eleven traits, were reduced to sixgroups after considering the strong correlationsthat existed between some traits. Thefibre characteristics that were retained formeasurement included length, length uniformity,strength, elongation, fineness ormaturity, and colour.For the six fibre quality componentsstudied, 50 QTL were identified thatmet permutation-based LOD thresholds(ranging between 3.2 and 4.0 for mostof the traits). Thirty additional suggestiveQTL (having a LOD value below thresholdbut above 2.5) were also taken into considerationafter comparing the resultsbetween the three populations or betweenthe present results and those reported inthe literature (Jiang et al., 1998; Kohel et al.,2001; Paterson et al., 2003; Mei et al., 2004).Table 2 summarizes the data generatedfrom the QTL analyses for the six traits ofinterest and the phenotypic effects of thedetected QTL. In general, the contributionof each QTL, measured as a percentageof explained variation of a given trait, wasvariable and in most cases fairly low. Forexample, for traits of economic importance,individual contributions varied from 4.8 to14.8 percent in the case of fibre length, 4.4to 21.3 percent for fibre strength and 4.6 to29.1 percent for colour reflectance.Overall, it was observed that these80 QTL partitioned as expected from thephenotypic values of the G. hirsutum andG. barbadense parents: a majority of positivealleles for length (12 of the 15 QTL),strength (8 of the 12 QTL) and fineness(13 of the 21 QTL) derived from theG. barbadense parent, while a majority ofpositive alleles for fibre colour (13 of the16 QTL) derived from the G. hirsutumparent (Table 2). Furthermore, the QTLdetected for the various traits often colocalizedwithin QTL-rich regions (Lacapeet al., 2005). In some cases, QTL detectionand mapping were in agreementbetween generations (BC 1 and BC 2 ) and,very interestingly, in 26 cases (33 percentof the 80 QTL) they confirmed the resultsreported in the literature, both for theposition of a QTL and for the sign of itsphenotypic effect. The most prominentcases of QTL consistently detected in thisstudy as well as in those of Paterson etal. (2003) and Kohel et al. (2001), i.e. indifferent crosses/populations, were found

Chapter 6 – Targeted introgression of cotton fibre quality quantitative trait loci using molecular markers 75Table 3Identification of the 19 targeted regions mapped on 15 different chromosomes and comprising oneor several co–localized fibre quality QTL from G. barbadense for introgression into a G. hirsutumgenetic backgroundCarrier chromosomeChromosomelength(cM)Target interval(cM)Target size(cM)Traitc14 197 28–57 29 Lengthc3 153 32–67 35 Length, fineness90–138 48 Length, strength, finenessc4 190 102–118 16 Finenessc22 139 112–139 27 Finenessc5 360 78–101 23 Strengthc6 296 137–144 7 Length, finenessc25 183 44–73 29 Length, strengthc16 168 65–117 52 Strength, fineness, colourc23 173 45–66 21 Strength (elongation –, colour –)113–135 22 Length, strengthc10 192 0–21 21 Fineness78–120 42 Length, fineness, colourc20 268 88–161 73 Elongation, finenessc26 195 67–143 76 Length (colour –)A01 233 16–54 38 Length171–209 38 Strengthc18 158 32–46 14 FinenessA03 271 209–234 25 Strength, uniformityTotal 3176 Total 636Note: All targeted QTL show a positive contribution from the G. barbadense allele, except for a few negative cases indicatedin brackets. The target region is defined as situated between the two loci flanking the QTL peak LOD value at a one LODconfidence interval.along chromosome 3 for QTL for fibrestrength and fineness, and chromosome 23for QTL for fibre strength and length.The chromosome regions carryingco-localized QTL (corresponding to asingle or to several traits measured on asingle or on several populations) whosepositive alleles derived from the G. barbadensedonor genome, were reduced to19 QTL-rich regions that were carriedby 15 different “carrier” chromosomes(Table 3). Altogether, the confidence intervals(one LOD) of the involved QTL-richregions delimited a total length of 636 cM(20 percent of the carrier genome), or11.5 percent of the total genome (Table 3).Eleven non-carrier chromosomes weredevoid of positive QTL, or harboured negative(positive alleles derived from the G.hirsutum alleles) QTL.MAS in the BC 3 and BC 4 generationsand allelic transmission throughoutgenerationsThe early selection of BC 3 and BC 4 plantsusing SSR markers that framed the 19 targetedregions of interest made it possibleto choose those plants that showed anallelic constitution with as many introgressedloci within the targeted regions aspossible. In total, 43 BC 3 plants out of 411(11.4 percent) and 37 BC 4 plants out of 450(8.2 percent) were retained based upon theinformation provided by the markers, i.e.without any phenotypic selection at thisstage. These plants were backcrossed to therecurrent parent (and self-pollinated in thecase of the BC 4 plants).The allelic transmission observed in thefour groups of BC 4 derived from fourdifferent BC 1 plants is given in Table 4.

76Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishTable 4Percentage of introgressed loci (I%), at the heterozygous state, of 37 BC 4 plants that were derivedfrom four different BC 1 plants (Nos. 3, 11, 16, and 27)PlantnumberBC 1 BC 2 BC 3 BC 4Numberof lociI% NumberofplantsNumberof lociI% Number Numberof plants of lociI% NumberofplantsNumberof lociI%globalI%target/non-targetNo. 3 646 55 1 479 14 1 467 8 9 456 5 10/4No. 11 654 64 1 446 28 1 403 13 1 408 10 29/6No. 16 681 67 2 464 31 3 420 15 21 428 9 25/6No. 27 668 63 1 471 26 1 433 15 6 439 10 25/6Mean 62 26 14 8 21/5Note: The number of plants and of loci analysed at each generation are given. At the BC 4 generation, the percentage ofintrogression is also differentiated between target and non-target (as defined in Table 3) regions.Moderate deviations were observed fromtheoretical transmission values (62, 26, 14and 8 percent compared with 50, 25, 12.5and 6.25 percent at the BC 1 , BC 2 , BC 3 andBC 4 stages, respectively), with a bias infavour of a higher rate of G. barbadenseallele transmission. This bias was probablydue to the selection pressure imposedat least in the BC 3 and BC 4 generations.Throughout the BC 1 and BC 2 generationsthat have undergone no deliberate selection,the introgression of G. barbadensealleles (at the heterozygous state) coveredthe complete genome fairly well, i.e.introgressed segments were found on allof the 26 chromosomes (not shown). Thisresult contradicts the findings of Jiang et al.(2000) who detected important deficienciesin donor (G. barbadense) allele transmissionin a population of 3 662 BC 3 plantsoriginating from 21 BC 1 plants.After combining the SSR and AFLPmarker data, it was observed that the introgressionrate differed between target andnon-target regions. When averaged overthe 37 BC 4 plants, the percentage of introgressedloci (8 percent genome-wide) wasmuch lower in the non-target regions(5 percent) than that reached within targetregions (21 percent) (Table 4). The differentBC 4 plants introgressed between three andsix QTL-rich target regions in differentcombinations. As an illustration of theselection pressure applied through the useof molecular markers, Figure 3 shows thegraphical genotype of two BC 4 individualsas well as that of the BC 1 plant (No. 16)from which these individuals were derived.The two BC 4 plants had a common BC 1ancestor but originated from two differentBC 2 plants. In this particular example,starting from a common BC 1 plant (No.16) which harboured 13 out of 19 possibleQTL-rich regions, the two BC 4 plants(Nos. 104 and 419) derived from it partly orcompletely retained respectively five (c16,c23top, c23bot, c25 and A03) and four (c6,c25, c26 and A01bot) genomic regions carryingfavourable alleles. The other regionscarrying QTL on c3, c4, c23, c20, A01 andA03, which had been introgressed andwere heterozygous in the BC 1 plant, hadreturned to the homozygous G. hirsutum/G. hirsutum state. The percentages of introgressedloci in target and non-target regionsin these two examples were 29 and 10 percent,and of 29 and 5 percent in the two BC 4plants (Nos. 104 and 419) respectively.This example shows that, at least insome cases, the process used was efficientin selecting for chromosomal regionsof interest (foreground selection), whileallowing the rest of the genome to returntowards that of the recurrent parent.

Chapter 6 – Targeted introgression of cotton fibre quality quantitative trait loci using molecular markers 77Figure 3Graphical genotypes (26 chromosomes) of a BC 1 plant (No. BC1/16) (upper panel) and of twoselected BC 4 plant (Nos. BC 4 /104 and BC 4 /419) (lower panel) derived from itBC 1/16c25 c16 c23topc26A01topA01botc3topc4c3botc6c23botc20A03BC 4 /104 BC 4/419c25 c16 c23topc25c26A01botc6c23botA03The two possible allelic forms, homozygous Gh/Gh and heterozygous Gh/Gb, are denoted in dark grey and black respectively.Regions in black are introgressed with G. barbadense alleles. Light grey areas indicate portions of unknown alleliccomposition. Boxed areas represent the localization of QTL-rich regions localized on 15 carrier chromosomes shown to theleft (11 non-carrier chromosomes are shown to the right). Arrows indicate the regions totally or partially introgressed.Fibre characteristics of BC 3 and BC 4generation plantsOwing to the limited number of individualsand the unbalanced frequenciesof genotypic classes in the BC 3 and BC 4material, significant marker-trait associationswere less frequent than observedfrom the BC 1 and BC 2 data. For example,markers mapped along five, nine and sixchromosome regions contributed (P=0.01),respectively, to length, strength or finenessvariation using BC 4 marker-trait data,as compared with 15, 12 and 21 from theBC 1 and BC 2 data (Table 2). However,the majority of significant associations,particularly those determined in the BC 4generation, were observed within previouslydetected regions (not shown). Usingfibre strength as an example, out of the eightstrength QTL-harbouring regions on chromosomesc3bot, c5, c16, c23sup, c23bot,c25, A01 and A03 identified from the combinedBC 1 and BC 2 data (Table 3), theBC 4 data confirmed significant marker-traitassociations in five of these regions, i.e. formarkers mapped on chromosomes c3bot,

78Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishc16, c23bot, c25 and A03. Furthermore,it is worth noting that the BC 4 plant No.104 presented in Figure 3, which had introgressedall these five regions, also displayedthe highest fibre strength value of its generation(39.1 g/tex, compared with 33.1 g/texfor the Guazuncho 2 parent). The concomitantintrogression of G. barbadense allelesdisplaying positive marker-trait associationsfor other fibre properties such as length orfineness was also observed. This translatedinto the development of different highlyvaluable BC progenies. These preliminaryresults suggest that the improvement ofG. hirsutum fibre properties through theintrogression of G. barbadense fibre QTLappears feasible.DiscussionIn an attempt to overcome the limitationsof conventional breeding for improvingcotton fibre quality through the use ofinterspecific hybridization, molecularmarkers were used in a MAS scheme toimprove the efficiency of introgressingfibre quality traits. The advanced backcross-QTL(AB-QTL) strategy (Tanksleyand Nelson, 1996) was used as this allowedconcomitant development of a genetic mapof the cotton genome and analysis of fibrequality QTL, and attempts to introgressfavourable alleles in an adequate recipientgenetic background (Figure 1).In contrast to monogenic characteristicssuch as disease and insect resistance,many important traits including yield andquality show continuous phenotypic variationand are governed by a number ofQTL. Cotton fibre quality is a complexconcept that involves a number of traits orcharacteristics. Each of these is under theinfluence of numerous QTL, indicatinga complex genetic determinism. Indeed,from the present results, at least six QTLgovern fibre uniformity and up to 21 QTLinfluence fibre fineness. When consideringsix traits that can account for fibrequality, a total of 80 QTL were detected(Table 2). This figure falls within the samerange as that found by Paterson et al.(2003). As some of these QTL co-localizedwithin the same chromosome region, bychoosing those QTL whose positive allelederived from the donor parent and had thestrongest effect on economically importantfibre characteristics, the number of targetregions to be introgressed was reduced to19 (Table 3). Nevertheless, this number ofQTL remains too high to identify a singleplant that would carry them all. Indeed, inthe authors’ experience, at the BC 3 stage,single plants carried a maximum of fiveregions of interest (eight if consideringregions only partially introgressed), whileat the BC 4 stage, this number was reducedto four (seven if considering regions onlypartially introgressed).At this stage of the MAS process, tworoutes are under way (Figure 1). The firstinvolves identifying the best BC 4 plants,i.e. those showing the highest amount offavourable QTL introgression, and thenfixing the favourable allele by self-pollination.Such BC 4 S 1 plants have been crossedwith other BC 4 S 1 plants of different ascentin order to pyramid as many QTL as possible(each contributing to different traits)within the same genome. Similarly, BC 4 S 1plants were used to pyramid various QTLresponsible for a given trait (“selectivepyramiding”). This latter strategy couldespecially apply to traits of commercialimportance, such as fibre strength or fineness.The second avenue involves repeatingthe backcrossing process until near isogeniclines differing only at a given QTL (QTL-NILs) are developed. Such plant materialcould prove useful not only to study the

Chapter 6 – Targeted introgression of cotton fibre quality quantitative trait loci using molecular markers 79effect of a single given QTL on the phenotypicvalue of a plant harbouring it, butalso in case the introgressed QTL is provento contribute significantly to the improvementof a given trait (Bernacchi et al., 1998).Also, QTL-NILs could be used as donormaterial for QTL pyramiding (Pelemanand van der Voort, 2003). Finally, an introgressionlibrary, i.e. a collection of NILs,will typically serve as primary plant materialfor QTL fine mapping and eventualQTL cloning (Salvi and Tuberosa, 2005).However successful marker-aided introgressionof genomic regions of interestmay be, only phenotypic analysis of plantmaterial stemming from the MAS process,including the assessment of its adaptabilityto any given set of local agronomic and ecologicalconditions, will allow validation ofthis procedure.ReferencesBasten, C., Weir, B. & Beng, Z-B. 1999. QTL cartographer, version 1.13. Dept. of Statistics, NorthCarolina State Univ., Raleigh, NC, USA.Bernacchi, D., Beck-Bunn, T., Emmatty, D., Eshed, Y., Inai, S., Petiard, V., Sayama, H., Uhlig, J.,Zamir, D. & Tanksley, S.D. 1998. Advanced backcross QTL analysis in tomato. II. Evaluation ofnear isogenic lines carrying single-donor introgressions for desirable wild QTL-alleles derives fromLycopersicon hirsutum and L. pimpinellifolium. Theor. Appl. Genet. 97: 170–180.Brubaker, C.L. & Wendel, J.F. 2001. RFLP diversity in cotton. In J.N. Jenkins & S. Saha, eds. Geneticimprovement of cotton, emerging technologies, pp 81-101. Enfield, NH, USA, Science Publisher Inc.Dekkers, J.C.M. & Hospital, F. 2002. The use of molecular genetics in the improvement of agriculturalpopulations. Nature Genet. 3: 22–32.Jiang, C., Chee, P., Draye, X., Morrell, P., Smith, C.W. & Paterson, A.H. 2000. Multilocus interactionsrestrict gene introgression in interspecific populations of polyploid Gossypium (cotton).Evolution 54: 798–814.Jiang, C., Wright, R., El-Zik, K.M. & Paterson, A.H. 1998. Polyploid formation created uniquevenues for response to selection in Gossypium (cotton). Proc. Nat. Acad. Sc. USA 95: 4419–4424.Kohel, R.J., Yu, J., Park, Y.-H. & Lazo, G. 2001. Molecular mapping and characterization of traitscontrolling fiber quality in cotton. Euphytica 121: 163–172Lacape, J.-M., Nguyen, T.-B., Thibivilliers, S., Courtois, B., Bojinov, B.M., Cantrell, R.G., Burr, B.& Hau, B. 2003. A combined RFLP-SSR-AFLP map of tetraploid cotton based on a Gossypiumhirsutum x Gossypium barbadense backcross population. Genome 46: 612–626.Lacape, J.-M., Nguyen, T.-B., Courtois, B., Belot, J.-L., Giband, M., Gourlot, J.-P., Gawryziak, G.,Roques, S. & Hau, B. 2005. QTL analysis of cotton fiber quality using multiple G. hirsutum x G.barbadense backcross generations. Crop Sci. 45: 123–140.Lander, E., Green, P., Abrahamson, J., Barlow, A., Daly, M.J., Lincoln, S. & Newburg, L. 1987.MapMaker: an interactive computer package for constructing primary linkage maps of experimentaland natural populations. Genomics 1:174–181.Mei, M., Syed, N.H., Gao, W., Thaxton, P.M., Smith, C.W., Stelly, D.M. & Chen, Z.J. 2004. Geneticmapping and QTL analysis of fiber-related traits in cotton (Gossypium). Theor. Appl. Genet. 108:280–291.Nguyen, T.-B., Giband, M., Brottier, P., Risterucci, A.-M. & Lacape, J.-M. 2004. Wide coverage oftetraploid cotton genome using newly developed microsatellite markers. Theor. Appl. Genet. 109:167–175.

80Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishPaterson, A.H., Saranga, Y., Menz, M., Jiang, C. & Wright, R.J. 2003. QTL analysis of genotype xenvironment interactions affecting cotton fiber quality. Theor. Appl. Genet. 106: 384–396.Peleman, J.D. & van der Voort, J.R. 2003. Breeding by design. Trends Plant Sci. 7: 330–334.Rong, J., Abbey, C., Bowers, J.E., Brubaker, C.L., Chang, C., Chee, P., Delmonte, T.A., Ding, X.,Garza, J.J., Marler, B.S., Park, C.-H., Pierce, G.J., Rainey, K.M., Rastogi, V.K., Schulze, S.R.,Trolinder, N., Wendel, J.F., Wilkins, T.A., Williams-Coplin, T.D., Wing, R.A., Wright, R.J., Zhao,X., Zhu, L. & Paterson, A.H. 2004. A 3347 locus genetic recombination map of sequence taggedsites reveals features of genome organization, transmission and evolution of cotton (Gossypium).Genetics 166: 389–417.Salvi, S. & Tuberosa, R. 2005. To clone or not to clone plant QTLs: present and future challenges.Trends in Plant Sci. 10: 297–304.Schön, C., Utz, H.F., Groh, S., Truberg, B., Openshaw, S. & Melchinger, A.E. 2004. Quantitativetrait locus mapping based on resampling in a vast maize test cross experiment and its relevance toquantitative genetics for complex traits. Genetics 167: 485–498.Tanksley, S.D. & Nelson, J.C. 1996. Advanced backcross QTL analysis: a method for the simultaneousdiscovery and transfer of valuable QTLs from unadapted germplasm into elite breeding lines.Theor. Appl. Genet. 92: 191–203.Wilkins, T.A. & Arpat, A. 2005. The cotton fiber transcriptome. Physiologia plantarum 124:295–300.

Chapter 7Marker-assisted selection incommon beans and cassavaMathew W. Blair, Martin A. Fregene, Steve E. Beebe and Hernán Ceballos

82Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishSummaryMarker-assisted selection (MAS) in common beans (Phaseolus vulgaris L.) and cassava(Manihot esculenta) is reviewed in relation to the breeding system of each crop and thebreeding goals of International Agricultural Research Centres (IARCs) and NationalAgricultural Research Systems (NARS). The importance of each crop is highlighted andexamples of successful use of molecular markers within selection cycles and breeding programmesare given for each. For common beans, examples are given of gene tagging forseveral traits that are important for bean breeding for tropical environments and aspectsconsidered that contribute to successful application of MAS. Simple traits that are taggedwith easy-to-use markers are discussed first as they were the first traits prioritized forbreeding at the International Center for Tropical Agriculture (CIAT) and with NARSpartners in Central America, Colombia and eastern Africa. The specific genes for MASselection were the bgm-1 gene for bean golden yellow mosaic virus (BGYMV) resistanceand the bc-3 gene for bean common mosaic virus (BCMV) resistance. MAS was efficientfor reducing breeding costs under both circumstances as land and labour savings resultedfrom eliminating susceptible individuals. The use of markers for other simply inheritedtraits in marker-assisted backcrossing and introgression across Andean and Mesoamericangene pools is suggested. The possibility of using MAS for quantitative traits such as lowsoil phosphorus adaptation is also discussed as are the advantages and disadvantages ofMAS in a breeding programme. For cassava, the use of multiple flanking markers for selectionof a dominant gene, CMD2 for cassava mosaic virus (CMV) resistance at CIAT andthe International Institute of Tropical Agriculture (IITA) as well as with NARS partners inthe United Republic of Tanzania using a participatory plant breeding scheme are reviewed.MAS for the same gene is important during introgression of cassava green mite (CGM) andcassava brown streak (CBS) resistance from a wild relative, M. esculenta sub spp. flabellifolia.The use of advanced backcrossing with additional wild relatives is proposed as a wayto discover genes for high protein content, waxy starch, delayed post-harvest physiologicaldeterioration, and resistance to whiteflies and hornworm. Other potential targets of MASsuch as beta carotene and dry matter content as well as lower cyanogenic potential are given.In addition, suggestions are made for the use of molecular markers to estimate averageheterozygosity during inbreeding of cassava and for the delineation of heterotic groupswithin the species. A final section describes the similarities and differences between theMAS schemes presented for the two crops. Differences between the species can be ascribedpartially to the breeding and propagation systems of common beans (seed propagated, selfpollinating)and cassava (clonally propagated, cross-pollinating). In addition, differencesin growth cycles, breeding methods, availability of genetic markers, access to selectionenvironments and the accompanying opportunities for phenotypic selection influence thedecisions in both crops of when and how to apply MAS. Recommendations are made forapplying MAS in breeding of both crops including careful prioritization of traits, markersystems, genetic stocks, scaling up, planning of crosses and the balance between MAS andphenotypic selection.

Chapter 7 – Marker-assisted selection in common beans and cassava 83Common beans: importance andgeneticsCommon beans (Phaseolus vulgaris L.)are the most important grain legume fordirect human consumption, especially inLatin America and eastern and southernAfrica. They are seed-propagated, true diploids(2n = 22) and have a relatively smallgenome (650 Mb) (Broughton et al., 2003).Originating in the Neotropics, commonbeans were domesticated in at least twomajor centres in Mesoamerica and theAndes (Gepts, 1988) and possibly in athird minor centre in the northern Andes(Islam et al., 2002). Wide DNA polymorphismis expressed between the twomajor gene pools. Mesoamerican beanstypically have small to medium size seedsand can be classed into four races that aredistinguished by randomly amplified polymorphicDNA (RAPD) polymorphisms(Beebe et al., 2000). Andean beans usuallyhave medium to large seeds, and landraceshave been classed into three races basedon plant morphology and agro-ecologicaladaptation (Singh, Gepts and Debouck,1991). These can be differentiated by microsatellites(M. Blair, unpublished data) butthe genetic distance among Andean races isnarrower than that among Mesoamericanraces (Beebe et al., 2001). A large numberof gene tagging studies have been conductedin common beans, predominantlywith RAPD markers, some of which havebeen converted subsequently to sequencecharacterized amplified regions (SCARs;reviewed most recently by Miklas et al.,2006).Beans display a wide range of growthhabits (Van Schoonhoven and Pastor-Corrales, 1987), from determinate bushtypes, to indeterminate upright or vinybush types, to vigorous climbers. Bushtypes are the most widely grown, and are arelatively short season crop, maturing in aslittle as 60 days from seeding in a tropicalclimate and yielding from 700 to 2 000 kg/ha on average. On the other hand, in smallholderagriculture where land is scarce,labour-intensive, high-yielding climbingbeans enjoy continuing or even expandingpopularity. Climbing beans can mature in100 to 120 days at mid-elevations, but candelay as long as ten months at higher elevationsand can produce the highest yields forthe crop, up to 5 000 kg/ha. These featureshave significant implications for breedingprogrammes. In bush types it is possibleto obtain up to three cycles per year inthe field, or even four cycles in greenhouseconditions. Breeding bush beans isthus quite agile with regard to advance ofgenerations, although seed harvest of individualplants is sometimes limited. Withclimbing beans, on the other hand, at bestit is possible to obtain two cycles per yearwith field grown plants, while managingclimbing beans in the greenhouse is logisticallydifficult. However, while bush beansproduce on average 20 to 50 seeds/plant,individual plants of climbing beans oftenproduce enough seeds to plant several rows(100 to 150 seeds).Beans are self-pollinating and thusbreeding methods for autogamous cropsare employed. Pedigree selection or someadaptation thereof is most common, andboth recurrent (Muñoz et al., 2004) andadvanced (or inbred) backcrossing (Sullivanand Bliss, 1983; Buendia et al., 2003; Blair,Iriarte and Beebe, 2003) have been used.Recurrent selection has also been employed(Kelly and Adams, 1987; Beaver et al.,2003) but seldom in a formal sense with adefined population structure. Singh et al.(1998) suggested a system that they calledgamete selection in which individual F 1plants of multiple parent crosses give rise

84Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishto families. This system takes advantage ofthe variability among F 1 plants that is createdbetween segregating parental plants.The choice of breeding method and itsadaptation to specific circumstances, thegrowth cycle of the crop in relation todifferent planting seasons, the access toselection environments and the accompanyingopportunities for phenotypicselection and the ease of implementing thespecific markers to be used will all influencethe decisions about where and howMAS will be most cost effective and usedto best advantage.MAS in bean breeding: experiences ofCIAT and NARSMolecular markers have been sought forboth simple and complex traits in beans,with an eye to eventual application in MAS.Tagging of genes and QTL in common beanand their application to MAS have beenreviewed previously (Kelly et al., 2003;Miklas et al., 2006). In the present chapter,some of the aspects that contribute to thesuccessful use of MAS are considered ingreater detail, referring to examples takenfrom bean breeding in the tropics at CIATand within NARS. Simple and complextraits are discussed separately, as they representtwo contrasting sorts of experience.Simple traitsBean golden yellow mosaic virus resistanceBean golden yellow mosaic virus (BGYMV)is a white fly-transmitted Gemini virus, anda major production limitation of beans inthe mid-to-low altitude areas of CentralAmerica, Mexico and the Caribbean.Host resistance to the virus is the mostpractical means of control, and any newvariety in these production areas must carryresistance. Studies on inheritance of resistancerevealed a major gene denominatedbgm-1 in breeding line A429 (Blair andBeaver, 1993) that originates in the Mexican(Durango race) accession “Garrapato” orG2402. Minor genes (Miklas et al., 2000c)as well as additional recessive and dominantresistance genes exist for the virus (Miklaset al., 2006). In most production areaswhere BGYMV exists, it is necessary topyramid genes for adequate disease control.Although lines developed in CIAT targetthese areas, BGYMV does not exist at levelsthat would permit selection under fieldconditions in Palmira, Colombia, at CIATheadquarters. Therefore, MAS was desirableto assure recovery of at least the mostimportant resistance genes. MAS has alsobeen employed in the Panamerican Schoolin Zamorano, Honduras, as a complementto field screening, to extend selection tosites and seasons with less disease pressure(J.C. Rosas, personal communication).A co-dominant RAPD marker wasidentified for the bgm-1 gene (Urrea etal., 1996) that was subsequently convertedto a SCAR marker named SR2 (CIAT,1997). The DNA fragment associated withbgm-1 gene has only been observed inone genotype other than G2402 and itsderivatives, and thus the polymorphismhas been very useful for recognizing thepresence of the gene in different geneticbackgrounds. This SCAR was evaluated onas many as 7 000 plants in a single sowing(CIAT, 2001; 2003). The uniqueness of themarker’s polymorphism and its reliabilityover laboratories, seasons and geneticbackgrounds have facilitated its wide use.More recently, a second SCAR (SW12.700)was developed from the W12.700 RAPDfor a QTL located on linkage group b04(Miklas et al., 2000c), and this has alsobeen incorporated into the breedingprogramme of CIAT. The combination ofbgm-1 and the QTL is expected to offer an

Chapter 7 – Marker-assisted selection in common beans and cassava 85Figure 1Examples of gel multiplexing for MAS of A) BGYMV and B) BCMV resistance genesLoad 1S alleleR alleleLoad 2S alleleR alleleLoad 3S alleleR alleleA) bgm-1 geneRS SCAR (Urrea et al., 1996/CIAT)Controls530bp 570bpGel 1Load 2 Load 1 Gel 2Load 3 Load 2 Load 1 Gel 2Load 3Load 2Load 1B) bc-3 geneROC11 SCAR (Johnson et al., 1997)intermediate level of resistance, while otherminor genes must be recovered throughconventional phenotypic selection to assurehigher resistance.Scaling up of MAS required the developmentof simple operational procedures inboth the field (tagging, tissue collection) andthe laboratory (DNA extraction, markerevaluation). For gamete selection strategiesin the field, individual, evenly-spaced plantsfrom segregating populations were markedwith numbered tags that were coated withparaffin to protect them until seed harvest.Leaf disks were sampled from young vegetativetissue with a paper hole puncherand placed directly into pre-numbered cellsof microtitre 96-well plates stored on ice,ready for grinding and extraction in thelaboratory. The implementation of MASfor bgm-1 and subsequently for SW12.700in the laboratory required substantial adaptationof standard protocols to establishhigh-throughput procedures. Grinding ofsamples in microtitre plates was accomplishedwith a block of 96 pegs that fitinto each well. Alkaline DNA extraction(Klimyuk et al., 1993) was employed withsuccess for both markers, and eventually itwas possible to multiplex the markers inboth the amplification and gel phases usingmultiple primer PCR and multiple loadingper gel wells (Figure 1A). With experienceand improved procedures, efficiency morethan doubled over a two-year period. MASwas often carried out before flowering todecide on a plant’s status as a carrier of theresistant allele for further use in crossing.Two small red seeded lines developed inthe Panamerican School using MAS havereached the stage of validation in Honduras(J.C. Rosas, personal communication) andshown resistance to the BGYMV strainsprevalent there. Resistance to BGYMV ofdrought tolerant lines selected at CIAT wasmaintained using MAS for one or moregenes, followed by field selection in CentralAmerica. Similarly, red mottled lines developedin CIAT with the aid of MAS showedfield resistance in the Caribbean and one ofthese lines from the red mottled advancedline for the Caribbean (RMC) series hasbeen released (Blair et al., 2006). MAS hasalso been an important element of maintainingBGYMV viral resistance in CIAT’sprogramme as other breeding objectivessuch as nutritional value have beenassumed, necessitating the inclusion of susceptibleparents in crosses with resistant

86Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishlines. MAS for this trait has also been practisedat the University of Puerto Rico andat the Biotechnology Institute of Cuba.Bean common mosaic virus and beancommon mosaic necrotic virusBean common mosaic virus (BCMV) andthe related necrotic strains (bean commonmosaic necrotic virus [BCMNV]) areaphid-transmitted potyviruses that arefound worldwide and are seed-borne fromseason to season. BCMNV resistance isvery important in Africa where necroticstrains are prevalent and has become arenewed priority for parts of the Caribbeanwhere necrotic strains have been discovered.BCMV is also endemic in the Andeanregion where it persists in farmer-saved seedand long-season climbing beans. Climbingbeans are grown in both intensive (trellised/staked monoculture) and extensive (intercroppingwith maize) farming systems. Inboth systems the need to protect the cropfrom easily transmitted viral diseases suchas BCMV or BCMNV is great; however,very few climbing beans have been bred forresistance to BCMV. A number of BCMV/BCMNV resistance genes have been taggedincluding the dominant I gene (with whichthe necrotic strains interact to producenecrosis) and the recessive bc-3, bc-2 andbc-1 2 genes (Haley, Afanador and Kelly,1994; Melotto, Afanador and Kelly, 1996;Johnson et al., 1997; Miklas et al., 2000a).The genes can be distinguished by inoculationwith different viral isolates, and a rangeof molecular marker tags are available foreach gene (reviewed in Kelly et al., 2003;Miklas et al., 2006). The dominant I genewas incorporated into a wide range of smallseeded bush beans at CIAT, while resistantbush beans of the bush bean resistant toblack root (BRB) series carrying recessivegenes were developed in the 1990s andhave been widely distributed as breedingparents. The need to reselect the recessivegenes with confidence from segregatingpopulations makes MAS a priority.CIAT started a collaborative project withthe Colombian national bean programmebased at the Colombian AgriculturalResearch Corporation (CORPOICA) in2002 to introgress BCMV resistance genesfrom BRB lines into local landraces andimproved genotypes of Andean climbingbeans (CIAT, 2002, 2003, 2004; Santana etal., 2004). During the breeding programmefor BCMV and over the course of fouryears, MAS was used extensively basedprimarily on the SCAR marker ROC11developed for the bc-3 gene (Johnson etal., 1997) and the SCAR marker SW13 forthe I gene (Melotto, Afanador and Kelly,1996) along with virus screening to confirmthe selection of resistant progeny.The programme was successful in movingbc-3 resistance into a background of creammottled and red mottled seed types forboth highland areas (known as Cargamantocommercial class) as well as mid-altitudeareas through triple-, double- and backcrosses.Although virus resistance was alsoscreened phenotypically, the frequency ofescape, the complex interaction of multiplegenes and the recessive nature of mostof these made MAS the best option forbreeding resistant varieties rapidly. In addition,as climbing bean breeding is a moretime-consuming and expensive endeavourthan bush bean breeding due to the longerseason, wider plant spacing and need forstaking material, MAS was also found to bea very effective measure to reduce breedingcosts and save on breeding nursery space.The implementation of MAS for BCMVwas based on a combination of the previouslydeveloped SCAR markers previouslymentioned and techniques developed at

Chapter 7 – Marker-assisted selection in common beans and cassava 87CIAT for the selection of BGYMV resistanceas discussed previously. Althoughmost BCMV and BCMNV resistance geneshad been tagged with SCAR markers,implementation required efforts to validateand scale up the use of the markers inapplied breeding programmes. Genotypingfor the ROC11 marker was carried out onadvanced lines given that this marker isdominant and in repulsion with the resistanceallele. In other words, the absence of aband was indicative of the presence of therecessive bc-3 allele and therefore it wasmore appropriate to evaluate after fixationof the alleles to homozygosity throughmass or pedigree selection with singleplant selections in the F 4 and F 5 generationwhen single plant rows were evaluatedfor the resistance gene marker. To determinewhether the advanced line continuedto segregate for the gene, alkaline DNAextraction was conducted on leaf discscollected from four leaflets from four individualplants per line using a hole-puncherrather than from a single plant per familyor advanced line. The presence or absenceof polymerase chain reaction (PCR) productswas evaluated for each genotype basedon scanned photographs or gel captureimagery of multiplexed gels (Figure 1B) topredict if the genotype contained the resistanceor the susceptible allele.Once optimized for parental genotypes,MAS was conducted on a large number ofprogeny rows. For example in 2003, morethan 4 000 advanced lines were evaluated forthe ROC11 marker for genotypes grown atthree sites within Colombia (CIAT-Darien,CIAT headquarters and CORPOICA-Rionegro). DNA was collected at all threesites and shipped successfully to the laboratoryin 96-well plate format as discussedabove. Both the ROC11 and SW13 markerswere single copy SCARs that did not produceextra bands and therefore were easyto multiplex. To facilitate the evaluationof markers on a large number of advancedlines, usually within two to three weeks,and increase the efficiency of MAS, severalinnovations were implemented: loading ofagarose gels (first with two and then threeloadings), increasing numbers of wells percomb (first 30-well and then 42-well combswere used), use of 384-well PCR plates andmultipipetor loading of gels. The resultingsavings decreased the time to PCR amplifyand load a gel by approximately 50 percentand increased the number of genotypes runper gel by 225 percent.The rapid increase in efficiency obtainedduring the application of the ROC11marker shows the advantages of testingnew markers in practical breeding programmes.The use and advantages of thesemolecular markers has been presented atan Organization of American States-sponsoredcourse in Colombia given in 2002and a Rockefeller Foundation-sponsoredcourse in Uganda given in 2003. Based onthis programme and the training courses,MAS for BCMV genes was initiated aspart of a recently approved Associationfor Strengthening Agricultural Research inEastern and Central Africa (ASARECA)project for three countries in eastern Africaand training of researchers from the Andeanregion has allowed more breeding linesfrom Peru to be screened (CIAT, 2004).Other examples of MAS for simplyinherited traitsSeveral pathogens, especially fungalpathogens, have co-evolved with the beanhost, and present a population structure(Andean/MesoAmerican) that mimicsthe major gene pools of bean (Pastor-Corrales, Jara and Singh, 1998). This isthe case with Phaeoisariopsis griseola, the

88Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishcausal agent of angular leaf spot (ALS),and Colletotrichum lindemutheanum,which induces anthracnose. In both cases,pathogen isolates tend to be more virulenton host genotypes of the same gene pool(Andean or Mesoamerican) and less so onhost genotypes from the contrary genepool. Resistance genes of utility to onehost gene pool thus tend to originate in theother gene pool and require introgressionfrom one gene pool to the other. MAS hasgreat potential for introgression as DNApolymorphisms are maximized in widecrosses across gene pools, and markers areavailable for this purpose for both ALS(Carvalho et al., 1998; Sartorato et al.,1999; Nietsche et al., 2000; Ferreira et al.,2000; Mahuku et al., 2004) and anthracnose(Young et al., 1998; Awale and Kelly 2001;Vallejo and Kelly, 2001).Other cases of wide crosses in whichMAS can be of use include those for theselection of genes for resistance to a storageinsect, the Mexican bean weevil (Zabrotessubfasciatus [Boheman]) derived from wildbean accessions from Mexico. Selection forresistance has also been achieved by analysisfor the active resistance agent, a seedprotein called arcelin, by either antibodyreaction or electrophoresis, but MAS is simplerand more efficient than either of theseanalyses that require protein extraction.Even wider crosses of common bean withPhaseolus acutifolius have recovered resistanceto common bacterial blight (causedby Xanthomonas axonopodis pv. phaseoli)(Muñoz et al., 2004) and markers have alsobeen developed for these resistance genes(Jung et al., 1997; Miklas et al., 2000b; Parket al., 1999; CIAT, unpublished data). Inthese cases also, the fact of deploying genesfrom relatively wide crosses favours maintaininga state of DNA polymorphism inrelation to the target genotypes.Complex multigenic traitsIn addition to the studies previously discussed,several attempts have been carriedout to tag quantitative trait loci (QTL)for abiotic stress tolerance or insect resistancein common bean, although most ofthese traits might better be described asoligogenic, as results usually suggest that alimited number of loci (from three to six)are involved in their genetic control.One example is tolerance to low soilphosphorus that was investigated in thelandrace G21212. Linkage group b08proved to be especially important to yieldunder low phosphorus, with as many asthree important and loosely linked QTL(Beebe, Velasco and Pedraza, 1999; Miklaset al., 2006). Interestingly, these same QTLwere linked to QTL for resistance to Thripspalmi Karny derived from the same source(Frei et al., 2005). This is a promising candidatefor applying MAS in the short term forabiotic stress tolerance, although anothernotable attempt was also made for droughttolerance breeding with MAS through ajoint programme between Michigan StateUniversity and the National Institutefor Forestry, Agriculture and LivestockResearch (INIFAP) in Mexico (Schneider,Brothers and Kelly, 1997).In theory, a breeder would prefermarkers for low heritability quantitativetraits that are difficult to select throughphenotypic selection. However, in general,markers for polygenic or oligogenic traitshave not moved into the application phase.The same problems that make phenotypicselection difficult apply in some degree toMAS. Multiple minor genes that are oftenassociated with poor heritability also implythat it is difficult to identify QTL withhighly significant effects and that merit theinvestment of MAS. Furthermore, goodgenome coverage is usually necessary to

Chapter 7 – Marker-assisted selection in common beans and cassava 89detect the QTL that explain the highestamount of genetic variability, and this hasbeen difficult to achieve in intragene poolcrosses in common beans.However, genetic analysis by markershas been very useful for revealing theinheritance of quantitative traits, especiallyphysiological traits, even when the markersinvolved did not result in application inMAS. Analysis of QTL was applied to roottraits of bean as they relate to absorptionof phosphorus from soil (Liao et al., 2004;Yan et al., 2005; Beebe et al., 2006). Thispermitted associating different physiologicaltraits to P uptake and estimating theirimportance in nutrient acquisition. Oncetraits are better understood, then an appropriateselection strategy can be devised, beit phenotypic or MAS. Thus, markers canbe useful to a breeding programme by elucidatingbasic plant mechanisms even ifthey are not applied directly in selection.Breeding schemes: adaptation toinclude MASThe eventual application of MAS requirescareful prioritization of traits and evenspecific genes for which markers are to besought, in light of the importance of thetrait and genes, and options for phenotypicselection. One should never assume thatMAS is necessarily superior to phenotypicselection, which for some traits may be aseffective and efficient as the use of molecularmarkers. However, if a gene is sufficientlyimportant in a breeding programme todemand that advanced lines have such agene (as in the case of the bgm-1 gene forvirus resistance in Central America), thereis probably some point in the selectionprocess at which MAS would be useful.Also, it is not necessary to select manygenes by MAS for it to be of great value.For example, if a single gene is segregatingand 50 percent of plants lack the gene inadvanced generations, an effective selectionwould eliminate half the population andincrease the subsequent efficiency of thebreeding programme by a factor of 2.Once markers are available, a key issue isdetermining the range of parental genotypeswithin which a marker is polymorphic andtherefore useful for selection. Markers ofgenes that originate from wider crosses (e.g.from different races, gene pools or species)will have a progressively greater chance ofbeing polymorphic among a range of parents(Figure 2) and therefore diagnostic forthe gene of interest. The example of bgm-1is again a good case in point as the resistanceallele and the SR2 marker are bothunique to the Durango gene pool and polymorphicin combinations across otherMesoamerican races as well as the Andeangene pool. In contrast, the ROC11 markerfor the bc-3 gene is only polymorphicacross gene pools and therefore not diagnosticfor the resistant allele.If a breeder has several potential parentsamong which to choose and these arecomparable with regard to other traits,it might be preferable to eliminate thosethat carry a band that would be confusedwith the linked marker and would resultin false positives. Conversely, if more thanone marker is available for a given gene,one might focus on those linked markersthat maintain polymorphism in the greaternumber of combinations. In some combinationsit might be informative to useboth linked markers simultaneously, bothto discern recombinants and to confirmmarkers.Several possible schemes for the introductionof MAS to different breedingschemes are represented in Figure 3. Abreeder must consider at what generationin the breeding programme selection

90Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishFigure 2Potential of MAS in crosses of varying genetic diversityMAS potentialMethodPurposeLow – due to poorpolymorphismNarrowelite x eliteMass/PedigreeElite crossesHigh – due to goodpolymorphismIntermediateInter-racialGamete/RecurrentGenepyramidingWidewild/secondary/tertiarygene poolsAdv. / Rec. / Cong.BackcrossIntrogressionPotential of MASWidth of crossesModified from Kelly, Schneider & Kolkman, 1999 and Singh, 1999by MAS will give the greatest cost/benefitratio. This would probably be early inthe breeding programme for the pedigreemethod or for gamete selection while itwould be later in the programme for bulkmethod or mass selection (Figure 3). In thecase of early generation selection, eliminationof plants without the gene(s) will avoidunproductive investment.Advantages and disadvantages ofMASMAS provides real advantages where theconditions are not favourable for phenotypicselection, for example, in the case ofBGYMV, which does not exist at epiphytoticlevels in CIAT. Indeed, bgm-1 behavesas a recessive gene, so phenotypic selectionin early generations would be inefficientin recovering the gene in the heterozygousstate.The same principle would apply to therecessive bc-3 gene, although the lack of amarker linked in coupling to this gene hasbeen a serious drawback and has limited theeffectiveness of MAS to advanced generationswhen the gene is fixed by inbreeding.In this case, early generation selection withMAS would be limited to negative selectionagainst homozygous dominant andheterozygous plants, and this eliminatespotentially useful allele-carrying genotypes.Indeed, MAS is impossible in generationssuch as the F 1 or BC 1 F 1 to the susceptibleparent when no homozygous recessiveplants exist at all.

Chapter 7 – Marker-assisted selection in common beans and cassava 91Figure 3Application of co-dominant and dominant markers during the breeding cyclein common beansMethodGeneration of crossing or inbreedingMass selection F 1 F 1:2 F 1:3 F 1:4 F 1:5 F 5:6Pedigree selection F 1 F 1:2 F 2:3 F 3:4 F 4:5 F 5:6 F 5:7Backcrossing BC 1 BC 2 BC 3 BC 4 BC 4 F 1:2Gamete selection BC x F 1:2 BC x F 1:3 BC x F 3:4MC 1 F 1:2 MC 1 F 1:2 MC 1 F 2:3 Adv. lineNote: Asterisks indicate co-dominant marker, while open circles indicate dominant marker in repulsion andclosed circles indicate dominant marker in coupling.In other cases where phenotypic selectionmethods are available, the advantageof MAS resides in its simplicity. This isthe case in the selection of arcelin, whichcan be achieved through protein extractionfollowed by antibody detection or electrophoresis,but both of these are laboriouswhile MAS can be applied more rapidly andwith much greater throughput. Similarly,markers for common blight resistance andanthracnose have the advantage of obviatingthe need for field inoculations that aresometimes ineffective if environmental conditionsare not favourable. The advantageof MAS is much greater if a single DNAextraction can serve for the evaluation ofseveral markers, as in the multiplexing ofbgm-1 and SW12.700 markers.In spite of attempts to apply MAS tocomplex traits, examples of successfulapplication are still limited to relativelysimple traits. This is contrary to someprevious expectations that markers wouldbenefit mostly traits of low heritability.However, experience has shown that theability to manipulate even one importantgene with confidence can make a breedingprogramme more efficient, if that gene ishighly desirable and valuable for advancedmaterials.Meanwhile the disadvantages of MAScompared with phenotypic selection arebased on effectiveness and cost considerations.The effectiveness of MAS is relativeto the ease of applying a given marker,its reliability and its level of linkage withthe gene of interest. Although molecularmarkers theoretically have a heritabilityof 1.0, variability among laboratories oramong runs within a laboratory makemarkers less than 100 percent reliable. Thisis especially true for RAPD markers forwhich band amplification is dependent onDNA concentration and quality, annealingtemperature and thermocycling conditions,Taq polymerase concentration andthe relative proportion of various otheringredients to the PCR cocktail. In comparison,SCAR markers are much morereliable and repeatable and therefore havehigher heritability than RAPD markers.Linkage distance between a marker for a

92Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishgene of interest and the actual locus itselfalso affects the reliability of a marker. Inturn, the type of cross (wide versus narrow)and parents involved (closely or distantlyrelated) affect the frequency of recombinationaround introgressed genes as well asthe level of polymorphism of the cross andwhether the marker will present distinctalleles for the desirable and undesirablecharacter states. In this regard, there is atradeoff as MAS is most effective whenthere is high polymorphism in the crossesbeing evaluated (Figure 2). However, thisis precisely the breeding situation in whichgene introgression is most difficult, timeconsumingand plagued by linkage drag,as is the case for interspecific or interspecific-derivedcrosses, hybridizations withwild or wild-derived genotypes and crossesbetween the Andean and Mesoamericangene pools. This issue is being addressedin beans with the development and mappingof microsatellite markers (Blair et al.,2003) that are much more polymorphicand useful for diagnosing the inheritance ofgenomic segments in narrow crosses. Thefirst application of microsatellite markersfor MAS in common beans was the selectionof arcelin based bruchid resistanceusing gene-derived simple sequence repeatsthat are diagnostic for the introgression ofalleles for resistance from wild beans intocultivated backgrounds (CIAT, 2004), butothers should also show promise.In terms of cost considerations, therelative costs of MAS versus phenotypicselection are relative to each trait and situation.The widely held perception that MASis expensive is often due to the ingredientsand time used to prepare DNA extractionsand PCR reactions, although these costshave been reduced by innovations suchas the alkaline DNA extraction technique(Klimyuk et al., 1993) that obviates the needfor organic solvents or expensive enzymesinvolved in other mini-preparation techniques(Afanador and Hadley, 1993). Whileexperienced labour was previously requiredfor DNA extraction at CIAT or in NARSbreeding programmes, the alkaline extractionmethod allows most laboratory steps tobe carried out even by untrained personnel.Furthermore, MAS costs can be reducedby miniaturization, especially in the PCRreaction (for example, use of 384-well PCRplates and small reaction volumes) and reuseof ingredients (for example plasticwareincluding pipette tips and microtitre platesas well as agarose from used gels). As previouslymentioned, multiplexing adds tothe efficiency and therefore reduces thedatapoint costs of MAS.Currently, MAS with SCAR markersand alkaline extraction at CIAT cost lessthan US$0.25 per datapoint. Therefore theexpense of MAS is now not as importantan issue as previously. In this regard,MAS sometimes has the advantage of beingimplemented in any generation and underboth field or greenhouse conditions, whilephenotypic selection often requires a separateplanting and specialized labour forinoculation, agronomic management andevaluations or scoring. However, in thefinal analysis, the most efficient and costeffective breeding programme will probablybe one that combines MAS and phenotypicselection in some optimal combination. Itis precisely the challenge of the breeder todefine that optimal combination.One last disadvantage of relying on MASis that it commits a breeder to a uniquegene(s) for a given trait. For example, theremight be multiple genes or gene combinationsfor resistance to a disease, or fora physiological trait such as root structure.To the extent that a breeder relieson MAS for selection, this excludes other

Chapter 7 – Marker-assisted selection in common beans and cassava 93possible genes and the use of other potentiallyuseful parents that do not share theDNA polymorphism that is used in MAS.On the other hand, phenotypic selectionwould permit recognizing different geneticoptions for a desired phenotype. Thus,MAS is most useful when it is applied totruly unique genes.Cassava: importance andgeneticsCassava is a perennial shrub but it isgenerally harvested as an annual crop at 10–11 months of age. Basically every part ofthe plant can be utilized. The starchy rootsare a valuable source of energy and can beboiled or processed in different ways forhuman consumption and different industrialpurposes such as starches, animal feedor alcohol (Ceballos et al., 2006). Cassavastorage roots are not tubers and thereforecannot be used for reproductive purposes;stems are the common planting materials.Cassava foliage is not widely exploited inspite of its high nutritive value (Buitrago,1990; Babu and Chatterjee, 1999). Foliageconsumption by humans is relativelycommon in certain countries of Africa,Asia and Latin America. The use of foliagefor animal feeding is generating increasedinterest in Asia.Cassava can be propagated by eitherstem cuttings or botanical seed. However,the former is the practice most widely usedby farmers for multiplication and plantingpurposes. Propagation from true seed occursunder natural conditions and is commonin breeding programmes. Occasionallybotanical seed is also used in commercialpropagation schemes (Rajendran et al.,2000).Cassava is monoecious and allogamous,with female flowers opening 10–14 daysbefore the male ones on the same branch.Pollination can be done manually in acontrolled way to produce full-sib familiesor else in polycross nurseries whereopen pollination takes place and, therefore,half-sib families are produced. Self-pollinationis feasible when using male andfemale flowers on different branches oron different plants of the same genotypes(Jennings and Iglesias, 2002). Someclones flower relatively early at four or fivemonths after planting whereas others onlydo so at eight to ten months after planting.As a result, the time required for the seedto mature, the growing cycle of the cropand the need to plant with the arrival of therains take about two years between a givencross being planned and the respective seedbecoming available. On average, betweenone and two seeds (out of the three possiblein the trilocular fruit) per pollinationare obtained (Kawano, 1980; Jennings andIglesias, 2002).Breeding objectivesProductivity plays a major role in industrialuses of cassava, whereas stability of productionis fundamental in the many regionswhere cassava is the main subsistence crop.Industrial uses of cassava require high drymatter content as the main quality trait forthe roots, whereas for human consumptionthe emphasis is on cooking quality,frequently even over productivity, as thedetermining trait. Stability of productionis associated with resistance or toleranceto major biotic and abiotic stresses, withthe emphasis varying with the target environment.Genetic resistance to the mostimportant diseases and pests and the prevalentabiotic stresses can be found in cassavagermplasm (Hillocks and Wydra, 2002;Bellotti et al., 2002; Belloti, 2002; Ceballoset al., 2004). Although cyanogenic glucosidesare found in every tissue except

94Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishthe cassava seed, most processing methodsallow a rapid release and elimination of thecyanide. Depending on the end use, highor low cyanide clones are preferred. Otherrelevant traits for the roots are dry matter,protein and carotenoid content (Chávez etal., 2005).Breeding schemesGenetic improvement of clonally propagatednon-inbred crops such as cassavais made possible by the fact that a superiorgenotype can be fixed at any stage inthe breeding scheme, even after a singlecross, the equivalent of an F 1 in commercialhybrids such as maize. Therefore, nonadditivegene actions including dominanceand epitasis become important componentsof the genetic variance to be manipulatedby the breeder (Jaramillo et al., 2005; Calleet al., 2005; Perez et al., 2005a). Large effectivebreeding population sizes are requiredto retain favourable dominant alleles andepistatic loci combination.As in most crop breeding activities, cassavagenetic improvement starts with theproduction of new recombinant genotypesderived from selected elite clones. Scientificcassava breeding began only a few decadesago, and the divergence between landracesand improved germplasm is not as wide asin other crops. Therefore, accessions forgermplasm bank collections from differentresearch institutions play a more relevantrole in cassava than in other crops that havebeen scientifically bred for longer periodsof time. Parental lines are selected basedmainly on their performance per se andlittle progress has been made to use generalcombining ability (Hallauer and MirandaFo, 1988) as a criterion for parental selection.Sexual seeds obtained by the differentcrossing schemes are germinated to initiatea new cycle of selection. The multiplicationrate of cassava planting material is lowas five to ten cuttings can be obtained fromone plant. This implies a lengthy selectionprocess, and in fact it takes about six yearsfrom the time the botanical seed is germinateduntil enough planting material isavailable for multilocation replicated trials.Table 1 illustrates a typical selectioncycle in cassava. It begins with the crossingof elite clones and finishes when the fewclones surviving the selection process reachthe stage of regional trials across severallocations. It should be emphasized thatthere is some variation among the fewcassava-breeding programmes in the worldwith respect to the number of genotypesTable 1Typical selection cycle in cassava beginning with the crossing of elite clones to the point when fewclones surviving the selection process reach the stage of regional trials across several locationsYear Activity Number Plants per genotype1-2 Crosses among elite clones planned, nurseriesplanted and pollinations made3 F 1 : Evaluation of seedlings from botanical seeds.Strong selection for African cassava mosaic virus(ACMV) in Africa.Up to 100 000100 000 a ; 50 0000 b ; 150 000 c4 Clonal evaluation trial (CET) 20 000–30 000 a, b 700 c 6–8 (1 rep, 1 location)5 Preliminary yield trial (PYT) 100 a ; 300 b ; 80 c 20–60 (3 reps, 1 location)6 Advanced yield trial (AYT) 25 a ; 100 b ; 20–25 c 100–500 (3 reps, 2–3location)7-9 Regional trials (RT) 5-30 a, b, c 500-4 000 (3 reps,3–4 locations)Figures for cassava breeding at a IITA (Ibadan, Nigeria); b CIAT (Cali, Colombia) and c CIAT and Rayong Field Research Stationfrom Department of Agriculture (Thailand).Source: adapted from Jennings and Iglesias, 2002.

Chapter 7 – Marker-assisted selection in common beans and cassava 95and plants representing them through thedifferent stages. Table 1 also provides an ideaof the selection pressures generally applied.Strong emphasis on highly heritabletraits (plant type, branching habits andreaction to diseases, harvest index and drymatter content) is applied during the earlyphases of selection (F 1 and CET), (Hahn,Howland and Terry, 1980; Hahn, Terry andLeuschner, 1980; Hershey, 1984; Kawano,2003; Ceballos et al., 2004). As the numberof plants representing each genotypeincreases, the weight of selection criteriashifts towards low heritability traits suchas root yield. The clones that show outstandingperformance in the regional trialsare released as new varieties and, eventually,incorporated as parents in the crossingnurseries. With that the selection cycle isfinished and a new one begins. The wholeprocess has the following characteristics(Ceballos et al., 2004):• the process is indeed phenotypic selectionbecause no family data are involved;• no data are collected in the early stagesof selection. Therefore, data regardinggeneral combining ability effects(∼ breeding value) are not available for abetter selection of parental materials;• there is no proper separation betweengeneral (GCA ∼ additive) and specific(SCA ∼ heterotic) combining abilityeffects. The outstanding performanceof selected materials is likely to dependon positive heterotic effects that cannotbe transferred to the progenies that aresexually derived from them;• no inbreeding is incorporated purposelyin the selection process. Therefore,large genetic loads are likely to remainhidden in cassava populations and usefulrecessive traits are difficult to detect;• several stages of selection are based onunreplicated trials. A large proportion ofgenotypes is eliminated without properevaluation.For the above-mentioned reasons, cassavabreeding is difficult, expensive andto a certain degree inefficient (Perez et al.,2005a; Cach et al., 2005a, b). Kawano et al.(1998) mention that, during a 14-year periodabout 372 000 genotypes derived from4 130 crosses were evaluated at the CIAT-Rayong Field Crop Research Center. Onlythree genotypes emerged from the selectionprocess to be released as official varieties.Similar experiences have been observedat the International Institute of TropicalAgriculture (IITA), CIAT-Colombia andBrazil. Therefore, the development andadaptation of molecular tools for cassavagenetic improvement offer importantadvantages to make the process more efficientand effective.MAS in cassava breedingCassava genetic improvement can be mademore efficient through the use of easilyassayable molecular genetic or DNAmarkers (MAS) that enable the preciseidentification of genotype without theconfounding effect of the environment,thereby increasing heritability. MAS canalso contribute to the efficient reduction oflarge breeding populations at the seedlingstage based upon “minimum selection criteria”.This is particularly important giventhe length of the growing cycle of cassavaand the expense involved in the evaluationprocess. Therefore, a pre-selection at the F 1phase (see Table 1) could greatly enhancethe efficiency of the CET experiments.The selection of progenies based on geneticvalues derived from molecular marker datasubstantially increases the rate of geneticgain, especially if the number of cycles ofevaluation or generations can be reduced(Meuwissen, Hayes and Goddard, 2001).

96Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishAnother application of MAS in cassavabreeding is reducing the length of timerequired for the introgression of traits fromwild relatives. Wild relatives are importantsources of genes for pest and diseaseresistance in cassava (Hahn, Howland andTerry, 1980; Hahn, Terry and Leuschner,1980; Chavarriaga et al., 2004), but theneed to reduce or eliminate undesirabledonor genome content and linkage drag canlengthen the process, making it unrealisticfor most breeders. Simulations by Stam andZeven (1981) indicate that markers couldreduce linkage drag and would reducethe number of generations required inthe backcross scheme. Hospital, Chevaletand Mulsant (1992) corroborated this inachieving a reduction of two backcrossgenerations with the use of molecularmarker selection. Frisch, Bohn andMelchinger (1999), through a simulationstudy, found that use of molecular markersfor the introgression of a single target allelesaved two to four backcross generations.They inferred that MAS had the potentialto reach the same level of recurrent parentgenome in generation BC 3 as reached inBC 7 without molecular markers.The decision to employ DNA-basedmarkers in cassava breeding is primarilybased on the heritability of a trait and theamount of genotypic variance explainedby the marker. There are many instances incassava breeding where h 2 is low or zero.Some examples are:• plant health traits where the pathogenor pest pressure is absent or low, suchas cassava mosaic disease (CMD) in theNew World tropics or cassava greenmite (CGM) during the wet season;• variable or erratic pest pressure, e.g. theCGM or diseases such as the cassavafrog skin disease (FSD);• evaluation based upon a single plant;• variable experimental fields and/or poormanagement resulting in large experimentalerrors;• traits that are affected by the stage ofplant growth or the part of the organused for tissue analysis, e.g. cyanogenicpotential.In the above-mentioned instances,having a marker(s) that explains a largeproportion of the genetic variance can accelerateprogress in breeding. Even where h 2 ismoderate or high, selection by markers canbe advantageous:• where different sources of genes existfor the trait that are indistinguishableby phenotype alone and pyramidingis difficult and time consuming, e.g.for different sources of resistance to adisease or pest;• where molecular tags that can be usedinexpensively and rapidly to identifydesirable genotypes early in thebreeding cycle exist, thereby eliminatingthe need to evaluate large numbers ofplants phenotypically, and obviating theconfounding effects of the environment.Markers may permit the efficientelimination of undesirable genotypesat the seedling stage. For example, thenumber of genotypes at the seedlingstage can be reduced by 50 percent if atrait is controlled by a single gene, orby 87.5 percent if controlled by threegenes;• for the introgression of useful genesfrom exotic germplasm into adaptedgene pools. MAS can be used to identifygenotypes that carry minimal amountsof flanking donor parent genome aroundthe gene of interest for faster backcrossing;• for definition of heterotic pools in agroup of germplasm accessions for moredirected crosses;

Chapter 7 – Marker-assisted selection in common beans and cassava 97• for definition of average heterozygosityin the selection of partially inbred linesfor tolerance to inbreeding;• for identification of the male parentin elite germplasm derived from polycrossesby fingerprinting. This tool isalso useful for checking the identityof different genotypes to eliminateduplication in germplasm collections.Best results are achieved when MAS iscombined with phenotypic data as comparedwith either approach independently(Hospital, Chevalet and Mulsant, 1992).Phenotypic data would reduce the cost ofgenotyping especially if phenotypic evaluationis conducted on early generations(Gimelfarb and Lande, 1994). This not onlyreduces the cost of MAS but also increasesits efficiency. Some examples of MAS incassava breeding conducted at an internationalcentre and national programmes aredescribed below.Molecular MAS for CMD resistance at anIARCAn ideal target for MAS is breeding fordisease resistance in the absence of thepathogen. This is the case of CMD in theAmericas, where the disease does not occur.CMD is a viral disease first reported byWarburg in 1894 in eastern Africa (quotedby Storey and Nichols, 1938). Several variantsof the disease (East Africa cassavamosaic virus [EACMV], South Africa cassavamosaic virus [SACMV], Indian cassavamosaic virus [ICMV]) have been reported(Swanson and Harrison, 1994) and areendemic in all cassava growing regions ofAfrica and southern India, where it is themost severe production constraint. Thewhite fly vector of CMD, Bemisia tabacibiotype A, does not colonize cassava inthe New World but recently a new biotypeof B. tabaci, biotype B (also referred to asB. argentifolia), has become widespreadin the Americas and has a wide host rangeincluding cassava (Polston and Anderson,1997), increasing the possibility that CMD,EACMV, SACMV, ICMV or a nativeAmerican gemini virus will become establishedon cassava in the neo-tropics. Thisis a frightening prospect for cassava productionin Latin America, considering thatmost Latin American cassava germplasm isvery susceptible to CMD (Okogbenin etal., 1998). The susceptibility of neo-tropicalgermplasm to CMD also limits the utilizationof germplasm from the crop’s centreof diversity in the neo-tropics for these keycassava production regions. Breeding forresistance to CMD in Latin America, wherethe disease does not exist and is unlikely tobe introduced due to very strict quarantinecontrols, requires the tools of MAS.Evaluations at IITA identified an excellentsource of resistance to CMD in someNigerian landraces (A.G.O. Dixon 1989,unpublished data), namely TME3, TME7,TME5, TME8, TME14 and TME28. Thisresistance is effective against all knownstrains of the virus, including the virulentUgandan variant (UgV) (Akano et al.,2002; CIAT, 2001). CIAT, in collaborationwith IITA in Ibadan, Nigeria, and withsupport from the Rockefeller Foundation,devel-oped several molecular markers forthis source of CMD resistance, revealedto be controlled by a single dominantgene designated as CMD2 (Akano et al.,2002). At least five markers tightly associatedto CMD2 have been developed, theclosest being RME1 and NS158 at distancesof four and seven cM respectively.The dominant nature of CMD2 and itseffectiveness against a wide spectrum ofviral strains makes its deployment veryappealing for protecting cassava againstthe actual or potential ravages of CMD

98Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishin both Africa and Latin America. CIATand IITA undertook a project to verifythe utility of these markers for MAS inbreeding CMD resistance by developingcrosses between the sources of TME3 andsusceptible varieties. A total of six families,ranging in size from 36–840 genotypes, anda total of 2 490 genotypes were used. Thecrosses were genotyped with two markersand also evaluated for CMD resistancein a high CMD pressure area in Nigeria.Results of the marker analysis and phenotypicevaluation of CMD resistance inthe field revealed that the markers RME1and NS158 SSR were excellent predictiontools for CMD resistance in some crosses(a prediction accuracy of 70–80 percent). Ina few families, however, the markers werenot polymorphic between the resistant andsusceptible parent and, therefore, were notuseful. This highlights the need to developmany markers around a gene of interest ina MAS programme and then to use thosemarkers to evaluate the parents and identifythe best markers for the different crosscombinations.Eighteen progenies from TME3 carryingthe CMD2 marker were established fromembryo axes and imported to CIAT fromIITA. They were crossed extensively to eliteparents. Seeds harvested from the crosseswere germinated in vitro from embryo axesaccording to standard protocols for cassava(Fregene et al., 1997, CIAT, 2002) to allowsharing the CMD resistant genotypes withcollaborators in Africa and India. Eachplantlet was multiplied after three to fourweeks of growth to obtain three to fiveplants. After another four weeks, leaves of allPhytosanitary conditions for the exchangeof cassava germplasm between Africa and Asiaare very stringent, but appropriately indexed invitro cultures of embryo axes are permitted forexperimental purposes.plants were removed for molecular analysisand the plants multiplied again to obtain10–20 plantlets. DNA isolation was by arapid mini preparation method developedfor rice (Nobuyuki et al., 2000). The DNAobtained is sufficient for 100 reactions andcan be held in the Costar plates for twomonths at –20 o C without any degradation.PCR amplification, polyacrylamide gelelectrophoresis (PAGE) or agarose gelanalysis of SSR markers NS158 and RME1were as described by Mba et al. (2001). Theversatility of spreadsheets makes them theappropriate software to handle the diverseinformation generated by MAS. Gel imagesfrom the marker analysis were entereddirectly into a spreadsheet that containsinformation on the parents, tissue cultureand greenhouse records, and subsequentphenotypic evaluation of the progenies.After molecular analysis, genotypes thatcarry the marker allele associated withCMD2 were further multiplied to obtainat least 30 plants. Ten plants were sent tothe greenhouse for hardening and latertransferred to the breeding programme forevaluation. Five plants were kept in vitro,while 15 plants were shipped to partners inIndia and Africa as shown in the flow chartfor MAS (Figure 4).To date, more than 50 000 progeny havebeen evaluated with CMD linked markersand resistant lines shared with nationalprogrammes in India or Africa, and alsoincorporated into the breeding scheme atCIAT. The cost of a single marker datapoint is US$0.30 and 32 000 samples can beprocessed in a year.MAS for CMD resistance at a NARSAlthough evaluation for CMD resistancein sub-Saharan Africa is relatively easy andmost areas have sufficient disease pressureto permit moderate to high heritability

Chapter 7 – Marker-assisted selection in common beans and cassava 99Figure 4Schematic representation of steps employed in breeding for resistance to the CMDin Latin America cassava gene poolsThe entire process from sexual seeds to tissue plants for shipment or transfer to thescreen house takes approximately three months.of resistance, overlapping outbreaks ofCGM, cassava bacterial blight (CBB), andCMD are common (Legg and Ogwal, 1998)and the need for modest-sized breedingpopulations make MAS for CMD resistancea powerful tool to accelerate cassavaimprovement even in Africa. A MAS andparticipatory plant breeding (PPB) projectwas initiated in 2003 with funding fromthe Rockefeller Foundation to improvethe resistance of local cassava varieties inthe United Republic of Tanzania to CMDand CGM and also to provide proof ofconcept for the use of MAS to acceleratecassava improvement. The United Republicof Tanzania is the fourth largest producerof cassava in Africa with average yieldsof about 8 tonnes/ha (FAO, 2001). This isbelow the continent’s average of 10 tonnes/ha, and well below the average yield of14 tonnes/ha of Africa’s (and the world’s)largest producer, Nigeria.The low yield in the United Republicof Tanzania is caused by many factors,including the susceptibility of commonlygrown varieties to major diseases and pestssuch as CMD and the cassava brown streakdisease (CBSD). The project crosses farmerpreferredgermplasm, by agro-ecology, toimproved introductions that are resistantto CMD and to CGM. Markers associatedwith resistance to CMD are used to reduce

100Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishFigure 5MAS scheme to improve local varieties of cassava in the United Republic of Tanzaniausing improved disease and pest resistant introductions from Latin AmericaYear1Improved introductions(≈90)Local varieties(≈60)Year2Crossing block(Polycross design)Year3Seedling trial(60 000 seedlings)MASYear4 Combining ability studiesSingle row trial(≈10 000 genotypes)Year5The scheme is now in its second year.Farmer participatory trials(≈ 600 genotypes)the population size and a small set of genotypeswith the “minimum criteria” forsuccessful cassava production are evaluatedin a single season in the corresponding agroecologyand then evaluated over two cyclesin collaboration with end-users (rural communitiesand cassava processors). Figure 5describes the scheme of the United Republicof Tanzanian MAS and PPB project. CMDresistant F 1 generated by MAS at CIATwere crossed to BC 1 derivatives of M. esculentasub spp. flabellifolia, showing goodresistance to CGM, to produce progeniesthat combine some CMD and CGMresistance (Kullaya et al., 2004). The progenieswere established from embryo axesas in vitro plants to aid shipment to Africa.Molecular markers associated with resistanceto CMD and phenotypic evaluationfor CGM resistance were used to screenand select progenies that combine resistanceto CMD and CGM. Resistant plants(300 genotypes and ten plants per genotype),were shipped to the United Republicof Tanzania as in vitro plantlets for useas improved parents. A selection basedon harvest index, a highly heritable trait,and total biomass was made and 80 genotypesselected. These were planted in thesecond year in a controlled crossing blocktogether with 54 local germplasm from theeastern and southern zones of the country.Emphasis was placed on local varieties with,or tolerance to, CBSD, which is a majordisease of cassava in coastal east Africa andMozambique. Over 40 000 crosses weremade between the improved genotypesand the local varieties producing more than60 000 seeds.Sexual seeds obtained from crossingimproved and local genotypes were plantedin the screen house and transferred tothe field 40 days after planting. Parentallines were also planted in the screen housefrom woody stakes. DNA was isolatedfrom parental lines using the rapid minipreparationmethod and evaluated withthe five markers associated with the

Chapter 7 – Marker-assisted selection in common beans and cassava 101CMD2 mediated resistance to CMD.Polymorphism in pair-wise combinationsof the parental lines was observed with atleast one of the five markers and will beused on the progeny. The phenotype ofthe progeny will be evaluated at three andsix months after planting for resistanceto CMD, CBSD and CGM. Markers arecurrently being tested for CGM resistanceand are being developed for resistance toCBSD; when their utility is confirmed,they will also be used on progenies.Using published broad sense heritabilityof 0.6 for CMD resistance (Hahn,Terry and Leuschner, 1980), it is expectedthat 24 000 symptomless genotypes willbe analysed with markers associated withresistance to CMD. The gain of MAS willbe the elimination of at least 38 400 (4800 x 8 plants) that would have been carriedto the single row trial stage (eightplant-rows per genotype), considering thatbreeders traditionally select 20 percent atthe seedling trial stage. This representsa reduction of about 4 ha at the CET. Ifmarkers can be used to select for resistanceto CGM and CBSD, then an additionalnumber of genotypes can be eliminatedfrom the CET leading to even greater savings.Using MAS for CMD alone wouldreduce the size of field trials by 50 percent.If additional second and third traits wereincluded, reductions could be as high as 75and 87.5 percent, respectively. Perhaps themost important advantage, however, comesfrom the increased genetic gain arising fromhigher heritabilities in these field evaluationswith fewer genotypes.MAS for transferring useful traits from wildrelatives of cassava into the cultivated genepoolWild Manihot germplasm offers a wealth ofuseful genes for the cultivated M. esculentaspecies, but its use in regular breedingprogrammes is restricted by linkage drag andthe long reproductive breeding cycle. Forexample, several accessions of M.esculentasub spp. flabellifolia, M. peruviana andM. tristis have high levels of proteins(Nichols, 1947; Asiedu et al., 1992; CIAT,2004). Low amylose content starch (3–5 percent) or waxy starch of relevance tothe cassava starch industry has also beenidentified in two wild relatives of cassava,namely M. crassisepala and M. chlorostricta.The only source of dramatically delayedpost-harvest physiological deterioration(PPD) has been identified in an interspecifichybrid between cassava and M. walkerae.The M. walkerae parent was collectedin Mexico and held at the WashingtonUniversity, St. Louis, United States ofAmerica (Bertram, 1993). It was broughtto CIAT in 1998 in an attempt to useit in improving PPD. Furthermore, theonly source of resistance to the cassavahornworm and the most widely deployedsource of resistance to CMD wereidentified in fourth backcross generationprogenies of M. glaziovii (Jennings, 1976;Chavarriaga et al., 2004). Moderate to highlevels of resistance to CGM, whitefliesand the cassava mealybug have been foundin interspecific hybrids of M. esculentasub spp. flabellifolia. The delayed PPDtrait and resistance to the pests weresuccessfully transferred to F 1 interspecifichybrids suggesting dominant or additivegene action of the gene(s) involved (CIAT,unpublished data).The long reproductive cycle and lengthytime required to develop new cassavavarieties (10–15 years) often discouragesthe use of wild species in most conventionalcassava breeding programmes. However,the use of molecular markers to introgressa single target region of the genome can

102Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishFigure 6Modified advanced backcross QTL scheme in the introgression of useful traits fromwild relatives into cassavaTo crossing block with elite or local varietiesTo crossing block with elite or local varietiessave between two to four backcrossgenerations (Frisch et al., 1999). Indeed,it has been shown in several crops thatthe “tremendous genetic potential” lockedup in wild relatives can be released moreefficiently through the aid of new tools ofmolecular genetic maps and the advancedbackcross QTL mapping scheme (ABC-QTL) (Tanksley and McCouch, 1997).For several years now molecular markertools and a modified ABC-QTL schemehave been tested in cassava at CIAT forthe introgression of useful genes from wildrelatives. The scheme entails generating BC 1crosses and carrying out QTL mappingfollowed by selection of genotypes carryingthe genome region of interest with minimumsegments of the donor genome (Figure 6).The modified ABC-QTL is currentlybeing used at CIAT to introgress genes forhigh protein content, waxy starch, delayedPPD, and resistance to whiteflies and thehornworm. The most advanced of theseMAS projects is the introgression of highprotein content from close wild relatives ofcassava. Two BC 1 families of between 250and 300 progenies were developed fromtwo accessions of M. esculenta sub spp.flabellifolia OW284-1 and OW231-3, andthe improved cassava variety from ThailandRayong 60 (MTAI 8 in the germplasmcollection). The BC 1 families were planted

Chapter 7 – Marker-assisted selection in common beans and cassava 103in a CET for evaluation of root proteincontent at ten months. The grand parentallines of the BC 1 population were genotypedwith over 800 simple sequence repeat (SSR)markers available for cassava and about 300polymorphic markers were identified. Thepolymorphic markers are being assayed inthe progenies after which QTL analysis willbe conducted using the phenotypic proteinand molecular marker data. Genotypes thathave QTL for protein and a minimum of thedonor parent genome will be selected andused for producing the BC 2 generation.For introgression of naturally occurringmutant granule-bound starch synthase(GBSSI) for waxy starch in wild relatives,a more targeted approach was taken.Sequencing of the glycosyl transferase regionof the GBSSI gene from the wild relativesand two cassava accessions identified foursingle nucleotide polymorphisms (SNPs)that differentiated the wild accessions fromcassava. Allele-specific molecular markersunique to these SNPs were developed forselection of these alleles in a breedingscheme.Genetic crosses were made betweenM. chlorosticta accession CW14-11 andMTAI8, and the resulting F 1 was backcrossedto MTAI8. The allele specificmarker will be used together with otheragronomic traits, particularly performance,to select for BC 1 that carry the mutantGBSS alleles for self-pollination to recoverthe waxy trait. The identification of naturalmutants in a key gene and development ofmarkers represent an innovative moleculartool to accelerate the introgression offavourable alleles from wild relatives intocassava. Backcross derivatives have alsobeen developed from M. walkerae (MWal001) for delayed post-harvest physiologicaldeterioration; from MNG11 (a BC 4derivative of M. glaziovii) for resistance tohornworm; and from M. esculenta sub spp.flabellifolia (FLA447-1) for resistance towhiteflies. Phenotypic and genetic mappingof these backcross populations are inprogress to be followed by identification ofQTL and selection of progenies to generatethe next generation. MAS will later be usedto combine these genes into progenitors foruse as parents in breeding which, togetherwith low cost marker technologies, willbe distributed extensively to nationalprogrammes in Africa, Asia and LatinAmerica to produce improved varieties.Marker-assisted estimation of averageheterozygosity during inbreeding of cassavaA principal use of molecular markers byprivate sector breeding companies is toaccelerate the development of inbred lines.Cassava genotypes are heterozygous andvery little inbreeding has been practisedto date. However, inbred lines arebetter as parents as they do not have theconfounding effect of dominance and carrylower levels of genetic load (undesirablealleles). Speed of inbreeding depends uponthe average heterozygosity of the originalparental lines, the homozygosity level ofthe selected genotypes at the end of theself-pollinating phase and the process ofselection of progenies to be self-pollinated(Scotti et al., 2000). Basically in theinbreeding process two events go together:phenotypically there is a decrease in vigour,which is correlated with the increased levelsof homozygosity. While the aim is to selectvigorous plants (tolerant to inbreeding), inthe process plants may be selected that areless homozgygous than the expected averagefor their generation. It is expected that thefirst few cycles of self-pollination will resultin a marked reduction of vigour (inbreedingdepression associated with the genetic loadof the parental lines); therefore, selection for

104Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishtolerance to inbreeding depression must beexerted. However, such selection is biasedby the differences in homozygosity levelsof segregating partially inbred genotypes.This highlights the need for a methodto measure the level of heterozygosityin these partially inbred individuals andto use this in a co-variance correction inthe selection of phenotypically vigorousgenotypes. Molecular markers can be usedto estimate the level of homozygosity ofa given plant, enabling selection of plantswith true tolerance to inbreeding.Molecular markers can identify regionsin the genome that are particularly relatedto the expression of heterosis and for measuringgenetic distances among inbred linesto direct crosses with higher probabilities ofhigh heterosis. Co-dominant SSR markerson a genome-wide basis are suitable forthis purpose. The effect of self-pollinationon vigour and heterozygosity was analysedin nine S 1 families, heterozygosity beingestimated in the S 1 families by 100 mappedSSR markers that cover over 80 percent ofthe cassava genome and plant vigour by dryroot yield and plant biomass. Results willassist in selecting the best performing andleast heterozygous plants during inbreedingby identifying superior partially inbredparental lines. Molecular markers couldalso be used to delineate heterotic groups incassava. Genetic resources of cassava havebeen characterized at the regional (Fregeneet al., 2003) and global (Hurtado et al.,2005) levels. Highly differentiated groupsof accessions were observed particularlyamong groups of materials from Guatemalaand Africa and they may represent heteroticpools. These groupings are being testedbased on molecular markers by geneticcrossing between and within the groupsas a first step to define heterotic patternsfor a more systematic improvement ofcombining ability via recurrent reciprocalselection.Other potential MAS targetsSeveral other traits for which MAS can beapplied to increase efficiency of breedinginclude:Beta-caroteneCIAT and a number of partners are involvedin a project to produce cassava varieties withhigher levels of β-carotene in yellow roots.This is one way of combating the deficiencyof this key micronutrient in areas wherecassava is a major staple. The experimentalapproach to increasing cassava β-carotenecontent includes conventional breeding andgenetic transformation. The discovery of awide segregation pattern of root colour intwo S 1 families from the Colombian landraceMCOL 72 (cross code AM 273) and MTAI8 (AM 320) was the basis for moleculargenetic analysis of β-carotene content incassava. Three markers, SSRY251, NS980and SSRY330, were found to be associatedwith β-carotene content. These are in thesame region of the genome and togetherexplain >80 percent of phenotypic variationfor β-carotene content in the populationused for this study. The homozygous stateof certain alleles of these markers translatesinto higher β-carotene content, suggestingthat breeding for this trait can benefit frommolecular markers to assist in combiningfavourable alleles in breeding populations.The work is continuing with the searchfor additional favourable alleles in yellowrootedgermplasm to give the best possiblephenotypic expression of the trait.Cyanogenic potentialA collaborative project between the SwedishUniversity of Agricultural Sciences (SLU),Uppsala, the Medical Biotechnology

Chapter 7 – Marker-assisted selection in common beans and cassava 105Laboratories (MBL), Kampala, and CIAT,is aimed at the genetic mapping of CNPin cassava. An S 1 family-AM 320, derivedfrom the bitter variety MTAI 8 is thebasis for the study. This family has beenevaluated for cyanogenic glucoside contentand has been genotyped with morethan 200 diversity array technology (DarT)markers at CAMBIA, Australia, and 150SSR markers at CIAT. The discovery ofmolecular markers for CNP will providea tool to select efficiently for low cyanogenicpotential in cassava. Also ongoing isthe genetic mapping of the two cytochromeP450 genes CYP79D1 and D2 that catalysethe rate-limiting step of the biosynthesisof the cyanogenic glucosides, linamarin inthe S 1 family AM 320. The group is alsolooking for an association with QTL forCNP. It is expected that markers associatedwith CNP will be identified at the end ofthe study.Dry matter contentFew key traits in cassava hold greaterpotential for increasing cost-effectivenessvia MAS than root dry matter content(DMC). This trait is usually measured atthe end of the growth cycle. A number ofgenetic and environmental effects influenceDMC. It is usually highest before the onsetof rains, but drops after the rains beginas the plant mobilizes starch from theroots for re-growth of leaves (Byrne, 1984).Defoliation from pest and disease attackscan lower DMC. Breeding programmeshave been quite successful in improvingDMC, especially for industrial markets.The entry point for developing markersassociated with DMC was recent diallelexperiments (Jaramillo et al., 2005; Calleet al., 2005; Pérez et al., 2005a, b; Cachet al., 2005b). Diallels, in this case madeup of 90 families, are an ideal methodto identify genes controlling DMC thatare useful in many genetic backgrounds.Estimates of general and specific combiningability (SCA and GCA, respectively) formany traits of agronomic interest werecalculated, with emphasis on DMC. Basedon GCA estimates, parents were selected togenerate larger-sized progenies for DMCmapping. Sizes of families in the originaldiallel experiment were about 30 progenies,which is rather small for genetic mapping.Parallel to the development of mappingpopulations was the search for markersassociated with DMC using two F 1 families,GM 312 and GM 313, selected from thediallel experiment having parents with highGCA for DMC.Initial marker analysis using bulked segregantanalysis led to the discovery of twomolecular genetic markers, SSRY160 andSSRY150, which explain about 30 and18 percent, respectively, of phenotypic variancefor DMC. These markers are beinganalysed on approximately 700 genotypesderived from 23 crosses with parents havinghigh GCA for DMC in order to confirmtheir utility across genetic backgrounds.Parallel to this, larger families are beingdeveloped from selected parents for QTLmapping of DMC.Disadvantages of MASPerhaps the greatest disadvantage of MAS isthe time and financial investment requiredto develop markers that are widely applicablefor traits of agronomic importance.Often a marker developed in one or afew related genotypes will not work forother genotypes in a breeding scheme dueto allelic effects. Furthermore, developmentof markers, particularly for QTL, iscomplicated by epistatic interactions andthe critical need for good quality phenotypicdata. Several ways around this

106Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishproblem have been proposed, such as theuse of candidate genes involved in the traitsdirectly as selectable markers without theneed for laborious gene tagging experiments.However, unravelling the geneticsand the development of markers for suchtraits is still many years down the road.New methods of association mapping andlinkage disequilibrium mapping that relyupon non-random association of candidategenes or markers on a high resolution mapwith a phenotype of interest in a non-structuredcollection of genotypes have beenused extensively in human medicine toidentify genes involved in disease (Cardonand Bell, 2001). Given the enormous difficultiesof quantitative mapping in humansand the success of association mapping,these methods have also been proposed asways around the problems in developingmarkers for low heritability traits in plants(Gaut and Long, 2003). The developmentof (partially) inbred cassava genetic stockswill certainly accelerate the application ofMAS for the genetic improvement of thecrop.ConclusionsGiven limited resources, further prioritizationof traits is needed for the developmentof markers if they do not already exist. Toppriority should be given to MAS for themost important pests and diseases prevalentin the region for which durable sources ofresistance genes exist. Priority should alsobe given to DMC as this is another traitthat, although having a high narrow senseheritability at the time of evaluation (usuallyafter the onset of the rains to permitplanting immediately thereafter), is significantlyaffected by non-genetic factors andis not as highly heritable. There are severalinitiatives to assist national programmesacquire new molecular tools to increase thecost-effectiveness of breeding. Prominentamong these are the “molecular breedingcommunities of practice” project of theGeneration Challenge Programme (GCP,www.generationcp.org) and the RockefellerFoundation-funded African MolecularMarker Network (AMMANET, www.africancrops.net/ammanet).Both have trainingprogrammes on molecular breeding that areopen to national programme scientists. TheCIAT cassava project has also developeda Web-based database resource includingprotocols, populations, and markers forMAS in cassava that can easily be accessedby national programmes (www.ciat.cgiar.org/mascas).Cassava and common beans: contrastsCassava and beans are similar with respectto the modest level of research input theyhave enjoyed over the past three to five decades.Both have been part of the researchagenda of CIAT and of the CGIAR fornearly thirty years, and especially beanshave benefited from inputs from laboratoriesand programmes in the UnitedStates of America and, to a lesser degree,Europe. However, research investments forhigh-scale genomics through marker developmentin these crops has been far less thanfor the “super crops” like maize, rice orsoybean that enjoy participation by the privatesector, but are more than minor orphancrops with local usage in the tropics.Yet biologically, these two crops arewidely contrasting. Cassava is a perennialversus beans, which are short-seasonannuals, although climbing beans at highaltitudes can be similar to cassava in growthcycle. Beans are an autogamous seed cropwhile cassava is an allogamous crop withvegetative propagation. Accompanying thislatter dichotomy are differences in geneaction. Beans present largely additive gene

Chapter 7 – Marker-assisted selection in common beans and cassava 107action, while cassava expresses importantcomponents of dominance and epistaticaction. Finally, cassava as a clonal cropcan fix heterotic combinations, while alack of genetic male sterility or apomixissystems in common bean have curtailedthe development of a hybrid industry forthis seed crop even though heterosis isobserved.In spite of their biological and otherdifferences, the results of several yearsexperience with MAS in beans and cassavaare surprisingly similar. In both crops,MAS is being employed principally tobolster phenotypic selection for diseaseresistance genes. Disease resistance is oftengoverned by relatively few genes, and phenotypicdata are obtained more easily. Onthe other hand, MAS for more complextraits has yet to find ready application.While there are candidates for such traitsin both crops (root bulking in cassava;low phosphorus or drought tolerance inbeans), the complexity of these traits hasmade the identification of reliable markersmore difficult and has delayed application.Obtaining reliable phenotypic datafor complex traits is especially difficult andis often the biggest bottleneck to eventualapplication of MAS. In the case of cassava,no inbred parents have been used to datefor the development of molecular markers,making the genetic analysis more difficult.However, some differences in the applicationof MAS for the two crops may benoted, arising from the form of reproductionof each crop. The time frame toselect cassava clones through multilocationaltrials is about six to seven years.During this period and with each step thenumber of genotypes is reduced as a resultof the selection exerted, but the genotype ofeach individual clone remains stable. In thecase of beans from the F 1 until stabilizationof pure lines there is an intense segregationprocess in the early generations whichtapers off in later generations. In both cropsMAS can be used in the early stages of theselection process but with different objectives.In cassava, MAS can help to selectearly on the clone that will ultimately bereleased, whereas in beans MAS is usedto “direct” the segregation process in themore desirable direction. Although mapswith significant saturation are available forboth crops, these have been constructedover several years, employing genotypes (inthe case of beans) from different gene poolswith wide polymorphism. A small proportionof these markers (often 20–30 percent)is polymorphic in other hybrid combinationsamong the genotypes within the samegene pool or race that have been created totag a specific trait. Thus, genome coverageis often still not optimal for the high qualityQTL analysis that is usually needed forcomplex traits.RecommendationsCareful prioritization of traits, markersystem and genetic stocks for MASThe limited resources available for cassava orbean research require a judicious allocationof efforts. In the past 10–20 years there hasbeen increased investment in molecularmarker research in both crops. However, aconsiderable proportion of that research wasdirected at demonstrating the usefulness ofdifferent techniques, e.g. RAPD, restrictionfragment length polymorphism (RFLP),amplified fragment length polymorphism(AFLP), etc. Over this period there hasbeen an ever-changing set of technologiesbut relatively little actual benefit derivedfrom their application. There is a tradeoffbetween being on the cutting edgewith the newest technologies and “stickingit out” with an “outdated” technology

108Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishuntil some benefit is extracted from it.On the other hand, far too much efforthas been expended in the identification ofmarkers for traits without carrying thesethrough to application. Often gene taggingis a component of a short-term project, anddoes not receive the necessary follow up inimplementation. In each case, the essentialquestion is: what are the key genes foreach crop? And once defined, which genesmerit the investment to develop molecularmarkers? For investments in molecularmarker development to yield results, itis important that traits are chosen forwhich molecular breeding has both a clearadvantage over field-based selection and isfeasible in the short to medium term. It isalso important that emphasis be given toselecting the relevant crosses, pedigrees andpopulations in which to practise MAS, andto have in place appropriate phenotypingstrategies for the confirmation of MASresults. In this regard, the use of parentalsurveys of many of the genotypes involvedin a given breeding programme is animportant first step in implementing MAS.Short- and long-term research relatedto MASThe present research structure that is normallybased on short-term projects, usuallyof three years’ duration, can seldom beexpected to deliver results of usable markersfor complex traits. Such short-term projectsthat seek to establish the basis for MASor to implement selection should limittheir objectives to simply inherited traits.On the other hand, longer-term fundingeither of a programmatic or successiveproject funding nature, must be obtainedto address more complex traits governedby QTL as these would normally requireat least two phases of three-year projects.The first phase might be expected to revealthe inheritance of a given trait, establishingthe location and numbers of QTL, while asecond phase would be required to validatethese over more environments and to findmarkers that are polymorphic over a widenumber of genotypes and therefore widelyuseful for breeding, as well as adapted torapid laboratory techniques. A mediumtolong-term investment likewise impliescareful prioritization of such traits, withregard to potential impact and the eventualneed for MAS. These reflections are basedupon presently available laboratory techniques,but as techniques for more detailedand widespread evaluation of loci and genotypesare developed (e.g. gene chips foranalysis of multiple loci), conclusions couldchange significantly.Scaling-up technologiesAfter the development of molecular markersfor a trait and their initial implementation,a period of scaling-up in use of the specificmarkers is necessary. Sometimes thisinvolves changes to MAS protocols, inthe marker detection technique or in themarkers themselves. Marker re-design hasbeen a common element of scaling-up exercisesand can involve something as simpleas changing a PCR fragment size to implementinga SNP assay for the actual sequencedifferences between alleles. Technologiesthat speed up the implementation processand lower the costs associated with scalingupare crucial to the success of MAS and areoften neglected.Development of markers that areuseful in a large number of crossesOften a marker developed for a particulartrait in one or a few related genotypes willnot work for other genotypes with highvalue of the trait due to differences in geneor allelic effects. Unravelling the genetics of

Chapter 7 – Marker-assisted selection in common beans and cassava 109major traits of agronomic interest even ina subset of elite parents used for breedingis beyond the resources available for beanand cassava research. Association mappingand linkage disequilibrium mapping,which rely upon non-random associationof candidate genes or markers on a high resolutionmap with a phenotype of interestin a non-structured collection of genotypes,have been proposed as a way aroundthis problem. Association mapping can beused to discover new marker-trait associationsor to validate associations that werefound through conventional genetic mapping.The GCP is facilitating associationmapping of traits of agronomic importancein cassava and beans with the goal of discoveringmore useful markers for a widerrange of genotypes.The need to strike a balance betweenMAS and field-based selectionOccasionally the question is raised: whichis better, MAS or conventional selection?This very question betrays a falsedichotomy that hinders progress. By itself,MAS is seldom an adequate selection tooland therefore must be combined withconventional phenotypic selection. Theobjective should be to develop the optimalbalance between conventional and molecularbreeding, and the “best” balance willbe unique to each situation, crop, selectionscheme, environment and opportunities fordifferent selection methods. More emphasisis needed on combined selection systems,rather than viewing MAS as a replacementfor phenotypic or field selection.ReferencesAfanador, L.K. & Hadley, S.D. 1993. Adoption of a mini-prep DNA extraction method for RAPDmarker analysis in common bean. Bean Improv. Coop. 35: 10–11.Akano, A.O., Dixon, A.G.O., Mba, C., Barrera, E. & Fregene, M. 2002. Genetic mapping of a dominantgene conferring resistance to cassava mosaic disease. Theor. Appl. Genet. 105: 521–525.Asiedu, R., Hahn, S.K., Bai, K.V. & Dixon, A.G.O. 1992. Introgression of genes from wild relativesinto cassava. In M.O. Akoroda & O.B. Arene, eds. Proc. 4 th . Triennial Symp. Internat. Soc. Trop.Root Crops-Africa Branch, pp. 89-91. Nigeria, ISTRC-AB/IDRC/CTA/IITA.Awale, H.E. &. Kelly, J.D. 2001. Development of SCAR markers linked to Co-42 gene in commonbean. Ann. Rep. Bean Improv. Coop. 44: 119–120.Babu, L. & Chatterjee, S.R. 1999. Protein content and amino acid composition of cassava tubers andleaves. J. Root Crops .25: 163–168.Beaver, J.S., Rosas, J.C., Myers, J., Acosta, J., Kelly, J.D., Nchimbi-Msolla, S.M., Misangu, R.,Bokosi, J., Temple, S., Aranud-Santana, E. & Coyne, D.P. 2003. Contributions of the bean/cowpea CRSP to cultivar and germplasm development in common bean. Field Crops Res. 82:87–102.Beebe, S., Velasco, A. & Pedraza, F. 1999. Marcaje de genes para rendimiento en condiciones de altoy bajo fósforo en las accesiones de frijol G21212 y BAT 881. VI Reunião Nacional de Pesquisa deFeijão. Salvador, Brazil.Beebe, S., Skroch, P.W., Tohme, J., Duque, M.C., Pedraza, F. & Nienhuis, J. 2000. Structure ofgenetic diversity among common bean landraces of Mesoamerican origin based on correspondenceanalysis of RAPD. Crop Sci. 40: 264–273.Beebe, S., Rengifo, J., Gaitan, E., Duque, M.C. & Tohme, J. 2001. Diversity and origin of Andeanlandraces of common bean. Crop Sci. 41: 854–862.

110Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishBeebe, S.E., Rojas, M., Yan, X., Blair, M.W., Pedraza, F., Muñoz, F., Tohme, J. & Lynch, J.P.2006. Quantitative trait loci for root architecture traits correlated with phosphorus acquisition incommon bean. Crop Sci. 46: 413–423.Bellotti, A.C. 2002. Arthropod pests. In R.J. Hillocks, J.M. Thresh & A.C. Bellotti, eds. Cassava:biology, production and utilization, pp. 209-235. Wallingford, UK, CABI Publishing.Bellotti, A.C., Arias B., Vargas, V.O., Reyes H.J.A. & Guerrero, J.M. 2002. Insectos y ácarosdañinos a la yuca y su control. In B. Ospina & H. Ceballos, eds. La yuca en el tercer milenio.Sistemas modernos de producción, procesamiento, utilización y comercialización, CIAT PublicationNo. 327, pp. 160-203. Cali, Colombia.Bertram, R.B. 1993. Application of molecular techniques resources of cassava (Manihot esculentaCrantz, Euphorfiabeae): interspecific evolutionary relationships and intraspecific characterization.University of Maryland, USA. (Ph.D. thesis).Blair, M.W. & Beaver, J.S. 1993. Inheritance of bean golden mosaic resistance from bean genotype A429. Ann. Rep. Bean Improv. Coop. 36: 143–144.Blair, M.W., Iriarte, G. & Beebe, S.E. 2003. QTL analysis of an Andean advanced backcross populationfor yield traits derived from wild P. vulgaris. Ann. Rep. Bean Improv. Coop. 46: 7–9.Blair, M.W., Pedraza, F., Buendía, H.F., Gaitán-Solís, E., Beebe, S.E., Gepts, P. & Tohme, J. 2003.Development of a genome-wide anchored microsatellite map for common bean (Phaseolus vulgarisL.) Theor. Appl. Genet. 107: 1362–1374.Blair, M.W., Beaver, J.S., Nin, J.C., Prophete, E. & Singh, S.P. 2006. Registration of PR9745–232and RMC-3 red-mottled dry bean germplasm lines with resistance to bean golden yellow mosaicvirus. Crop Sci. 46: 1000–1002.Broughton, W.J., Hernandez, G., Blair, M.W., Beebe, S.E., Gepts, P. & Vanderleyden, J. 2003. Beans(Phaseolus spp.)–model food legumes. Plant & Soil 252: 55–128.Buendia, H.F., Beebe, S.E., Blair, M.W., Teran, H. & Pedraza, F. 2003. Identificación de marcadoresmoleculares asociados a genes de rendimiento en una población RC2F3.7 de frijol común Phaseolusvulgaris L. (DOR390 x G19892). Fitotecnia Colombiana 3: 57–64.Buitrago, A.J.A. 1990. La yuca en la alimentación animal, p. 446. Cali, Colombia, CIAT.Byrne, D. 1984. Breeding cassava. Plant Breeding Rev. 2: 73–134.Cach, N.T., Lenis, J.I., Perez, J.C., Morante, N., Calle, F. & Ceballos, H. 2005a. Inheritance of relevanttraits in cassava (Manihot esculenta Crantz) for sub-humid conditions. Plant Breed. 125(2):177–182.Cach, N.T., Perez, J.C., Lenis, J.I., Calle, F., Morante, N. & Ceballos, H. 2005b. Epistasis in theexpression of relevant traits in cassava (Manihot esculenta Crantz) for sub-humid conditions. J.Heredity 96: 586–592.Calle, F., Perez, J.C. Gaitán, W., Morante, N., Ceballos, H., Llano, G. & Alvarez, E. 2005. Diallelinheritance of relevant traits in cassava (Manihot esculenta Crantz) adapted to acid-soil savannas.Euphytica 144: 177–186.Cardon, L.R. & Bell, J.I. 2001. Association study design for complex disease. Nature Revs. Genet.2: 91–99.Carvalho, G.A., Paular, T.J., Alzate-Marin, A.L., Nietsche, S., Barros, E.G. & Moreira. M.A. 1998.Herança da resistência da linhagem AND 277 de feijoeiro-comun à raça 63-23 de Phaeoisariopsisgriseola e identificação de marcador RAPD ligado ao gene de resistência. Fitopatol. Bras. 23:482–485.

Chapter 7 – Marker-assisted selection in common beans and cassava 111Ceballos, H., Iglesias, C.A., Pérez, J.C. & Dixon, A.G.O. 2004. Cassava breeding: opportunities andchallenges. Plant Mol. Biol. 56: 503–515.Ceballos, H., Fregene, M., Lentini, Z.J., Sanchez, T., Puentes, Y.I., Perez, J.C., Rosero, A. & Tofiño,A.P. 2006. Development and identification of high-value cassava clones. Acta Hort. 703:63–70.Chavarriaga, P., Prieto, S., Herrera, C.J., López, D., Bellotti, A. & Tohme, J. 2004. Screening transgenicsunveils apparent resistance to hornworm (E. ello) in the non-transgenic, African cassavaclone 60444. In A. Alves & J. Tohme, eds. Adding value to a small-farmer crop. Proc. 6 th . Internat.Scientific Mtg. Cassava Biotech. Network, Book of Abstracts, p.4. Cali, Colombia, CIAT.Chávez, A.L., Sánchez, T., Jaramillo, G., Bedoya, J.M., Echeverry, J., Bolaños, E.A., Ceballos, H. &Iglesias, C.A. 2005. Variation of quality traits in cassava roots evaluated in landraces and improvedclones. Euphytica 143: 125–133.CIAT (International Center for Tropical Agriculture). 1997. Assessing and enhancing agrobiodiversitythrough biotechnology. Annual Report, 1994-1997, pp. 63–65. Cali, Colombia.CIAT. 2001. Bean improvement for sustainable productivity, input use efficiency and poverty alleviation.Annual Report, pp. 37–38. Cali, Colombia.CIAT. 2002. Bean improvement for sustainable productivity, input use efficiency and poverty alleviation.Annual Report, pp. 104–107. Cali, Colombia.CIAT. 2003. Bean improvement for sustainable productivity, input use efficiency and poverty alleviation..Annual Report, pp. 32–33 and 67–75. Cali, Colombia.CIAT. 2004. Bean improvement for sustainable productivity, input use efficiency and poverty alleviation,Annual Report, pp. 84 and 115–117. Cali, Colombia.FAO. 2001. Food Outlook. December 1999. Rome.Ferreira, C.F., Borém, A., Carvalho, G.A., Nietsche, S., Paula, T.J., de Barros, E.G. & Moreira,M.A. 2000. Inheritance of angular leaf spot resistance in common bean and identification of aRAPD marker linked to a resistance gene. Crop Sci. 40: 1130–1133.Fregene, M., Angel, F., Gomez, R., Rodriguez, F., Chavarriaga, P., Roca, W., Tohme, J., &Bonierbale, M. 1997. A molecular genetic map of cassava (Manihot esculenta Crantz). Theor. Appl.Genet. 95: 431–441.Fregene, M., Suarez, M., Mkumbira, J., Kulembeka, H., Ndedya, E., Kulaya, A., Mitchel, S.,Gullberg, U., Rosling, H., Dixon, A. & Kresovich, S. 2003. Simple sequence repeat (SSR) diversityof cassava (Manihot esculenta Crantz) landraces: genetic structure in a predominantly asexuallypropagated Crop Theor. Appl. Genet. 107: 1083–1093.Frei, A., Blair, M.W., Cardona, C., Beebe, S.E., Gu, H. & Dorn, S. 2005. Identification of quantitativetrait loci for resistance to Thrips palmi Karny in common bean (Phaseolus vulgaris L.). CropSci. 45: 379–387.Frisch, M., Bohn, M. & Melchinger, A.E. 1999. Comparison of selection strategies for markerassistedbackcrossing of a gene. Crop Sci. 39: 1295–1301.Gaut, B.S. & Long, A.D. 2003. The lowdown on linkage disequilibrum. Plant Cell 15: 1502–1506.Gepts, P. 1988. A middle American and an Andean common bean gene pool. In P. Gepts, ed. Geneticresources of Phaseolus beans, pp. 375-390. Dordrecht, Netherlands, Kluwer Academic Publishers.Gimelfarb, A. & Lande, R. 1994. Simulation of marker-assisted selection in hybrid populations.Genetical Res. 63: 39–47.Hahn, S.K., Howland, A.K. & Terry, E.R. 1980. Correlated resistance to cassava to mosaic and bacterialblight diseases. Euphytica 29: 305–311.

112Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishHahn, S.K., Terry, E.R. & Leuschner, K. 1980. Breeding cassava for resistance to cassava mosaic disease.Euphytica 29: 673–683.Haley, S.D., Afanador, L. & Kelly, J.D. 1994. Identification and application of a random amplifiedpolymorphic DNA marker for the I gene (potyvirus resistance) in common bean. Phytopathology84: 157–160.Hallauer, A.R. & Miranda Fo., J.B. 1988. Quantitative genetics in maize breeding, second edition.USA, Iowa State University Press.Hershey, C.H. 1984. Breeding cassava for adaptation to stress conditions: development of a methodology,pp. 303-314. Proc. 6 th Internat. Symp. Soc. Trop. Root Crops, Lima, Peru.Hillocks, R.J. & Wydra, K. 2002. Bacterial, fungal and nematode diseases. In R.J. Hillocks, J.M.Thresh & A.C. Bellotti, eds. Cassava: biology, production and utilization, pp. 261–280. USA andUK, CABI Publishing.Hospital, F. & Charcosset, M. 1997. Marker-assisted introgression of quantitative loci. Genetics 147:1469–1485.Hospital, F, Chevalet, C. & Mulsant, P. 1992. Using markers in gene introgression breeding programs.Genetics 132: 1119–1210.Hurtado L., P.X., Ospina, C., Marín, J., Buitrago R.Ch., Fregene, M.A., De Vicente, M.C.,Wongtiem, P., Killian, A. & Wenzl, P. 2005. Comparación de dos marcadores moleculares:Microarreglos-microsatélites para el análisis de la diversidad estructural de yuca Manihot esculentaCrantz [resumen] [CD-ROM]. In Congreso Asociación Colombiana de Fitomejoramientoy Producción de Cultivos (9, 2005, Palmira, Colombia). [Resúmenes] [CD-ROM]. CorporaciónColombiana de Investigación Agropecuaria (CORPOICA), Cali, CO. p. 138.Ikeda, N., Bautista, N.S., Yamada, T., Kamijima, O. & Ishii, T. 2001. Ultra-simple DNA extractionmethod for marker-assisted selection using microsatellite markers in rice. Plant Mol. Bio. Reporter19: 27–32.Islam, F.M.A., Basford, K.E., Redden, R.J., Gonzalez, A.V., Kroonenberg, P.M. & Beebe, S. 2002.Genetic variability in cultivated common bean beyond the two major gene pools. Genetic Resourcesand Crop Evolution 49(3): 271–283.Jaramillo, G., Morante, N., Pérez, J.C., Calle, F., Ceballos, H., Arias, B. & Bellotti, A. C. 2005. Diallelanalysis in cassava adapted to the midaltitude valleys environments. Crop Sci. 45: 1058–1063.Jennings, D.L. 1976. Breeding for resistance to African cassava mosaic. African cassava mosaic.Report Interdiscipl. Workshop IDRC 071e, pp. 39–44. Muguga, Kenya.Jennings, D.L. & Iglesias, C. 2002. Breeding for crop improvement, pp. 149–166. In R.J. Hillocks,J.M. Thresh & A.C. Bellotti, eds. Cassava: biology, production and utilization. UK, CABIPublishing.Johnson, W.C., Guzman, P., Mandala, D., Mkandawire, A.B.C., Temple, S., Gilbertson, R.L. &Gepts, P. 1997. Molecular tagging of the bc-3 gene for introgression into Andean common bean.Crop Sci. 37: 248–254.Jung, G., Skroch, P.W., Coyne, D.P., Nienhuis, J., Aranud-Santana E., Ariyarathne, H.M.,Kaeppler, S.M. & Bassett, M. 1997. Molecular marker-based genetic analysis of tepary beanderivedcommon bacterial blight resistance in different development stages of common bean. J.Am.. Soc. Hort. Sci. 122: 329–337.Kawano, K. 1980. Cassava. In W.R. Fehr & H.H. Hadley, eds. Hybridization of crop plants, pp 225-233. Madison, WI, USA, ASA and CSSA.

Chapter 7 – Marker-assisted selection in common beans and cassava 113Kawano, K. 2003. Thirty years of cassava breeding for productivity- biological and social factors forsuccess. Crop Sci. 43: 1325–1335.Kawano K., Narintaraporn, K., Narintaraporn, P., Sarakarn, S., Limsila, A., Limsila, J., Suparhan,D., Sarawat, V. & Watananonta, W. 1998. Yield improvement in a multistage breeding programfor cassava. Crop Sci. 38: 325–332.Kelly, J.D. & Adams, M.W. 1987. Phenotypic recurrent selection in ideotype breeding of pinto beans.Euphytica 36: 69–80.Kelly, J.D., Schneider, K.A. & Kolkman, J.M. 1999. Breeding to improve yield. In S.P. Singh, ed.Common bean improvement in the twenty-first century. Developments in plant breeding, Vol. 7,pp. 185–222. Dordrecht, Netherlands, Kluwer Academic Publishers.Kelly, J.D, Gepts, P., Miklas, P.N. & Coyne, D.P. 2003. Tagging and mapping of genes and QTL andmolecular-marker assisted selection for traits of economic importance in bean and cowpea. FieldCrops Res. 82: 135–154.Klimyuk, V.I., Carroll, B.J., Thomas, C.M. & Jones, J.D.G. 1993. Alkali treatment for rapid preparationof plant material for reliable PCR analysis. Plant J. 3: 493–494.Kullaya, A., Mtunda, K., Kulembeka, H., Ferguson, M., Marin, J., Ospina, C., Barrera, E., Jarvis,A., Morante, M., Ceballos, H., Tohme, J. & Fregene, M. 2004. Molecular marker-assisted andfarmer participatory improvement of cassava germplasm for farmer/market preferred traits inTanzania. 6 th. Internat. Scientific Mtg. Cassava Biotech. Network. Cali, Colombia, CIAT.Legg, J.P. & Ogwal, S. 1998. Changes in the incidence of African cassava mosaic virus disease andthe abundance of its whitefly vector along south-north transects in Uganda. J. Appl. Entom. 122:169–178.Liao, H., Yan, X., Rubio, G., Beebe, S.E., Blair, M.W. & Lynch, J.P. 2004. Basal root gravitropismand phosphorus acquisition efficiency in common bean. Funct. Plant Biol. 31: 1–12.Mahuku, G., Montoya, C., Henríquez, M.A., Jara, C., Terán, H. & Beebe, S. 2004. Inheritance andcharacterization of angular leaf spot resistance gene in common bean accession G 10474 and identificationof an AFLP marker linked to the resistance gene. Crop Sci. 44: 1817–1824.Mba, R.E.C., Stephenson, P., Edwards, K., Melzer, S., Nkumbira, J., Gullberg, U., Apel, K., Gale,M., Tohme, J. & Fregene, M. 2001. Simple sequence repeat (SSR) markers survey of the cassava(Manihot esculenta Crantz) genome: towards a SSR-based molecular genetic map of cassava.Theor. Appl. Genet. 102: 21–31.Melotto, M., Afanador, L. & Kelly, J.D. 1996. Development of a SCAR marker linked to the I genein common bean. Genome 39: 1216–1219.Meuwissen, T.H.E., Hayes, B.J. & Goddard, M.E. 2001. Prediction of total genetic value usinggenome-wide dense marker maps. Genetics 157: 1819–1829.Miklas, P.N., Larsen, R.C., Riley, R. & Kelly, J.D. 2000a. Potential marker-assisted selection forbc-1 2 resistance to bean common mosaic potyvirus in common bean. Euphytica 116(3):211–219.Miklas, P.N., Smith, J.R., Riley, R., Grafton, K.F., Singh, S.P., Jung, G. & Coyne, D.P. 2000b.Marker-assisted breeding for pyramided resistance to common bacterial blight in common bean.Ann. Rep. Bean Improv. Coop. 43: 39–40.Miklas, P.N., Stone, V., Daly, M.J., Stavely, J.R., Steadman, J.R., Bassett, M.J., Delorme, R. &Beaver, J.S. 2000c. Bacterial, fungal, and viral disease resistance loci mapped in a recombinantinbred common bean population (‘Dorado’/XAN 176). J. Am. Soc. Hort. Sci. 125: 476–481.

114Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishMiklas, P.N., Kelly, J.D., Beebe, S.E. & Blair, M.W. 2006. Common bean breeding for resistanceagainst biotic and abiotic stresses: from classical to MAS breeding. Euphytica 147: 105–131.Muñoz, L.C., Blair, M.W., Duque, M.C., Roca, W. & Tohme, J. 2004. Level of introgression in interspecific(Phaseolus vulgaris x P. acutifolius) congruity-backcross lines. Crop Sci. 44: 637–645.Nichols, R.F.W. 1947. Breeding cassava for virus resistance. East Afr. Agric. J. 12: 184–194.Nietsche, S., Borém, A., Carvalho, G.A., Rocha, C., Paula, T.J., de Barros, E.G. & Moreira, M.A.2000. RAPD and SCAR markers linked to a gene coferring resistance to angular leaf spot incommon bean. J. Phytopathol. 148: 117–121.Nobuyuki, I., Bautista, N.S., Yamada, T., Kamijima, O. & Ishi, T. 2000. Ultra simple DNA extractionmethod for marker-assisted selection using microsatellite markers in rice. Plant Mol. Biol.Reporter 18: 1–6.Okogbenin, E., Porto, M.C.M & Dixon, A.G.O. 1998. Influence of planting season on incidence andseverity of African cassava mosaic disease in the subhumic zone of Nigeria. In I.J. Ekanayake, ed.Root crops and poverty alleviation, pp. 338–392. Ibadan, Nigeria, IITA.Park, S.O., Coyne, D.P., Mutlu, N., Jung, G. & Steadman, L.R. 1999. Confirmation of molecularmarkers and flower color associated with QTL for resistance to common bacterial blight incommon beans. J. Amer. Soc. Hort. Sci. 124: 519–526.Pastor-Corrales M.A., Jara, C. & Singh, S. 1998. Pathogenic variation in, source of, and breedingfor resistance to Phaeoisariopsis griseola causing angular leaf spot in common bean. Euphytica 103:161–171.Pérez, J.C., Ceballos, H., Jaramillo, G., Morante, N., Calle, F., Arias, B. & Bellotti, A.C. 2005a.Epistasis in cassava (Manihot esculenta Crantz) adapted to mid-altitude valley environment. CropSci. 45: 1491–1496.Pérez, J.C., Ceballos, H., Calle, F., Morante, N., Gaitán, W., Llano, G. & Alvarez, E. 2005b.Within-family genetic variation and epistasis in cassava (Manihot esculenta Crantz) adapted to theacid-soils environment. Euphytica 145: 77–85.Polston, J.E. & Anderson, P.K. 1997. The emergence of whitefly transmitted geminiviruses in tomatoin the western hemisphere. Plant Disease 81: 1357–1369.Rajendran, P.G., Ravindran, C.S., Nair, S.G. & Nayar, T.V.R. 2000. True cassava seeds (TCS) forrapid spread of the crop in non-traditional areas. Thiruvananthapuram, Kerala, India, CentralTuber Crops Research Institute. 22 pp.Santana, G.E., Blair, M.W., Morales, F., Mahuku, G., Jara, C. & Castaño, M. 2004. Uso de tecnicasclasicas y avanzadas para identificar genotipos de frijol resistentes a antracnosis y mosaico común.Fitotecnia Colombiana 4: 44–54.Sartorato, A., Nietsche, S., Barros, E.G. & Moreira, M.A. 1999. SCAR marker linked to angular leafspot resistance gene in common bean. Bean Improv. Coop. 42: 23–24.Schneider, K.A., Brothers, M.E. & Kelly, J.D. 1997. Marker-assisted selection to improve droughttolerance in common bean. Crop Sci. 37: 51–60.Scotti, C., Pupilli, F., Salvi, S. & Arcioni, S. 2000. Variation in Vigot and in RFLP-estimated heterozygosityby selfing tetraploid alfalfa: new perspectives for the use of selfing in alfalfa breeding.Theor. Appl. Genet. 101: 120–125.Shull, G.F. 1952. Beginnings of the heterosis concept. In J.W. Gowen, ed. Heterosis, pp. 14–48. Ames,IA, USA, Iowa State College Press.

Chapter 7 – Marker-assisted selection in common beans and cassava 115Singh, S.P. 1999. Integrated genetic improvement. In S.P. Singh, ed. Common bean improvementin the twenty-first century. Developments in plant breeding, Vol. 7, pp. 133–166. Dordrecht,Netherlands, Kluwer Academic Publishers.Singh, S.P., Gepts, P. &. Debouck, D.G. 1991. Races of common bean (Phaseolus vulgaris, Fabaceae).Econ. Bot. 45: 379–396.Singh, S.P., Cardona, C., Morales, F.J., Pastor-Corrales, M.A. & Voysest, O. 1998. Gamete selectionfor upright carioca bean with resistance to five diseases and a leafhopper. Crop Sci. 38: 666–672.Stam, P., & Zeven, A.C. 1981. The theoretical proportion of donor genome in near-isogenic lines ofself-fertilizers bred by backcrossing. Euphytica 30: 227–238.Storey, H.H. & Nichols, R.F.W. 1938. Studies on the mosaic disease of cassava. Ann. Appl. Bio. 25:709–806.Sullivan, J.G. & Bliss, F.A. 1983. Expression of enhanced seed protein content in inbred backcrosslines of common bean. J. Am. Soc. Hort. Sci. 108: 787–791.Swanson, M.M. & Harrison, B.D. 1994. Properties, relationships and distribution of cassava mosaicgeminiviruses. Trop. Sci. 34: 15–25.Tanksley, S.D. & McCouch, S.R. 1997. Seed banks and molecular maps: Unlocking genetic potentialfrom the wild. Science 277: 1063–1066.Urrea, C.A., Miklas, P.N., Beaver, J.S. & Riley. R.H. 1996. A codominant randomly amplified polymorphicDNA (RAPD) marker useful for indirect selection of BGMV resistance in common bean.J. Am. Soc. Hort. Sci. 121: 1035–1039.Vallejo, V. & Kelly, J.D. 2001. Development of a SCAR marker linked to Co-5 gene in common bean.Ann. Rept. Bean Improv. Coop. 44: 121–122.Van Schoonhoven, A. & Pastor-Corrales, M.A. 1987. Standard system for the evaluation of beangermplasm. Cali, Colombia, CIAT.Yan, X., Liao, H., Beebe, S.E., Blair, M.W. & Lynch, J.P. 2005. Molecular mapping of QTLs associatedwith root hairs and acid exudation as related to phosphorus uptake in common bean. Plant &Soil 265: 17–29.Young, R.A., Melotto, M., Nodari, R.O. & Kelly, J.D. 1998. Marker assisted dissection of the oligogenicanthracnose resistance in the common bean cultivar G2333. Theor. Appl. Genet. 96: 87–94.

Chapter 8Marker-assisted selection in maize:current status, potential, limitationsand perspectives from the privateand public sectorsMichel Ragot and Michael Lee

118Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishSummaryMore than twenty-five years after the advent of DNA markers, marker-assisted selection(MAS) has become a routine component of some private maize breeding programmes.Line conversion has been one of the most productive applications of MAS in maize breeding,reducing time to market and resulting in countless numbers of commercial products.Recently, applications of MAS for forward breeding have been shown to increase significantlythe rate of genetic gain when compared with conventional breeding. Costs associatedwith MAS are still very high. Further improvements in marker technologies, data handlingand analysis, phenotyping and nursery operations are needed to realize the full benefitsof MAS for private maize breeding programmes and to allow the transfer of provenapproaches and protocols to public breeding programmes in developing countries.

Chapter 8 – Marker-assisted selection in maize 119IntroductionThe ability to identify genetic componentsof traits, particularly quantitative traits, inMendelian factors, and to monitor or directtheir changes during breeding through theuse of DNA-based markers has createdmuch enthusiasm. Claims were sometimesmade that marker-assisted selection(MAS) would rapidly replace phenotypicselection and dramatically reduce the timerequired to develop commercial varieties(Mazur, 1995). At the turn of this century,phenotypic selection was still theapproach on which maize breeding programmesmostly relied to develop new andimproved cultivars while MAS had contributedto advances in introgression, orbackcross breeding (Ragot et al., 1995; Ho,McCouch and Smith, 2002; Ribaut, Jiangand Hoisington, 2002; Morris et al., 2003).Overly optimistic statements and exaggeratedpromises about the power of MAS toimprove complex traits created excessivelyhigh and largely unfulfilled hopes andprompted a wave of cautious and sometimespessimistic views (Melchinger, Utzand Schön, 1998; Young, 1999; Goodmanand Carson, 2000; Bernardo, 2001).Recently, multinational corporationswith large maize breeding programmesreported the routine and successful use ofMAS (Johnson, 2004; Niebur et al., 2004;Eathington, 2005; Crosbie et al., 2006).Rates of genetic gain twice as high as thoseachieved through conventional breedingwere reported for MAS in maize. Accountswere also given of a number of MAS-derivedsingle-cross (i.e. simple) hybrids being currentlyon the market. Although too littleis known about the methods (e.g. breedingschemes, mathematical algorithms) andtools (e.g. marker technologies, computerprograms, databases) used to develop thesehybrids, these results have raised confidencein the ability of MAS to increasethe rate of genetic gain over what can beachieved through conventional breeding.As technologies evolve and marker genotypesbecome less expensive, MAS becomesincreasingly within the reach of developingcountries. Whenever necessary, transfer ofmethods or tools from private companiesto developing countries should be madepossible while preserving the commercialinterests of the companies concerned,thereby contributing to increasing the rateof genetic gain where it is most needed.Much has happened in maize breedingsince Stuber and Moll (1972) first reportedthat selection for grain yield in maize hadresulted in changes in allele frequenciesat several isozyme loci throughout thegenome. In so doing, they essentially laidthe grounds for MAS in maize. Indeed,if phenotypic selection could produce achange in marker allele frequencies, thenwhy could deliberately altering markerallele frequencies at specific loci not producepredictable phenotypic changes forone or several traits?The objectives of this chapter are toprovide the scientific community anddecision-makers with information on thecurrent status of MAS in maize breedingprogrammes, including the major stepsthat led to it, and to provide suggestionsto developing countries for deploying thetechnology and methods involved in an efficient,cost-effective and realistic manner.How has MAS been used by theprivate sector to improve themaize crop?Applications of DNA markers in privatemaize breeding programmes startedin the 1980s with the identification ofDNA clones used to detect restrictionfragment length polymorphisms (RFLPs)

120Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishin the nuclear genome. As described below,the methods used to detect RFLPs wereincompatible with the magnitude, speedand efficiency of all but a few aspects ofselection in maize breeding programmes.Gradually, however, the methods used todetect DNA polymorphisms and to createmeaningful information from DNA markerand phenotypic data sets have evolved to thepoint where they are routine componentsof some maize breeding programmes in theprivate sector.Selection occurs at various stages inmaize breeding programmes. The firstopportunity arises when choosing inbredlines to mate as parents of new populations.In some programmes, all such inbreds aregenotyped systematically at DNA markerloci (Smith and Smith, 1992). If the markerloci are sufficiently close on genetic orphysical maps then reasonably good inferencesmay be made about the inbred’shaplotype. Such information is used toestablish identity, resolve disagreementsrelated to germplasm ownership and acquisition,enforce laws intended to encouragegenetic diversity of the hybrids and avoidusing inbreds that contain transgenes whichmay violate regulatory considerations andrestrictions. These selection practices, whileadmittedly not conventional MAS, haveled to improvements in the maize crop byenabling more informed stewardship anddeployment of genetic resources and byproviding a degree of protection of intellectualproperty and related investments inmaize breeding.Unquestionably, the most pervasive anddirect use of MAS in maize by the privatesector has been with backcrossing oftransgenes into elite inbred lines, the directparents of the commercial hybrids (Ragotet al., 1995; Crosbie et al., 2006). Currently,the most widely deployed transgenes andcombinations thereof (i.e. gene stacks) arefor resistance to herbicides or insects (e.g.Ostrinia and Diabrotica). As the commercialmaize crop of any region, maturityzone, market or country is not yet uniformor homogeneous for any transgene,maize breeders have elected to developnear-isogenic versions (transgenic and nontransgenic)of elite inbreds and commercialhybrids in order to satisfy combinations oflicensing agreements, agronomic practices,regulatory requirements, market demandsand product development schemes. Thishas required companies to have twoparallel maize breeding programmes, transgenicand non-transgenic. In this manner,marker-assisted backcrossing (MABC) oftransgenes, and to a lesser degree, of nativegenes and quantitative trait loci (QTL) forother traits, has expedited the developmentof commercial hybrids.More recently, marker-assisted recurrentselection (MARS) schemes and infrastructurehave been developed for “forwardbreeding” of native genes and QTL forrelatively complex traits such as diseaseresistance, abiotic stress tolerance and grainyield (Ribaut and Betrán, 1999; Ragot et al.,2000; Ribaut, Jiang and Hoisington, 2000;Eathington, 2005; Crosbie et al., 2006).Simulation studies suggested that MAScould be effective for such traits under certainconditions (Edwards and Page, 1994;Gimelfarb and Lande, 1994), but the initialempirical attempts at such selection werenot successful (Stromberg, Dudley andRufener, 1994; Openshaw and Frascaroli,1997; Holland, 2004; Moreau, Charcossetand Gallais, 2004) except in the specialcase of sweetcorn (Edwards and Johnson,1994; Yousef and Juvik, 2001). The successreported for sweetcorn is due to the fact thatthe genetic base of sweetcorn is extremelynarrow relative to dent or flint maize; thus

Chapter 8 – Marker-assisted selection in maize 121predicted gains and extrapolations acrosspopulations are more reliable. Also, phenotypicanalyses of many traits in a sweetcornbreeding programme are extremely expensivebecause they involve processing largevolumes of grain; therefore, MAS would berelatively inexpensive and effective undersuch circumstances. However, subsequentdevelopments in technology, refinementsin analytical methods and improvementsin experimental designs have been assembledinto a process that has shown promisefor some reference populations of dentmaize (Ragot et al., 2000; Johnson, 2004;Crosbie et al., 2006) as improvement ingrain yield from MAS often exceeded thatfrom non-MAS approaches. Presumably,such results will lead to the developmentof new and superior inbred lines and commercialhybrids in a cost-effective manner.While the impact of such MAS has not yetbeen fully realized in the maize crop, themethods have been employed to variousdegrees by programmes in the private sectorthat have the necessary infrastructure.The potential for MAS to contribute toimprovements in the maize crop shouldincrease in parallel with our understandingof the relationships among genomes, theenvironment and phenotypes. Candidatetransgenes will be developed on a regularbasis and their contributions to maizeimprovement will be realized in the mostefficient manner with MAS. Likewise, theidentification of candidate native genes andtheir gene products and functions, andof other DNA sequences (e.g. miRNA,matrix attachment and regulatory regions),will improve the power of methods such asassociation mapping and genome scans toassess their genotypic value in the contextof defined reference populations of significanceto maize breeding (Thornsberryet al., 2001; Rafalski, 2002; Niebur et al.,2004; Varshney, Graner and Sorrels, 2005).Beyond its use in MARS schemes, thisinformation might make it reasonable toreconsider ideas such as methods for predictinghybrid performance that may havebeen limited by the amount and type ofinformation and by the design of the experimentwhen they were initially evaluated(Bernardo, 1994).methodology and design ofbreeding programmes supportedby MASAs expected, private sector maize programmesfocus entirely on inbred-hybridbreeding schemes intended to develop eliteinbred lines that enable the profitable productionof commercial F 1 hybrids. To alarge extent, MAS breeding programmesuse the same designs and methods knownto maize breeders for decades and genericdescriptions of these have been published(Hallauer and Miranda, 1981; Sprague andDudley, 1988; Bernardo, 2002). When MASis included in the breeding programme, thesignificant differences are, of course, theavailability of genotypic data at differentstages of selection and some knowledgeof the relationships between the genotypicand phenotypic data sets for the referencepopulation(s) in the target environment(s).In contrast to conventional breedingschemes, the methods and design ofinfrastructure needed to support MAShave been the areas of greatest change.In order to utilize MAS, companieshad to make significant investments toassemble or modify various aspects ofinfrastructure such as methods to detectDNA polymorphism, manage information,or analyse and track samples, softwareto relate genotype with phenotype, andoff-season or continuous nurseries. Thesecomponents had to be integrated with each

122Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishother and with breeding activities, whichmeant that scientists needed to learn howand when MAS provided a comparativeadvantage over other methods.MAS: enabling methods, tools andinfrastructurePerhaps the component of infrastructure ingreatest need of development was related tothe acquisition of genotypic data (i.e. DNAmarkers). Although the concept of associatingmarkers with quantitative traits wasnot new (Sax, 1923), the discovery reportedby Stuber and Moll (1972) was very significant.Stuber and Moll (1972) described forthe first time associations between molecularmarkers and quantitative traits whileprevious associations had been based onmorphological markers (Sax, 1923). Theadvantages of molecular over morphologicalmarkers soon became obvious anddetailed descriptions of these advantageswere published by Tanksley et al. (1989)and Stuber (1992).Two of these advantages are of particularimportance. First, molecular marker genotypescan usually be obtained from anyplant tissue, even from young seedlings orkernels, while morphological markers frequentlyrequire the observation of whole,mature plants. Selection can therefore occurearlier in the plant’s cycle when usingmolecular markers than when using morphologicalmarkers. The ability to conductearly selection, possibly before flowering,can have a tremendous impact on the rateof genetic gain of a breeding programmeand therefore constitutes a very significantadvantage of molecular over morphologicalmarkers.Second, molecular markers are neutralmarkers. They are not affected by environmentalor growing conditions. Theyare not affected by the genetic backgroundeither, nor do they affect phenotypes. Theexpression of morphological traits, by contrast,can be dependent on environmentalor growing conditions. In addition, epistasicinteractions are often observed amongmorphological marker loci or betweenmorphological marker loci and the geneticbackground. These epistatic interactionsprevent distinguishing all genotypes associatedwith morphological markers and furtherlimit the number of morphological markersthat can be studied simultaneously.Although isozyme markers had manyadvantages over morphological markers,the lack of a sufficient number of polymorphicloci limited their use for MAS(Goodman et al., 1980). Nevertheless, isozymemarkers are still used for qualitycontrol during seed production.RFLPs (Botstein et al., 1980) are basedon DNA polymorphisms detected throughrestriction nuclease digestions followedby DNA blot hybridizations. The abundanceand high level of polymorphism ofRFLPs, especially in maize, allowed theconstruction of extensive maize geneticmaps (Helentjaris et al., 1986; Burr et al.,1988; Hoisington, 1989; Coe et al., 1995;Davis et al., 1999) as well as the identificationand mapping of many QTL.Being robust, reproducible and codominant,RFLPs are perfectly suited forgenetic studies as well as for MAS applications.Their two main disadvantagesare the large quantities of DNA required,and the difficulty to miniaturize and automate.Nevertheless, RFLPs were quicklyadopted and represented the marker systemof choice for many plant species includingmaize throughout the 1980s and duringmuch of the 1990s.The development of the polymerasechain reaction (PCR) (Saiki et al., 1988)turned out to be a major breakthrough in

Chapter 8 – Marker-assisted selection in maize 123molecular marker technology. PCR-basedmarkers require little DNA, allowing samplingof young seedlings and very earlyselection and thereby optimization ofbreeding schemes. PCR-based marker protocolsare very amenable to automation andminiaturization and improvements to protocolsresulted in considerable reductionsin both cost and time required to producedata points. The first two PCR-basedmarker systems were random amplified polymorphicDNA (RAPDs), and amplifiedfragment length polymorphisms (AFLPs).Detailed descriptions and critical assessmentsof these two systems can be found inWelsh and McClelland (1990), Williams etal. (1990), Penner et al. (1993), Ragot andHoisington (1993), Skroch and Nienhuis(1995) and Jones et al. (1997) for RAPDs,and in Vos et al. (1995), Jones et al. (1997)and Castiglioni et al. (1999) for AFLPs.They are also described in other chaptersof this book.Simple sequence repeats (SSRs) or microsatellitesrapidly became the marker ofchoice in maize, almost entirely displacingRFLPs and previously developed PCRbasedmarker systems. Polymorphism ofSSRs is due to variable numbers of shorttandem repeats, often two or three basepairs in length and usually flanked byunique regions (Tautz, 1989). SSRs arevery reproducible (Jones et al., 1997) andco-dominant (Shattuck-Eidens et al., 1990;Senior and Heun, 1993; Senior et al., 1996)and are therefore very suitable for maizeMAS applications.Many additional variations of PCRbasedmarker systems have been developedand a thorough review can be found inMohan et al. (1997).All the DNA-based marker systems describedto date are gel-based systems, a majorconstraint for automation. Single nucleotidepolymorphisms (SNPs) (Lindblad-Toh etal., 2000) can be revealed in many waysincluding allele-specific PCR, primerextension approaches, or DNA chips, all ofwhich are not gel-based. SNPs can generallybe scored as co-dominant markers, exceptin the case of insertion-deletion polymorphisms.Although allelic diversity at SNPs isusually limited to two alleles, this limitationcan be offset by the abundance of SNPs andthe analysis of haplotypes, combinations ofgenotypes at several neighbouring SNPs.Haplotype analyses increase informativeness(Ching et al., 2002), although at someexpense because two to four SNPs have tobe genotyped where one SSR sufficed. SNPgenotyping can be highly miniaturized andautomated, thereby reducing the cost andallowing the production of very large numbersof data points. With genetic mapscontaining several thousand mapped SNPs,these have become the marker of choice forprivate maize MAS programmes.DNA marker technology has been adynamic and often expensive componentof the infrastructure needed for MAS. Forexample, one corporation indicated havingspent tens of millions of United Statesdollars to develop an automated systemfor detecting RAPDs, a technology thatwas never suited for MAS in a large maizebreeding programme. Later, another corporationspent an even greater amount ofmoney to acquire technology for matrixassistedlaser desorption/ionization time offlight (MALDI-TOF) analysis of amplifiedDNA fragments. These technologies wereeither rapidly replaced or never used. Suchdecisions would have bankrupted mostnational maize programmes or a coupleof centres belonging to the ConsultativeGroup on International AgriculturalResearch (CGIAR). Fortunately, this areaof infrastructure has matured somewhat

124Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishand become more stable so that start-upand operating costs, while still high forsome programmes, are more predictable.Statistical methods and related softwarehave also been areas of significant development,especially for the detection anddescription of putative QTL. QTL, whichare nothing more than associations betweenmarkers and traits, were first describedusing simple association tests betweentrait values and marker genotypes (Stuberand Moll, 1972). These tests consider eachmarker locus independently and neitherrequire nor take advantage of the existenceof genetic maps. Statistical methods havebeen developed that take advantage of theexistence of genetic maps (see review byManly and Olson, 1999). These statisticalmethods, simple interval mapping (Landerand Botstein, 1989) and composite intervalmapping (Jansen, 1993; Zeng, 1993, 1994),test the existence of associations betweenhypothetical marker genotypes and traitvalues at several points in intervals betweenpairs of adjacent marker loci on the geneticmap, allowing the positioning of QTL onthese genetic maps. All of the previousmethods are based on single QTL models.Other statistical methods have been developedthat simultaneously test the presenceof several QTL in the genome (Kao, Zengand Teasdale, 1999).Many software packages are availablefor QTL mapping and based on one orseveral of the statistical methods developedto date. No two packages are exactly alikeand all have specific strengths and weaknesseswith respect to particular situations,making it sometimes beneficial to use morethan one package to perform QTL mappinganalyses. The software packages mostcommonly used for QTL mapping in maizeinclude QTL Cartographer (Basten, Weirand Zeng, 1994), MapQTL (van Ooijen andMaliepaard, 1996), and PLABQTL (Utz andMelchinger, 1996). All of these only handlebi-allelic populations, while MCQTL(Jourjon et al., 2005) also performs QTLmapping in multi-allelic situations, includingbi-parental populations made fromsegregating parents, or sets of bi-parental,bi-allelic populations.More recently, methods based onBayesian analysis (Jansen, Jannink andBeavis, 2003; Gelman et al., 2004) andassociation (Varshney, Graner and Sorrels,2005) or in silico mapping (Parisseaux andBernardo, 2004) have been proposed asmore powerful and refined approaches toassess the relationships between genotypeand phenotype that are needed for MAS.Methods of Bayesian analysis should beless affected by the uncertainties of QTLeffects and locations and produce betterestimates of those parameters in MAS.Association mapping approaches are particularlyuseful to validate the relevanceof genes and alleles in specific germplasmsuch as that used by maize breeders. Insilico mapping takes advantage of the pedigreerelationships among individuals tostructure the population used to establishmarker-trait associations. This approach,which is highly complex due to the populationstructure resulting from pedigreebreeding, is particularly appropriate formaize where data across many years andenvironments are available for large setsof related individuals. Certainly, as theannotation of genomes gradually improves,such methods will be common componentsof breeding programmes. Currently, theapplications of methods such as associationmapping for MAS are hindered by thefact that a very low percentage of the genesin crop plants have a function assigned tothem on the basis of direct experimentation.However, this impoverished situation

Chapter 8 – Marker-assisted selection in maize 125is being enriched through a variety ofprojects on functional genomics.In sharp contrast to the many methodsand software packages developed for QTLidentification and mapping, little has beenpublished for MAS. This paucity of informationon MAS tools most likely reflectsboth the low level of activity in the publicsector and the fully proprietary nature ofdevelopments in the private sector.In parallel with advancements in DNAtechnology and statistical methods, privatesector programmes have enhanced the capabilitiesand capacities of their continuousnurseries. Such nurseries have been usedfor decades by programmes in both the privateand public sectors. In order to conductMAS to its greatest advantage, continuousnurseries had to be managed, equipped andstaffed in new ways so that the plants completetheir life cycle as quickly as possibleand that the genotypic data (and sometimessome phenotypic data) needed for MASmay be collected at each sexual generation.Three to four sexual generations per yearmay be completed at such nurseries.These activities and the continuous collectionof both genotypic and phenotypicdata in the target environment and theirintegrated analyses create huge data setsthat must be analysed quickly and relatedto other large extant data sets. Data managementand bioinformatics for breedershave therefore become critical componentsof the infrastructure needed to use MAS.Prior to the advent of MAS, some largeprivate breeding programmes had establisheda group of dedicated data managersto assist with research and marketing, andwith the arrival of genomics and MASthe need for such dedicated specialists hasincreased greatly.Once the basic infrastructure had beenestablished to complement the activitiesof maize breeders, programmes wereready to implement several basic aspectsof MAS; many of which are derived fromwell established methods and principles ofmaize breeding.MAS-based breedingSelection occurs at various stages in maizebreeding programmes. The first opportunityfor selection is the choice of inbredlines to mate as parents of new populations.Prior to the advent of DNA markerdata, the selection of such parents wouldbe based on a combination of phenotypicassessments, pedigree information, breedingrecords and chance (Hallauer and Miranda,1981; Sprague and Dudley, 1988). In someprogrammes today, all such inbreds aregenotyped systematically at DNA markerloci. Depending on the resources andobjectives, the degree of genotyping mayrange from a low density of marker loci(e.g. SNPs in candidate genes) to higherdensity whole genome scans (Varshney,Graner and Sorrels, 2005). These genotypicdata, alone or integrated with phenotypicinformation, may reveal novel aspects ofmaize gene pools, heterotic groups, haplotypeevolution, gene content and parentsused in MAS for specific target environments(Fu and Dooner, 2002; Niebur et al.,2004; Crosbie et al., 2006). When properlyintegrated with phenotypic informationand functional genomics, genotypic dataof inbred lines should allow breeders tochoose parents that, when mated, shouldprovide populations or gene pools enrichedfor the more desirable combinations offavourable alleles. Such a starting point isa huge advantage in plant breeding becauseit increases the probability of selectingprogeny that are superior to the parentsand that approximate a predicted optimumgenotype.

126Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishMABC is certainly the form of MASwith the most immediate and obvious benefitsfor maize breeding. MABC is usedfor three main purposes: selection of transgenes(or of native DNA sequences of themaize genome, whether genes or QTL),elimination of unwanted regions of thedonor-parent genome linked to the transgeneand selection of unlinked regions ofthe recurrent-parent genome. With theexception of DNA markers and transgenes,these have been the same goals of backcrossbreeding since the inception of that methoddecades ago (Fehr, 1987). Of course, DNAmarkers enable breeders to identify progenythat contain the desired recombinantchromosomes and donor-parent genome ina more direct manner. Also, MABC facilitatesthe process of combining more thanone transgene in a given inbred line (e.g.“gene or trait stacking or pyramiding”).This reduces the number of generationsneeded to reach certain stages of a breedingprogramme and reduces the time needed toproduce commercial hybrids for the market.Generic MABC schemes suitable for maizebreeding programmes have been describedin detail for single genes (Hospital, Chevaletand Mulsant, 1992; Ragot et al., 1995;Frisch, Bohn and Melchinger, 1999a, 1999b;Frisch and Melchinger, 2001a; Hospital,2001; Ribaut, Jiang and Hoisington, 2002),for QTL (Hospital and Charcosset, 1997;Bouchez et al., 2002) and for gene stacks(Frisch and Melchinger, 2001b). Versions ofsuch schemes have been used in maize breedingprogrammes in the private sector, oftenat their continuous nurseries (Ragot et al.,1995). Most recently, MABC has also beenadopted as a tool to develop sets of nearisogeniclines (NILs) for genomics research(Peleman and van der Voort, 2003).Theoretical and simulation studies havebeen conducted to identify the most efficientMABC protocols. Parameters mostcommonly studied include the number ofindividuals genotyped at each generation,the number of markers used, relative selectionpressure for recombination around thetarget locus or global recovery of recurrentparent genome and the number of individualsselected at any generation. Optimalvalues for each of the above depend onthe objective of the MABC approach interms of quality (required level of recurrentparent genome recovery), speed(fastest possible conversion or set numberof generations) and resources (unlimitedor limited). While the fastest and highestquality MABC approaches have themost expensive protocols, less intensiveapproaches can result in significant timesavings and quality improvements whencompared with conventional backcrossingapproaches and at a fraction of the cost ofthe most expensive MABC protocols.Frisch, Bohn and Melchinger (1999b)showed that to minimize linkage dragaround the target locus (loci), selection ofrecombination events close to the targetlocus (loci) should be conducted in theearly backcross generations. Frisch andMelchinger (2001a) and Ribaut, Jiang andHoisington (2002) further demonstratedthat minimizing linkage drag around thetarget locus requires very large numbersof individuals (possibly hundreds) to begenotyped. Hospital and Charcosset (1997)proposed a selection scheme based onselecting a single individual to be backcrossed.By contrast, Frisch and Melchinger(2001a) proposed selecting several individualsand determining the family sizeof their backcross progeny based on theindividuals’ genotypes. By using varyingrather than constant numbers of individualsor markers at the different backcrossgenerations, it was shown that the number

Chapter 8 – Marker-assisted selection in maize 127of marker data points required could bereduced and thus the efficiency of MABCimproved (Hospital, Chevalet and Mulsant,1992; Frisch, Bohn and Melchinger, 1999b).Several studies also showed that using alimited number of markers on non-carrierchromosomes was sufficient to recovermore that 95 percent of the recurrent parentgenome in three or fewer backcross generations(Hospital, Chevalet and Mulsant,1992; Visscher, Haley and Thompson, 1996;Servin and Hospital, 2002).One of the most important lessons fromthe various theoretical and simulation studiesof MABC is that the effects of thedifferent MABC parameters are not independentof each other. With maize, largebackcross populations can be generatedfrom a single plant when that plant is usedas the male and recurrent parent plants areused as females. Marker systems in maizeare also such that very large amounts ofmarker data can be generated on plantsbefore flowering. Potential MABC protocolsare almost endless in maize andidentifying the most efficient is only possibleon a case-by-case basis. For example,while achieving almost complete recovery ofthe recurrent parent’s genome is necessaryfor registering backcross-derived lines andhybrids in many European countries, partialrecovery might be sufficient to improvethe agronomic performance of varieties indeveloping countries. The optimal MABCprotocols for these two strikingly differentobjectives will be very different. Protocolsfor the first objective will involve backgroundselection and the use of backgroundmarkers very close to the target locus (loci).Protocols for the second objective mightinvolve markers for the target locus (loci)only, while relying on successive backcrossgenerations to recover an adequate amountof recurrent parent genome.Successful examples of MABC in maizeinclude backcrossing of transgenes (Ragotet al., 1995), and QTL for insect resistance(Willcox et al., 2002), flowering maturity(Ragot et al., 2000; Bouchez et al.,2002) and grain yield (Ho, McCouch andSmith, 2002).Methods of “forward breeding” withDNA markers have also been proposed andimplemented by maize breeding programmes.As with the pedigree-based methods ofmaize breeding favoured by the private sector,many of the “new” methods that utilizegenetic data from DNA markers integratedwith phenotypic data are essentially a formof recurrent selection, a method that hasbeen in use for several decades (Hallauerand Miranda, 1981). The key advantages ofthe new versions of recurrent selection are,of course, the availability of genetic data forall progeny at each generation of selection,the integration of genotypic and phenotypicdata, and the rapid cycling of generations ofselection and information-directed matingsat continuous nurseries.At least two distinct forms of forwardbreeding with MAS have been proposed:single large-scale MAS (SLS-MAS) (Ribautand Betrán, 1999) and MARS (Edwardsand Johnson, 1994; Lee, 1995; Stam, 1995).A key difference between the methods isthat SLS-MAS employs DNA markers atonly one generation and attempts to retaingenetic variation in regions of the genomeunlinked to the DNA markers, whileMARS uses markers at each generation,exhausting genetic variation in mostregions of the genome. Versions of bothSLS-MAS and MARS have been used bybreeding programmes in the private sector(Johnson, 2004; Eathington, 2005; Crosbieet al., 2006).SLS-MAS is of particular interest inpedigree breeding as it consists of screening

128Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishand selecting individuals at a few lociat early generations, usually F 2 or F 3 ,(Eathington, Dudley and Rufener, 1997),using large populations (Ribaut and Betrán,1999). Individuals displaying homozygousfavourable genotypes at the loci of interestare selected and self-pollinated while othersare discarded. Self-pollinated progeny ofthe selected plants then proceed normallythrough subsequent steps of pedigreebreeding. Screening large populations isnecessary to ensure that genetic diversity ismaintained at regions not under genotypicselection, thereby allowing furtherphenotypic selection to be conducted.Loci at which marker selection operatescan be QTL as described by Ribaut andBetrán (1999). SLS-MAS is thus limitedby issues such as the precision of the QTLparameters (position, effect), and relevanceof the QTL across environments or genepools. SLS-MAS can also be conducted forgenes, eliminating many of the limitationspertaining to QTL. Although a powerfulapproach adopted in several species (barley,soybean, sunflower, wheat) to enrichbreeding populations at a few loci (Crosbieet al., 2006), SLS-MAS does not appear tohave been widely implemented in maizebreeding programmes.MARS targets all traits of importancein a breeding programme and for whichgenetic information can be obtained.Genetic information is usually obtainedfrom QTL analyses performed on experimentalpopulations and comes in the formof maps of QTL with their correspondingeffects. If the QTL mapping analysisis conducted based on a bi-parental population,the sign of the effect at each QTLindicates which of the two parents carriedthe favourable allele at that QTL. As bothparents often contribute favourable alleles,the ideal genotype is a mosaic of chromosomalsegments from the two parents. Thisassumes that the goal is to obtain individualswith as many accumulated favourable allelesas possible, a different goal from that ofmarker-assisted population improvement asstudied elsewhere (Lande and Thompson,1990; Gimelfarb and Lande, 1994; Gallais,Dillmann and Hospital, 1997; Hospital,Chevalet and Mulsant, 1997; Knapp, 1998;Moreau et al., 1998; Xie and Xu, 1998).Population improvement schemes aregenerally based on the random matingof selected individuals while the schemeproposed here is based on directed recombinationbetween specific individuals. Asreported by Stam (1995), the ideal genotype,defined as the mosaic of favourablechromosomal segments from two parents,will usually never occur in any F n populationof realistic size. It is, however, possibleto design a breeding scheme to produceor approach this ideal genotype based onindividuals of the experimental population.This breeding scheme could involve severalsuccessive generations of crossing individuals(Stam, 1995; Peleman and van der Voort,2003) and would therefore constitute whatis referred to as MARS or genotype construction.This idea can be extended tosituations where favourable alleles comefrom more than two parents (Stam, 1995;Peleman and van der Voort, 2003).Van Berloo and Stam (1998, 2001) andCharmet et al. (1999) developed computersimulations around this idea and assessedthe relative merits of marker-assistedgenotype construction over phenotypicselec-tion. MARS was simulated in anexperimental population where QTL hadbeen mapped. Index (genetic) values werecomputed for each individual based onits genotypes at QTL-flanking markers(van Berloo and Stam, 1998, 2001). Allsimulation studies of MARS found that

Chapter 8 – Marker-assisted selection in maize 129it was generally superior to phenotypicselection in accumulating favourable allelesin one individual (van Berloo and Stam,1998, 2001; Charmet et al., 1999). MARSappeared to take better advantage of thegenetic diversity present in the populationsto which it was applied than phenotypicselection. Simulation research conductedby van Berloo and Stam (2001) showedthat MARS was between 3 and almost20 percent more efficient than phenotypicselection. The advantage of MARS overphenotypic selection was greater when thepopulation under selection was larger ormore heterozygous (BC 1 s or F 2 s vs. RILs,recombinant inbred lines, or DHs, doubledhaploids). Although van Berloo andStam (2001) limited their simulations topopulations of up to 200 individuals, theirresults seem to indicate that the relativeadvantage of marker-assisted over phenotypicselection would keep increasing aspopulation size increased. The same simulationstudies showed that the advantageof marker-assisted over phenotypic selectionwas larger when dominant QTL wereinvolved in the selection index, or whentrait heritability was low in the case ofselection for a single trait (van Berloo andStam, 1998, 2001). These latter observationsare of little relevance to most commercialmaize breeding programmes, the goalof which is generally the development ofinbred lines improved for several traitsthat will be later combined into superiorhybrid varieties. They should, however,increase the appeal of MARS approachesfor breeding programmes aimed at developingopen-pollinated varieties.Simulations have also addressed theimpact of the amount and quality ofQTL information on selection efficiency.Simulation and empirical studies (Beavis,1994, 1999) showed that QTL mappingexperiments based on segregatingpopulations of less than 500 individualsgenerally revealed only a subset of all QTLaffecting the complex traits segregating inthese populations. Quantitative trait lociinformation used in subsequent MARSwas therefore necessarily incomplete.Van Berloo and Stam (2001) showed thatthe relative advantage of MARS overphenotypic selection decreased rapidlywhen the fraction of the total genotypicvariance explained by the QTL included inthe selection index decreased. By contrast(van Berloo and Stam, 1998; Charmet et al.,1999), the efficiency of MARS seems to berather robust to the well-documented (Lee,1995) uncertainty of QTL genetic locations.The use of genotypic information at markersflanking the QTL possibly explains thisobservation.The cost efficiency of MARS was alsoinvestigated through simulation (Moreauet al., 2000; Xie and Xu, 1998). Whensimulating selection for a single trait,Moreau et al. (2000) found that, irrespectiveof the heritability of the trait, MARS wasalways more cost efficient than phenotypicselection if the cost of genotyping was lessthan that of evaluating one individual inone plot. When simulating simultaneousselection for multiple traits, Xie and Xu(1998) found that MARS was more costefficient than phenotypic selection if thecost of genotyping was less than that ofphenotyping one individual for all traits.These studies were based on a singlegeneration of MARS. Also, they did nottake into consideration any factors besidesgenotyping and phenotyping costs, althoughfactors influencing the length of a selectioncycle or the number of cycles that can becompleted in a year can obviously affect therelative economic merits of marker-assistedand phenotypic selection.

130Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishIn contrast to the abundance of QTLmapping reports, very few accounts ofMARS experiments are found in theliterature. Moreau, Charcosset and Gallais(2004) compared phenotypic, markeronly,and combined recurrent selection forgrain yield and grain moisture at harvestover several cycles and years in maize.Combined selection was based both onphenotypic and marker information whilemarker-only selection was based on markerinformation only. Both the marker-onlyand the combined selection methodsconstitute MARS approaches. Severalcombinations of these three methods ofselection were applied to the segregatingpopulation that served to map the QTLused in marker-based selection indices.Over the six years of the experiment,two cycles of phenotypic selection, twocycles of combined selection, one cycle ofcombined selection followed by two cyclesof marker-only selection, and one cycle ofmarker-only selection were conducted inparallel. A reassessment of the positionsand effects of QTL was conducted after thefirst cycle for the three schemes containingmultiple cycles. All MARS methods weremore efficient than phenotypic selectionto increase the frequency of favourablealleles at QTL. Nevertheless, Moreau,Charcosset and Gallais (2004) reportedno significant difference between markerassistedand phenotypic selection on themultitrait performance index, although allMARS methods resulted in genetic gain forboth grain yield and grain moisture whilephenotypic selection resulted in genetic gainfor grain yield but an unfavourable evolutionof grain moisture. This disappointingresult was tentatively explained by thehigh heritability of the traits, favourable tophenotypic selection, while the percentageof total phenotypic variance explained bythe QTL detected for both traits was onlyabout 50 percent. One very encouragingresult of this experiment, although Moreau,Charcosset and Gallais (2004) failed topresent it as such, was that the first cycleof marker-only selection was as efficientas phenotypic or combined selection indelivering genetic gain. Two conclusionscan be drawn from this observation.First, the QTL identified in the initialexperimental population were in generalnot artefacts. Second, selection pressureapplied at these QTL, and aimed at fixingalleles identified as favourable, resulted ina change in performance of the selectedpopulation in the desired direction whencompared with the initial population.A similar experiment, although basedsolely on marker-only recurrent selection,was reported by Openshaw and Frascaroli(1997). They conducted MARS in maizesimultaneously for four traits, for each ofwhich about ten QTL had been identified.They showed that genetic gain had beenachieved in the first cycle of MARS, butthat later cycles did not result in any gain.Possible explanations given for these resultsincluded uncertainties about QTL parameters(location and effect), interaction effects(epistasis, genetic x environment interaction),and the fact that selection was basedon single markers rather than chromosomalsegments (Openshaw and Frascaroli, 1997).Recent communications from severalprivate MARS research programmes (Ragotet al., 2000; Eathington, 2005; Crosbieet al., 2006) revealed large-scale successfulapplications in maize. Accounts weregiven of commercial maize hybrids forwhich at least one of the parental lineswas derived through MARS. Eathington(2005) and Crosbie et al. (2006) reportedthat the rates of genetic gain achievedthrough MARS were about twice those

Chapter 8 – Marker-assisted selection in maize 131of phenotypic selection in some referencepopulations. Marker-only recurrent selectionschemes have been implemented fora variety of traits including grain yield andgrain moisture (Eathington, 2005), or abioticstress tolerance (Ragot et al., 2000), andmultiple traits are being targeted simultaneously.Selection indices were apparentlybased on 10 to probably more than 50 loci,these being either QTL identified in theexperimental population where MARS wasbeing initiated, QTL identified in otherpopulations, or genes. Marker genotypesare generated for all markers flanking QTLincluded in the selection indices (Ragotet al., 2000). Plants are genotyped at eachcycle and specific combinations of plantsare selected for crossing, as proposed byvan Berloo and Stam (1998). Several, probablythree to four, cycles or MARS areconducted per year using continuous nurseries.In maize, early versions of suchschemes have been tested and implemented(Johnson, 2004; Crosbie et al., 2006).Results reported in these recent communicationsabout private MARS experiments(Ragot et al., 2000; Eathington, 2005) are insharp contrast to those in earlier publications(Openshaw and Frascaroli, 1997; Moreau,Charcosset and Gallais, 2004). Several factorscan explain these discrepancies:• Size of the populations submitted to selectionat each cycle. Given reports thatincreasing population size should resultin higher genetic gain through MARS(van Berloo and Stam, 2001,) it is likelythat populations submitted to selectionin private programmes are rather large,larger than the 160 and 300 individualsreported respectively by Openshaw andFrascaroli (1997) and Moreau, Charcossetand Gallais (2004).• Use of flanking versus single markers.The use of flanking markers for QTLunder selection allows better predictionof the genotype at the QTL than whenusing single markers. When single markersare used, recombination events thatoccur between the marker and the QTLlead to loss of linkage between the markerand the QTL much faster than whenflanking markers are used, thereby rapidlyreducing the predictive power of thesingle marker.• Early selection, pre-flowering. The abilityto select plants before flowering ensuresoptimal mating schemes as the genotypesof plants being selfed or intercrossed arefully known. However, this is not thecase when selection cannot take placebefore flowering and involves intercrossingselfed progenies of selected plants, thegenotypes of which might have driftedsignificantly from those of their genotypedparents.• Number of generations per year. To theauthors’ knowledge, none of the simulationor experimental studies of MARS hasassessed the effects of cycle length on itsefficiency despite its direct relationshipto the rate of genetic gain. In maize,cycle length can be reduced three- to sixfoldwhen using marker-only recurrentselection compared with phenotypicrecurrent selection. Consequently, markeronlyrecurrent selection will be superiorto phenotypic selection as soon as thegenetic gain achieved through one cycleof MARS is, respectively, more than athird or a sixth of that achieved throughone cycle of phenotypic selection. Privatemaize breeding programmes have accessto off-season nurseries. Furthermore,they have often established efficientcontinuous nurseries where three to fourgenerations of maize can be grown peryear. The use of such nurseries allowsthem to carry MARS continuously, i.e.

132Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishwith up to four cycles per year, whereasphenotypic recurrent selection is limitedto one cycle per year at most. The impacton the rate of genetic gain of such animplementation of MARS might be verypositive even if MARS did not presentany advantage over phenotypic selectionon a per-cycle basis.• Cost of marker data points. Large privatecompanies have made considerable effortsto reduce both the cost of marker datapoints and the cost of experimental fieldplots. The ratio of cost of marker datapoint to cost of experimental field plot ismost likely lower in large private breedingprogrammes than in most public researchlaboratories or small private programmes,potentially leading to different views onthe economic efficiency of MARS.Marker-based and phenotypic selectioncan be mobilized in many different ways,with respect to each other, in markerassistedbreeding schemes. Marker andphenotypic information can be used eithersimultaneously or sequentially. Selectionof parents for breeding populations canbe made using marker information alone,phenotypic information alone, or a combinationof each. Selection of individuals in abackcross programme can be made on thesole basis of either marker or phenotypicinformation, or using both. Advancementof individuals in a line development programmecan also be made at each generationon the basis of either marker informationonly, phenotypic information only, or acombination of each. In order to maximizethe rate of genetic gain it is likely thatMAS breeding schemes such as MABCand MARS will involve generations ofmarker-only selection conducted at continuousnurseries. The advent of improvedmethods of producing doubled haploidswill certainly further influence the waymarker-based and phenotypic selection aremobilized with respect to each other.In spite of the development ofmarker-only selection and regardless ofthe underlying technology and breedingscheme, high-quality phenotyping remainsvital and without substitute at severalstages; but it may become more focused.Phenotypic evaluation remains the ultimatescreen before any cultivar is released.MAS-derived lines and hybrids that meetphenotypic requirements are selected forfurther evaluation and selection on the basisof their phenotypic value, while those thatdo not are discarded. Phenotypic evaluationis also critical to establish marker-traitassociations or perform the candidate genevalidations required to conduct MAS.Here, high quality phenotyping is necessary.Phenotyping protocols will thereforelikely be different from those commonlyused for phenotypic selection. Experimentsmay be conducted that involve side-by-sidecomparisons of different treatments such aswater stress or nitrogen fertilization levelsto dissect complex traits into their componentsand facilitate the elucidation of theirgenetic basis.Enhancements of such approachesto maize breeding will be based on theincorporation of improved methods ofproducing doubled haploid inbred lines,information from functional genomics andby learning how to incorporate favourablenative genetic variation systematically afterMAS has reduced the genetic variationin the original reference populations tounacceptable levels.Advantages and limitations ofMAS in maize breeding programmesAdvantages of MASFor private breeding programmes, MAShas offered several attractive features, most

Chapter 8 – Marker-assisted selection in maize 133of which are related to time and resourceallocations.MABC clearly provides the informationneeded to reduce the number of generationsof backcrossing, to combine (i.e.“stack”) transgenes, “native” genes or QTLinto one inbred or hybrid quickly, and tomaximize the recovery of the recurrentparent’s genome in the backcross-derivedprogeny. In several private breeding programmes,MABC has enabled the numberof backcrossing generations needed torecover 99 percent of the recurrent parentgenome to be reduced from six to three,reducing the time needed to develop a convertedvariety by one year (Crosbie et al.,2006; Ragot et al., 1995). As a line derivedby MABC can be made to be very similarto the original non-converted line, mostof its attributes, including agronomic performance,can be assumed to be equal orsimilar to those of the original line. Onlylimited phenotyping is therefore necessaryto verify these assumptions, comparedwith the extensive multiyear phenotypingrequired when backcrossing is conductedwithout markers. One or two years can besaved with MABC during post-conversionphenotyping when compared with conventionalbackcrossing, resulting in an overalltime advantage of MABC over conventionalbackcrossing of up to three years.In many situations, the greatest advantagesand profits are realized by those whoare first to the market with their products.Also, for reasons related to the practicesof seed production or legal aspects of cropregistration procedures, it may be quiteimportant to be able to produce nearisogenicversions of inbreds and hybrids;MABC provides such ability at a higherprobability.By contrast with MABC, SLS-MAS andMARS do not necessarily decrease the timeneeded to develop inbred lines. The useof MARS might actually increase it. Theadvantage of SLS-MAS and MARS residesin their ability to increase the rate of geneticgain (Eathington, 2005), which potentiallyresults in higher performing lines andhybrids than can be developed through phenotypicselection only. Both SLS-MAS andMARS increase the frequency of favourablealleles in the population of selectedindividuals. The difference between thetwo approaches is that SLS-MAS operateson few loci while MARS operateson many. When SLS-MAS or MARS areused, the effective size of the populationon which selection operates is increased,either directly for SLS-MAS or indirectlythrough several consecutive generations forMARS when compared with phenotypicpedigree selection. This increase in effectivepopulation size permits the applicationof a greater selection intensity and henceproduces a higher genetic gain. SLS-MASand MARS can also be seen as pre-selectionsteps if conducted prior to phenotypicselection and therefore improve the chancesof evaluating genotypes with a higher frequencyof favourable alleles phenotypicallybecause the truly undesirable portion of thepopulation may have been eliminated priorto phenotyping. Phenotypic selection cantherefore be conducted with higher selectionintensity than would be possible ifno pre-selection had taken place, resultingpotentially in additional genetic gain.Alternatively, the resources used forphenotyping can be allocated differentlybased on whether individuals have beenpre-selected or not with MAS. MASschemes for forward breeding should enablebreeding programmes to reallocate or focusresources for phenotypic evaluation in thetarget environment. For example, if DNAmarkers are linked to genes for resistance

134Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishto a disease or insect then it should bepossible initially to select resistant progenyin the absence of the disease or insectby using the DNA data at continuousnurseries. The selected progeny couldthen be evaluated using relatively moreexpensive bioassays with the pest(s) in thetarget environment. This shift in resourcesis inherent to MARS schemes for complextraits (Edwards and Johnson, 1994; Johnson,2004; Crosbie et al., 2006). By enrichingpopulations through rapid cycles of MARSat continuous nurseries, breeders shouldderive a higher frequency of progeny withfavourable alleles and haplotypes that arethen evaluated in the target environment.Without MARS, resources for evaluation inthe target environment would be diluted bythe inclusion of too many progeny with anundesirable genetic constitution.Concerns about reduced genetic diversityamong commercial maize hybrids anddepletion of genetic diversity in gene poolsused in breeding may be partially alleviatedby successful implementations ofMAS. MABC may revive interest in usingessentially untapped maize exotic germplasmas a source of favourable alleles forimprovement of elite varieties. Very smalland targeted chromosomal segments ofexotic origin can be introgressed into eliteinbred lines with limited risk of carryingalong undesirable characteristics. Suchan approach could be beneficial in maizealthough no accounts of its implementationhave been reported despite the manyyears as reports of its successful use intomato (Tanksley et al., 1996; Bernacchi etal., 1998a, b; Robert et al., 2001), rice (Xiaoet al., 1998), and soybean (Concibio et al.,2003). MARS, in turn, may also contributeto increasing genetic diversity among commercialmaize hybrids because, by focusingon selecting specific recombination events,it will result in the development of genuinelynew genomic rearrangements. AsQTL identified in any experiment representonly a fraction of the loci responsiblefor the phenotypes of complex traits, onecan assume that breeding programmes indifferent private companies will conductMARS based on their different geneticmodels and select for different genomicrearrangements. As a result, hybrids ofsimilar and high performance might bedeveloped that are based on different setsof favourable alleles at different loci, representingdistinct “genetic solutions” andcontributing to increased genetic diversityin farmers’ fields.An indirect but important advantage ofMAS and its underlying information andtechnology relates to intellectual property.Some maize breeding programmes havecreated a form of wealth through their collectionand knowledge of maize germplasm.Significant investments have been made inmaize breeding as exemplified by the billionsof United States dollars that wereused to purchase a few private programmesbetween 1995 and 2005. Protecting andmaximizing returns on such investmentshave always been important but are nowof greater concern. Information from MASshould be advantageous for addressingissues concerning ownership and derivationof germplasm, relatedness among germplasmand for the formation of some claimsin patents and similar documents.Perhaps one of the greatest advantagesof MAS is that, for the first time, maizebreeders have the means of learning someof the genetic details about germplasm andthe response to selection. Some maize programmesin the private sector have startedthis process (Niebur et al., 2004). As realfunctions become associated with the manycandidate genes and other DNA sequences,

Chapter 8 – Marker-assisted selection in maize 135the opportunities for learning about andunderstanding the response to selectionwill increase dramatically. It may then bepossible to ameliorate some of the limitationsof MAS and truly breed by design.Limitations of MASWhile not truly an inherent limitation ofthe methods involved, one unavoidablelimitation of MAS is the cost of assemblingand integrating the necessary infrastructureand personnel. These can be substantial andbeyond the means of many programmes.For such programmes, implementation ofMAS could lead to a delusional or unbalancedreallocation of resources from vitalactivities such as high-quality phenotypicevaluation and selection in the target environment.Currently, only the largest maizebreeding programmes in a given market orregion have the scale of sales and diversityof products that can justify and supportMAS and withstand some of the financialburdens of establishing and replacing componentsof the system (e.g. changes in themethods and platforms for detecting DNApolymorphisms).Some inherent limitations to MAS arerelated to the estimates of QTL positionand genetic effects and the rates of falsepositives and negatives. Confidence intervalsfor QTL are typically 10–15 cM; agenetic region that should not be a majorbarrier for implementing MAS althoughit could become a limitation to achievinggenetic gain by preventing the selection ofdesired recombination events. The adventof association mapping and a growing poolof candidate genes should provide someresources needed to minimize problemsrelated to the estimation of QTL position.The genetic effects of QTL are overestimatedfor many reasons, some of whichare linked to experimental designs forphenotyping or population developmentwhile others are inherent to the process ofQTL detection (Lee, 1995; Beavis, 1998;Melchinger, Utz and Schön, 1998; Holland,2004). In addition, genetic effects relatedto epistasis are either poorly estimated orignored by programmes in the private sector(Holland, 2001; Crosbie et al., 2006).Such assessments of genetic effects willinflate predictions of genetic gain. Therelative merit of MAS will depend on thenature of predictions, actual results andcosts of alternative methods.A possible limitation of MAS withmaize is the structure and content of variousgene pools. Examples of maize genepools would include European flint anddent germplasm, United States dents andvarious heterotic groups within each ofthese and other larger pools. Surveys withDNA markers have established differencesamong such groups of germplasm (Smithand Smith, 1992; Niebur et al., 2004). Therelatively allele-rich maize gene pools coupledwith genetic heterogeneity for manytraits will hinder the ability to extrapolateinformation about genotype-phenotyperelationships across gene pools. Such transferof information is expected to be moresuccessful in relatively homogeneous andless diverse maize gene pools (e.g. sweetcornor popcorn) and with self-pollinatedplant species (Lee, 1995). There have beenundocumented reports of a few alleles atQTL that have relatively universal geneticeffects across a relatively broad range ofmaize populations and target environments,but details of such genetic factorshave not been publicly disclosed (Crosbieet al., 2006). More resources will need to bedevoted to discovering where genetic informationcannot be easily extrapolated acrossgene pools or even populations within agene pool compared with situations where

136Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishit could. Although this should not impactthe economic efficiency of MABC or forwardbreeding, it could affect the overallcost efficiency of MAS.Finally, the efficacy for MAS in relativelycomplex populations such as syntheticsand open-pollinated varieties (OPVs) hasnot been investigated. Compared with thebi-allelic populations used in the privatesector, such populations are likely to havemore than two alleles at a given locus. Also,unlike the simple bi-allelic populations,allele frequency should be an importantcomponent of predictions with such populations.Therefore, there should be moregenetic effects and interactions to considerwhen making predictions based on MASwith OPVs and synthetics.In the future, successful implementationof MAS in maize may lead to more frequentproblems related to limited genetic variation.The emphasis of aggressive private sectormaize breeding programmes on crossesbetween elite, related inbred lines to createsegregating source populations has led toconcerns about the depletion of geneticdiversity in such gene pools and the abilityto enhance such gene pools with highquality genetic variation (Niebur et al.,2004). Such concerns, which existed priorto the deployment of DNA markers andMAS in maize, are likely to increase asMAS becomes more prevalent. If MAS inforward breeding schemes is as effective asreported, then alleles and haplotypes mayapproach fixation more rapidly (Crosbieet al., 2006). At that point, breedingprogrammes will need to repeat theprocess of calibrating genotype-phenotyperelationships in a slightly different arrayof reference populations to start the nextmetacycle of MAS (Johnson, 2004).There is much anticipation for the futureof MAS as genic sequences become themarker loci, functional information is discoveredfor the many candidate genes andgene products are assessed for their potentialas useful sources of information inbreeding programmes (Varshney, Granerand Sorrels, 2005; Lee, 2006). Certainly,these huge sets of raw data will contributeto progress. Eventually, other sources ofgenetic variation unrelated to the primaryDNA sequence such as DNA methylationwill be evaluated for their influence on genotype-phenotyperelationships. Currently,epigenetic variation is mostly ignored fromthat assessment although it is well knownthat much of the maize genome may bemethylated (Kaeppler, 2004) and may bemore dynamic than predicted by currentgenetic models and mechanisms (Fu andDooner, 2002). Also, the influences of noncodingsequences such as small interferingRNA (siRNA), matrix attachment regionsand long-distance regulatory sequenceshave yet to be considered for their effectson genetic variation and estimates of geneticvalues used in MAS (Lee, 2006).Most of the early limitations of MAS,due to the availability or cost of genotypicdata, have been overcome. However,the availability or cost of high-qualityphenotypic information is becomingone of the major limitations of MAS.During the past 20 years, developmentof new technologies and automation andminiaturization of laboratory procedureshave contributed to reducing the cost ofmarker data points as well as the timeneeded to produce them. Large-scalemarker laboratories produce marker datapoints at less than a tenth of the cost of20 years ago. By contrast, neither cost northe time required to produce phenotypicdata has changed much, if at all, in thesame timeframe. As the establishment ofmarker-trait associations and ultimately

Chapter 8 – Marker-assisted selection in maize 137the success of MAS depends on access tohigh-quality phenotypic data, means willhave to be found to decrease the cost ofphenotypic information while maintainingor increasing its quality. Alternatively,a greater proportion of budgets needsto be devoted to collecting phenotypicinformation.Achievements of maize breedingprogrammes with MASIn some important ways, maize breeding hasgradually changed since the mid 1990s withthe advent of genomics. Genetic principleswere always an important component ofmodern maize breeding and now geneticinformation of various types is seeping intobreeding schemes. MAS is the connectionbetween the growing pool of geneticinformation and actual plant breeding.Establishing and enhancing this connectionhave been important achievements.For the simplest breeding scenario,programmes in the private sector havedemonstrated that MABC is an effective androutine method to backcross one or moretransgenes into established elite inbred lines,the direct parents of commercial hybrids.Hybrids with effective combinationsof transgenes have been very successfulin the market. Consequently, MAS hasaccelerated the delivery of some productsto the market; an important achievement incompetitive economies.Programmes in the private sector have alsodemonstrated a sufficient degree of efficacyof MAS methods to secure protection ofintellectual property in patents. Methods,ideas and linkage relationships have beenincluded in claims of patents or patentapplications related to MAS (e.g. US5 492547 1996; US6 455 758B1 2002; US2005/0144664A1 2005; WO2005/000006A22005; WO2005/014858A2 2005), or theestablishment of marker-trait associations(e.g. US5 746 023P 1998; US6 368 806B12002; US6 399 855B1 2002). Given themagnitude of the investments made in maizebreeding by the private sector, receivingsuch a legal position may be a valuableachievement for the owner of the patent.The efficacy of MAS for forwardbreeding of complex traits has yet to befirmly established. Positive results fromcalibration studies have been reported,but although accounts of MAS-derivedcommercial varieties have been made(Eathington, 2005), the impact on actualbreeding and the development of newcommercial hybrids has not been disclosedto a significant extent (Johnson, 2004;Niebur et al., 2004; Crosbie et al., 2006). Atthis point in time, it is therefore too early tomake a definitive and databased assessmentof this aspect of MAS.The history and cost of the geneticgain achieved through MAS will certainlyvary among target environments. In someregions of the world, such as the centralUnited States, maize breeding achievedsteady genetic gains in grain yield forseveral consecutive decades prior tothe advent of MAS (Duvick, Smith andCooper, 2004). Nevertheless, the cost perunit gain has increased as more resourcesare needed for phenotypic evaluation inmore environments (Smith et al., 1999).However, the advent of applied genomicsand the discovery of many genes and genefunctions, coupled with MAS, could reducethe dependence on costly phenotypicinformation for breeding. In regions wherebiotic and abiotic stress factors are moreimportant than in the central Unites States,MAS may be very effective. Ultimately, thevalue and achievements of MAS will dependon the ecological and socio-economiccontext of the target environment.

138Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishCollaboration between theprivate and public sectors in MASand maize improvementThe increased investments in maizebreeding, expected returns on investmentand concerns regarding intellectual propertyby the private sector have made it moredifficult for corporations to collaboratewith external parties of any kind. Suchfactors hinder the exchange of informationand material that is common in collaborativeprojects. Nevertheless, around the world,the private and public sectors still manageto collaborate through various mechanismsand at different levels in the pursuit ofmaize improvement. Such collaborationinvolves interactions among multinationalcorporations, philanthropic foundations,national and subnational governments,universities and individuals. Majorcategories of collaboration include socialprogrammes and institutions, research anddevelopment, and education.In many regions of the world, privatesector maize breeding would not have grownwithout some critical social programmesand institutions. For example, legislationrelated to intellectual property, transferof capital and material, and regulatoryapproval of biotechnical innovations inmaize improvement have been importantcomponents of legal systems that haveencouraged financial investment in maizebreeding. The stability of these systemsand the rule of law have contributed tothe long-term gains in selection. Also,long-term crop subsidy programmes insome regions have provided an element ofsecurity for investments in maize researchand development by the private sector(Troyer, 2004; Crosbie et al., 2006). In thosesame regions, MAS has been deployedinitially and on the largest scale for maizebreeding.With respect to research and development,there is a long history of effectivecollaboration between the public and privatesectors in maize breeding. While suchinteraction continues in the era of MAS,the nature of the collaboration has changedwith the growth and development of thebreeding programmes in the private sector.Initially, collaboration was absolutelyvital for the private sector because breedingprogrammes in the public sector wereimportant, or the sole, sources of the inbredlines used directly by the private sector toproduce commercial hybrids or to sourcepopulations from which elite inbreds werederived. Also, the inbred lines from thepublic sector were usually provided on anunrestricted basis and without paymentsof royalties or licensing fees. Public breedingprogrammes continue to develop eliteinbred lines, occasionally in collaborationwith the private sector (e.g. the GermplasmEnhancement of Maize programme in theUnited States; Pollak, 2003). However, thedirect impact of contemporary public germplasmvaries greatly among regions andgradually, in many regions of the world,the private sector has become the primarysource of elite maize inbred lines and commercialhybrids.In addition to germplasm, most or allof the critical concepts, methods and basictechnologies have their origins in the publicsector (Niebur et al., 2004; Troyer, 2004;Crosbie et al., 2006). The private sector,with its unique ability to concentrate capitalthrough various mechanisms (e.g. profitsfrom products or licence fees, venture capitaland financial markets), is in the bestposition to allocate resources quickly toassess, modify and apply new developmentsin MAS and ancillary areas of maizeimprovement across large geographicaland political regions of a market zone. As

Chapter 8 – Marker-assisted selection in maize 139described in previous sections, cost-effectiveMAS requires several components of anintegrated infrastructure, some features ofwhich have had a relatively high rate of renovationand replacement (e.g. methods ofdetecting DNA polymorphism), and thereforerequired substantial financial resources.Competing corporations and the potentialfor profit provide the necessary motivationfor such investments (Troyer, 2004; Crosbieet al., 2006). To the authors’ knowledge,such financial mechanisms either do notexist or are limited in the public sector.Collaboration between the public andprivate sectors in MAS for maize may bestrongest in basic genetics and genomeannotation. In order for MAS to reachits full potential, it may be necessary toacquire a much better understandingof gene function and products. For anyplant species, only a small percentage ofgenes and other DNA sequences have afunction defined through direct experimentation(Lee, 2006). Discoveries in plantgene function will occur in many laboratoriesaround the world and, ultimately, thedevelopment groups in the private sectorwill have the necessary concentration ofresources and sense of purpose to assemblethe relatively raw basic information intotools and products from MAS. The maizenuclear genome, with tens of thousandsof genes and many other important DNAsequences, is mostly a “black box” withrespect to understanding the role of thesein mediating phenotypes in response toenvironmental cues. Such understanding, apotential key to MAS and maize improvement,can only occur through informal andformal collaboration between the publicand private sectors investigating a broadarray of plant species.Examples of collaborative researchbetween the public and private sectorsrelevant to MAS in maize include attemptsto select for hybrid yield (Stromberg,Dudley and Rufener, 1994), QTLmapping and selecting of hybrid yield(Eathington, Dudley and Rufener, 1997)and grain quality (Laurie et al., 2004), thedevelopment of the IBM population ofrecombinant inbred lines, and mappinggenomic regions that include the vgt1locus in maize (Lee et al., 2002; Salvi etal., 2002). National collaborative researchprogrammes such as Génoplante in Franceand GABI in Germany, as well as severalprojects within the European Commissionsponsoredframework programmes, areadditional examples of such collaboration.Certainly, other collaborative projectsbetween the public and private sectorshave been conducted in maize MAS buttheir proprietary nature prevents publicdisclosure.Future collaborative research activitiesin maize MAS could assume many forms.In most regions of the world, the privatesector has the obvious superiority in termsof infrastructure needed for genotyping,phenotyping and data analysis. Theseresources are mostly devoted to the directpursuit of products and profits. That pursuitmay also be the greatest disadvantage of theprivate sector because such a focus limitsthe attention devoted to many interestingyet seemingly ancillary observations ofgenotype-phenotype relations in MAS.Some components of that infrastructurecould possibly be made accessible to thepublic sector as “in-kind” contributionsto collaborative or service-related projectsin regions that are unlikely to emerge asimportant markets for the private sector orfor phenotypes and germplasm that are notof direct interest to the private sector.Education and training are alsoimportant areas in which the public and

140Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishprivate sectors should collaborate. With theadvent of MAS, there has been an obviousneed for maize breeders in the privatesector to become familiar with all aspectsof the process, and the public sector hasdeveloped several new short courses andtraining sessions in MAS-related concepts(Niebur et al., 2004; Crosbie et al., 2006).Such knowledge is now considered astandard component of recent graduatetraining. However, while new students mayhave an adequate grasp of the theoreticalaspects of MAS, their lack of exposure tothe private sector’s advanced infrastructurerepresents a gap in their education. Thissituation is similar to that of students witha new degree in engineering who joinadvanced engineering and design groups inother industries: the private sector’s capacityto concentrate and focus capital often leadsto advanced infrastructure that does notexist in the public sector. In such situations,new students have to navigate a rathersteep learning curve before they becomeproductive members of their new group.To reduce the slope of the learning curve,the private sector could provide internshipsto graduate students or to professors whoteach plant breeding courses. It is unlikelythat the public sector will have the resourcesto duplicate or exceed some features of theinfrastructure that has been developed formaize MAS in the private sector. Therefore,for some aspects of education, it will be toeveryone’s benefit to find ways to worktogether.private sector perspectives onMAS for maize improvementThe development of molecular markersin the 1980s provided the first tools todissect the genetic basis of traits andselect individuals based on their predictedgenetic value. Back in these early days, theavailability of genetic information was alimiting factor. Today’s landscape is verydifferent as advances in applied genomicsand laboratory technology have providedthe tools to generate genetic information forall traits of interest. Gene similarities andsynteny across genomes mean that muchof the information generated on any plantspecies has relevance to other plant species.The speed at which genetic informationbecomes available never ceases to increase.Rather than its availability, it is the abilityto handle and utilize genetic informationthat is becoming the limiting factor forMAS. New and improved informationtechnology and bioinformatics capabilitiestherefore need to be developed that connectthe growing wealth of genetic informationwith maize breeding programmes whereknowledge about the genetic basis of traitsand allelic variation at these loci is translatedinto varieties.QTL and gene mapping will remainkey for the generation and use of geneticinformation. As sequencing of cerealgenomes including maize progresses,physical mapping of cloned genes willbecome a powerful alternative to statisticalapproaches. Characterization of allelicdiversity at loci of interest can proceedfrom analyses of bi-parental populations orassociation studies. An effective alternativeis the use of sets of NILs, or introgressionline (IL) libraries (Peleman and van derVoort, 2003). As NILs developed arounda specific locus differ only by the alleleat this locus, and because most traits ofagronomic interest in maize are quantitative,phenotypic differences among such NILs areexpected to be rather small. High precisionphenotyping will not only be required butwill be critical for the evaluation of suchmaterial (Peleman and van der Voort, 2003).Private corporations have realized the need

Chapter 8 – Marker-assisted selection in maize 141for such high precision phenotyping ascan be seen from their active recruiting oftrait-specific phenotyping scientists oftenlocated in targeted areas where the trait ofinterest can be more easily measured (e.g.positions dedicated to drought toleranceand located in arid regions of the world).In order to further the implementationof MAS in breeding, increased numbersof marker data points will be required.Private corporations have established orare developing the capacity to producehundreds of millions of data points peryear in service laboratories, distinct fromresearch units. Besides, smaller “biotech”companies are developing technologies thatcould reduce the cost of each marker datapoint to a mere few United States cents.Moving to marker systems that are notbased on gels is permitting the automation ofmost laboratory steps. Data points are beingproduced around the clock with laboratorytechnicians working in shifts. Here again,private companies are actively recruitinghighly qualified technology specialists aswell as laboratory managers whose role ismore to optimize the running of productionplants than dwell on the science. Beyondlaboratories, plant handling is becoming abottleneck to high-throughput protocols.High-throughput facilities have to beestablished and equipped at continuousnursery sites potentially to handle millionsof plants per year.There is little doubt that the largestprivate maize breeding programmes areinvesting very heavily in the implementationof MAS. Unless regulatory issues changedramatically, MABC will remain thepreferred means of delivering transgenesto the market. Faster MABC protocols willalways represent a potential commercialadvantage in an area where competition isfierce and a one-year advantage may meanmuch on the market. Most recent investmentshave been directed at implementing MARSin breeding. The size of the investmentin this approach seems to suggest thatprivate corporations have more insight intoits benefits compared with conventionalbreeding than has been reported publicly.Genotype-driven breeding should alsoallow faster development of specializedvarieties as the maize market becomes moreand more fragmented based on end-use ofthe harvest: animal feed (silage or grain),ethanol, dry or wet milling. Favourablealleles for traits of interest are likely tobe spread across more than two linestherefore requiring the assembly of allelesfrom many different sources in a singleinbred line. Proposals have been made toachieve such goals (Peleman and van derVoort 2003), although software tools todetermine the optimal breeding schemesare not yet available to generate these“ideal” genotypes.Maize breeding is likely to changemore in the coming 10 or 20 years thanit has over the past 50. Developing newhybrids efficiently now requires integratingdata from many sources, sometimesbeyond maize, generating high-qualitygenotypic and phenotypic data neededfor the construction of “ideal” genotypes,and finally selecting phenotypically thebest individuals from populations ofmarker-assisted-derived materials. Manystakeholders beyond maize breeders nowtake an active part in the development ofnew varieties and therefore breeding willincreasingly become the responsibility ofgroups of individuals with complementaryskills than stand-alone breeders. Trainingof all to understand and challenge thecontribution of others will be critical tooperating multidisciplinary breeding teamsefficiently.

142Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishMAS for maize improvement indeveloping countriesA rapid analysis of the implementationof MAS in private maize breeding programmespoints to three elements as beingof particular importance: availability ofhigh-quality phenotypic data, access tolow-cost molecular marker data points andaccess to reliable continuous nurseries. Theimportance of high-quality phenotypicanalyses has been clearly recognized bygroups in the private sector (Niebur et al.,2004; Crosbie et al., 2006). Implementationof MAS in maize breeding requires largeamounts of marker data points to be generated.Private groups have spent mucheffort developing technologies and platformsto achieve cost-efficient genotyping.Simultaneously, highly efficient continuousnurseries have been established in tropicalenvironments or local greenhouses.By contrast, maize breeding for developingcountries is rather fragmented.National agricultural research institutionsand international centres of the CGIARsuch as the International Maize and WheatImprovement Center (CIMMYT) focusmuch of their efforts on poor farmers andunderserved regions. Private maize breedingprogrammes are also established ina number of developing countries. Dueto the large up-front costs of assemblinginfrastructure and personnel for genotyping,it is unlikely that individual nationalmarker laboratories could produce datapoints in a cost-efficient manner. However,regional facilities serving the needs of severalnational programmes and supportedby local laboratories that could processsamples (processing samples could be aseasy as taking and air-drying them) andprovide information in a timely manner,would probably be very sustainable alternatives.Such a regional molecular servicelaboratory has been established recentlyin Nairobi, Kenya, in a joint effort bytwo CGIAR centres, CIMMYT and theInternational Livestock Research Institute(ILRI) and Kenya’s Agricultural ResearchInstitute (KARI), under the CanadianInternational Development Agency(CIDA)-funded Biosciences eastern andcentral Africa (BecA) platform, to providetechnical access and training for Africanmaize breeders (Delmer, 2005). Such afacility could be an excellent componentof a comprehensive maize breeding effortif it is possible to establish and maintainhigh-quality personnel and facilities for allof the other aspects of maize breeding inkey target environments. However, withouthigh-quality capabilities in phenotypicevaluation and selection, molecular laboratorieswill be worthless. Research projectsinvolving large-scale (transnational) phenotypicevaluations of key genetic materialand focused on specific traits (tolerance tobiotic or abiotic stresses) should providegenetic information that is both locallyrelevant and broadly applicable (geographicallyand in terms of germplasm). Suchprojects would also spread the cost of phenotypingacross all participants but wouldonly be successful with effective transnationalcoordination.Private companies running MAS inmaize could contribute to its implementationin developing countries in severalways. First, they could make some of theirgenetic information available, thereby addingto that already available in the publicdomain. Much information is being generatedin the private sector on traits ofimportance to developing countries such asdisease resistance (e.g. grey leaf spot, northerncorn leaf blight, Fusarium stalk and earrots), drought tolerance and nitrogen useefficiency. After validation of its relevance

Chapter 8 – Marker-assisted selection in maize 143to the germplasm and environments of targetareas, this genetic information could beused to select efficiently for specific traitsthrough MAS. Second, private companiescould provide access to some of their genotypingor nursery platforms. Genotypingsamples for MAS projects in developingcountries would not substantially disruptprivate companies’ own research if conductedin periods of lower activity, andwould provide these MAS projects withmarker data points for as low a cost as possible.Third and probably most critically,private companies could train scientistsfrom developing countries on the principles,mechanics and logistics of applyingand implementing MAS in maize. Scientistsin private maize breeding groups havealready identified many of the pitfalls andovercome many of the hurdles linked withthe implementation of MAS. Transfer ofthis knowledge to scientists from developingcountries would help them immenselyto design marker-based breeding schemesadapted to their sets of constraints.Beyond their contribution to the implementationof MAS in maize in developingcountries, private companies could, in verysimilar ways, contribute to MAS programmesin other species of importance todeveloping countries but remote from theircore interests. Synteny and gene conservationacross species should allow some of themaize genetic information to be transferableto other species. Technology platforms andbreeding approaches developed for MASin maize should be good models for othercrops and some might be directly usable.Mechanisms or organizations need to beput in place for these transfers of knowledgeand technologies to occur from privatemaize MAS programmes to other crops indeveloping countries. Private programmeswill likely not drive these transfers butmight be very willing to contribute or bedirectly involved in specific projects providedadequate frameworks exist.Public–private partnerships will needto be established to manage intellectualproperty issues related to the transfers ofinformation, material or technologies fromprivate companies to developing countries(Naylor et al., 2004). The AfricanAgricultural Technology Foundation(AATF) is one initiative that has beenestablished to deal with such issues. Severalprivate corporations with major investmentsin MAS in maize have agreed toprovide access to germplasm and knowledgefor African countries (Naylor et al.,2004; Delmer, 2005).As with the private sector in Europeand North America, it will be necessaryto provide regular and easy access to educationand training in maize MAS as thephenotypes and population structures arelikely to differ from those encountered byprogrammes in the private sector in relativelyhigh-input production environments.Also, and in common with the changes inthe private sector, some reorganization orrestructuring of public sector programmesmay be warranted with the advent of morespecialized roles for some personnel.Understanding the genetic basis of traitsand cloning and sequencing the underlyinggenes will not have an impact on poorfarmers unless translated into varietiesthrough breeding. Implementing MASrequires significant investments in bothpeople and infrastructures. Some of themost promising marker-based breedingschemes (e.g. MARS), take about as longas conventional breeding schemes todevelop improved varieties and thereforerequire long-term funding commitments.Funding of practical crop improvementhas declined for several years, particularly

144Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishin the international public sector (Knight,2003), and as a result investments havefavoured research at the expense ofpractical applications (Naylor et al., 2004).Whether current funding mechanismsbased on short-term (two to five years)grants are adequate to allow maize or anyother breeding programmes in developingcountries to benefit from the much neededadvantages of MAS is questionable.ReferencesBasten, C.J., Weir, B.S. & Zeng, Z-B. 1994. Zmap--a QTL cartographer. In C. Smith, J.S. Gavora, B.Benkel, J. Chesnais, W. Fairfull, J.P. Gibson, B.W. Kennedy & E.B. Burnside, eds. Proc. 5th Wrld.Congr. Genet. Appl. Livest. Prodn., Vol. 22, pp. 65–66. Guelph, Ontario, Canada, University ofGuelph.Beavis, W.D. 1994. The power and deceit of QTL experiments: lessons from comparative QTLstudies. Proc. 49 th Ann. Corn and Sorghum Ind. Res. Conf., pp. 250–266. Washington, DC, USA,American Seed Trade Association.Beavis, W.D. 1998. QTL analyses: power, precision and accuracy. In H. Paterson, ed. Molecular dissectionof complex traits, pp. 143–162. Boca Raton, FL, USA, CRC Press.Beavis, W.D. 1999. Integrating the central dogma with plant breeding. Abstr. Plant & Anim. Genome7: 28.Bernacchi, D., Beck-Bunn, T., Emmatty, D., Eshed, Y., Inai, S., Lopez, J., Petiard, V., Sayama,H., Uhlig, J., Zamir, D. & Tanksley, S. 1998a. Advanced backcross QTL analysis of tomato. II.Evaluation of near-isogenic lines carrying single-donor introgressions for desirable wild QTL allelesderived from Lycopersicon hirsutum and L. pimpinellifolium. Theor. Appl. Genet. 97: 170–180.Bernacchi, D., Beck-Bunn, T., Eshed, Y., Lopez, J., Petiard, V., Uhlig, J., Zamir, D. & Tanksley, S.1998b. Advanced backcross QTL analysis in tomato. I. Identification of QTLs for traits of agronomicimportance from Lycopersicon hirsutum. Theor. Appl. Genet. 97: 381–397.Bernardo, R. 1994. Prediction of maize single-cross performance using RFLPs and information fromrelated hybrids. Crop Sci. 34: 20–25.Bernardo, R. 2001. What if we knew all the genes for a quantitative trait in hybrid crops ? Crop Sci.41: 1–4.Bernardo, R. 2002. Breeding for quantitative traits in plants. Woodbury, MN, USA, Stemma Press.Botstein, D., White, R.L., Skolnick, M. & Davis, R.W. 1980. Construction of genetic linkage map inhuman using restriction fragment length polymorphisms. Am. J. Human Genet. 32: 314–331.Bouchez, A., Hospital, F., Causse, M., Gallais, A. & Charcosset, A. 2002. Marker-assisted introgressionof favorable alleles at quantitative trait loci between maize elite lines. Genetics 162: 1945–1959.Burr, B., Burr, F.A., Thompson, K.H., Albertsen, M.C. & Stuber, C.W. 1988. Gene mapping withrecombinant inbreds in maize. Genetics 118: 519–526.Castiglioni, P., Ajmone-Marsan, P., van Wijk, R. & Motto, M. 1999. AFLP markers in a molecular linkagemap of maize: codominant scoring and linkage group distribution. Theor. Appl. Genet. 99: 425–431.Charmet, G., Robert, N., Perretant, M.R., Gay, G., Sourdille, P., Groos, C., Bernard, S. & Bernard,M. 1999. Marker-assisted recurrent selection for cumulating additive and interactive QTL’s inrecombinant inbred lines. Theor. Appl. Genet. 99: 1143–1148.Ching, A., Caldwell, K.S., Jung, M., Dolan, M., Smith, O.S., Tingey, S., Morgante, M. & Rafalski,A. 2002. SNP frequency, haplotype structure and linkage disequilibrium in elite maize inbred lines.BMC Genetics 3: 19.

Chapter 8 – Marker-assisted selection in maize 145Coe, E., Hancock, D., Kowalewski, S. & Polacco, M. 1995. Gene list and working maps. MaizeGenet. Coop. Newsl. 69: 191–267.Concibio, V.C., La Vallee, B., McLaird, P., Pineda, N., Meyer, J., Hummer, L., Yang, J., Wu, K. &Delannay, X. 2003. Introgression of a quantitative trait locus for yield from Glycine soja into commercialsoybean cultivars. Theor. Appl. Genet. 106: 575–582.Crosbie, T.M., Eathington, S.R., Johnson, G.R., Edwards, M., Reiter, R., Stark, S., Mohanty, R.G.,Oyervides, M., Buehler, R.E., Walker, A.K., Dobert, R., Delannay, X., Pershing, J.C., Hall, M.A.& Lamkey, K.R. 2006. Plant breeding: past, present, and future. In K.R. Lamkey & M. Lee, eds.Plant Breeding: The Arnel R. Hallauer Internat. Symp. Ames, IA, USA., Blackwell Publishing.Davis, G.L., McMullen, M.D., Baysdorfer, C., Musket, T., Grant, D., Staebell, M., Xu, G., Polacco,M., Koster, L., Melia-Hancock, S., Houchins, K., Chao, S. & Coe Jr., E.H. 1999. A maize mapstandard with sequenced core markers, grass genome reference points and 932 expressed sequencetagged sites (ESTs) in a 1736-locus map. Genetics 152: 1137–1172.Delmer, D.P. 2005. Agriculture in the developing world: connecting innovations in plant research todownstream applications. Proc. Natl. Acad. Sci. USA 102: 15739–15746.Duvick, D.N., Smith, J.S.C. & Cooper, M. 2004. Long-term selection in a commercial hybrid maizebreeding program. In J. Janick, ed. Plant Breeding Rev. 24: 109–152.Eathington, S.R. 2005. Practical applications of molecular technology in the development of commercialmaize hybrids. Proc. 60 th Ann. Corn and Sorghum Seed Res. Conf. Washington, D.C.,USA, American Seed Trade Association.Eathington, S.R., Dudley, J.W. & Rufener, G.K. 1997. Usefulness of marker-QTL associations inearly generation selection. Crop Sci. 37: 1686–1693.Edwards, M. & Johnson, L. 1994. RFLPs for rapid recurrent selection. Proc. Symp. Analysis ofMolecular Marker Data, pp. 33–40. Corvallis, OR, USA, American Society of HorticulturalScience and Crop Science Society of America.Edwards, M.D. & Page, N.J. 1994. Evaluation of marker-assisted selection through computer simulation.Theor. Appl. Genet. 88: 376–382.Fehr, W.R. 1987. Principles of cultivar development. New York, USA, Macmillan.Frisch, M. & Melchinger, A.E. 2001a. Marker-assisted backcrossing for introgression of a recessivegene. Crop Sci. 41: 1485–1494.Frisch, M. & Melchinger, A.E. 2001b. Marker-assisted backcrossing for simultaneous introgressionof two genes. Crop Sci. 41: 1716–1725.Frisch, M., Bohn, M. & Melchinger, A.E. 1999a. Minimal sample size and optimal positioning of flankingmarkers in marker-assisted backcrossing for transfer of a target gene. Crop Sci. 39: 967–975.Frisch, M., Bohn, M. & Melchinger, A.E. 1999b.Comparison of selection strategies for markerassistedbackcrossing of a gene. Crop Sci. 39: 1295–1301.Fu, H. & Dooner, H.K. 2002. Intraspecific violation of genetic co-linearity and its implications inmaize. Proc. Natl. Acad. Sci. USA 99: 9573–9578.Gallais, A., Dillmann, C. & Hospital, F. 1997. An analytical approach of marker assisted selectionwith selection on markers only. In R. Krajewski & Z. Kaczmarek, eds. Advances in biometricalgenetics. Proc.10 th . Mtg. EUCARPIA Section Biometrics in Plant Breeding, pp. 111–116. Poznan,Institute of Plant Genetics, Polish Academy of Sciences.Gelman, A., Carlin, J.B., Stern, H.S. & Rubin, D.B. 2004. Bayesian data analysis, 2nd edition. BocaRaton, FL, USA, Chapman and Hall/CRC.

146Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishGimelfarb, A. & Lande, R. 1994. Simulation of marker-assisted selection in hybrid populations.Genet. Res. 63: 39–47.Goodman, M.M. & Carson, M.L. 2000. Reality vs. myth: corn breeding, exotics, and genetic engineering.Proc. 55 th Ann. Corn and Sorghum Res. Conf., pp. 149–172. Washington, D.C., USA,American Seed Trade Association.Goodman, M.M., Stuber, C.W., Newton, K. & Weissinger, H.H. 1980. Linkage relationships of 19enzyme loci in maize. Genetics 96: 697–710.Hallauer, A.R. & Miranda, J.B. 1981. Quantitative genetics in maize breeding. Ames, IA, USA, IowaState University Press.Helentjaris, T., Slocum, M., Wright, S., Schasfer, A. & Nienhuis, J. 1986. Construction of geneticlinkage maps in maize and tomato using restriction fragment length polymorphisms. Theor. Appl.Genet. 72: 761–769.Ho, J.C., McCouch, S.R. & Smith, M.E. 2002. Improvement of hybrid yield by advanced backcrossQTL analysis in elite maize. Theor. Appl. Genet. 105: 440–448.Hoisington, D. 1989. Working linkage maps. Maize Genet. Coop. Newsl. 63: 141–151.Holland, J.B. 2001. Epistasis and plant breeding. In J. Janick, ed. Plant Breeding Rev. 21: 29–32.Holland, J.B. 2004. Implementation of molecular markers for quantitative traits in breeding programs– challenges and opportunities. In T. Fischer, N. Turner, J. Angus, L. McIntyre, M.Robertson, A. Borrell & D. Lloyd, eds. New directions for a diverse planet. Proc. 4 th Internat. CropSci. Congr., Brisbane, Australia (available at www.cropscience.org.au/icsc2004).Hospital, F. 2001. Size of donor chromosome segments around introgressed loci and reduction oflinkage drag in marker-assisted backcross programs. Genetics 158: 1363–1379.Hospital, F. & Charcosset, A. 1997. Marker-assisted introgression of quantitative trait loci. Genetics147: 1469–1485.Hospital, F., Chevalet, C. & Mulsant, P. 1992. Using markers in gene introgression breeding programs.Genetics 132: 1199–1210.Jansen, R.C. 1993. Interval mapping of multiple quantitative trait loci by using molecular markers.Theor. Appl. Genet. 85: 252–260.Jansen, R.C., Jannink, J-L. & Beavis, W.D. 2003. Mapping quantitative trait loci in plant breedingpopulations: use of parental haplotype sharing. Crop Sci. 43: 829–834.Johnson, G.R. 2004. Marker-assisted selection. In J. Janick, ed. Plant Breeding Rev. 24: 293–310.Jones, C.J., Edwards, K.J., Castaglione, S., Winfield, M.O., Sala, F., van de Wiel, C., Bredemeijer,G., Vosman, B., Matthes, M., Daly, A., Brettschneider, R., Bettini, P., Buiatti, M., Maestri, E.,Malcevschi, A., Marmiroli, N., Aert, R., Volkaert, G., Rueda, J., Linacero, R., Vazquez, A. &Karp, A. 1997. Reproducibility testing of RAPD, AFLP and SSR markers in plants by a networkof European laboratories. Mol. Breed. 3: 381–390.Jourjon, M-F., Jasson, S., Marcel, J., Ngom, B. & Mangin, B. 2005. MCQTL: multi-allelic QTLmapping in multi-cross design. Bioinformatics 21: 128–130.Kao, C-H., Zeng, Z-B. & Teasdale, R. 1999. Multiple interval mapping for quantitative trait loci.Genetics 152: 1203–1216.Kaeppler, S. 2004. Biotechnology: new horizons. In C.W. Smith, J. Betrán, & E.C.A. Runge, eds. Corn:origin, history, technology, and production, pp. 399–425. Hoboken, NJ, USA, John Wiley & Sons.Knapp, S.J. 1998. Marker-assisted selection as a strategy for increasing the probability of selectingsuperior genotypes. Crop Sci. 38: 1164–1174.

Chapter 8 – Marker-assisted selection in maize 147Knight, J. 2003. Crop improvement: a dying breed. Nature 421: 568–570.Lande, R. & Thompson, R. 1990. Efficiency of marker-assisted selection in the improvement ofquantitative traits. Genetics 124: 743–756.Lander, E.S. & Botstein, D. 1989. Mapping Mendelian factors underlying quantitative traits usingRFLP linkage maps. Genetics 121: 185–199.Laurie, C.C., Chasalow, S.D., LeDeaux, J.R., McCarroll, R., Bush, D., Hauge, B., Lai, C., Clark,D., Rocheford, T.R. &. Dudley, J.W. 2004. The genetic architecture of response to long-term artificialselection for oil concentration in the maize kernel. Genetics 168: 2141–2155.Lee, M. 1995. DNA markers in plant breeding programs. Adv. Agron. 55: 265–344.Lee, M. 2006. The phenotypic and genotypic eras of plant breeding. In K.R. Lamkey & M. Lee, eds.Plant Breeding: The Arnel R. Hallauer Internat. Symp. Ames, IA, USA, Blackwell Publishing.Lee, M., Sharopova, N., Beavis, W.D., Grant, D., Katt, M., Blair, D. & Hallauer, A.R. 2002.Expanding the genetic map of maize with the intermated B73 x Mo17 (IBM) population. PlantMol. Biol. 48: 453–461.Lindblad-Toh, K., Winchester, E., Daly, M., Wang, D., Hirschhorn, J.N., Laviolette, J.P., Ardie, K.,Reich, D.E., Robinson, E., Sklar, P., Shah, N., Thomas, D., Fan, J.B., Gingeras, T., Warrington,J., Patil, N., Hudson, T.J. & Lander, E.S. 2000. Large-scale discovery and genotyping of singlenucleotidepolymorphisms in the mouse. Nature Genet. 24: 381–386.Manly, K.F. & Olson, J.M. 1999. Overview of QTL mapping software and introduction to MapManager QT. Mamm. Genome 10: 327–334.Mazur, B.J. 1995. Commercializing the products of plant biotechnology. Trends Biotech. 13:319–323.Melchinger, A.E., Utz, H.F. & Schön, C.C. 1998. Quantitative trait locus (QTL) mapping using differenttesters and independent population samples in maize reveals low power of QTL detectionand large bias in estimates of QTL effects. Genetics 149: 383–403.Mohan, M., Nair, S., Krishna, T.G., Yano, M., Bhatia, C.R. & Sasaki, T. 1997. Genome mapping,molecular markers and marker-assisted selection in crop plants. Mol. Breed. 3: 87–103.Moreau, L., Charcosset, A. & Gallais, A. 2004. Experimental evaluation of several cycles of markerassistedselection in maize. Euphytica 137: 111–118.Moreau, L., Charcosset, A., Hospital, F. & Gallais, A. 1998. Marker-assisted selection efficiency inpopulations of finite size. Genetics 148: 1353–1365.Moreau, L., Lemarié, S., Charcosset, A. & Gallais, A. 2000. Economic efficiency of one cycle ofmarker-assisted selection. Crop Sci. 40: 329–337.Morris, M., Dreher, K., Ribaut, J-M. & Khairallah, M. 2003. Money matters (II): costs of maizeinbred line conversion schemes at CIMMYT using conventional and marker-assisted selection.Mol. Breed. 11: 235–247.Naylor, R.L., Falcon, W.P., Goodman, R.M., Jahn, M.M., Sengooba, T., Tefera, H. & Nelson, R.J.2004. Biotechnology in the developing world: a case for increased investments in orphan crops.Food Policy 29: 15–44.Niebur, W.S., Rafalski, J.A., Smith, O.S. & Cooper, M. 2004. Applications of genomics technologiesto enhance rate of genetic progress for yield of maize within a commercial breeding program.In T. Fischer, N. Turner, J. Angus, L. McIntyre, M. Robertson, A. Borrell & D. Lloyd, eds. Newdirections for a diverse planet. Proc. 4 th Internat. Crop Sci. Congr., Brisbane, Australia. (available atwww.cropscience.org.au/icsc2004).

148Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishOpenshaw, S. & Frascaroli, E. 1997. QTL detection and marker-assisted selection for complex traitsin maize. Proc. 52 nd Ann. Corn and Sorghum Res. Conf., pp. 44–53. Washington, DC, AmericanSeed Trade Association.Parisseaux, B. & Bernardo, R. 2004. In silico mapping of quantitative trait loci in maize. Theor. Appl.Genet. 109: 508–514.Peleman, J.D. & van der Voort, J.R. 2003. Breeding by design. Trends Plant Sci. 7: 330–334.Penner, G.A., Bush, A., Wise, R., Kim, W., Domier L., Kasha, K., Laroche, A., Scoles, G., Molner,S.J. & Fedak, G. 1993. Reproducibility of random amplified polymorphic DNA (RAPD) analysisamong laboratories. PCR Meth. Appl. 2: 341–345.Pollak, L.M. 2003. The history and success of the public-private project on germplasm enhancementof maize (GEM). Adv. Agron. 78: 45–87.Rafalski, A. 2002. Applications of single nucleotide polymorphisms in crop genetics. Curr. Opin.Plant Biol. 5: 94–100.Ragot, M. & Hoisington, D.A. 1993. Molecular markers for plant breeding: comparisons of RFLPand RAPD genotyping costs. Theor. Appl. Genet. 86: 975–984.Ragot M., Biasiolli, M., Delbut, M.F., Dell’Orco, A., Malgarini L., Thevenin, P., Vernoy, J., Vivant,J., Zimmermann, R. & Gay, G. 1995. Marker-assisted backcrossing: a practical example. In A.Bervillé & M. Tersac, eds. Les Colloques, No. 72, Techniques et utilisations des marqueurs moléculaires,pp. 45–56. Paris, France, INRA.Ragot, M., Gay, G., Muller, J-P. & Durovray, J. 2000. Efficient selection for the adaptation to theenvironment through QTL mapping and manipulation in maize. In J-M. Ribaut & D. Poland, eds.Molecular approaches for the genetic improvement of cereals for stable production in water-limitedenvironments, pp. 128–130. México, D.F., CIMMYT.Ribaut, J-M. & Betrán, J. 1999. Single large-scale marker-assisted selection (SLS-MAS). Mol. Breed.5: 531–541.Ribaut, J-M., Jiang, C. & Hoisington, D. 2002. Efficiency of a gene introgression experiment bybackcrossing. Crop Sci. 42: 557–565.Ribaut, J-M., Edmeades, G., Perotti, E. & Hoisington, D. 2000. QTL analyses, MAS results, andperspectives for drought-tolerance improvement in tropical maize. In J-M. Ribaut & D. Poland,eds. Molecular approaches for the genetic improvement of cereals for stable production in water-limitedenvironments, pp. 131–136. México, D.F., CIMMYT.Robert, V.J.M., West, M.A.L., Inai, S., Caines, A., Arntzen, L., Smith, J.K. & St-Clair, D.A. 2001.Marker-assisted introgression of blackmold resistance QTL alleles from wild Lycopersicon chesmaniito cultivated tomato (L. esculentum) and evaluation of QTL phenotypic effects. Mol. Breed. 8:217–233.Saiki, R.K., Gelfand, D.H., Stoffel, S., Scharf, S., Higuchi, R.H., Horn, G.T., Mullis, K.B. & Erlich,H.A. 1988. Primer-directed enzymatic amplification of DNA with a thermostable DNA polymerase.Science 239: 487–491.Salvi, S., Tuberosa, R., Chiapparino, E., Maccaferri, M., Veillet, S., van Beuningen, L., Isaac, P.,Edwards, K. & Phillips, R.L. 2002. Towards positional cloning of Vgt1, a QTL controlling thetransition from the vegetative to the reproductive phase in maize. Plant Mol. Biol. 48: 601–613.Sax, K. 1923. The association of size differences with seed coat pattern and pigmentation in Phaseolusvulgaris. Genetics 8: 552–560.Senior, M.L. & Heun, M. 1993. Mapping maize microsatellites and polymerase chain reaction confir-

Chapter 8 – Marker-assisted selection in maize 149mation of the target repeats using a CT primer. Genome 36: 884–889.Senior, M.L., Chin, E.C.L., Lee, M., Smith, J.S.C. & Stuber, C.W. 1996. Simple sequence repeatmarkers developed from maize sequences found in the GENEBANK database: Map construction.Crop Sci. 36: 1676–1683.Servin, B. & Hospital, F. 2002. Optimal positioning of markers to control genetic background inmarker-assisted backcrossing. J. Hered. 93: 214–217.Shattuck-Eidens, D.M., Bell, R.N., Neuhausen, S. & Helentjaris, T. 1990. DNA sequence variationwithin maize and melon: Observations from polymerase chain reaction and direct sequencing.Genetics 126: 207–217.Skroch, P. & Nienhuis, J. 1995. Impact of scoring error and reproducibility of RAPD data on RAPDbased estimates of genetic distance. Theor. Appl. Genet. 91: 1086–1091.Smith, J.S.C. &. Smith, O.S. 1992. Fingerprinting crop varieties. Adv. Agron. 47: 85–140.Smith, O.S., Hoard, K., Shaw, F. & Shaw, R. 1999. Prediction of single-cross performance. In J. G.Coors & S. Pandey, eds. The genetics and exploitation of heterosis in crops, pp. 277–285. Madison,WI, USA, American Society of Agronomy.Sprague, G.F. &. Dudley, J.W. 1988. Corn and corn improvement, 3rd edition. Madison, Wisconsin,USA, American Society of Agronomy.Stam, P. 1995. Marker-assisted breeding. In J.W. Van Ooijen & J. Jansen, eds. Biometrics in plantbreeding: applications of molecular markers. Proc. 9 th Mtg. EUCARPIA Section on Biometrics inPlant Breeding, pp. 32–44. Wageningen, Netherlands, CPRO-DLO.Stromberg, L.D., Dudley, J.W. & Rufener, G.K. 1994. Comparing conventional early generationselection with molecular marker assisted selection in maize. Crop Sci. 34: 1221–1225.Stuber, C.W. 1992. Biochemical and molecular markers in plant breeding. In J. Janick, ed. PlantBreeding Rev. pp 37–61.Stuber, C.W. & Moll, R.H. 1972. Frequency changes of isozyme alleles in a selection experiment forgrain yield in maize (Zea mays L.). Crop Sci. 12: 337–340.Tanksley, S.D., Young, N.D., Patterson, A.H. & Bonierbale, M.W. 1989. RFLP mapping in plantbreeding: New tools for an old science. Biotech. 7: 257–264.Tanksley, S.D., Grandillo, S., Fulton, T.M., Zamir, D., Eshed, Y., Petiard, V., Lopez, J. & Beck-Bunn, T. 1996. Advanced backcross QTL analysis in a cross between an elite processing line oftomato and its wild relative L. pimpinellifolium. Theor. Appl. Genet. 92: 213–224.Tautz, D. 1989. Hypervariability of simple sequences as a general source for polymorphic DNA.Nucl. Acids Res. 17: 6463–6471.Thornsberry, J.M., Goodman, M.M., Doebley, J., Kresovich, S., Nielsen, D. & Buckler, E.S. 2001.Dwarf8 polymorphisms associate with variation in flowering time. Nature Genet. 28: 286–289.Troyer, A.F. 2004. Persistent and popular germplasm in seventy centuries of corn evolution. In C.W.Smith, J. Betrán & E.C.A. Runge, eds. Corn: origin, history, technology, and production, pp. 133–232. Hoboken, NJ, USA, John Wiley & Sons.Utz, H.F. & Melchinger, A.E. 1996. PLABQTL: a program for composite interval mapping of QTL.J. Quant. Trait Loci 2: 1.van Berloo, R. & Stam, P. 1998. Marker-assisted selection in autogamous RIL populations: a simulationstudy. Theor. Appl. Genet. 96: 147–154.van Berloo, R. & Stam, P. 2001. Simultaneous marker-assisted selection for multiple traits in autogamouscrops. Theor. Appl. Genet. 102: 1107–1112.

150Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishvan Ooijen, J.W. & Maliepaard, C. 1996. MapQTL version 3.0: software for the calculation of QTLpositions on genetic maps. Plant Genome IV, San Diego, CA, USA (available at www.intl-pag.org/4/abstracts/p316).Varshney, R.K., Graner, A. & Sorrels, M.E. 2005. Genomics-assisted breeding for crop improvement.Trends Plant Sci. 10: 621–630.Visscher, P.M., Haley, C.S. & Thompson, R. 1996. Marker-assisted introgression in backcross breedingprograms. Genetics 144: 1923–1932.Vos P., Hogers, R., Bleeker, M., Reijans, M., van de Lee, T., Hornes, M., Frijters, A., Pot, J.,Peleman, J., Kuiper, M. & Zabeau, M. 1995. AFLP: a new technique for DNA fingerprinting.Nucl. Acids Res. 23: 4407–4414.Welsh, J. & McClelland, M. 1990. Fingerprinting genomes using PCR with arbitrary primers. Nucl.Acids Res. 18: 7213–7218.Willcox, M.C., Kairallah, M.M., Bergvinson, D., Crossa, J., Deutsch, A., Edmeades, G.O.,González-de-León, D., Jiang, C., Jewell, D.C., Mihm, A J., Williams, W.P. & Hoisington, D.2002. Selection for resistance to southwestern corn borer using marker-assisted and conventionalbackcrossing. Crop Sci. 42: 1516–1528.Williams, J.G.K., Kubelik, A.R., Livak, K.J., Rafalski, J.A. & Tingey, S.V. 1990. DNA polymorphismsamplified by arbitrary primers are useful as genetic markers. Nucl. Acids Res. 18:6531–6535.Xiao, J., Li, J., Grandillo, S., Nag Ahn, S., Yuan, L., Tanksley, S.D. & McCouch, S.R. 1998.Identification of trait-improving quantitative trait loci alleles from a wild rice relative, Oryzarufipogon. Genetics 150: 899–909.Xie, C. & Xu, S. 1998. Efficiency of multistage marker-assisted selection in the improvement of multiplequantitative traits. Heredity 80: 489–498.Young, N.D. 1999. A cautiously optimistic view for marker-assisted breeding. Mol. Breed. 5:505–510.Yousef, G.G. & Juvik, J.A. 2001. Comparison of phenotypic and marker-assisted selection formquantitative traits in sweet corn. Crop Sci. 41: 645–655.Zeng, Z-B. 1993. Theoretical basis for separation of multiple linked gene effects in mapping quantitativetrait loci. Proc. Natl. Acad. Sci. USA 90: 10972–10976.Zeng, Z-B. 1994. Precision mapping of quantitative trait loci. Genetics 136: 1457–1468.

Chapter 9Molecular marker-assisted selectionfor resistance to pathogensin tomatoAmalia Barone and Luigi Frusciante

152Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishSummarySince the 1980s, the use of molecular markers has been suggested to improve the efficiencyof releasing resistant varieties, thus overcoming difficulties met with classical breeding. Fortomato, a high-density molecular map is available in which more than 40 resistance genesare localized. Markers linked to these genes can be used to speed up gene transfer and pyramiding.Suitable PCR markers targeting resistance genes were constructed directly on thesequences of resistance genes or on restriction fragment length polymorphisms (RFLPs)tightly linked to them, and used to select resistant genotypes in backcross schemes. In somecases, the BC 5 generation was reached, and genotypes that cumulated two homozygousresistant genes were also obtained. These results supported the feasibility of using markerassistedselection (MAS) in tomato and reinforcing the potential of this approach for othergenes, which is today also driven by the development of new techniques and increasingknowledge about the tomato genome.

Chapter 9 – Molecular marker-assisted selection for resistance to pathogens in tomato 153INTRODUCTIONTomato (Solanum lycopersicum, formerlyLycopersicon esculentum) is one of the mostwidely grown vegetable crops in the world.It is used as a fresh vegetable and can alsobe processed and canned as a paste, juicesauce, powder or as a whole. World volumehas increased approximately 10 percentsince 1985, reflecting a substantial increasein dietary use of the tomato. Nutritionally,tomato is a significant source of vitaminsA and C. Furthermore, recent studies haveshown the importance of lycopene, a majorcomponent of red tomatoes, which hasantioxidant properties and may help toprotect against cancer and heart disease(Rao and Agarwal, 2000).One of the main constraints to tomatocultivation is damage caused by pathogens,including viruses, bacteria, nematodes andfungi, which cause severe losses in production.The control of pathogen spreadmainly involves three strategies: husbandrypractices, application of agrochemicals anduse of resistant varieties. Husbandry techniquesgenerally help to restrict the spreadof pathogens and their vectors as well as tokeep plants healthy, thus allowing pathogenattack to be limited. Chemical controlgives good results for some pathogens, butpoor results against others, such as bacteria,and has practically no effect on viruses.Moreover, reducing chemical treatmentslowers the health risks to farmers andconsumers. Therefore, in order to achievesustainable agriculture and obtain highquality,safe and healthy products, the useof resistant varieties is one of the principaltools to reduce pathogen damage.Since the early part of the twentiethcentury, breeding for disease resistance hasbeen a major method for controlling plantdisease. Varieties that are resistant or tolerantto one or a number of specific pathogensTable 1List of pathogen resistances present in tomatobreeding lines, varieties and F 1 hybrids obtainedthrough conventional breedingVirusBeet curly top virus (BCTV)Tobacco mosaic virus (TMV)Tomato mosaic virus (ToMV)Tomato yellow leaf curl virus (TYLCV)Tomato spotted wilt virus (TSWV)BacteriaCorynebacterium michiganensePseudomonas solanacearumPseudomonas syringae pv. tomatoNematodesMeloidogyne spp.FungiAlternaria alternata f. sp. lycopersiciAlternaria solaniCladosporium fulvumFusarium oxysporum f. sp. lycopersiciFusarium oxysporum f. sp. radicis-lycopersiciPhytophthora infestansPyrenochaeta lycopersiciStemphylium solaniVerticillium dahliaeModified from Laterrot (1996) and updated as reported inthe text.are already available for many crops, andhybrids with multiple resistance to severalpathogens exist and are currently used invegetable production. In tomato, geneticcontrol of pathogens is a very useful practicewith most resistance being monogenicand dominant. Various sources of resistancehave been used in traditional breedingprogrammes, and resistant breeding lines,varieties and F 1 hybrids have been developedwith varying stability and levels of expression(Table 1) (Laterrot, 1996; Gardner andShoemaker, 1999; Scott, 2005).MARKER-ASSISTED BREEDING FORPATHOGEN RESISTANCEAlthough conventional plant breedinghas had a significant impact on improvingtomato for resistance to importantdiseases, the time-consuming process of

154Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishTable 2Resistance genes mapped on the tomato molecular mapPathogen Gene 1 ChromosomallocationReferenceVirusAlfalfa mosaic virus (AMV) Am 6 Parrella et al., 2004Cucumber mosaic virus Cmr 12 Stamova and Chetelat, 2000(CMV)Potato virus Y (PVY) pot-1 3 Parrella et al., 2002Tomato mottle virus 2 genes 6 Griffiths and Scott, 2001(ToMoV)Tobacco mosaic virus (TMV) Tm-1, Tm2a 2, 9 Young and Tanksley, 1988; Levesque etal., 1990Tomato spotted wilt virus(TSWV)Sw5 9 Stevens, Lamb & Rhoads, 1995Tomato yellow leaf curlvirus (TYLCV)BacteriaTy-1 (Q), Ty-2 6, 11 Zamir et al., 1994; Chagué et al., 1997;Hanson et al., 2000Clavibacter michiganensis Cm1.1- Cm 10.1 (Q) 1, 6, 7, 8, 9, 10 Sandbrink et al., 1995QTLs 5, 7, 9 van Heusden et al., 1999Rcm2.0 (Q), Rcm5.1 (Q) 2, 5 Kabelka, Franchino & Francio, 2002Pseudomonas syringae pv.tomatoPrfPto66Salmeron et al., 1996Martin et al., 1993Ralstonia solanacearum Bw 1, Bw 3, Bw 4, Bw 5 (Q) 6, 10, 4, 6 Danesh et al., 2004; Thoquet et al., 1996Xanthomonas campestris Bs4 5 Ballvora et al., 2001pv vesicatoriarx-1, rx-2, rx-3 1 Yu et al., 1995NematodesGlobodera rostochiensis Hero 4 Ganal et al., 1995Meloidogyne spp. Mi, Mi-3, Mi-9 6, 12, 6 Williamson et al., 1994; Yaghoobi et al.,1995; Ammiraju et al., 2003FungiAlternaria alternata f. sp.lycopersiciAsc 3 van der Biezen, Glagotlkaya & Overduin,1995QTLs 2a, 2c, 3, 9, 12 2, 3, 9, 12 Robert et al., 2001Alternaria solani EBR-QTLs All Foolad et al., 2002; Zhang et al., 2003Cladosporium fulvum Cf-1, Cf-2, Cf-4, Cf-5, Cf-9 1, 6, 1, 6, 1 Balint-Kurti et al., 1994; Jones et al.,1993Fusarium oxysporum f. sp.radicis- lycopersiciFrl 9 Vakalounakis et al., 1997Fusarium oxysporum f. sp.lycopersiciI1, I2, I3 7, 11, 7 Bournival, Vallejos and Scott, 1990;Sarfatti et al., 1991; Tanksley andCostello, 1991; Ori et al., 1997Leveillula taurica Lv 12 Chunwongse et al., 1994Oidium lycopersicon Ol-1, ol-2, Ol-3, Ol-4 6, 4, 6, 6 Huang et al., 2000; Bai et al., 2004;De Giovanni et al., 2004Ol-qtl1, Ol-qtl2, Ol-qtl3 6, 12 Bai et al., 2003Phytophthora infestans lb1-lb12 (Q) All Brouwer, Jones and St. Clair, 2004Ph-1, Ph-2, Ph-3 7, 10, 9 Moreau et al., 1998; Chunwongse et al.,2002Pyrenochaeta lycopersici py-1 3 Doganlar et al., 1998Stemphylium spp. Sm 11 Behare et al., 1991Verticillium dahliae Ve1, Ve2 9 Diwan et al., 1999; Kawchuck et al., 20011Q in parenthesis, QTL and qtl indicate quantitative trait loci. Recessive resistance genes are reported with small letters.

Chapter 9 – Molecular marker-assisted selection for resistance to pathogens in tomato 155making crosses and backcrosses, and theselection of the desired resistant progeny,make it difficult to respond adequately tothe evolution of new virulent pathogens.Moreover, several interesting resistances aredifficult to use because the diagnostic testsoften cannot be developed due to the challengeposed by inoculum production andmaintenance. In addition, where symptomsare detectable only on adult plants and/orfruits, diagnostic tests can be particularlyexpensive and difficult to perform.Since the 1980s, the use of molecularmarkers has been suggested as a tool forbreeding many crops, including tomato.In the last two decades, molecular markershave been employed to map and tag majorgenes and quantitative trait loci (QTL)involved in monogenic and polygenicresistance control, known respectively asvertical and horizontal resistance. To date,more than 40 genes (including many singlegenes and QTL) that confer resistanceto all major classes of plant pathogens havebeen mapped on the tomato molecular map(Table 2) and/or cloned from Solanaceousspecies, as reported by Grube, Radwanskiand Jahn (2000). Since then, other resistancegenes together with resistance geneanalogues (RGAs), which are structurallyrelated sequences based on the proteindomain shared among cloned R genes(Leister et al., 1996), have been added to themap. A molecular linkage map of tomatobased on RGAs has also been constructedin which 29 RGAs were located on nine ofthe 12 tomato chromosomes (Foolad et al.,2002; Zhang et al., 2002). Several RGA lociwere found in clusters and their locationscoincided with those of several knowntomato R genes or QTL. This map providesa basis for further identifying and mappinggenes and QTL for disease resistance andwill be useful for MAS.In fact, independently of the type ofmarker used for selection, by making itpossible to follow the gene under selectionthrough generations rather than waitingfor phenotypic expression of the resistancegene, markers tightly linked to resistancegenes can greatly aid disease resistanceprogrammes. In particular, genetic mappingof disease resistance genes has greatlyimproved the efficiency of plant breedingand also led to a better understanding of themolecular basis of resistance.DNA marker technology has been usedin commercial plant breeding programmessince the early 1990s, and has proved helpfulfor the rapid and efficient transfer ofuseful traits into agronomically desirablevarieties and hybrids (Tanksley et al., 1989;Lefebvre and Chèvre, 1995). Markers linkedto disease resistance loci can now be usedfor MAS programmes, thus also allowingseveral resistance genes to be cumulated inthe same genotype (“pyramiding” of resistancegenes), and they may be also usefulfor cloning and sequencing the genes. Intomato, several resistance genes have beensequenced to date, among them Cf-2, Cf-4, Cf-5, Cf-9, Pto, Mi, I2, and Sw5. Thesecloned R genes now provide new tools fortomato breeders to improve the efficiencyof breeding strategies, via MAS. AlthoughMAS is still not used routinely for improvingdisease resistance in many importantcrops (Michelmore, 2003), it is being usedby seed companies for improving simpletraits in tomato (Foolad and Sharma, 2005).Furthermore, while the deep knowledgeof the tomato genome and the availabilityof a high-density molecular map for thisspecies (Pillen et al., 1996) should providefurther opportunities to acceleratebreeding through MAS, the time-consumingand expensive process of developingmarkers associated with genes of inter-

156Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishest and the high cost of genotyping largepopulations has and will continue to limitthe use of MAS in most tomato breedingprogrammes.The potential of MAS to speed up thebreeding of tomato using molecular markerslinked to various resistance genes hasbeen examined in the authors’ laboratory.The two main goals of the research wereto find the most suitable markers, andto test the feasibility of MAS for pyramidingresistance genes in tomato “elite”lines selected for their good processingqualities.STRATEGIES FOR GENE TRANSFER ANDPYRAMIDINGSix tomato genotypes carrying variousresistance genes (Table 3) were crossed withtomato “elite” lines previously selected foryield and quality but lacking resistancetraits. Each resistant genotype was crossedinitially with each “elite” tomato line andvarious backcross schemes were then carriedout starting from different F 1 hybrids.At each backcross generation the screeningof resistant genotypes was performed usingmolecular markers linked to the resistancegenes and DNA extracted from youngleaves at seedling stage. Only resistantplants were then transplanted and grown inthe greenhouse. At flowering, crosses weremade with the recurrent parent to obtainthe subsequent generations.As the efficiency of MAS depends on theavailability of polymerase chain reaction(PCR)-based markers highly linked to theresistance gene to be selected, for eachresistance gene the most suitable markersystem was investigated. For this purpose,three different strategies were undertaken.The first involved searching PCR markersalready available in the literature andverifying their usefulness on the geneticmaterial used. The second consisted ofdesigning PCR primers from the sequenceof cloned genes reported in the GeneBankdatabase of the National Center forBiotechnology Information (www.ncbi.nlm.nih.gov/Genbank), while the thirdinvolved designing PCR primers fromRFLP markers tightly linked to resistancegenes. This last strategy was made possibleby the availability of sequences of variousmapped tomato RFLPs in the SolGenesdatabase (www.sgn.cornell.edu).In most cases, the results were obtainedusing cleaved amplified polymorphicsequences (CAPS; Konieczyn and Ausubel,1993), which require one PCR reaction followedby restriction digest of the amplifiedfragment. In three cases (markers linked togenes Mi, Sw5 and Tm2a), the primers andenzymes used were those reported in the literature(Williamson et al., 1994; Folkertsmaet al., 1999; Sobir et al., 2000). In the caseof gene py-1, the procedure reported inthe literature (Doganlar et al., 1998) wasTable 3Tomato genotypes used as resistant parents in the backcross breeding schemes. For each genotyperesistant genes are reportedGenotype Resistance gene PathogenMomor Frl, Tm2a, Ve Fusarium oxysporum f. sp. radicis-lycopersici, TMV, Verticillium dahliaeMotelle I2, Mi, Ve Fusarium oxysporum f. sp lycopersici, Meloidogyne spp., VerticilliumdahliaeOkitzu I2, Tm2a Fusarium oxysporum f. sp lycopersici, TMVOntario Pto Pseudomonas syringaePyrella py-1 Pyrenochaeta lycopersiciStevens Sw5 TSWV

Chapter 9 – Molecular marker-assisted selection for resistance to pathogens in tomato 157Figure 1Number of generations reached in backcross schemes carried out between four susceptiblerecurrent genotypes and five resistant donor genotypes5432MomorOkitzuOntarioPyrellaStevens10Sel 8 137 PI15 AD17simplified, enabling a faster and cheapermarker system, i.e. a sequence characterizedamplified region (SCAR; Kawchuk,Hachey and Lynch, 1998) marker, whichonly requires one PCR reaction to detectpolymorphism between the resistant andthe susceptible genotypes, to be set up.(Barone et al., 2004).The second strategy was followed todesign primers and enzymes suitable fortargeting three resistance genes (I2, Pto andVe2). This strategy allowed gene-assistedselection to be achieved through the simplePCR procedure. Finally, the third strategywas applied in the case of one CAPSmarker targeting the resistance gene Frl;it was derived from one RFLP tomatomarker (TG101) linked to the gene (Fazio,Stevens and Scott, 1999).The markers found were used to selectresistant genotypes in backcross breedingschemes, while the process itself allowedthree generations to be screened annually.At present, for some cross combinations,the BC 5 generation has been reached, forothers the BC 2 -BC 3 (Figure 1). Where aBC 5 generation was already available, thebreeding programme continued by selfingBC 5 resistant genotypes. In all other casesthe backcross programme will continue upto the fifth backcross generation. At theend of each backcross scheme, the resistantBC 5 F 3 genotypes, selected through molecularmarker analysis, will also be testeddirectly for resistance by inoculating thepathogen and monitoring signs of disease.This will allow verification that no linkagebreakage and loss of resistance geneoccurred.This procedure was already adopted inthe case of one backcross scheme aimedat transferring a resistance gene to tomatospotted wilt virus (TSWV) to the susceptiblegenotype AD17 (Langella et al., 2004).The in vivo test performed on F 1 BC 5 ,F 2 BC 5 and, F 3 BC 5 generations confirmedthe introgression of the resistance trait andrevealed that the resistance gene Sw5 was

158Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishFigure 2Breeding scheme used for thepyramiding of two resistanthomozygous genes (Sw5 and Pto) in thesusceptible genotype PI15PI15 x Stevens Sw5/Sw5F1BC 1 Sw5/-F1BC 2 Sw5/-F1BC 3 Sw5/-F1BC 4 Sw5/-XPI15 x Ontario Pto/PtoF1BC 1 Pto/-F1BC 2 Pto/-F1BC 3 Pto/-F1BC 4 Pto/-F 1 (F 1BC 4 x F 1BC 4 ) Sw5/-, Pto/-XF 2 (F 1BC 4 x F 1BC 4 ) Sw5/Sw5, Pto/Ptofixed at the homozygous stage at the F 3 BC 5generation.Finally, besides the transfer of one resistancegene to each susceptible genotype, acrossing scheme was undertaken to accumulatetwo or three resistance genes in thesame genotype. In this case, the decisionwas made to stop the backcross scheme atthe BC 3 or BC 4 generation as both parentallines were cultivated varieties and hencegenetically very similar, and therefore therecovery of the recurrent genome couldbe satisfactory. F 1 BC 4 hybrids carrying thesame genetic background in the recurrentparent have been intercrossed, followingthe breeding scheme shown in Figure 2.At the end of each F 1 BC 4 x F 1 BC 4 crossand after selecting the genotypes carryingall the resistant alleles at the heterozygouslevel, one or two selfing generations will becarried out to fix all the resistant genes atthe homozygous level.This strategy has already started in somecases and the first homozygous multiresistantgenotypes have been obtained. Alsoavailable are two F 2 genotypes out of 52 analysedplants, obtained by intercrossing theF 1 BC 4 progeny from PI15 x Stevens withthe F 1 BC 4 progeny from PI15 x Ontario(Table 4). This F 2 generation exhibited twogenotypes carrying both resistant genesSw5 and Pto at the homozygous level aswell as 29 genotypes carrying both genes atthe heterozygous level.The work reported here on transferringresistance genes among tomato genotypesdemonstrates the usefulness of MAS forimproving traditional breeding strategies.The contribution of molecular markerslinked to resistance genes was very efficientin reducing the time and space necessaryfor selection, enabling both early screeningfor resistance and reduced numbers ofgenotypes to be transplanted. The mostchallenging work was the search for suitablemarkers, which often required bothconsiderable time and financial resources.Different strategies were used successfullyto find the most suitable markers to performMAS for transferring eight resistance genesinto superior tomato genotypes; such strategiescould be repeated in tomato for manyother genes due to advanced molecularknowledge of the genome of this species.PERSPECTIVESThe availability of PCR-based markersfor many resistance genes allows MAS forbiotic resistance in tomato to be appliedsuccessfully in any laboratory without theneed for highly sophisticated techniques.

Chapter 9 – Molecular marker-assisted selection for resistance to pathogens in tomato 159Table 4Resistant heterozygous plants obtained in some cross combinations realized for pyramiding ofresistance genesCrossPyramidedgenesGenerationAnalysed plant(number)Resistant plant(number)(137 x Momor) F 1 BC 3 xTm2a – Frl(137 x Ontario) F 1 BC 3 Ve – PtoF 1 50 7(AD17 x Okitzu) F 1 BC 4 xTm2a – Frl(AD17 x Stevens) F 1 BC 4Sw5F 1 24 5(AD17 x Ontario) F 1 BC 4 x(AD17 x Stevens) F 1 BC 4Pto – Sw5 F 1 24 6(PI15 x Stevens) F 1 BC 4 x(PI15 x Ontario) F 1 BC 4Sw5 - Pto F 2 52 29Indeed, once a marker has been set up,its use on large populations for resistancescreening is then routine. Technical facilitiesare today available for screening manysamples simultaneously and also costs forequipment are decreasing. In addition, therapid development of new molecular techniques,combined with the ever-increasingknowledge about the structure and functionof resistance genes (Hulbert et al.,2001), will help to identify new molecularmarkers for MAS, such as single nucleotidepolymorphisms (SNPs). Moreover, thanksto the International Solanaceae GenomeProject (SOL), sequencing of the tomatogenome is in progress, and in a few yearsthis will enhance information on resistancegenes. This in turn will facilitate thedevelopment of molecular markers fromtranscribed regions of the genome, therebyallowing large-scale gene-assisted selection(GAS) to be achieved.Over the coming years, selection forpathogen resistance in tomato will beunderpinned by research aimed at: mappingother resistance genes for new pathogens;developing PCR-based functional markers(Andersen and Ludderstedt, 2003); designingthe most suitable breeding schemes(Peleman and van der Voort, 2003), especiallyfor transferring QTL resistances;large-scale screening through automation;allele-specific diagnostics (Yang et al.,2004); and DNA microarrays (Borevitzet al., 2003). In effect, the combination ofnew knowledge and new tools will lead tochanges in the strategies used for breedingby exploiting the potential of integrating“omics” disciplines with plant physiologyand conventional plant breeding, a processthat will drive the evolution of MASinto genomics-assisted breeding (Morganteand Salamini, 2003; Varshney, Graner andSorrells, 2005).ACKNOWLEDGEMENTSThe authors wish to thank Mr MarkWalters for editing the manuscript andAngela Cozzolino for her technical assistance.This research was partially financedby the Ministry of Agriculture andForestry of Italy in the framework ofproject PROMAR (Plant Protection by theUse of Molecular Markers). Contributionno. 119 from the Department of Soil, Plantand Environmental Sciences.REFERENCESAmmiraju, J.S.S., Veremis, J.C., Huang, X., Roberts, P.A. & Kaloshian, I. 2003. The heat-stableroot-nematode resistance gene Mi-9 from Lycopersicon peruvianum is localized on the short armof chromosome 6. Theor. Appl. Genet. 106: 478–484.

160Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishAndersen, J.R. & Lubberstedt, T. 2003. Functional markers in plants. Trends Plant Sci. 8: 554–560.Bai, Y., Huang, C-C., van der Hulst, R., Meijer-Dekens, F., Bonnema, G. & Lindhout, P. 2003.QTLs for tomato powdery mildew resistance (Oidium lycopersici) in Lycopersicon parviflorumG1.1601 co-localize with two qualitative powdery mildew resistance genes. Mol. Plant - MicrobeInteract. 16: 169–176.Bai, Y., van der Hulst, R., Huang, C-C., Wei, L., Stam, P. & Lindhout, P. 2004. Mapping Ol-4, agene conferring resistance to Oidium neolycopersici and originating from Lycopersicon peruvianumLA2172, requires multi-allelic, single locus markers. Theor. Appl. Genet. 109: 1215–1223.Balint-Kurti, P.J., Dixon, M.S., Jnes, D.A., Norcott, K.A. & Jones, J.D.G. 1994. RFLP linkageanalysis of the Cf-4 and Cf-9 genes for resistance to Cladosporium fulvum in tomato. Theor. Appl.Genet. 88: 691–700.Ballvora, A., Pierre, M., van den Ackerveken, G., Chornack, S., Rossier, O., Ganal, M., Lahaye,T. & Bonas, U. 2001 Genetic mapping and functional analysis of the tomato Bs4 locus governingrecognition of the Xanthomonas campestris pv.vesicatoria AvrBs4 protein. Mol. Plant-MicrobeInteract. 14: 629–638.Barone, A., Langella, R., Ercolano, M.R., Di Matteo, A. & Monti, L. 2004 Strategies for pathogenresistance gene transfer in tomato with the aid of molecular markers. In S.G. Pandalai, ed. Recentresearch developments in genetics and breeding, pp. 211–222. Kerala, India, Research Signpost.Behare, J., Laterrot, H., Sarfatti, M. & Zamir, D. 1991. RFLP mapping of the Stemphylium resistancegene in tomato. Mol. Plant-Microbe Interact. 4: 489–492.Borevitz, J.O., Liang, D., Plouffe, D., Chang, H-S., Zhu, T., Weigel, D., Berry, C.C., Winzeler, E. &Chory, J. 2003. Large-scale identification of single-feaqture polymorphisms in complex genomes.Genome Res. 13: 513–523.Bournival, B.L., Vallejos, C.E. & Scott, J.W. 1990. Genetic analysis of resistances to race 1 and 2 ofFusarium oxysporum f. sp. lycopersici from the wild tomato Lycopersicon pennellii. Theor. Appl.Genet. 79: 641–645.Brouwer, D.J., Jones, E.S. & St Clair, D.A. 2004. QTL analysis of quantitative resistance toPhytophthora infestans (late blight) in tomato and comparisons to potato. Genome 47: 475–492.Chagué, V., Mercier, J.C., Guénard, M., de Courcel, A.G.L. & Vedel, F. 1997. Identification ofRAPD markers linked to a locus involved in quantitative resistance to TYLCV in tomato by bulkedsegregant analysis. Theor. Appl. Genet. 95: 671–677.Chunwongse, J., Bunn, T.B., Crossman, C., Jing, J. & Tanksley, S.D. 1994. Chromosomal localizationand molecular-marker tagging of the powdery mildew resistance gene (Lv) in tomato. Theor.Appl. Genet. 89: 76–79.Chunwongse, J., Chunwongse, C., Black, L. & Hanson, P. 2002. Molecular mapping of the Ph-3gene for late blight resistance in tomato. J. Hort. Sci. & Biotech. 77: 281–286.Danesh, D., Aarons, S., McGill, G.E. & Young, N.D. 1994. Genetic dissection of oligenic resistanceto bacterial wilt in tomato. Mol. Plant-Microbe Interact. 7: 464–471.De Giovanni, C., Dell’Orco, P., Bruno, A., Ciccarese, F., Lotti, C. & Ricciardi, L. 2004. Identificationof PCR-based markers (RAPD, AFLP) linked to a novel powdery mildew resistance gene (ol-2) intomato. Plant Sci. 166: 41–48.Diwan, N., Fluhr, R., Eshed, Y., Zamir, D. & Tanksley, S.D. 1999. Mapping of Ve in tomato: a geneconferring resistance to broad-spectrum pathogen, Verticillium dahliae race 1. Theor. Appl. Genet.98: 315–319.

Chapter 9 – Molecular marker-assisted selection for resistance to pathogens in tomato 161Doganlar, S., Dodson, J., Gabor, B., Beck-Bunn, T., Crossman, C. & Tanksley, S.D. 1998 Molecularmapping of the py-1 gene for resistance to corky root rot (Pyranochaeta lycopersici) in tomato.Theor. Appl. Genet. 97: 784–788.Fazio, G., Stevens, M.R., & Scott, J.W. 1999. Identification of RAPD markers linked to fusariumcrown and root rot resistance (Frl) in tomato. Euphytica 105: 205–210.Folkertsma, R.T., Spassova, M.I., Prins, M., Stevens, M.R., Hille, J. & Goldbach, R.W. 1999.Construction of a bacterial artificial chromosome (BAC) library of Lycopersicon esculentum cv.Stevens and its application to physically map the Sw5 locus. Mol. Breed. 5: 197–207.Foolad, M.R. & Sharma, A. 2005. Molecular markers as selection tools in tomato breeding. ActaHort. 695: 225–240.Foolad, M.R., Zhang, L.P., Khan, A.A., Nino-Liu, D. & Lin, G.Y. 2002. Identification of QTLs forearly blight (Alternaria solani) resistance in tomato using backcross populations of a Lycopersiconesculentum x L. hirsutum cross. Theor. Appl. Genet. 104: 945–958.Ganal, M.W., Simon, R., Brommonschenkel, S., Arndt, M., Tanksley, S.D. & Kumar, A. 1995.Genetic mapping of a wide spectrum nematode resistance gene (Hero) against Globodera rostochiensisin tomato. Mol. Plant-Microbe Interact. 8: 886–891.Gardner, R.G. & Shoemaker, P.B. 1999. “Mountain Supreme” early blight-resistant hybrid tomatoand its parents NC EBR-3 and NC EBR-4 tomato breeding lines. Hort. Sci. 34: 745–746.Griffiths, P.D. & Scott, J.W. 2001. Inheritance and linkage of tomato mottle virus resistance genesderived from Lycopersicon chilense accession LA 1932. J. Am. Soc. Hort. Sci. 126: 462–467.Grube, R.C., Radwanski, E.R. & Jahn, M. 2000. Comparative genetics of disease resistance withinthe Solanaceae Genetics 155: 873–887.Hanson, P.M., Bernacchi, D., Green, S., Tanksley, S.D. & Muniy-Appa, V. 2000. Mapping a wildtomato introgression associated with tomato leaf curl virus resistance in a cultivated tomato line. J.Am. Soc. Hort. Sci. 125: 15–20.Huang, C.C., van de Putte, P.M., Haanstra-van der Meer, J.G., Mejer-Dekens, F. & Lindhout, P.2000. Characterization and mapping of resistance to Oidium lycopersicum in two Lycopersiconhirsutum accessions: evidence for close linkage of two Ol-genes on chromosome 6. Heredity 85:511–520.Hulbert, S.H., Webb, C.A., Smith, S.M. & Sun, Q. 2001. Resistance gene complexes: evolution andutilization. Ann. Rev. Phytopathol. 39: 285–312.Jones, D.A., Dickinson, M.J., Balint-Kurt, P.J., Dixon, M.S. & Jones, J.D.G. 1993. Two complexresistance loci revealed in tomato by classical and RFLP mapping of Cf-2, Cf-4, Cf-5, and Cf-9genes for resistance to Cladosporium fulvum. Mol. Plant-Microbe Interact. 6: 348–357.Kabelka, E., Franchino, B. & Francio, D.M. 2002. Two loci from Lycopersicon hirsutum LA407confer resistance to strains of Clavibacter michiganense subsp. michiganensis. Phytopath. 92:504–510.Kawchuck, L.M., Hachey, J. & Lynch, D.R. 1998. Development of sequence characterized DNAmarkers linked to a dominant Verticillium wilt resistance gene in tomato. Genome 41: 91–95.Kawchuk, L.M., Hachey, J., Lynch, D.R., Kulcsar, F., van Rooijen, G., Waterer, D.R., Robertson,A., Kokko, E., Byers, R., Howar, R.J., Fischer, R. & Prufer, D. 2001. Tomato Ve disease resistancegenes encode cell surface-like receptors. Proc. Nat. Acad. Sci. USA 98: 6511–6515.Konieczyn, A. & Ausubel, F.M. 1993. A procedure for mapping Arabidopsis mutations using codominantecotype-specific PCR-based markers. Plant J. 4: 403–410.

162Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishLangella, R., Ercolano, M.R., Monti, L., Frusciante, L. & Barone, A. 2004 Molecular markerassisted transfer of resistance to TSWV in tomato elite lines. J. Hort. Sci. & Biotech. 79: 806–810Laterrot, H. 1996. Breeding strategies for disease resistance in tomatoes with emphasis to the tropics:current status and research challenges, pp.126–132. Proc.1 st Internat. Symp. Trop. Tomato Dis.Recife, Brazil.Lefebvre, V. & Chèvre, A.M. 1995. Tools for marking plant disease and pest resistance genes: a review.Agronomie 15: 3–19.Leister, D., Ballvora, A., Salamini, F. & Gebhardt, C. 1996. PCR-based approach for isolating pathogenresistance genes from potato with potential for wide application in plants. Nature Genet. 14:421–429.Levesque, H., Vedel, F., Mathieu, C. & de Courcel, A.G.L. 1990. Identification of a short rDNAspacer sequence highly specific of a tomato line containing Tm-1 gene introgressed fromLycopersicon hirsutum. Theor. Appl. Genet. 80: 602–608.Martin, G.B., Brommonschenkel, S., Chunwongse, J., Frary, A., Ganal, M.W., Spivey, R., Wu, T.,Earle, E.D. & Tanksley, S.D. 1993. Map-based cloning of a protein kinase gene conferring diseaseresistance in tomato. Science 262: 1432–1436.Michelmore, R.W. 2003. The impact zone: genomics and breeding for durable disease resistance. Curr.Opin. Plant Biol. 6397–6404.Moreau, P., Thoquet, P., Olivier J., Laterrot, H. & Grimsley, N.H. 1998. Genetic mapping of Ph-2, asingle locus controlling partial resistance to Phytophthora infestans in tomato. Mol. Plant-MicrobeInteract. 11: 259–268.Morgante, M. & Salamini, F. 2003. From plant genomics to breeding practice. Curr. Opin. Biotechnol.14: 214–219.Ori, N., Eshed, Y., Paran, I., Prestino, G., Aviv, D., Tanksley, S.D., Zamir, D. & Fluhr, R. 1997. TheI2C family from the wilt disease resistance locus I2 belongs to the nucleotide binding, leucine-richrepeat superfamily of plant resistance genes. Plant Cell 9: 521–532.Parrella, G., Ruffel, S., Moretti, A., Morel, C., Palloix, A. & Caranta, C. 2002. Recessive resistancegenes against potyviruses are localized in colinear genomic regions of the tomato (Lycopersiconspp.) and pepper (Capsicum spp.) genomes. Theor. Appl. Genet. 105: 855–861.Parrella, G., Moretti, A., Gognalons, P., Lesage, M-L., Marchoux, G., Gebre-Selassie, K. & Carata,C. 2004. The Am gene controlling resistance to alfalfa mosaic virus in tomato is located in the clusterof dominant resistance genes on chromosome 6. Phytopath. 94: 345–350.Peleman, J.D. & van der Voort, J.R. 2003. Breeding by design. Trends Plant Sci. 8: 330–334.Pillen, K., Pineda, O., Lewis, C.B. & Tanksley, S.D. 1996. Status of genome mapping tools in thetaxon Solanaceae. In A. Paterson, ed. Genome mapping in plants, pp. 281–308. Austin, TX, USA,Landes.Rao, A.V. & Agarwal, S. 2000. Role of antioxidant lycopene in cancer and heart disease. J. Am. Coll.Nutr. 19: 563–569.Robert, V.J.M., West, M.A.L., Inai, S., Caines, A., Arntzen, L., Smith, J.K. & St. Clair, D.A. 2001.Marker-assisted introgression of blackmold resistance QTL alleles from wild Lycopersicon cheesmaniito cultivated tomato (L. esculentum) and evaluation of QTL phenotypic effects. Mol. Breed.8: 217–233.Salmeron, J.M., Oldroyd, G.E.D., Rommens, C.M.T., Scofield, S.R., Kim, H.S., Lavello, D.T.,Dahlbeck, D. & Staskawicz, B.J. 1996. Tomato Prf is a member of the leucine-rich repeat class of

Chapter 9 – Molecular marker-assisted selection for resistance to pathogens in tomato 163plant disease resistance genes and lies embedded within the tomato Pto kinase gene cluster. Cell86: 123–133.Sandbrink, J.M., van-Ooijen, J.W., Purimahua, C.C., Vrielink, M., Verkerk, R., Zabel, P. &Lindhout, P. 1995. Localization of genes for bacterial canker resistance in Lycopersicon peruvianumusing RFLPs. Theor. Appl. Genet. 90: 444–450.Sarfatti, M., Abu-Abied, M., Katan, J. & Zamir, D. 1991. RFLP mapping of I1, a new locus intomato conferring resistance to Fusarium oxysporum f. sp. lycopersici race 1. Theor. Appl. Genet.80: 22–26.Scott, J.W. 2005. Perspectives on tomato disease resistance breeding: past, present, and future. ActaHort. 695: 217–224.Sobir, Ohmori T., Murata, M. & Motoyoshi, F. 2000. Molecular characterization of the SCAR markerstightly linked to the Tm-2 locus of the genus Lycopersicon. Theor. Appl. Genet. 101: 64–69.Stamova, B.S. & Chetelat, R.T. 2000. Inheritance and genetic mapping of cucumber mosaic virusresistance introgressed from Lycopersicon chilense into tomato. Theor. Appl. Genet. 101: 527–537.Stevens, M.R., Lamb, E.M. & Rhoads, D.D. 1995. Mapping the Sw-5 locus for tomato spotted wildvirus resistance in tomatoes using RAPD and RFLP analyses. Theor. Appl. Genet. 90: 451–456.Tanksley, S.D. & Costello, W. 1991. The size of L. pennellii chromosome 7 segment containing the I-3gene in tomato breeding lines measured by RFLP probing. Rep. Tomato Genet. Coop. 41: 60.Tanksley, S.D., Young, N.D., Paterson, A.H. & Bonierbale, M.W. 1989. RFLP mapping in plantbreeding: new tools for an old science. BioTech. 7: 257–264.Thoquet, P., Olivier, J., Sperisen, C., Rogowsky, P., Laterrot, H. & Grimsley, N.H. 1996.Quantitative trait loci determining resistance to bacterial wilt in tomato cultivar Hawaii 7996. Mol.Plant-Microbe Interact. 9: 826–836.Vakalounakis, D.J., Laterrot, H., Moretti, A., Ligoxigakis, E.K. & Smardas, K. 1997. Linkagebetween Fr-1 (Fusarium oxysporum f.sp. radici-lycopersicy resistance) and Tm-2 (tobacco mosaicvirus resistance-2) loci in tomato (Lycopersicon esculentum). Ann. Appl. Biol. 130: 319–323.van der Biezen, E.A., Glagotlkaya, T. & Overduin, B. 1995. Inheritance and genetic mapping ofresistance to Alternaria alternata f. sp. lycopersici in Lycopersicon pennellii. Mol. Gen. Genet. 247:453–461.van Heusden, A.W., Koornneef, M., Voorrips, R.E., Bruggemann, W., Pet, G., Vrielink-vanGinkel, R., Chen, X. & Lindhout, P. 1999. Three QTLs from Lycopersicon peruvianum confer ahigh level of resistance to Clavibacter michiganensis subsp. michiganensis. Theor. Appl. Genet. 99:1068–1074.Varshney, R.K., Graner, A. & Sorrells, M.E. 2005. Genomics-assisted breeding for crop improvement.Trends Plant Sci. 10: 621–630.Williamson, V.M., Ho, J-Y., Wu, F.F., Miller, N. & Kaloshian, I. 1994. A PCR-based marker tightlylinked to the nematode resistance, Mi, in tomato. Theor. Appl. Genet. 87: 757–763.Yaghoobi, J., Kaloshian, I., Wen, Y. & Willimson, V.M. 1995. Mapping of a new nematode resistancelocus in Lycopersicon peruvianum. Theor. Appl. Genet. 91: 457–464.Yang, W., Bai, X., Kabelka, E., Eaton, C., Kamoun, S., van der Knapp, E. & Francis, D. 2004.Discovery of single nucleotide polymorphisms in Lycopersicon esculetum by computer aided analysisof expressed sequence tags. Mol. Breed. 14: 21–34.Young, N.D. & Tanksley, S.D. 1988. Use of isogenic lines and simultaneous probing to identify DNAmarkers tightly linked to the Tm-2 gene in tomato. Genetics 120: 579–589.

164Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishYu, Y.G., Wang, J.F., Stall, R.E. & Allejos, C.E. 1995. Genomic localization of tomato genes that controla hypersensitivity reaction to Xanthomonas campestris pv vesicatoria (Doidge) dye. Genetics141: 675–682.Zamir, D., Ekstein-Michelson, I., Zakay, U., Navot, N., Zeidan, M., Sarfatti, M., Eshed, Y., Harel,E., Pleban, T., van-Oss, H., Kedar, N., Rabinowitch, H.D. & Czosnek, H. 1994. Mapping andintrogression of a tomato yellow leaf curl virus tolerance gene, Ty-1. Theor. Appl. Genet. 88:141–146.Zhang, L.P., Khan, A., Nino-Liu, D. & Foolad, M.R. 2002. A molecular linkage map of tomato displayingchromosomal locations of resistance gene analogs based on a Lycopersicon esculentum xLycopersicon hirsutum cross. Genome 45: 133–146.Zhang, L.P., Lin, G.Y., Nino-Liu, D. & Foolad, M.R. 2003. Mapping QTLs conferring early blight(Alternaria solani) resistance in a Lycopersicon esculentum x L. hirsutum cross by selective genotyping.Mol. Breed. 12: 3–19.

Section IIIMarker-assisted selectionin livestock – case studies

Chapter 10Strategies, limitations andopportunities for marker-assistedselection in livestockJack C.M. Dekkers and Julius H.J. van der Werf

168Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishSummaryThis chapter reviews the principles, opportunities and limitations for detection ofquantitative trait loci (QTL) in livestock and for their use in genetic improvementprogrammes. Alternate strategies for QTL detection are discussed, as are methods forinclusion of marker and QTL information in genetic evaluation. Practical issues regardingimplementation of marker-assisted selection (MAS) for selection in breed crosses and forselection within breeds are described, along with likely routes towards achieving that goal.Opportunities and challenges are also discussed for the use of molecular information forgenetic improvement of livestock in developing countries.

Chapter 10 – Strategies, limitations and opportunities for marker-assisted selection in livestock 169IntroductionSince the 1970s, the discovery of technologythat enables identification and genotypingof large numbers of genetic markers, andresearch that demonstrated how this technologycould be used to identify genomicregions that control variation in quantitativetraits and how the resulting QTL could beused to enhance selection, have raised highexpectations for the application of gene-(GAS) or marker-assisted selection (MAS)in livestock. Yet, to date, the application ofGAS or MAS in livestock has been limited(see e.g. review by Dekkers, 2004 and thecase study chapters that follow). However,recent further advances in technology,combined with a substantial reduction inthe cost of genotyping, have stimulatedrenewed interest in the large-scale applicationof MAS in livestock.Successful application of MAS inbreeding programmes requires advances inthe following five areas:• Gene mapping: identification and mappingof genes and genetic polymorphisms.• Marker genotyping: genotyping of largenumbers of individuals for large numbersof markers at a reasonable cost for bothQTL detection and routine applicationfor MAS.• QTL detection: detection and estimationof associations of identified genes andgenetic markers with economic traits.• Genetic evaluation: integration of phenotypicand genotypic data in statisticalmethods to estimate breeding values ofindividuals in a breeding population.• MAS: development of breeding strategiesand programmes for the use of moleculargenetic information in selection and matingprogrammes.This chapter outlines the main strategiesfor the application of MAS in livestock andidentifies and discusses the limitations andopportunities for successful MAS in commercialbreeding programmes. It concludesby discussing limitations and opportunitiesfor applying MAS in developing countries.Markers and linkagedisequilibriumOver the past decades, a substantial number ofalternate types of genetic markers have becomeavailable to study the genetic architecture oftraits and for their use in MAS, includingrestriction fragment length polymorphisms(RFLPs), microsatellites, amplified fragmentlength polymorphisms (AFLPs) and singlenucleotide polymorphisms (SNPs). Detailedinformation on these markers can be foundelsewhere in this publication. Althoughalternate marker types have their ownadvantages and disadvantages, dependingon their abundance in the genome, degreeof polymorphism, and ease and cost ofgenotyping, what is crucial for their usefor both QTL detection and MAS is theextent of linkage disequilibrium (LD) thatthey have in the population with loci thatcontribute to genetic variation for the trait.Linkage disequilibrium relates to dependenceof alleles at different loci and is centralto both QTL detection and MAS. Thus, athorough understanding of LD and of thefactors that affect the presence and extent ofLD in populations is essential for a discussionof both QTL detection and MAS.Linkage disequilibriumConsider a marker locus with alleles M andm and a QTL with alleles Q and q that ison the same chromosome as the marker,i.e. the marker and the QTL are linked. Anindividual that is heterozygous for bothloci would have genotype MmQq. Allelesat the two loci are arranged in haplotypeson the two chromosomes of a homologous

170Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishpair that each individual carries. An individualwith genotype MmQq could havethe following two haplotypes: MQ/mq,where the / separates the two homologouschromosomes. Alternatively, it could carrythe haplotypes Mq/mQ. This alternativearrangement of linked alleles on homologouschromosomes is referred to as themarker-QTL linkage phase. The arrangementof alleles in haplotypes is importantbecause progeny inherit one of the twohaplotypes that a parent carries, barringrecombination.The presence of linkage equilibrium(LE) or disequilibrium relates to the relativefrequencies of alternative haplotypes inthe population. In a population that is inlinkage equilibrium, alleles at two loci arerandomly assorted into haplotypes. In otherwords, chromosomes or haplotypes thatcarry marker allele M are no more likelyto carry QTL allele Q than chromosomesthat carry marker allele m. In technicalterms, the frequency of the MQ haplotypesis equal to the product of the populationallele frequency of M and the frequencyof Q. Thus, if a marker and QTL are inlinkage equilibrium, there is no value inknowing an individual’s marker genotypebecause it provides no information on QTLgenotype. If the marker and QTL arein linkage disequilibrium, however, therewill be a difference in the probability ofcarrying Q between chromosomes thatcarry M and m marker alleles and, therefore,a difference in mean phenotype betweenmarker genotypes would also be expected.The main factors that create LD in apopulation are mutation, selection, drift(inbreeding), and migration or crossing. SeeGoddard and Meuwissen (2005) for furtherbackground on these topics. The mainfactor that breaks down LD is recombination,which can rearrange haplotypes thatDisequilibriumFigure 1Break-up of LD over generationsLD is continuously eroded byrecombination10.90.80.70.60.50.40.30.20.100r=.5r=.2r=.1r=.05Generationr=.01rM Qexist within a parent in every generation.Figure 1 shows the effect of recombination(r) on the decay of LD over generations.The rate of decay depends on the rate ofrecombination between the loci. For tightlylinked loci, any LD that has been createdwill persist over many generations but, forloosely linked loci (r > 0.1), LD will declinerapidly over generations.Population-wide versus within-family LDAlthough a marker and a linked QTL maybe in LE across the population, LD willalways exist within a family, even betweenloosely linked loci. Consider a double heterozygoussire with haplotypes MQ/mq(Figure 2). The genotype of this sire isidentical to that of an F 1 cross betweeninbred lines. This sire will produce fourtypes of gametes: non-recombinants MQand mq and recombinants Mq and mQ.As non-recombinants will have higher frequency,depending on the recombinationrate between the marker and QTL, thissire will produce gametes that will be inLD. Furthermore, this LD will extendover a larger distance (Figure 1), becauseit has undergone only one generation ofrecombination. This specific type of LD,mqr=.0015 10 15 20 25

Chapter 10 – Strategies, limitations and opportunities for marker-assisted selection in livestock 171Figure 2Within-family marker-QTL LDWithin-family LD among progeny of a sireLinkage phase can differ between familiesSire 1 Sire 2 Sire 3 Sire 4M Q M q M Q M qmqmQhowever, only exists within this family;progeny from another sire, e.g. an Mq/mQsire, will also show LD, but the LD is inthe opposite direction because of the differentmarker-QTL linkage phase in thesire (Figure 2). On the other hand, MQ/mQ and Mq/mq sire families will not bein LD because the QTL does not segregatein these families. When pooled acrossfamilies these four types of LD will canceleach other out, resulting in linkage equilibriumacross the population. Nevertheless,the within-family LD can be used to detectQTL and for MAS provided the differencesin linkage phase are taken into account, aswill be demonstrated later.mQmqQTL detection and types ofmarkers for MASApplication of molecular genetics forgenetic improvement relies on the abilityto genotype individuals for specific geneticloci. For these purposes, three types ofobservable genetic loci can be distinguished,as described by Dekkers, 2004:• direct markers: loci for which the functionalpolymorphism can be genotyped;• LD-markers: loci in population-wide LDwith the functional mutation;• LE-markers: loci in population-widelinkage equilibrium with the functionalmutation but which can be used for QTLdetection and MAS based on withinfamilyLD.For these alternate types of markers, differentstrategies are appropriate to detectQTL in livestock populations. These aresummarized in Table 1 and will be describedin more detail. Strategies for QTL detectionin livestock differ from those used in plantsbecause of the lack of inbred lines.QTL detection using LD markers withincrossesCrossing two breeds that differ in allele and,therefore, haplotype frequencies, createsextensive LD in the crossbred population.This LD extends over large distancesTable 1Summary of strategies for QTL detection in livestockType of population Within crosses Outbred populationF2/BackcrossAdvancedintercrossHalf- or full-sibfamiliesExtendedpedigreeNon-pedigreed populationsampleType of markers LD markers LE markers LD markersGenome coverage Genome-wide Genome-wide Candidate gene Genome-wideregionsMarker density Sparse Denser Sparse More dense Few loci DenseType of LD used Population-wide LD Within-family LD Population-wide LDNumber of generations ofrecombination used formapping1 >1 1 >1 >>1Extent of LD around QTL Long Smaller Long Smaller SmallMap resolution Poor Better Poor Better High

172Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishbecause it has undergone only one generationof recombination in the F 2 (Figure 1).Thus, although these markers may be inLE with QTL within the parental breeds,they will be in partial LD with the QTLin the crossbred population if the markerand QTL differ in frequency between thebreeds. This population-wide LD enablesdetection of QTL that differ between theparental breeds based on a genome scanwith only a limited number of markersspread over the genome (~ every 15 to20 cM). This approach has formed the basisfor the extensive use of F 2 or backcrossesbetween breeds or lines for QTL detection,in particular in pigs, poultry and beef cattle(see Andersson, 2001 for a review). Theextensive LD enables detection of QTLthat are some distance from the markersbut also limits the accuracy (map resolution)with which the position of the QTLcan be determined.More extensive population-wide LD isalso expected to exist in synthetic lines,i.e. lines that were created from a cross inrecent history. These can be set up on anexperimental basis through advanced intercrosslines (Darvasi and Soller, 1995) orbe available as commercial breeding lines.Depending on the number of generationssince the cross, the extent of LD will haveeroded over generations and will, therefore,span shorter distances than in F 2 populations(Figure 1). This will require a moredense marker map to scan the genome withequivalent power as in an F 2 but will enablemore precise positioning of the QTL.QTL detection using LE markers inoutbred populationsAs linkage phases between the markerand QTL can differ from family to family,use of within-family LD for QTL detectionrequires QTL effects to be fitted on awithin-family basis, rather than across thepopulation. Similar to F 2 or backcrosses,the extent of within-family LD is extensiveand, thus, genome-wide coverage is providedby a limited number of markers butsignificant markers may be some distancefrom the QTL, resulting in poor map resolution.Thus, LE markers can be readilydetected on a genome-wide basis usinglarge half-sib families, requiring only sparsemarker maps (~15 to 20 cM spacing). Manyexamples of successful applications of thismethodology for detection of QTL regionsare available in the literature, in particularfor dairy cattle, utilizing the large paternalhalf-sib structures that are available throughextensive use of artificial insemination (seeWeller, Chapter 12).QTL detection using LE markers canalso be applied to extended pedigrees bymodelling the co-segregation of markersand QTL (Fernando and Grossman, 1989).These approaches use statistical modelsthat are described further in the sectionon genetic evaluation using LE markers.Depending on the number of generationswith phenotypes and marker genotypesthat are included in the analysis, map resolutionwill be better than with analysis ofhalf-sib families because multiple rounds ofrecombination are included in the data set.QTL detection using LD markers inoutbred populationsThe amount and extent of LD that existsin the populations that are used for geneticimprovement are the net result of all forcesthat create and break down LD and are,therefore, the result of the breeding andselection history of each population, alongwith random sampling. On this basis, populationsthat have been closed for manygenerations are expected to be in linkageequilibrium, except for closely linked loci.

Chapter 10 – Strategies, limitations and opportunities for marker-assisted selection in livestock 173Thus, in those populations, only markersthat are tightly linked to QTL may show anassociation with phenotype (Figure 1), andeven then there is no guarantee because ofthe chance effects of random sampling.There are two strategies to find markersthat are in population-wide LD with QTL(see Table 1):• evaluating markers that are in, or closeto, genes that are thought to be associatedwith the trait of interest (candidategenes);• a genome scan using a high-densitymarker map, with a marker every 0.5 to2 cM.The success of both approaches obviouslydepends on the extent of LD in thepopulation. Studies in human populationshave generally found that LD extends overless than 1 cM. Thus, many markers areneeded to obtain sufficient marker coveragein human populations to enable detectionof QTL based on population-wide LD.Opportunities to utilize population-wideLD to detect QTL in livestock populationsmay be considerably greater because of theeffects of selection and inbreeding. Indeed,Farnir et al. (2000) identified substantialLD in the Dutch Holstein population,which extended over 5 cM. Similar resultshave been observed in other livestock species(e.g. in poultry, Heifetz et al., 2005).The presence of extensive LD in livestockpopulations is advantageous for QTLdetection, but disadvantageous for identifyingthe causative mutations of theseQTL; with extensive LD, markers that aresome distance from the causative mutationcan show an association with phenotype.The candidate gene approach utilizesknowledge from species that are rich ingenome information (e.g. human, mouse),effects of mutations in other species, previouslyidentified QTL regions, and/orknowledge of the physiological basis oftraits, to identify genes that are thoughtto play a role in the physiology of thetrait. Following mapping and identificationof polymorphisms within the gene,associations of genotype at the candidategene with phenotype can be estimated(Rothschild and Plastow, 1999).Whereas the candidate gene approachfocuses on LD within chosen regions of thegenome, recent advances in genome technologyhave enabled sequencing of entiregenomes, including of several livestock species;the genomes of the chicken and cattlehave been sequenced and public sequencingof the genome of the pig is under way.In addition, sequencing has been usedto identify large numbers of positions inthe genome that include SNPs, i.e. DNAbase positions that show variation. Forexample, in the chicken, over 2.8 millionSNPs were identified by comparing thesequence of the Red Jungle Fowl with thatof three domesticated breeds (InternationalChicken Polymorphism Map Consortium,2004). This, combined with reducing costsof genotyping, now enables detection ofQTL using LD-mapping with high-densitymarker maps.QTL detection using combined LDand linkage analysis in outbredpopulationsAs markers may not be in complete LDwith the QTL, both population-wideassociations of markers with QTL and cosegregationof markers and QTL withinfamilies can be used to detect QTL. Usingthese combined properties of being bothLD and LE markers, methods have beendeveloped to combine LD and linkageinformation. These methods are furtherexplored under genetic evaluation modelsin what follows.

174Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishIncorporating marker informationin genetic evaluation programmesThe value of genotypic information forpredicting the genetic merit of animals isdependent on the predictive ability of themarker genotypes. The three types of molecularloci described previously differ not onlyin methods of detection but also in methodsof their incorporation in genetic evaluationprocedures. Whereas direct and, to alesser degree, LD markers, allow selectionon genotype across the population, use ofLE markers must allow for different linkagephases between markers and QTL fromfamily to family, i.e. LE markers are familyspecific and family specific information mustbe derived. As discussed later in this chapter,this makes LE markers a lot less attractivefor use in breeding programmes. In thissection, the different types of models thathave been proposed for genetic evaluationbased on marker information are describedand this is followed by a brief description ofsome practical issues regarding implementationof such methods and the likely routestowards achieving that goal.Modelling QTL effects in geneticevaluationBy using QTL information in genetic evaluation,in principle, part of the assumedpolygenic variation is substituted by a separateeffect due to a genetic polymorphismat a known locus. This has the immediateeffect of having a much better handle onthe Mendelian sampling process, as phenotypicco-variance can be evaluated basedon specific genetic similarity rather thanon an average relationship. For example,on average two full sibs share 50 percentof their alleles, but at a specific locus it isnow possible to know whether these fullsibs carry exactly the same complete genotype(both paternal and maternal alleles arein common), or actually have a completelydifferent genotype. The actual degree ofsimilarity of full sibs at a QTL can thusvary between 0 and 1. This additional informationhelps to better evaluate the geneticmerit due to specific QTL, and to betterpredict offspring that do not yet have phenotypicmeasurements.A number of different approaches havebeen described to accommodate markerinformation in genetic evaluation. Roughly,these methods can be distinguishedthrough their modelling of the QTL effectand through the type of genetic markerinformation used. The QTL effect can bemodelled as random or fixed, while themolecular information comes from LE, LDor direct markers.With a fixed QTL model, regression ongenotype probabilities would be used ingenetic evaluation to account for the effectof QTL polymorphisms. In the simplestadditive QTL model, suitable for estimatingbreeding values, simple regressionscould be included on the probability of carryingthe favourable mutation. Regressioncan be on known genotypes (class variables),or probabilities can be derived forungenotyped animals in a general complexpedigree (Kinghorn, 1999). A fixed QTLmodel is sensible if few alleles are knownto be segregating, and where dominanceand/or epistasis are important. The modelalso assumes effects being the same acrossfamilies. The effects of various genotypescould be fitted separately, giving power toaccount for dominance and epistasis in caseof multiple QTL. For selection purposes,a fixed QTL effect, if additive, would beadded to the polygenic estimated breedingvalues (EBVs), similar to breed effects inacross-breed evaluations. The advantage ofa fixed QTL model is the limited number ofeffects that need to be fitted.

Chapter 10 – Strategies, limitations and opportunities for marker-assisted selection in livestock 175Alternatively, QTL effects could bemodelled as random effects, with eachindividual having a different QTL effect.Co-variances are based on the probabilityof QTL alleles being identical by descentrather than on numerator relationshipsas in the usual animal model with polygeniceffects. With full knowledge aboutsegregation, this would effectively fit allfounder alleles as different effects. Therandom QTL model was first describedby Fernando and Grossman (1989), wherefor each animal both the paternal and thematernal allele were fitted. Without loss ofinformation, these effects can be collapsedinto one genotypic effect for each animal(Pong-Wong et al., 2001). The randomQTL model makes no assumptions aboutnumber of alleles at a QTL and it automaticallyaccommodates possible interactioneffects of QTL with genetic background(families or lines). Therefore, the randomQTL model is less reliant on assumptionsabout homogeneity of QTL effects. Therandom QTL model is a natural extensionto the usual mixed model and seems thereforea logical way to incorporate genotypeinformation into an overall genetic evaluationsystem. These models result in EBVsfor QTL effects along with a polygenicEBV. The total EBV is the simple sum ofthese estimates. One of the main computationallimitations of this method, however,is the large number of equations that mustbe solved, which increases by two peranimal for each QTL that is fitted. Thus,the number of QTL regions that can beincorporated is limited.Genetic evaluation using direct markersWhen the genotype of an actual functionalmutation is available, no pedigree informationis needed to predict the genotypiceffect, as QTL genotypes are measureddirectly. When there is only a small numberof alleles, the number of specific genotypesis limited. In genetic evaluation, it wouldseem appropriate to treat the genotypeeffect as a fixed effect, i.e. the assumptionis that genotype differences are the samein different families and herds or flocks.Such assumptions might be reasonable for abi-allelic QTL model in a relatively homogeneouspopulation. Alternatively, randomQTL models could be used with differenteffects for different founder alleles, or evenQTL by environment interactions. In bothfixed and random QTL models, genotypeprobabilities can be derived for individualswith missing genotypes.Genetic evaluation using LE markersWhen the genotype test is not for the geneitself, but for a linked marker, QTL probabilitiesderived from marker genotypeswill be affected by the recombination ratebetween marker and QTL and by theextent of LD between the QTL and markeracross the population. If LD betweenthe QTL and a linked marker only existswithin families, marker effects or, at a minimum,marker-QTL linkage phase must bedetermined separately for each family. Thisrequires marker genotypes and phenotypeson family members. If linkage betweenthe marker and QTL is loose, phenotypicrecords must be from close relatives ofthe selection candidate because associationswill erode quickly through recombination.With progeny data, marker-QTL effectsor linkage phases can be determined basedon simple statistical tests that contrast themean phenotype of progeny that inheritedalternate marker alleles from the commonparent. A more comprehensive approach isbased on Fernando and Grossman’s (1989)random QTL model, where marker informationfrom complex pedigrees can be used

176Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishto derive co-variances between QTL effects,yielding best linear unbiased prediction(BLUP) of breeding value for both polygenicand QTL effects. Random effects ofpaternal and maternal QTL alleles are addedto the standard animal model with randompolygenic breeding values. The varianceco-variancestructure of the random QTLeffects, also known as the gametic relationshipmatrix (GRM), is based on probabilitiesof identity by descent (IBD), and is nowderived from co-segregation of markers andQTL within a family. Probabilities of IBDderived from pedigree and marker data linkQTL allele effects that are expected to beequal or similar, therefore using data fromrelatives to estimate an individual’s QTLeffects. For example, if two paternal halfsibsi and j have inherited the same paternalallele for markers that flank the QTL (withrecombination rate r), they are likely IBDfor the paternal QTL allele and the correlationbetween the effects of their paternalQTL alleles will be (1-r) 2 . The method isappealing, but computationally demandingfor large-scale evaluations, especially whennot all animals are genotyped and complexprocedures must be applied to derive IBDprobabilities.Genetic evaluation using LD markersMost QTL projects have moved towardsfine mapping where the final result is amarker or marker haplotype in LD withthe QTL, if not the direct mutation. Ahaplotype of marker alleles close enoughto the putative QTL is likely to be inLD with QTL alleles. Such a marker testprovides information about QTL genotypeacross families, and is in a sense notvery different from a direct marker. Themost convenient way to include genotypicinformation from marker haplotypesin genetic evaluation systems is throughthe random QTL model. In their originalpaper, Fernando and Grossman (1989)derived IBD from genotype data on singlemarkers and recombination rates betweenmarker and QTL. However, the randomQTL model is more versatile, and co-variancesbased on IBD probabilities can alsouse information beyond pedigree, based onLD. The latter can be derived from markeror haplotype similarity, e.g. based on anumber of marker genotypes surroundinga putative QTL. Meuwissen and Goddard(2001) proposed using both linkage and LDinformation to derive IBD-based co-variances(termed LDL analysis). Lee and vander Werf (2005) showed that with densermarkers, the value of linkage information,and therefore pedigree, reduces. Hence,when QTL positions become more accuratelydefined, genetic information fromclose markers (within a few cM) can beused increasingly to derive LD-based IBDprobabilities, thereby defining co-variancesbetween random QTL effects without theneed for a family structure or informationthrough pedigree.Lee and van der Werf (2006) have shownthat LD information results in a very denseGRM. Genetic evaluation, which is usuallybased on mixed model equations that arerelatively sparse, is currently not feasiblecomputationally for the LDL method fora large number of individuals and alternativemodels are needed. One approach isto model population-wide LD by simplyincluding the marker genotype or haplotypeas a fixed effect in the animal modelevaluation, as suggested by Fernando(2004). An advantage of modelling population-wideLD effects as fixed rather thanrandom is that fewer assumptions aboutpopulation history are needed. A disadvantageis that estimates are not “BLUPed”,i.e. regressed towards a mean depending on

Chapter 10 – Strategies, limitations and opportunities for marker-assisted selection in livestock 177the amount of information that is availableto estimate their effects. This will be importantif some of the genotype or haplotypeeffects cannot be estimated with substantialaccuracy because the number of individualswith that genotype or haplotype is limited.Haplotype effects could also be fitted asrandom, but more development is neededin this area.Whole genome approach for geneticevaluation using high-density LD markersWith more and more QTL being discovered,the polygenic component will slowly bereplaced by multiple QTL effects, the inheritanceof each being followed by markerbrackets or more generally by informationon haplotypes. Nejati-Javaremi, Smith andGibson (1997) presented the concept of thetotal allelic relationship, where the co-variancebetween two individuals was derivedfrom allelic identity by descent, or by state(based on molecular marker information),with each location weighted by the varianceexplained by that region. This approachcontrasts with the average relationshipsderived from pedigree that are used inthe numerator relationship matrix. Nejati-Javaremi, Smith and Gibson (1997) showedthat using total allelic relationship resultedin a higher selection response than pedigreebased relationships, because it moreaccurately accounts for the variation in theadditive genetic relationships between individuals.Therefore, the gain of followinginheritance at specific genome locationscontributes to more accurate genetic evaluation,and is able to deal more specificallywith within and between loci interactionsand with specific modes of inheritance atdifferent QTL.When large-scale marker genotypingbecomes cheap and available to breedersat low cost, this approach could even beused for non-detected QTL and geneticevaluation could be based on a “wholegenome approach” (Meuwissen, Hayesand Goddard, 2001). In this approach,marker haplotypes are fitted as independentrandom effects for each, e.g. 1 cM region ofthe genome. In the work by Meuwissen,Hayes and Goddard (2001), variances associatedwith each haplotype were eitherassumed to be equal for each chromosomalregion or estimated from the datausing Bayesian procedures with alternateprior distributions. In essence, this procedureestimates breeding values for eachhaplotype, and EBVs of individuals arecomputed by simply summing EBVs forthe haplotypes that they contain.Using this procedure, Meuwissen,Hayes and Goddard (2001) demonstratedthrough simulation, that for populationswith an effective population size of 100 anda spacing of 1 or 2 cM between informativemarkers across the genome, sufficient LDwas present to predict genetic values withsubstantial accuracy for several generationsbased on associations of marker haplotypeswith phenotype on as few as 500 individuals.It should be noted that, in the approachproposed by these authors, no polygeniceffect is included since all regions of thegenome are included in the model. It may,however, be useful to include a polygeniceffect because LD between markers andQTL will not be complete for all regions.In addition, this model assumes that haplotypeeffects are independent within andacross regions. Incorporating IBD probabilitiesto model co-variances betweenhaplotypes within a region as in Meuwissenand Goddard (2000), and by incorporatingco-variances between adjacent regionscaused by LD between regions, could leadto further improvements but would alsolead to increasing computational demands.

178Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishIn general, for the purpose of increasedgenetic change of economically importantquantitative traits, and in the contextof well recorded and efficient breedingprogrammes, there is no need to haveknowledge of functional mutations sincenearby markers will have a high predictivevalue about genetic merit. Moreover,the benefit from the extra investment andtime spent on finding functional mutationsmight be superseded by the geneticchange that can be made in the breedingprogramme in the meantime.Implementation of marker-assisted geneticevaluationIt is important to note that, for most ofthe gene marker tests currently on themarket, integration with existing systemsfor genetic evaluation is not obvious. Thisis because the gene testing is either fora Mendelian characteristic, or it predictsphenotypic differences for traits that arenot the same as those in current geneticevaluation. Moreover, breeders would notonly be interested in more accurate EBVsbased on gene markers, but they wouldalso want to know the actual QTL genotypesfor their breeding animals. Thisinformation on individual genotype willbecome less relevant if more gene testsbecome available and if testing becomescheaper and more widespread. This mightstill take some years. Thus, as gene markertesting is gradually introduced, it is morelikely to create additional selection criteriato consider and it will take sometime before QTL information is seamlesslyand optimally integrated in existing geneticevaluation programmes. In particular, ifgenetic evaluation is based on informationfrom many different breeding units, suchas in cattle or sheep, genotyping informationwill initially be available for only asmall proportion of the breeding animals,possibly not justifying a total overhaul ofthe system for genetic evaluation. Simplead hoc procedures where QTL effects areestimated and presented separately as additionaleffects are initially a more likelyroute to implementation.Solutions for fixed QTL genotypeeffects, along with genotype probabilitiesas outputs of genetic evaluation, mightbe interesting to breeders and, comparedwith random QTL effects, may be morelikely to be presented and used separatelyfrom polygenic EBVs. This would alsobe the case for genotypic information onMendelian characters, where there is nopolygenic component.Incorporating MAS in selectionprogrammesMolecular information can be used toenhance both the processes of integratingsuperior qualities of different breeds andwithin-breed selection. These strategies arefurther described below.Between-breed selectionCrossing breeds results in extensive LD,which can be capitalized upon using MASin a number of ways. If a large proportionof breed differences in the trait(s) ofinterest are due to a small number of genes,gene introgression strategies can be used.If a larger number of genes is involved,MAS within a synthetic line is the preferredmethod of improvement.Marker-assisted introgressionIntrogression of the desirable allele at atarget gene from a donor to a recipient breedis accomplished by multiple backcrossesto the recipient, followed by one or moregenerations of intercrossing. The aim ofthe backcross generations is to produce

Chapter 10 – Strategies, limitations and opportunities for marker-assisted selection in livestock 179individuals that carry one copy of the donorQTL allele but that are similar to the recipientbreed for the rest of the genome. The aim ofthe intercrossing phase is to fix the donorallele at the QTL. Marker information canenhance the effectiveness of the backcrossingphase of gene introgression strategies by: (i)identifying carriers of the target gene(s)(foreground selection); and (ii) enhancingrecovery of the recipient genetic background(background selection). The effectiveness ofthe intercrossing phase can also be enhancedthrough foreground selection on thetarget gene(s). If the target gene cannot begenotyped directly, carrier individuals canbe identified based on markers that flankthe QTL at

180Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishbetween lines or breeds. Thus, marker-QTLassociations identified in the F 2 generationcan be selected for several generations, untilthe QTL or markers are fixed or the disequilibriumdisappears. Zhang and Smith(1992) evaluated the use of markers in sucha situation with selection on BLUP EBV.Although both studies considered the idealsituation of a cross with inbred lines, therewill be opportunities to utilize a limitednumber of markers to select for favourableQTL regions that are detected incrosses between breeds, thereby enhancingthe development of superior synthetics.Pyasatian, Fernando and Dekkers (2006)investigated use of the whole genomeapproach of Meuwissen, Hayes andGoddard (2001) for MAS in a cross byincluding all markers as random effectsin the model for genetic evaluation. Theyshowed that this resulted in substantiallygreater responses to selection than selectionon identified QTL regions only. Due to themuch greater LD, whole genome selectionin a cross can be accomplished with a muchsmaller number of markers compared withthe number required for whole genomeselection in an outbred population.Within-breed selectionThe procedures described previously forincorporating markers in genetic evaluationresult in estimates of breeding values associatedfor QTL, together with estimates ofpolygenic breeding values. Alternatively, ifmolecular data are not incorporated intogenetic evaluations, as will be the casefor more ad hoc approaches and for genetests for Mendelian characteristics, separateselection criteria will be available thatcapture the molecular information. The followingthree selection strategies can then bedistinguished (Dekkers, 2004):• select on the QTL information alone;• tandem selection, with selection on QTLfollowed by selection on polygenic EBV;• selection on the sum of the QTL andpolygenic EBV.Selection on QTL or marker informationalone ignores information that isavailable on all other genes (polygenes)that affect the trait and is expected to resultin the lowest response to selection unlessall genes that affect the trait are includedin the QTL EBV. This strategy does not,however, require additional phenotypesother than those that are needed to estimatemarker effects, and can be attractivewhen phenotype is difficult or expensiveto record (e.g. disease traits, meat quality,etc.). Selection on the sum of the QTL andpolygenic EBV is expected to result in maximumresponse in the short term, but maybe suboptimal in the longer term becauseof losses in polygenic response (Gibson,1994). Indexes of QTL and polygenic EBVcan be derived that maximize longer-termresponse (Dekkers and van Arendonk,1998) or a combination of short- and longertermresponses (Dekkers and Chakraborty,2001). However, if selection is on multipleQTL and emphasis is on maximizingshorter-term response, selection on the sumof QTL and polygenic EBV is expected tobe close to optimal. Optimizing selectionon a number of EBVs, indexes and genotypes,while also considering inbreedingrate and other practical considerations isnot a trivial task. Kinghorn, Meszaros andVagg (2002) have proposed a mate selectionapproach that could be used to handlesuch problems, and it can be expected thatwith more widespread use of genotypicinformation for a larger number of regions,specific knowledge about individual QTLbecomes less interesting and will simplycontribute to prediction of whole EBV orwhole genotype.

Chapter 10 – Strategies, limitations and opportunities for marker-assisted selection in livestock 181Meuwissen and Goddard (1996) publisheda simulation study that looked at themain characteristics determining efficiencyof MAS using LE markers. They foundthat MAS could improve the rate of geneticimprovement up to 64 percent by selectingon the sum of QTL and polygenic EBV.Their work also demonstrated that MAS ismainly useful for traits where phenotypicmeasurement is less valuable because of:(i) low heritability; (ii) sex-limited expression;(iii) availability only after sexualmaturity; and (iv) necessity to sacrifice theanimal (e.g. slaughter traits). Selection ofanimals based on (most probable) QTLgenotype will allow earlier and more accurateselection, increasing the short- andmedium-term selection response.Most simulation studies have assumedcomplete marker genotype information butin practice only a limited number of individualswill be genotyped. However, inan advanced breeding programme withcomplete information on phenotype andpedigree information, marker and QTLgenotype probabilities could be derivedfor un-genotyped animals and genotypingstrategies could be optimized to achievea high value for the investments made.Marshall, Henshall and van der Werf (2002)looked at strategies to minimize genotypingcost in a sheep breeding programme. Closeto maximal gain could be achieved whengenotyping was undertaken only for highranking males and animals whose markergenotype probability could not be derivedwith enough certainty based on informationon relatives. Marshall, van der Werfand Henshall (2004) also looked at progenytesting of sires to determine family-specificmarker-QTL phase within a breedingnucleus. Again, testing of a limited numberof males provided a lot of information aboutphase for several generations of breedinganimals, as progeny tested sires have relationshipswith descendants. However, inbreeding programmes for more extensiveproduction systems (beef, sheep), pedigreerecording is often incomplete and only asmall proportion of animals are genotyped.Moreover, these genotyped animals are notnecessarily the key breeding animals. Theutility of linked markers will be even morelimited if pedigree relationships cannotbe used to resolve genotype probabilitiesand marker-QTL phase of un-genotypedindividuals.A second point of caution is that manystudies on MAS have taken a single-traitapproach and shown that genetic markerscould have a large impact on responses fortraits that are difficult to improve by phenotypicselection. However, within thecontext of a multitrait breeding objective,the overall impact of such markers on thebreeding goal may be less because a greaterresponse for one trait often appears at theexpense of another. For example, geneticmarkers for carcass traits improve the abilityto select (i.e. earlier, with higher accuracy)for such traits, but selection emphasisfor other traits is reduced. Therefore, theoverall effect of MAS on the breeding programmewill generally be much smallerthan predicted for single trait MAS-favourablecases. The main effects of MAS wouldbe to shift the selection response in favourof the marked traits, rather than achievingmuch additional overall response. Hence,while it will be easier to select for carcassand disease resistance, further improvementfor these traits will be at the expenseof genetic change for production traits(growth, milk).The impact of MAS on the rate ofgenetic gain may be limited in conventionalbreeding programmes (ranging upto perhaps 10 percent extra gain) unless

182Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishthe variation in profitability is dominatedby traits that are hard to measure.However, new technologies often lead toother breeding programme designs beingcloser to optimal. Genotypic informationhas extra value in the case of early selectionand where within-family variance can beexploited, which is particularly the case inprogrammes where reproductive technologiesare used. Reproductive technologiesusually lead to early selection and moreemphasis on between-family selection.DNA marker technology and reproductivetechnologies are therefore highly synergisticand complementary (van der Werfand Marshall, 2005) and gene markers havemuch more value in such programmes.Gene marker information is also clearlyvaluable in introgression programmes, asdemonstrated by simulation (Chaiwong etal., 2002; Dominik et al., 2006) as well as inpractice (Nimbkar, Pardeshi and Ghalsasi,2005). Yet, although these examples arefavourable to the value of gene markerinformation, the added value of MAS stillrelies heavily on a high degree of trait andpedigree recording.Opportunities for MAS indeveloping countriesComplete phenotypic and pedigree informationis often only available in intensivebreeding units. Therefore, in the context oflow input production systems, some questionscan be raised concerning the validityand practicality of the simulation studiesdescribed above, and it would be moredifficult to realize the value of markerinformation. It would be harder and moreexpensive to determine the linkage phase inthe case of using linked markers. Moreover,even if the genetic marker were a direct orLD marker, its effect on phenotype wouldhave to be estimated for the populationand the environment in which it is used.This would require phenotypes and genotypeson a sample of a rather homogeneouspopulation to avoid spurious associationsthat could result from unknown populationstratification. Therefore, a gene markerfor a QTL is likely to be most successfulin an environment with intensive pedigreeand performance recording. Nevertheless,in low input environments, direct andLD markers will be more useful than LEmarkers because the latter require routinerecording of phenotypes and genotypes toestimate QTL effects within families.In addition to MAS within local breeds,several other strategies for breed improvementcould be pursued in developingcountries, including gene introgression andMAS within synthetic breeds. This wouldbe most advantageous for introducing specificdisease resistance alleles into breedswith improved production characteristicsto make them more tolerant to the environmentsencountered in developing countries.Gene introgression is, however, a long andexpensive process and only worthwhilefor genes with large effects. MAS withinsynthetic breeds, e.g. a cross between localand improved temperate climate breeds,can allow development of a breed that isbased on the best of both breeds (e.g. Zhangand Smith, 1992). Because of the extensiveLD within the cross, a limited number ofmarkers would be needed. Care should,however, be taken to avoid the impactof genotype x environment interactions ifMAS is implemented in a more controlledenvironment.

Chapter 10 – Strategies, limitations and opportunities for marker-assisted selection in livestock 183ReferencesAndersson, L. 2001. Genetic dissection of phenotypic diversity in farm animals. Nature Revs. Genet.2: 130–138.Chaiwong, N., Dekkers, J.C.M., Fernando, R.L. & Rothschild, M.F. 2002. Introgressing multipleQTL in backcross breeding programs of limited size. Proc. 7 th Wld. Congr. Genet. Appl. Livest.Prodn. Electronic Communication No. 22: 08. Montpellier, France.Darvasi, A. & Soller, M. 1995. Advanced intercross lines: an experimental population for fine geneticmapping. Genetics 141: 1199–1207.Dekkers, J.C.M. & van Arendonk, J.A.M. 1998. Optimizing selection for quantitative traits withinformation on an identified locus in outbred populations, Genet. Res. 71: 257–275.Dekkers, J.C.M. 2004. Commercial application of marker- and gene-assisted selection in livestock:strategies and lessons. J. Anim. Sci. 82: E313–E328.Dekkers J.C.M. & Chakraborty, R. 2001. Potential gain from optimizing multi-generation selectionon an identified quantitative trait locus. J. Anim. Sci. 79: 2975–2990.Dominik, S., Henshall, J., O’Grady, J. & Marshall, K.J. 2007. Factors influencing the efficiency of amarker assisted introgression program in merino sheep. Genet. Sel. Evol. In press.Farnir, F., Coppieters, W., Arranz, J.-J., Berzi, P., Cambisano, N., Grisart, B., Karim, L., Marcq,F., Moreau, L., Mni, M., Nezer, C., Simon, P., Vanmanshoven, P., Wagenaar, D. & Georges, M.2000. Extensive genome-wide linkage disequilibrium in cattle. Genome Res. 10: 220–227.Fernando, R.L. 2004. Incorporating molecular markers into genetic evaluation. Session G6.1. Proc.55th Meeting of the European Association of Animal Production. 5–9 September 2004, Bled,Slovenia.Fernando, R.L. & Grossman, M. 1989. Marker-assisted selection using best linear unbiased prediction.Genet. Sel. Evol. 21: 467–477.Gibson, J.P. 1994. Short-term gain at the expense of long-term response with selection of identifiedloci, Proc. 5 th Wld. Cong. Genet. Appl. Livest. Prodn. CD-ROM Communication No. 21: 201–204.University of Guelph, Canada.Goddard, M.E. & Meuwissen, T.H.E. 2005. The use of linkage disequilibrium to map quantitativetrait loci. Austr. J. Exp. Agric. 45: 837–845.Heifetz, E.M., Fulton, J. E., O’Sullivan, N., Zhao, H., Dekkers, J.C.M. & Soller, M. 2005. Extentand consistency across generations of linkage disequilibrium in commercial layer chicken breedingpopulations. Genetics 171: 1173–1181.Hospital, F. & Charcosset, A. 1997. Marker-assisted introgression of quantitative trait loci. Genetics147: 1469–1485.International Chicken Polymorphism Map Consortium. 2004. A genetic variation map for chickenwith 2.8 million single nucleotide polymorphisms. Nature 432: 717–722.Kinghorn, B.P. 1999. Use of segregation analysis to reduce genotyping costs. J. Anim. Breed. Genet.116: 175–180.Kinghorn, B.P., Meszaros, S.A. & Vagg, R.D. 2002. Dynamic tactical decision systems foranimal breeding, Proc. 7 th Wld. Congr. Genet. Appl. Livest. Prodn. Communication No. 23-07.Montpellier, France.Koudandé, O.D., Iraqi, F., Thomson, P.C., Teale, A.J. & van Arendonk, J.A.M. 2000. Strategiesto optimize marker-assisted introgression of multiple unlinked QTL. Mammal. Genome 11:145–150.

184Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishLande, R. & Thompson, R. 1990. Efficiency of marker-assisted selection in the improvement ofquantitative traits. Genetics 124: 743–56.Lee, S.H. & van der Werf, J.H.J. 2005. The role of pedigree information in combined linkage disequilibriumand linkage mapping of quantitative trait loci in a general complex pedigree. Genetics169: 455–466.Lee, S.H. & van der Werf, J.H.J. 2006. An efficient variance component approach implementing anaverage information REML suitable for combined LD and linkage mapping with a general complexpedigree. Genet. Sel. Evol. 38: 25–43.Marshall, K., Henshall, J. & van der Werf, J.H.J. 2002. Response from marker-assisted selectionwhen various proportions of animals are marker typed: a multiple trait simulation study relevantto the sheep meat industry. Anim. Sci. 74: 223–232.Marshall, K.J., van der Werf J.H.J. & Henshall, J. 2004. Exploring major gene - marker phase typingstrategies in marker assisted selection schemes. Anim. Sci. 78: 213–227.Meuwissen, T.H.E. & Goddard, M.E. 1996. The use of marker haplotypes in animal breedingschemes. Genet. Sel. Evol. 28: 161–176.Meuwissen, T.H.E. & Goddard, M.E. 2000. Fine mapping of quantitative trait loci using linkage disequilibriawith closely linked marker loci. Genetics 155: 421–430.Meuwissen, T.H.E. & Goddard, M.E. 2001. Prediction of identity by descent probabilities frommarker haplotypes. Genet. Sel. Evol. 33: 605–634.Meuwissen, T.H.E., Hayes, B. & Goddard, M.E. 2001. Prediction of total genetic value usinggenome-wide dense marker maps. Genetics 157: 1819–1829.Nejati-Javaremi, A., Smith, C. & Gibson, J.P. 1997. Effect of total allelic relationship on accuracyand response to selection. J. Anim. Sci. 75: 1738–1745.Nimbkar, C., Pardeshi, V. & Ghalsasi, P. 2005. Evaluation of the utility of the FecB gene to improvethe productivity of Deccani sheep in Maharashtra, India. pp. 145–154. In H.P.S. Makkar & G.J.Viljoen, eds. Applications of gene-based technologies for improving animal production and health indeveloping countries. Netherlands, Springer.Pong-Wong, R., George, A.W., Woolliams, J.A. & Haley, C.S. 2001. A simple and rapid method forcalculating identity-by-descent matrices using multiple markers. Genet. Sel. Evol. 33: 453–471.Pyasatian, N., Fernando, R.L. & Dekkers, J.C.M. 2006. Genomic selection for composite line developmentusing low density marker maps. In Proc. 8 th Wrld. Congr. Genet. Appl. Livest. Prodn.,Paper 22–65. Belo Horizonte, Brazil.Rothschild, M.F. & Plastow, G.S. 1999. Advances in pig genomics and industry applications.AgBiotechNet. 1: 1–7.van der Werf, J.H.J. & Marshall, K. 2005. Combining gene-based methods and reproductive technologiesto enhance genetic improvement of livestock in developing countries, pp. 131–144. InH.P.S Makkar & G.J. Viljoen, eds. Applications of gene-based technologies for improving animalproduction and health in developing countries. Netherlands, Springer.Zhang, W. & Smith, C. 1992. Computer simulation of marker-assisted selection utilizing linkage disequilibrium.Theor. Appl. Genet. 83: 813–820.

Chapter 11Marker-assisted selectionin poultryDirk-Jan de Koning and Paul M. Hocking

186Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishSummaryAmong livestock species, chicken has the most extensive genomics toolbox available fordetection of quantitative trait loci (QTL) and marker-assisted selection (MAS). The uptakeof MAS is therefore not limited by technical resources but mostly by the priorities andfinancial constraints of the few remaining poultry breeding companies. With the cost ofgenotyping decreasing rapidly, an increase in the use of direct trait- single nucleotide polymorphism(SNP)-associations in MAS can be predicted.

Chapter 11 – Marker-assisted selection in poultry 187Current status of chickenbreeding programmesPoultry production has been the fastestgrowing livestock industry over the lastdecades especially in middle- and lowincomecountries (Taha, 2003). In 2001,poultry production accounted for 70 milliontonnes of poultry meat and 47 milliontonnes of eggs (Arthur and Albers, 2003).Among poultry, chicken account for 85 percentof meat production and 96 percent ofegg production (Bilgili, 2001; Arthur andAlbers, 2003; Taha, 2003). While chickenshave been domesticated and selectedfor thousands of years, modern poultrybreeding started during the 1950s. One ofthe most notable features is the diversificationbetween chickens bred for meatproduction (broilers) and those bred fortable egg production (layers). This is aresult of the negative genetic correlation inchicken between growth and reproductivetraits. Within breeds, there is a separationinto male and female lines that are crossedto produce commercial hybrids. In broilers,male lines are selected for growth and carcassquality whereas in female lines lessemphasis is placed on growth and moreon reproductive traits such as egg productionand hatchability. In table egg-layingchickens, male lines are selected for high eggproduction and high egg weight whereas infemale lines selection may emphasize rateof lay with less attention to egg size. Inboth broiler and layer lines the primaryselection goal is the improvement of feedefficiency and economic gain.Significant heterosis for fitness traits inpoultry is well established and all commercialpoultry (chickens, turkeys and ducks) arehybrids that are produced in a selection andmultiplication pyramid that is illustrated inFigure 1. Crossing male and female linesmaximizes heterosis at the grandparentand parent levels of the hierarchy, andallows traits that have been geneticallyimproved in different lines to be combinedin the commercial birds. The power of thisstructure to deliver large economic gains inchickens is a result of their high reproductiverate and short generation interval and isclearly illustrated by this example of anegg-laying improvement programme. Evengreater numerical efficiency is possiblein broilers: a single pen containing tenfemales and one male at the nucleus levelmight produce 150 great-grandparentsafter selection (line D of Figure 1); thesewill produce 50 female offspring each or7 500 grandparents in a year and thesegrandparents will generate 375 000 femaleparent stock during the succeeding year.These hybrid parent females will eachproduce over 130 male and female offspringand generate nearly 50 million commercialbroilers or 70 000 tonnes of meat. Thefigure illustrates the rapidity with whichgenetic improvement at the nucleus levelcan be disseminated to commercial flocksand the fact that relatively few pure-linebirds are needed to produce very largenumbers of commercial layers.The existence of this breeding structureresults in rapid transmission of geneticchange to commercial flocks (about fouryears), including traits that might beimproved by MAS. Conversely, undesirablegenetic change can also be disseminatedvery quickly to a very large number ofbirds. In practice, far more birds are keptat the nucleus level than shown in Figure 1where the numbers presented are purely forillustrative purposes.Status of functional genomics inchickenAmong the various livestock species, chickenhas the most comprehensive genomic tool-

188Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishFigure 1Multiplication pyramid for a four-strain commercial hybrid layerGenerationMale linesFemale linesTimePure lineselectionA1 ♂x 10 ♀B10 ♂x 100 ♀C10 ♂x 100♀D100 ♂x 1 000♀Year 0GrandparentsA200 ♂XB2 000 ♀C2 000 ♂XD20 000 ♀Year 1ParentsAB40 000 ♂XCD400 000 ♀Year 2CommercialhybridsABCD32.2 millionYears3 to 4The boxes in bold are the multiplication (crossing) phase. The top line of the boxes indicates the lowest level of the pure-line(nucleus) selection stage. The numbers in the boxes refer to the minimal numbers of birds in line A, and corresponding numbersin lines B, C and D, that are required to generate hybrid laying hens at the commercial level and the numbers that result fromthis process. The time scale is indicated on the right hand side of the figure assuming a generation interval of one year.Adapted from Bowman, 1974.box. The chicken genome consists of 39pairs of chromosomes: eight cytologicallydistinct macrochromosomes, the sexchromosomes Z and W and 30 pairs ofcytologically indistinguishable microchromosomes.Linkage maps were developedinitially using three separate mappingpopulations (Bumstead and Palyga, 1992;Crittenden et al., 1993; Groenen et al., 1998)that were later merged to provide a consensusmap with 1 889 markers (Groenenet al., 2000). A good overview of the consensuslinkage map and the cytogeneticmap can be found in the First Report ofChicken Genes and Chromosomes 2000 andits successor in 2005 (Schmid et al., 2000,2005). All chicken maps can be viewed atwww.thearkdb.org.More recently, the chicken genomebecame the first livestock genome to besequenced with a six-fold coverage (sixfull genome equivalents) (Hillier et al.,2004). The chicken genome sequence can bebrowsed via a number of Web sites, whichare summarized at www.chicken-genome.org/resources/databases.html. The genomesequence effort was accompanied by partialsequencing of three distinct poultry breeds(a broiler, a layer and a Chinese Silky), toidentify SNPs between and among theseand the reference sequence of the RedJungle Fowl. This resulted in an SNP mapconsisting of about 2.8 million SNPs (Wonget al., 2004). The chicken polymorphismdatabase (ChickVD) can be browsed at:http://chicken.genomics.org.cn/index.jsp

Chapter 11 – Marker-assisted selection in poultry 189(Wang et al., 2005). The SNP map willfacilitate the development of genome-wideSNP assays, containing between 5 000 and20 000 SNPs per assay.For the study of gene expression, thereare various complementary DNA (cDNA)microarrays available, varying from targetedarrays (immune, neuroendocrine,embryo) to whole genome generic arrays.Recently, a whole-genome Affymetrixchip was developed in collaboration withthe chicken genomics community (www.affymetrix.com and www.chicken-genome.org/resources/affymetrix-faq1.htm).Altogether, this provides a very comprehensivetoolbox to study the functionalgenomics of chicken, whether this be anindividual gene or the entire genome.Current uptake of MAS in chickenImplementation of MAS requires knowledgeof marker-trait associations basedon QTL and candidate gene studies, andideally from studies of the underlyinggenetic mechanisms. There have been alarge number of QTL studies in chickencovering a wide range of traits includinggrowth, meat quality, egg production, diseaseresistance (both infectious diseasesand production diseases) and behaviour.These studies have recently been reviewed(Hocking, 2005). A total of 27 papersreported 114 genome-wide significant QTLfrom experimental crosses largely involvingWhite Leghorn and broiler lines. A summaryof the QTL that have been detected ispresented in Table 1. While the abundanceof QTL would indicate ample opportunityfor MAS in chicken, it must be notedthat nearly all studies were carried out inexperimental crosses and hence the resultsdo not reflect QTL within selected populations.However, these results do providea good starting point to search for QTLwithin commercial populations, as demonstratedfor growth and carcass traits wheremany published QTL also explained variationwithin a broiler dam line (de Koninget al., 2003; de Koning et al., 2004). To theauthors’ knowledge, there are no otherQTL studies within commercial lines ofpoultry in the public domain. Of the QTLfrom experimental crosses, only a smallnumber has been followed up by fine mappinganalyses and the responsible genemutation has only been described for somedisease resistance QTL (Liu et al., 2001a, b;Liu et al., 2003).A good example of how QTL mappingcombined with functional studies canidentify functional variants is for Marek’sdisease. Marek’s disease (MD) is an infec-Table 1Quantitative traits and chromosomal locations in experimental chicken crossesTrait Chromosome (number of QTL) Total QTL Number of papersBehaviour/fear 1,2(3),3,4(2),7,10,27, E22 11 5Body fat 1(2),3,5,7(2),15,28 8 3Body weight 1(7),2(4),3(4),4(5),5,8(2),11,12,13(2),27(3)Z(2) 32 9Carcass quality 1(2),2,3,4(2).5(2),6(2),7(3),8(2),9,13(2),27,Z(2) 21 1Disease resistance 1(4),2(2),3(2),4(2),5(5),6(2),7,8,14,18,27,Z 23 10Egg number 8,Z(2) 3 1Egg quality 2,11,Z 3 2Egg weight 1,2,3,4(3),14,23,Z 9 3Feed intake 1,4 2 2Sexual maturity Z(2) 2 2Source: Hocking, 2005.

190Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishtious viral disease caused by a memberof the herpes virus family and costs thepoultry industry about US$1 000 millionper annum. An F 2 cross between resistantand susceptible lines was challenged experimentallyand genotyped, providing the datafor a QTL analysis that resulted in a totalof seven QTL for susceptibility to MD(Vallejo et al., 1998; Yonash et al., 1999).Subsequently, the founder lines of the F 2cross were used for a micro-array studyto identify genes that were differentiallyexpressed between the two lines followingartificial infection. Fifteen of these geneswere mapped onto the chicken genomeand two of them mapped to a QTL regionfor resistance to MD (Liu et al., 2001a). Atthe same time, protein interaction studiesbetween a viral protein (SORF2) and achicken splenic cDNA library revealed aninteraction with the chicken growth hormone(GH) (Liu et al., 2001b). This ledto the detection of a polymorphism in theGH gene that was associated with differencesin the number of tumours betweenthe susceptible and the resistant line (Liu etal., 2001b). GH coincided with a QTL forresistance and was differentially expressedbetween founder lines (Liu et al., 2001a).Alongside the various genome scans forQTL, a large number of candidate genestudies have been carried out. The majorityof studies summarized in Table 2 havebeen conducted on White Leghorn strainsand have utilized restriction fragmentlength polymorphisms (RFLPs), SNPs orsingle strand conformation polymorphisms(SSCPs). These techniques require boththat the gene is known and that the experimenteris able to sequence part of the geneto detect polymorphisms that distinguishthe experimental lines.Candidate gene studies have been usedin two ways. First, candidate genes maybe used merely as a marker for a trait(typically disease) based on prior knowledgeand, second, and much less often,to search for the mutation within a genethat is associated with phenotypic variationin a trait. Currently, potential (candidate)genes for a QTL may be obtained froma knowledge of physiology (Dunn et al.,2004) or comparative linkage maps (i.e.locating genes that are in the location of theQTL based on common areas of the generichgenomes of different species, usuallyhuman and mouse). There are likely to bemany more of the second type of candidategene studies as information from large-scalegene expression and proteomic experimentsbegin to suggest novel gene candidates fortraits of commercial and biological importance.It should also be noted that thereis good evidence that genetic variation isnot limited to genomic DNA: associationsbetween polymorphisms in mitochondrialgenes and MD resistance, body weight andegg shell quality were reported by Li et al.(1998a, b).Despite great enthusiasm for breedingcompanies to be involved in functionalgenomics research in poultry, there arevery few applications of MAS in commercialpoultry breeding. One existingexample is the use of blood group markersto improve resistance to MD where selectionof haplotypes B 21 and B 12 based onconventional serological tests has beenwidely used (McKay, 1998). In discussionswith the industry it is clear that mostinterest is in QTL or candidate genes forresistance to diseases like MD or ascites, agenetic condition associated with pulmonaryhypertension, leading to mortalityin fast growing birds. There is also considerableinterest among breeders of layerlines for egg quality, especially egg shellquality because of its importance for food

Chapter 11 – Marker-assisted selection in poultry 191Table 2Association of candidate genes with quantitative traits in poultryTrait Chromosomes 1 Gene symbols ReferencesAge at first egg 1,2,3 GH, NPY, ODC Feng et al., 1997; Dunn et al., 2004;Parsanejad et al., 2004Disease resistance (E. coli) 16 MHC1, MHC4, TAP2 Yonash et al., 1999Disease resistance (MD 2 ) 1,NK GH, LY6E Kuhnlein et al., 1997; Liu et al., 2001a,band 2003Disease resistance (Sal 3 ) 4,6,7,16,19,1,17,NKTNC, PSAP, NRAMP1 4 ,MHC1,CASP1, IAP1, TLR4,TLR5Double yolked eggs 10 GNRHR Dunn et al., 2004Hu et al., 1997; Lamont et al., 2002;Leveque et al. 2003; Liu and Lamont,2003; Iqbal et al., 2005Egg production Z,1,20 GHR, GH, PEPCK Feng et al., 1997; Kuhnlein et al., 1997;Parsanejad et al., 2003Egg weight 1 IGF1 Nagaraja et al., 2000Eggshell quality 1,3,20 IGF1, ODC, PEPCK Nagaraja et al., 2000; Parsanejad et al.,2003, 2004Body fat 1,1,5,Z GH, IGF1, TGFβ3, GHR Feng et al., 1998; Fotouhi et al., 1993; Liet al., 2003; Zhou et al., 2005Feed efficiency 3,20 ODC, PEPCK Parsanejad et al., 2003 and 2004Body weight/carcass quality 1,3,5,Z,1,1,1 IGF1, ODC, TGFβ3, GHR,APOA2, PIT1Organ weight (spleen) 3,5,32 TGFβ2, TGFβ3, TGFβ4 5 Li et al., 2003Feng et al., 1998; Li et al., 2003; Jiang etal., 2004; Parsanejad et al., 2004; Li et al.,2005; Zhou et al., 2005Skeletal traits 1,3,5,32 IGF1, TGFβ2, TGFβ3, TGFβ4 5 Li et al., 2003; Zhou et al., 20051NK = gene has not yet been assigned to a chromosome.2Marek’s Disease.3Salmonellosis.4Now known as Slc11a1.5TGFβ4 in the paper is now known to be TGFβ1.safety. For production traits such as growthand egg numbers, breeders make sufficientprogress using traditional selectionmethods, and they expect little improvementfrom MAS for such traits unlessmarkers can be used to increase the accuracyof selection. Nonetheless, among breedersof broiler stock there is interest in markersfor traits that are difficult to measure suchas feed efficiency and meat quality in additionto disease resistance.Potential for MAS in chickenThe technical aspects and potentialimplications of implementing MAS inlivestock are discussed in Chapter 10and Dekkers (2004), and van der Beekand van Arendonk (1996) evaluated thetechnical aspects of MAS in poultrybreeding. A review of the potential of MASin poultry is provided by Muir (2003) butthis includes many of the technical issuesthat are common across livestock species.This chapter therefore focuses on poultryspecificissues, and readers are referredto Chapter 10 or Muir (2003) for a morecomprehensive overview of applicationsand limitations of MAS.Muir (2003) identified two cases whereMAS could increase the selection intensityin poultry breeding: (i) traits that are measuredlater in life or are costly to measure(such as egg production and feed efficiencyfor broiler breeders); and (ii) selectionwithin full-sib families for sex-limitedtraits (e.g. male chicks for egg production).Accuracy of selection can also be improvedvia MAS when selecting between full-sibfamilies for sex-limited traits and traits thatcannot be measured directly on one or both

192Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishsexes and/or have a low heritability (e.g.egg production, disease resistance, carcassquality and welfare traits).Limiting factors for application ofMAS (Muir, 2003) include biological factors(reproductive capacity) and manytheoretical considerations related to theeffectiveness of MAS (e.g. diverting selectionpressure from polygenes to a singlemarked gene), which are generally applicableto MAS in livestock (Dekkers, 2004;Chapter 10). One of the concerns of Muir(2003) is the expected lack of major QTLfor traits that have been under selectionfor many generations (following simulationresults). However, recent QTL studieswithin commercial lines of pigs (Evanset al., 2003; Nagamine et al., 2003) andpoultry (de Koning et al., 2003, 2004) havedemonstrated that many sizeable QTL arestill segregating in commercial populationsdespite decades of selection.There is strong academic interest inchicken genomics outside agriculture from,among others, developmental biologists andevolutionary geneticists, and this has contributedgreatly to the development of thecurrent functional genomics toolbox availablefor chicken. Among livestock species,chickens are best placed to pioneer newapproaches where QTL studies are complementedby gene expression studies (Liuet al., 2001a) or where they become fullyintegrated within “genetical genomics” (deKoning, Carlborg and Haley, 2005; deKoning and Haley, 2005).If poultry breeders decide to embraceMAS, one of the main questions iswhether they are prepared to re-structuretheir breeding programmes aroundMAS or implement these around theircurrent breeding strategies. Adopting theterminology of Dekkers (2004), there arethree levels of MAS: gene-assisted selection(GAS) where the functional mutation andits effects are known; linkage disequilibriumMAS (LD-MAS) where a marker (ormarker haplotypes) is in population-widedisequilibrium with a QTL; and linkageequilibrium MAS (LE-MAS) where markersare in Hardy-Weinberg equilibrium withthe QTL at the population level, but linkagedisequilibrium exists within families. Afourth type of MAS that was recentlyproposed is “genome-wide MAS” (GW-MAS), where dense markers (i.e. SNPs)across the genome are used to predictthe genetic merit of an individual withouttargeting any individual QTL or measuring(expensive) phenotypes on every generation(Meuwissen, Hayes, and Goddard, 2001).Integrating current evaluations with MASis most straightforward for GAS and LD-MAS because the QTL effect can be includedin routine evaluations as a fixed effect(Chapter 10). LE-MAS, on the other hand,requires extensive genotyping and fairlycomplicated statistical procedures (Wang,Fernando and Grossman, 1998), while GW-MAS reduces the genome to a “black-box”but does not require selection of QTLusing arbitrary thresholds. Furthermore,the dense marker information requiredfor GW-MAS may dispense with oftenfaulty pedigree records because all pedigreeinformation is encoded in the genome-widegenotypes.In terms of quantitative genetic theory,there are ongoing developments in thetools required to detect and evaluate QTLin arbitrary pedigrees, moving away fromstrictly additive-dominance models toepistasis and parent-of-origin effects (Liu,Jansen and Lin, 2002; Shete and Amos,2002). At the same time, the technology toanalyse more than 10 000 SNPs in a singleassay is available, and a cost of as little asUS$0.02 per genotype is likely for chicken

Chapter 11 – Marker-assisted selection in poultry 193SNPs in the near future. However, the finemapping and characterization of identifiedQTL remain costly and time-consumingprocesses and are often restricted to themost promising QTL, resulting in hundredsof QTL that will never make it past thestage of mapping to a 30 cM confidenceinterval.While current research and developmentsin poultry functional genomics are relevantto all four possible applications of MAS tolivestock, poultry breeders need to decideat what level they want to exploit molecularinformation and for which traits.The emerging picture is that breedersare more comfortable with known genemutations as this provides an easy route toimplementation as well as knowledge aboutthe underlying biology. Furthermore, thereis concern that the marker-trait linkagewill break down over a relatively few generationsof selection in large commercialflocks. While candidate gene studies wouldprovide the quickest route to implementation,fine mapping and characterization ofQTL (e.g. using expression studies) mayreveal gene variants that are not obviouscandidate genes for quantitative traits.Potential for MAS in poultry indeveloping countriesOwing to the relatively low value ofsingle animals, the high reproductive ratein poultry and good portability of eggsor day-old hatchlings, the concentrationof resources is very high in the poultrybreeding industry and all poultry breedingis privately owned. Fifty years ago therewere many primary breeders in each andevery industrialized country, but not solong ago there were only 20 breeding companiesworldwide. Today, three groups ofprimary breeders dominate the internationallayer market. Equally, in the chickenmeat industry, there are four major playersin broiler breeding worldwide (Flockand Preisinger, 2002). The concentrationprocess is probably now complete, and thepresent players are sufficient to meet theglobal supply for 700 000 million eggs asfinal products. A similar trend is expectedin the pig industry, where internationalbreeding companies of hybrid products areincreasing their market share (Preisinger,2004). For large-scale farming of broilersand layers in developing countries thereare additional challenges with regard toheat stress and potential disease pressure.With increasing poultry productionin developing countries, breeding companiesmay give priority to using breedingand molecular tools to address these additionalchallenges. While chickens are veryefficient in converting grain into valuablemeat and egg protein, and smallholderchicken production can be valuable forsustaining the livelihoods of farmers in thedeveloping world, this type of poultry productionwould require robust dual-purpose(meat and egg) birds, rather than specializedbroiler and layer lines. It is unlikelythat the commercial breeders will developsuch lines but there may be scope fornational or international research organizationsto do so. Any MAS would haveto be done at the institutional level wherethe line is developed and would necessitateprior knowledge of trait-marker associationsat the farm level. The implementationof whole genome SNP approaches to farmlevel recording might facilitate progress inthis area but the challenges, both practicaland theoretical, are formidable.Concluding remarksAmong livestock species, chicken haveby far the most comprehensive genomictoolbox. However, uptake of MAS will

194Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishdepend strongly on whether the industrywishes to supplement its current selectionprogramme with a known gene variantor whether it is prepared to restructurebreeding programmes around MAS.Compared with, for instance, the dairy cattleindustry, the poultry breeding communitymay be slower to embrace emerging complexapproaches to MAS. This is somewhatsurprising because the closed structure ofthe poultry breeding pyramid offers muchbetter protection of intellectual propertythan the dairy cattle industry where semenfrom highest ranking bulls is available forall. On the other hand, the fact that bloodgroups have been used to select for resistanceto MD suggests that poultry breedershave some experience and skills in thistype of selection. Poultry breeding companiescontribute significantly to poultrygenomics research but may not be fullyconvinced about the economic feasibilityof MAS. To implement MAS successfully,a company must tackle the problems ofidentifying the traits to select and theireconomic significance, the lack of currentknowledge of the genes or markers associatedwith these traits, and their associationwith other economic selection criteria. Thecurrent “toolbox” provides the means toanswer some of these questions but thereare obvious concerns about human andcapital resources and the potential lossof gains in other traits in a competitivemarket. Coupled with these reservationsmust be the very evident success of currentbreeding programmes in achieving manydesirable commercial goals.AcknowledgementsThe Roslin Institute is supported by a corestrategic grant from the Biotechnologyand Biological Sciences Research Council(BBSRC). The authors gratefully acknowledgeinformation on breeding objectives from representativesof the poultry breeding industry.ReferencesArthur, J.A. & Albers, G.A.A. 2003. Industrial perspective on problems and issues associated withpoultry breeding. In W. M. Muir & S. E. Aggrey, eds. Poultry genetics, breeding and biotechnology,pp. 1–12. Wallingford, UK, CABI Publishing.Bilgili, S.F. 2001. Poultry products and processing in the international market place. Proc. Internat. Anim.Agric. & Food Sci. Conf. Indianapolis, IN, USA (available at www.fass.org/fass01/pdfs/Bilgili.pdf).Bowman, J.C. 1974. An introduction to animal breeding. The Institute of Biology’s Studies in BiologyNo 46. London, Edward Arnold.Bumstead, N. & Palyga, J. 1992. A preliminary linkage map of the chicken genome. Genomics 13:690–697.Crittenden, L.B., Provencher, L., Sanatangello, L., Levin, I., Abplanalp, H., Briles, R.W., Briles,W.E. & Dodgson, J.B. 1993. Characterization of a Red Jungle Fowl by White Leghorn backcrossreference population for molecular mapping of the chicken genome. Poultry Sci. 72: 334–348.de Koning, D.J., Carlborg, O. & Haley. C.S. 2005. The genetic dissection of immune response usinggene-expression studies and genome mapping. Vet. Immun. & Immunopath. 105: 343–352.de Koning, D.J. & Haley, C.S. 2005. Genetical genomics in humans and model organisms. TrendsGenetics 21: 377–381.de Koning, D.J., Haley, C.S., Windsor, D., Hocking, P.M., Griffin, H., Morris, A., Vincent, J. &Burt, D.W. 2004. Segregation of QTL for production traits in commercial meat-type chickens.Genetic. Res. 83: 211–220.

Chapter 11 – Marker-assisted selection in poultry 195de Koning, D.J., Windsor, D., Hocking, P.M., Burt, D.W., Law, A., Haley, C.S., Morris, A., Vincent,J. & Griffin, H. 2003. Quantitative trait locus detection in commercial broiler lines using candidateregions. J. Anim. Sci. 81:1158–1165.Dekkers, J.C.M. 2004. Commercial application of marker- and gene-assisted selection in livestock:strategies and lessons. J. Anim. Sc. 82 E-Suppl: E313–E328.Dunn, I.C., Miao, Y.W., Morris, A., Romanov, M.N., Wilson, P.W. & Waddington, D. 2004. A studyof association between genetic markers in candidate genes and reproductive traits in one generationof a commercial broiler breeder hen population. Heredity 92: 128–134.Evans, G.J., Giuffra, E., Sanchez, A., Kerje, S., Davalos, G., Vidal, O., Illan, S., Noguera, J.L.,Varona, L., Velander, I., Southwood, O.I., de Koning, D.J., Haley, C.S., Plastow, G.S. &Andersson, L. 2003. Identification of quantitative trait loci for production traits in commercial pigpopulations. Genetics 164: 621–627.Feng, X.P., Kuhnlein, U., Aggrey, S.E., Gavora, J.S. & Zadworny, D. 1997. Trait association ofgenetic markers in the growth hormone and the growth hormone receptor gene in a White Leghornstrain. Poultry Science 76: 1770–1775.Feng, X.P., Kuhnlein, U., Fairfull, R.W., Aggrey, S.E., Yao, J. & Zadworny, D. 1998. A geneticmarker in the growth hormone receptor gene associated with body weight in chickens. J. Heredity89: 355–359.Flock, D.K. & Preisinger, R. 2002. Breeding plans for poultry with emphasis on sustainability. 7thWorld Congress on Genetics Applied to Livestock Production, 19–23 August, 2002, Montpellier,France, Session 24: Sustainable breeding plans in developed countries. Communication 24-02.Fotouhi, N., Karatzas, C.N., Kuhnlein, U. & Zadworny, D. 1993. Identification of growth-hormoneDNA polymorphisms which respond to divergent selection for abdominal fat-content inchickens. Theor. Appl. Genet. 85: 931–936.Groenen, M.A., Cheng, H.H., Bumstead, N., Benkel, B.F., Briles, W.E., Burke, T., Burt, D.W.,Crittenden, L.B., Dodgson, J., Hillel, J., Lamont, S., de Leon, A.P., Soller, M., Takahashi, H. &Vignal, A. 2000. A consensus linkage map of the chicken genome. Genome Research 10: 137–147.Groenen, M.A., Crooijmans, R.P., Veenendaal, A., Cheng, H.H., Siwek, M. & van der Poel, J.J.1998. A comprehensive microsatellite linkage map of the chicken genome. Genomics 49: 265–274.Hillier, L.W., Miller, W., Birney, E., Warren, W., Hardison, R.C., Ponting, C.P., Bork, P., Burt,D.W., Groenen, M.A., Delany, M.E., Dodgson, .J.B., Chinwalla, A.T., Cliften, P.F., Clifton,S.W., Delehaunty, K.D., Fronick, C., Fulton, R.S., Graves, T.A., Kremitzki, C., Layman, D.,Magrini, V., McPherson, J.D., Miner, T.L., Minx, P., Nash, W.E., Nhan, M.N., Nelson, J.O.,Oddy, L.G., Pohl, C.S., Randall-Maher, J., Smith, S.M., Wallis, J.W., Yang, S.P., Romanov,M.N., Rondelli, C.M., Paton, B., Smith, J., Morrice, D., Daniels, L., Tempest, H.G., Robertson,L., Masabanda, J.S., Griffin, D.K., Vignal, A., Fillon, V., Jacobbson, L., Kerje, S., Andersson,L., Crooijmans, R.P., Aerts, J., van der Poel, J.J., Ellegren, H.. Caldwell, R.B., Hubbard, S.J.,Grafham, D.V., Kierzek, A.M., McLaren, S.R., Overton, I.M., Arakawa, H., Beattie, K.J.,Bezzubov, Y., Boardman, P.E., Bonfield, J.K., Croning, M.D., Davies, R.M., Francis, M.D.,Humphray, S.J., Scott, C.E., Taylor, R.G., Tickle, C., Brown, W.R., Rogers, J., Buerstedde, J.M.,Wilson, S.A., Stubbs, L., Ovcharenko, I., Gordon, L., Lucas, S., Miller, M.M., Inoko, H., Shiina,T., Kaufman, J., Salomonsen, J., Skjoedt, K., Wong, G.K., Wang, J., Liu, B., Wang, J., Yu, J.,Yang, H., Nefedov, M., Koriabine, M., Dejong, P.J., Goodstadt, L., Webber, C., Dickens, N.J.,Letunic, I., Suyama, M., Torrents, D., Von, M.C., Zdobnov, E.M., Makova, K., Nekrutenko, A.,

196Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishElnitski, L., Eswara, P., King, D.C., Yang, S., Tyekucheva, S., Radakrishnan, A., Harris, R.S.,Chiaromonte, F., Taylor, J., He, J., Rijnkels, M., Griffiths-Jones, S., Ureta-Vidal A., Hoffman,M.M., Severin, J., Searle, S.M., Law, A.S., Speed, D., Waddington, D., Cheng, Z., Tuzun, E.,Eichler, E., Bao, Z., Flicek, P., Shteynberg, D.D., Brent, M.R., Bye, J.M., Huckle, E.J., Chatterji,S., Dewey, C., Pachter, L., Kouranov, A., Mourelatos, Z., Hatzigeorgiou, A.G., Paterson, A.H.,Ivarie, R., Brandstrom, M., Axelsson, E., Backstrom, N., Berlin, S.,Webster, M.T., Pourquie, O.,Reymond, A., Ucla, C., Antonarakis, S.E., Long, M.J.J., Emerson, J.J Betran, E., Dupanloup,I., Kaessmann, H., Hinrichs, A.S., Bejerano, G., Furey, T.S., Harte, R.A., Raney, B., Siepel, A.,Kent, W.J., Haussler, D., Eyras, E., Castelo, R., Abril, J.F., Castellano, S., Camara, F., Parra, G.,Guigo, R., Bourque, G., Tesler, G., Pevzner, P.A., Smit, A., Fulton, L.A., Mardis, E.R. & Wilson,R.K. 2004. Sequence and comparative analysis of the chicken genome provide unique perspectiveson vertebrate evolution. Nature 432: 695–716.Hocking, P.M. 2005. Review of QTL results in chicken. Wld. Poultry Sci. J. 61: 215–226.Hu, J.X., Bumstead, N., Barrow, P., Sebastein, G., Olien, L., Morgan, K. & Malo, D. 1997. Resistanceto salmonellosis in the chicken is linked to NRAMP1 and TNC. Genome Res. 7: 693–704.Iqbal, M., Philbin, V.J., Withanage, G.S.K., Wigley, P., Beal, R.K., Goodchild, M.J., Barrow,P., McConnell, I., Maskell, D.J., Young, J., Bumstead, N., Boyd, Y. & Smith A.L. 2005.Identification and functional characterization of chicken Toll-like receptor 5 reveals a fundamentalrole in the biology of infection with Salmonella enterica serovar typhimurium. Infect. & Immunity73: 2344–2350.Jiang, R., Li, J.,Qu, Li, H. & Yang, N. 2004. A new single nucleotide polymorphism in the chickenpituitary- specific transcription factor (POULF1) gene associated with growth rate. Anim. Genet.35: 344–346.Kuhnlein, U., Ni, L., Weigend, S., Gavora, J.S., Fairfull, W. & Zadworny, D. 1997. DNA polymorphismsin the chicken growth hormone gene: response to selection for disease resistance andassociation with egg production. Anim. Genet. 28: 116–123.Lamont, S.J., Kaiser, M.G. & Liu, W. 2002. Candidate genes for resistance to Salmonella enteritidiscolonization in chickens as detected in a novel genetic cross. Vet. Immun. & Immunopath. 87:423–428.Leveque, G., Forgetta, V., Morroll, S., Smith, A.L., Bumstead, N., Barrow, P., Loredo-Osti, J.C. ,Morgan, K. & Malo, D. 2003. Allelic variation in TLR4 is linked to susceptibility to Salmonellaenterica serovar typhimurium infection in chickens. Infect. & Immunity 71: 1116–1124.Li, S., Aggrey, S.E., Zadworny, D., Fairfull, W. & Kuhnlein, U. 1998a. Evidence for a geneticvariation in the mitochondrial genome affecting traits in White Leghorn chickens. J. Hered. 89:222–226.Li, S., Zadworny, D., Aggrey, S.E. & Kuhnlein, U. 1998b. Mitochondrial pepck: A highly polymorphicgene with alleles co- selected with Marek’s disease resistance in chickens. Anim. Genet. 29:395–397.Li, H., Deeb, N., Zhou, H., Mitchell, A.D., Ashwell, C.M. & Lamont, S.J. 2003. Chicken quantitativetrait loci for growth and body composition associated with transforming growth factor-betagenes. Poult. Sci. 82: 347–356.Li, H., Deeb, N., Zhou, H., Ashwell, C.M. & Lamont, S.J. 2005. Chicken quantitative trait loci forgrowth and body composition associated with the very low density apolipoprotein-II gene. Poult.Sci. 84: 697–703.

Chapter 11 – Marker-assisted selection in poultry 197Liu, W. & Lamont, S.J. 2003. Candidate gene approach: Potential association of caspase-1, inhibitorof apoptosis protein-1, and Prosaposin gene polymorphisms with response to Salmonella enteritidischallenge or vaccination in young chicks. Anim. Biotech. 14: 61–76.Liu, H.C., Cheng, H.H., Tirunagaru, V., Sofer, L. & Burnside, J. 2001a. A strategy to identify positionalcandidate genes conferring Marek’s disease resistance by integrating DNA microarrays andgenetic mapping. Anim. Genet. 32: 351–359.Liu, H.C., Kung, H.J., Fulton, J.E., Morgan, R.W. & Cheng. H.H. 2001b. Growth hormone interactswith the Marek’s disease virus SORF2 protein and is associated with disease resistance inchicken. Proc. Natl. Acad. Sci. USA 98: 9203–9208.Liu, H.C., Niikura, M., Fulton, J.E. & Cheng. H.H. 2003. Identification of chicken lymphocyteantigen 6 complex, locus E (LY6E, alias SCA2) as a putative Marek’s disease resistance gene via avirus-host protein interaction screen. Cytogenet. & Genome Res. 102: 304–308.Liu, Y., Jansen, G.B. &. Lin, C.Y. 2002. The covariance between relatives conditional on geneticmarkers. Genet. Select. Evoln. 34: 657–678.McKay, J.C. 1998. A poultry breeder’s approach to avian neoplasia. Avian Pathology 27: S74–S77.Meuwissen, T.H.E., Hayes, B.J. & Goddard, M.E. 2001. Prediction of total genetic value usinggenome-wide dense marker maps. Genetics 157: 1819–1829.Muir, W.M. 2003. Incorporating molecular information in breeding programmes: applications andlimitations. In W.M. Muir & S.E. Aggrey, eds. Poultry Genetics, Breeding and Biotechnology, pp.549–562. Wallingford, UK, CABI Publishing.Nagamine, Y., Haley, C.S., Sewalem, A. & Visscher, P.M. 2003. Quantitative trait loci variation forgrowth and obesity between and within lines of pigs (Sus scrofa). Genetics 164: 629–635.Nagaraja, S.C., Aggrey, S.E., Yao, J., Zadworny, D., Fairfull R.W. & Kuhnlein, U. 2000. Trait associationof a genetic marker near the IGF-I gene in egg-laying chickens. J. Heredity 91: 150–156.Parsanejad, R., Torkamanzehi, A., Zadworny, D. & Kuhnlein, U. 2003. Alleles of cytosolic phosphoenolpyruvatecarboxykinase (PEPCK): trait association and interaction with mitochondrialPEPCK in a strain of white leghorn chickens. Poultry Sci. 82: 1708–1715.Parsanejad, R., Praslickova, D., Zadworny, D. & Kuhnlein, U. 2004. Ornithine decarboxylase: haplotypestructure and trait associations in White Leghorn chickens. Poultry Sci. 83: 1518–1523.Preisinger, R. 2004. Internationale Tendenzen der Tierzüchtung und die Rolle der Zuchtunternehmen.Züchtungskunde 76(6): 395–402.Schmid, M., Nanda, I., Guttenbach, M., Steinlein, C., Hoehn, M., Schartl, M., Haaf, T., Weigend,S., Fries, R., Buerstedde, J.M., Wimmers, K., Burt, D.W., Smith, J., A’Hara, S., Law, A., Griffin,D.K., Bumstead, N., Kaufman, J., Thomson, P.A., Burke, T., Groenen, M.A., Crooijmans, R.P.,Vignal, A., Fillon, V., Morisson, M., Pitel, F.,Tixier-Boichard, M.T., Ladjali-Mohammedi, K.,Hillel, J., Maki-Tanila, A., Cheng, H.H., Delany, M.E., Burnside, J. & Mizuno, S. 2000. Firstreport on chicken genes and chromosomes 2000. Cytogenet. & Cell Genet. 90: 169–218.Schmid, M., Nanda, I., Hoehn, H., Schartl, M., Haaf, T., Buerstedde, J.M., Arakawa, H., Caldwell,R.B., Weigend, S., Burt, D.W., Smith, J., Griffin, D.K., Masabanda, J.S., Groenen, M.A.,Crooijmans, R.P., Vignal, A., Fillon, V., Morisson, M., Pitel, F., Vignoles, M., Garrigues, A.,Gellin, J., Rodionov, A.V., Galkina, S.A., Lukina, N.A., Ben-Ari, G., Blum, S., Hillel, J., Twito,T., Lavi, U., David, L., Feldman, M.W., Delany, M.E., Conley, C.A., Fowler, V.M., Hedges, S.B.,Godbout, R., Katyal, S., Smith, C., Hudson, Q., Sinclair, A. & Mizuno, S. 2005. Second reporton chicken genes and chromosomes 2005. Cytogen.& Genome Res. 109: 415–479.

198Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishShete, S. & Amos, C.I. 2002. Testing for genetic linkage in families by a variance componentsapproach in the presence of genomic imprinting. Am. J. Human Genet. 70: 751–757.Taha, F.A. 2003. The poultry sector in middle income countries and its feed requirements: the case ofEgypt. Outlook Report No. WRS03-02, pp. 3–14. Washington, DC, Economic Research Service,USDA.Vallejo, R.L., Bacon, L.D., Liu, H.C., Witter, R.L., Groenen, M.A., Hillel, J. & Cheng, H.H. 1998.Genetic mapping of quantitative trait loci affecting susceptibility to Marek’s disease virus inducedtumors in F2 intercross chickens. Genetics 148: 349–360.van der Beek, S. &. van Arendonk, J.A.M. 1996. Marker-assisted selection in an outbred poultrynucleus. Anim. Sc. 62: 171–180.Wang, T., Fernando, R.L. & Grossman, M. 1998. Genetic evaluation by best linear unbiased predictionusing marker and trait information in a multibreed population. Genetics 148: 507–515.Wang, J., He, X., Ruan, J., Dai, M., Chen, J., Zhang, Y., Hu, Y., Ye, C., Li, S., Cong, L., Fang, L.,Liu, B., Li, S., Wang, J., Burt, D.W., Wong, G.K., Yu, J., Yang, H. & Wang, J. 2005. ChickVD: asequence variation database for the chicken genome. Nucl. Acid Res. 33: D438–D441.Wong, G.K., Liu, B., Wang, J., Zhang, Y., Yang, X., Zhang, Z., Meng, Q., Zhou, J., Li, D., Zhang,J., Ni, P., Li, S., Ran, L., Li, H., Zhang, J., Li, R., Li, S., Zheng, H., Lin, W., Li, G., Wang, X.,Zhao, W., Li, J., Ye, C., Dai, M., Ruane, J., Zhou, Y., Li, Y., He, X., Zhang, Y., Wang, J., Huang,X., Tong, W., Chen, J., Ye, J., Chen, C., Wei, N., Li, G., Dong, L., Lan, F., Sun, Y., Zhang, Z.,Yang, Z., Yu, Y., Huang, Y., He, D., Xi, Y., Wei, D., Qi, Q., Li, W., Shi, J., Wang, M., Xie, F.,Wang, J., Zhang, X., Wang, P., Zhao, Y., Li, N., Yang, N., Dong, W., Hu, S., Zeng, C., Zheng,W., Hao, B., Hillier, L.W., Yang, S.P., Warren, W.C., Wilson, R.K., Brandstrom, M., Ellegren, H.,Crooijmans, R.P., van der Poel, J.J., Bovenhuis, H., Groenen, M.A., Ovcharenko, I., Gordon,L., Stubbs, L., Lucas, S., Glavina, T., Aerts, A., Kaiser, P., Rothwell, L., Young, J.R., Rogers,S., Walker, B.A., van, H.A., Kaufman, J., Bumstead, N., Lamont, S.J., Zhou, H., Hocking,P.M., Morrice, D., de Koning, D.J., Law, A., Bartley, N., Burt, D.W., Hunt, H., Cheng, H.H.,Gunnarsson, U., Wahlberg, P., Andersson, L., Kindlund, E., Tammi, M.T., Andersson, B.,Webber, C., Ponting, C.P., Overton, I.M., Boardman, P.E., Tang, H., Hubbard, S.J., Wilson,S.A., Yu, J., Wang, J. & Yang, H. 2004. A genetic variation map for chicken with 2.8 million singlenucleotidepolymorphisms. Nature 432: 717–722.Yonash, N., Bacon, L.D. Witter, R.L. & Cheng. H.H. 1999. High resolution mapping and identificationof new quantitative trait loci (QTL) affecting susceptibility to Marek’s disease. Anim. Genet.30: 126–135.Zhou, H., Mitchell, A.D., McMurtry, J.P., Ashwell, C.M. & Lamont, S.J. 2005. Insulin-like growthfactor-1 gene polymorphism associations with growth, body composition, skeleton integrity, andmetabolic traits in chickens. Poult. Sci. 84: 212–219.

Chapter 12Marker-assisted selectionin dairy cattleJoel Ira Weller

200Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishSummaryConsidering the long generation interval, the high value of each individual, the very limitedfemale fertility and the fact that nearly all economic traits are expressed only in females, itwould seem that cattle should be a nearly ideal species for application of marker-assistedselection (MAS). As genetic gains are cumulative and eternal, application of new technologiesthat increase rates of genetic gain can be profitable even if the nominal annualcosts are several times the value of the nominal additional annual genetic gain. Completegenome scans for quantitative trait loci (QTL) based on the granddaughter design havebeen completed for most commercial dairy cattle populations, and significant acrossstudyeffects for economic traits have been found on chromosomes 1, 3, 6, 9, 10, 14 and20. Quantitative trait loci associated with trypanotolerance have been detected in a crossbetween the African N’Dama and the Boran breeds as the first step in the introgressionof these genes into breeds susceptible to trypanosomosis. In dairy cattle, the actual DNApolymorphism has been determined twice, for QTL on BTA 6 and BTA 14. In both casesthe polymorphism caused a non-conservative amino acid change, and both QTL chieflyaffect fat and protein concentration. Most theoretical studies have estimated the expectedgains that can be obtained by MAS to be in the range of a 5 to 20 percent increase in therates of genetic gain obtained by traditional selection programmes. Applied MAS programmeshave commenced for French and German Holsteins. In both programmes geneticevaluations including QTL effects are computed by variants of marker-assisted best linearunbiased prediction (MA-BLUP).

Chapter 12 – Marker-assisted selection in dairy cattle 201IntroductionCompared with other agricultural species,dairy cattle are unique in terms of thevalue of each animal, their long generationinterval and the very limited fertilityof females. Thus unlike plant and poultrybreeding, most dairy cattle breeding programmesare based on selection within thecommercial population. Similarly, detectionof quantitative trait loci (QTL) and markerassistedselection (MAS) programmes aregenerally based on analysis of existing populations.The specific requirements of dairycattle breeding have led to the generationof very large data banks in most developedcountries, which are available for analysis.In this chapter, dairy cattle breeding programmesin the developed and developingcountries are reviewed and compared. Theimportant issues in the application of MASare then outlined. These include economicconsiderations based on phenotypic selection,the current status of cattle markermaps, methods to detect QTL and to estimateQTL effects and location suitablefor dairy cattle, the current state of QTLdetection in dairy cattle, methods to incorporateinformation from genetic markersin genetic evaluation systems, methods toidentify the actual polymorphisms responsiblefor observed QTL and description ofthe reported results, methods and theoryfor MAS in dairy cattle, the current statusof MAS and, finally, the future prospectsfor MAS in dairy cattle.Dairy cattle breeding programmesin developed countriesIn most developed countries, dairy cattlebreeding programmes are based on the“progeny test” (PT) design. The PT isthe design of choice for moderate to largedairy cattle populations, including theUnited States Holsteins, which includeover ten million animals. An example ofthe Israeli PT design is given in Figure 1.Figure 1The Israeli Holstein breeding programme50 candidatebulls100 elite cows200 elite cows3 foreign bulls4 local bulls30 000cows120 000 cows20 bulls5 000daughtersrecords ondaughters5 bulls are selected45 bulls are culled

202Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishThis population consists of approximately120 000 cows of which 90 percent aremilk recorded. Approximately 20 bullsare used for general service. Each yearabout 300 elite cows are selected as bulldams. These are mated to the two to fourbest local bulls and an equal number offoreign bulls to produce approximately 50bull calves for progeny testing. At the ageof one year, the bull calves reach sexualmaturity, and approximately 1 000 semensamples are collected from each young bull.These bulls are mated to approximately30 000 first parity cows to produce about5 000 daughters, or 100 daughters peryoung bull. Gestation length for cattle isnine months. Thus the young bulls areapproximately two years old when theirdaughters are born, and are close to fourwhen their daughters calve and begintheir first lactation. At the completion oftheir daughters’ first lactations, most ofthe young bulls are culled. Only four tofive are returned to general service, and asimilar number of the old proven sires areculled. By this time the selected bulls areapproximately five years old.Various studies have shown that ratesof genetic gain by a PT scheme are about0.1 to 0.2 genetic standard deviations ofthe selection index per year (Nicholasand Smith, 1983; Israel and Weller, 2000).The PT was devised to take advantageof the nearly unlimited fertility of males.However, compared with breeding schemesfor other species, the PT has several majorweaknesses. First, for a PT system to beeffective, the population must include atleast several tens of thousands of animalswith recording on production traits andpaternity. Inaccurate recording can significantlyreduce rates of genetic gain (Israeland Weller, 2000). Second, generation intervals,especially along the sire-to-dam andsire-to-sire paths, are much longer thanthe biological requirements. The increasein generation interval reduces genetic gainper year. As artificial insemination (AI)institutes generally pay a premium pricefor male calves of elite cows, these cows areoften given preferential treatment in orderto increase their genetic evaluations (Powelland Norman, 1988). The small number ofbulls actually used for general service, andthe even smaller number of bulls used asbull sires, tends to reduce the effective populationsize, which increases inbreeding anddecreases genetic variance in the population.The effective population size of the UnitedStates Holstein population with ten millioncows has been estimated at about 100(Farnir et al., 2000). Finally, there is virtuallyno selection along the dam-to-dam path.Generally, 70-80 percent of healthy femalecalves produced are used as replacements.Various studies have suggested thatselection intensities along the dam-to-dampath could be increased by application ofmultiple ovulation and embryo transfer(MOET) and sexed semen. Costs of bothtechnologies are still prohibitively high tobe applied to the entire population, asshown below. To overcome this problemfor MOET, Nicholas and Smith (1983)proposed a “nucleus” breeding scheme. Innucleus schemes, the selection populationconsists of several hundred individuals, andbulls are not progeny tested. Instead, bullsare selected based on the genetic evaluationsof their dams and sisters, which shortens thegeneration interval on the sire-to-dam andsire-to-sire paths, but reduces the reliabilitiesof the genetic evaluations. Dams of bulls andcows are selected based chiefly on their ownproduction records, and MOET is appliedto increase the number of progeny per dam.As the selection population consists of onlyseveral hundred individuals, MOET costs

Chapter 12 – Marker-assisted selection in dairy cattle 203are manageable if costs are spread overthe entire national dairy industry. Ratesof genetic gain within the nucleus are thushigher than can be obtained by a nationalPT design. This gain is transferred to thegeneral population through the use of bullsfrom the nucleus population. In addition tothe greater overall rate of genetic gain, thenucleus scheme has the advantage that it isnecessary to collect data on a much smallerpopulation, which should reduce costsand increase accuracy. The disadvantagesof MOET are that overall costs and ratesof increase of inbreeding will be greaterunless steps are taken to reduce inbreeding.However, these steps will also slightlydecrease rates of genetic gain. In practice, nocountry has replaced its standard PT schemewith a nucleus breeding programme.Dairy cattle breeding indeveloping countriesThe genus Bos includes five to seven species,of which Bos taurus and Bos indicusare the most widespread and economicallyimportant. B. taurus is the main dairycattle species, and is found generally intemperate climates. Several tropical andsubtropical cattle breeds are the result ofcrosses between B. taurus and B. indicus,which interbreed freely. In the tropics,cows need at least some degree of toleranceto environmental stress due to poornutrition, heat and disease challenge tosustain relatively high production levels(Cunningham, 1989). Tropical breeds areadapted to these stresses but have low milkyield, whereas more productive temperatebreeds cannot withstand the harsh tropicalconditions, to the point of not being able tosustain their numbers (de Vaccaro, 1990).Furthermore, most tropical countries aredeveloping countries, which lack systematiclarge-scale milk and pedigree recording.A number of studies have been conductedon crosses between imported andlocal breeds in the tropics. Generally, the F 1B. taurus x B. indicus crosses are economicallysuperior to either of the purebredstrains (FAO, 1987). The heterosis effectof the F 1 cross is due to genes for diseaseresistance from the local parent, and genesfor milk production from the importedstrain (Smith, 1988; Cunningham, 1989).However, this heterosis is lost in futuregenerations if the F 1 is backcrossed to eitherparental strain. Madalena (1993) presentedan F 1 continuous replacement schemeto capitalize on its superiority. Recently,Kosgey, Kahi and van Arendonk (2005)proposed a closed adult nucleus MOETscheme to increase milk production intropical crossbred cattle.Economic considerations inapplying MAS to dairy cattleFor any new technique to be economicallyviable, overall gains must be greater thanoverall costs. This also applies to usingMAS within a dairy cattle breeding programme.However, unlike investment innew equipment, genetic gains never “wearout”, i.e. breeding is unique in that geneticgains are cumulative and eternal. Thus, asshown by Weller (1994, 2001) investmentsin MAS or other techniques that enhancebreeding programmes are economicallyviable even if “nominal” costs are greaterthan “nominal” gains.For example, consider an ongoing breedingprogramme with a constant rate of geneticgain per year. Assume that the annual rate ofgenetic gain has a nominal economical valueof V. The cumulative discounted returns toyear T, R v , will be a function of the nominalannual returns, the discount rate, d, theprofit horizon, T, and the number of yearsfrom the beginning of the programme until

204Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishfirst returns are realized, t. R v is computed asfollows (Hill, 1971):Rvtrd- r= V(1 - rT + 1d2d)( T - t + 1) r-1-r{1}where r d = 1/(1+d). For example, with d =0.08, T = 20 years, and t = 5 years, R v =32.58V. That is, the cumulative returns areequal to nearly 33 times the nominal annualreturns. For an infinite profit horizon,Equation {1} reduces to:RVrtv= =2 2 t–2(1 – rd) d (1 + d)and R v = 124.04V.The value of nominal annual geneticgain will now be compared with the annualcosts of a breeding programme, assuming afixed nominal cost per year. Costs, unlikegenetic gain, only have an effect in the yearthey occur. Assuming that annual costs areequal during the length of the breedingprogramme, and that first costs occur in theyear after the base year, C T , the net presentvalue of the total costs of the breeding programmeis computed as follows:CTCcrd(1 − r=1−rdTd)where C c = annual costs of the breedingprogramme. Using the same values for Tand d, C T = 9.82C c . Thus, with a profithorizon of 20 years, cumulative profit ispositive if V > 0.31C c . For an infinite profithorizon, C T = 12.5C c , and profit will bepositive if V > 0.1C c .Therefore, a breeding programme can beprofitable even if the nominal annual costsare several times the value of the nominalannual genetic gain. For example, considerVdT + 1d{2}{3}the United States of America dairy cattlepopulation, which consists of about tenmillion cows. Annual genetic gain is about100 kg milk per year. The value of a 1 kggain in milk production has been estimatedat US$0.1 (Weller, 1994). Thus, the nominalannual value of a 10 percent increase in therate of genetic gain (10 kg per year) is:V = (10 kg per cow per year)(US$0.1 per kg)(10 000 000 cows) =US$10 000 000 per year{4}The cumulative value with a profithorizon of 20 years and an 8 percent discountrate would be US$326 million, andbreak-even annual costs for a technologythat increases annual genetic gain by 10 percentare US$32 million per year. Thus, itwould be profitable to spend quite a lot fora relatively small genetic gain.The value of genetic gain to a specificbreeding enterprise will generally be lessthan the gain to the general economy. Thisis because most of the gains obtained bybreeding will be passed on to the consumers.Brascamp, van Arendonk and Groen (1993)considered the economic value of MASbased on changes in returns from semensales for a breeding organization operatingin a competitive market. In this case,a breeding firm that adopts a MAS programmecan increase its returns either byincreasing its market share or increasing themean price of a semen dose. Although thevalue of genetic gain will be less, relativelysmall changes in genetic merit can result inlarge changes in market share.current status of marker mapsin cattleCattle have 29 pairs of autosomes and onepair of sex chromosomes. All the autosomesare acrocentric, and map units are

Chapter 12 – Marker-assisted selection in dairy cattle 205scored from the centromere. Chromosomesare denoted with the prefix “BTA” (B.taurus). Similar to other mammals, thebovine DNA includes 3×10 9 base pairs(bp), and the map length is approximately3 000 cM. The human genome is estimatedto encode 20 000–25 000 protein-codinggenes (International Human GenomeSequencing Consortium, 2004), and it canbe assumed that the number of genes inother mammals, including cattle, should bequite similar. Thus, a single map unit, onaverage, includes approximately eight genesand one million bp.As in other animal species, microsatellitesare still the marker of choice for mapconstruction due to their prevalence andhigh polymorphism. Although singlenucleotide polymorphisms (SNPs) aremuch more prevalent, genetic maps basedon SNPs are still in the future. Morethan 50 000 SNPs have been identified inhumans, but only several thousand havebeen validated in cattle (www.afns.ualberta.ca/Hosted/Bovine%20Genomics/), andrates of polymorphism are generallyunknown. With the completion of the sixfoldcoverage of the bovine genome bythe Bovine Genome Sequencing Project atBaylor College of Medicine (www.hgsc.bcm.tmc.edu/projects/bovine/) many moreSNPs will be identified.Several genetic maps are available on theinternet. The United States Meat AnimalResearch Center (MARC) (www.marc.usda.gov/) includes thousands of markers,chiefly microsatellites. The ArkDB databasesystem, hosted at Roslin Institute,includes data from several published maps(www.thearkdb.org/). The CommonwealthScientific and Industrial Research Organization(CSIRO) livestock industries cattlegenome marker map is built upon dataprovided by the University of Sydney’scomparative location database (www.livestockgenomics.csiro.au/perl/gbrowse.cgi/cattlemap/). This map combined all publicly-availablemaps into a single integratedmap that currently includes 9 400 markers.Methods of QTL detectionsuitable for commercial dairycattle populationsDetection of QTL requires generation oflinkage disequilibrium (LD) between thegenetic markers and QTL. In plants, this isgenerally accomplished by crosses betweeninbred lines but, for the reasons noted inthe introduction, this is not a viable optionfor dairy cattle in developed countries, inwhich all analyses must be based on analysisof the existing population. Detection ofQTL in developing countries is consideredbelow. For advanced commercial populations,the “daughter” and “granddaughter”designs, which make use of the existenceof large half-sib families, are most appropriatefor QTL analysis (Weller, Kashi andSoller, 1990). These designs are presentedin Figures 2 and 3.Both designs are similar to the backcrossdesign for crosses between inbred lines inthat only the alleles of one parent are followedin the progeny. Thus, similar to thebackcross design, dominance cannot beestimated. These designs differ from crossesbetween inbred lines in that several familiesare analysed in which the linkage phasebetween QTL and genetic markers maydiffer. In addition, any specific QTL will beheterozygous in only a fraction of the familiesincluded in the analysis. Thus, QTLeffects must be estimated within families,and these designs are therefore less powerfulper individual genotyped than designsbased on crosses between inbred lines.The granddaughter design has theadvantage of greater statistical power per

206Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishFigure 2The daughter designMAmaM?m?A?a?Only a single family is shown, although in practice several families will be analysed jointly. The sireis assumed to be heterozygous for a QTL and a linked genetic marker. The two alleles of the markerlocus are denoted “M” and “m”, and the two alleles of the QTL are denoted “A” and “a”. Alleles ofmaternal origin are denoted by question marks.individual genotyped. As each genotypeis associated with multiple phenotypicrecords, the power per individual genotypedin the granddaughter design can befour-fold the power of the daughter design(Weller, Kashi and Soller, 1990). The disadvantageof this design is that the appropriatedata structure (hundreds of progeny testedbulls, sons of a limited number of sires) isfound only in the largest dairy cattle populations.Both daughter and granddaughterdesigns are less powerful per individualgenotyped than designs based on analysisof inbred lines. Furthermore, the half-sibdesigns have the disadvantage that progenywith the same genotype as the sire are uninformative,because the progeny could havereceived either paternal allele.Additional experimental designs havealso been proposed. Coppieters et al.(1999) proposed the “great-granddaughterdesign”. One of the disadvantages of thegranddaughter design is that the number ofprogeny-tested sons of most sires is too lowto obtain reasonable power to detect QTLof moderate effects. Coppieters et al. (1999)proposed that power can be increased byalso genotyping progeny-tested grandsonsof the grandsire. Inclusion of the grandsonsis complicated by the fact that there isanother generation of meiosis between thegrandsire and his grandson.A significant drawback of all the designsconsidered above is that they give noindication of the number of QTL allelessegregating in the population or their rela-

Chapter 12 – Marker-assisted selection in dairy cattle 207Figure 3The granddaughter designMAmaMA??ma??The grandsire is assumed to be heterozygous for a QTL and a linked genetic marker. As in Figure 2, only a single familyis shown. The two alleles of the marker locus are denote “M” and “m”, and the two alleles of the QTL are denoted “A”and “a”. Alleles of maternal origin are denoted by question marks. Genotypes are not listed for the granddaughtersbecause they were not genotyped.tive frequencies. To answer this question,Weller et al. (2002) proposed the “modifiedgranddaughter design” presented inFigure 4. Assume that a segregating QTLfor a trait of interest has been detectedand mapped to a short chromosomalsegment using either a daughter or a granddaughterdesign. Consider the maternalgranddaughters of a grandsire with a significantcontrast between his two paternalalleles. This grandsire will be denoted the“heterozygous grandsire”. Each maternalgranddaughter will receive one allele fromher sire, who is assumed to be unrelated tothe heterozygous grandsire, and one allelefrom her dam, who is a daughter of theheterozygous grandsire. Of these granddaughters,one-quarter should receive thegrandpaternal QTL allele with the positiveeffect, one-quarter should receive the

208Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishFigure 4The modified granddaughter designQ1 Q2 M1M2H1H2Q1M1Q2M2H3H4H1 Q1 H2 M1 H3 M2Q2H4Only alleles for the QTL are shown. Alleles originating in the heterozygous grandsire are termed “Q1” and Q2”. Alleles originatingin the grand-dams are termed “M1” and “M2”. Alleles originating in the sires are termed “H1”, “H2”, “H3” and “H4”.negative grandpaternal QTL allele, and halfshould receive neither grandpaternal allele.In the third case, the granddaughter receivedone of the QTL alleles of her grand-dam,the mate of the heterozygous grandsire.These grand-dams can be considered arandom sample of the general populationwith respect to the allelic distribution of theQTL. All genetic and environmental effectsnot linked to the chromosomal segment inquestion are assumed to be randomly distributedamong the granddaughters, or areincluded in the analysis model. Thus, unlikethe daughter or granddaughter designs, it ispossible to compare the effects of the twograndpaternal alleles with the mean QTLpopulation effect.Assuming that the QTL is “functionallybiallelic” (i.e. there are only two alleleswith differential expression relative to thequantitative trait), and that allele origincan be determined in the granddaughters,the relative frequencies of the two QTLalleles in the population can be determinedby comparing the mean values of thethree groups of granddaughters for thequantitative trait. Using the modifiedgranddaughter design it is also possible toestimate the number of alleles segregatingin the population, and to determine if thesame alleles are segregating in differentcattle populations. Weller et al. (2002)estimated the frequency of the QTL allelethat increases fat and protein concentrationon BTA6 in the Israeli Holstein populationas 0.69 and 0.63, relative to fat and proteinpercent, by the modified granddaughterdesign. This corresponded closely to thefrequency of 0.69 estimated for the Y581allele of the ABCG2 gene for cows bornduring the same time period (Cohen-Zinderet al., 2005).

Chapter 12 – Marker-assisted selection in dairy cattle 209Methods to estimate QTL effectsand location in dairy cattleIf a significant effect on a quantitative traitis associated with a genetic marker, thedifference between the means of markergenotype classes will be a biased estimateof the QTL effect due to recombinationbetween the QTL and the genetic marker.Weller (1986) first demonstrated that maximumlikelihood (ML) methodology couldbe used to obtain estimates of QTL locationand effect unbiased by recombination,while Lander and Botstein (1989) proposedinterval mapping, based on ML for a QTLbracketed between two markers. Haley andKnott (1992) and Martinez and Curnow(1992) proposed an interval mappingmethod based on non-linear regression,which was easier to apply than ML. Theirmethods are not directly applicable to halfsibdesigns because, as noted previously,linkage relationships between the QTL andthe genetic markers will be different acrossfamilies, and in some families the commonancestor will be homozygous for the QTL.Furthermore, if multiple QTL alleles aresegregating in the population, or if theobserved effect is due to several tightlylinked QTL, the magnitude of the effectwill also differ across families.A method suitable for interval mappingthat accounts for these problems has beendeveloped by Knott, Elsen and Haley (1996)and has been applied to nearly all daughterand granddaughter design analyses. Theirmethod is a modification of the non-linearregression method, and assumes a singleQTL location for all families, but estimatesa separate QTL effect for each family. Thismethod has the advantage that, unlike ML,it can readily deal with missing and uninformativegenotypes for some markers.Mackinnon and Weller (1995) proposed anML method to estimate both QTL locationand effect for half-sib designs under theassumption that only two QTL alleles aresegregating in the population. Using thismethod it is also possible to estimate QTLgenotype of the common parent of eachfamily. However, these determinations areaccurate only for relatively large QTL. Themethod of Mackinnon and Weller (1995) ismore difficult to apply than the method ofKnott, Elsen and Haley (1996), and has notcome into general usage.Lander and Botstein (1989) proposedthe LOD-score (logarithm of the odds tothe base 10) drop-off method to estimateconfidence intervals for QTL location, butseveral studies have shown that this seriouslyunderestimate the actual value (e.g.Darvasi et al., 1993). The non-parametricbootstrap method (Visscher, Thompsonand Haley, 1996) was found to be moreaccurate, but tends to overestimate confidenceintervals. Bennewitz, Reinsch andKalm (2003) proposed improvements to thebootstrap method that result in shorter butstill unbiased confidence intervals.Most studies to detect QTL in dairycattle have considered many markers andmultiple traits. In some studies nearly theentire genome was analysed, which raises aserious problem with respect to the appropriatethreshold to declare significance. Ifnormal point-wise significance levels of 5 or1 percent are used, many marker-trait combinationswill show “significance” by chance.While this is a problem for all QTL genomescans, it is even more severe for dairy cattlein which multiple half-sib families are analysed,in addition to multiple markers andtraits. Several solutions to this problem havebeen proposed, none of which is completelysatisfactory. The only solution to deal adequatelywith both multiple traits and familiesin addition to multiple markers is the falsediscovery rate (Weller et al., 1998).

210Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishThe QTL effects derived from eitherdaughter or granddaughter by ML or nonlinearregression will still be biased forseveral reasons. First, the usual assumptionsof interval mapping, a single QTLsegregating within the marker interval andno QTL in adjacent intervals, often donot reflect reality. Second, the dependantvariable is generally an “adjusted” record,either daughter yield deviations (DYD;VanRaden and Wiggans, 1991) or geneticevaluations. Israel and Weller (1998) demonstratedthat QTL effects derived fromanalysis of either genetic evaluations, yielddeviations or DYD will be underestimated.In addition to this downward bias, thereare two sources of upward bias for QTLeffects. First, the direction of the effectsis generally arbitrary, and therefore absolutevalues are retained and all effects are>0. Third, only the effects deemed “significant”are retained, and this is a selectedsample (Georges et al., 1995). Bayesiananalysis methods that account for bias ofQTL effect due to selection have recentlybeen developed by Weller, Schlezinger andRon (2005).current status of QTL detectionin dairy cattleGenome scans by the granddaughterdesign have been completed for Holsteinsfrom Canada (Nadesalingam, Plante andGibson, 2001), the Netherlands (Spelmanet al., 1996; Schrooten et al., 2000), France(Bennewitz, et al., 2003a; Boichard et al.,2003), Germany (Bennewitz, et al., 2003a;Kuhn et al., 2003a), New Zealand (Spelmanet al., 1999), and the United States (Georgeset al., 1995; Ashwell et al., 1996, 1997,1998a, 1998b, 2004; Ashwell, Van Tasselland Sonstegard, 2001; Zhang et al., 1998;Ashwell and Van Tassell, 1999; Heyen etal., 1999); Finnish Ayrshires (Vilkki et al.,1997; Viitala et al., 2003; Schulman et al.,2004); French Normande and Montbeliardecattle (Boichard et al., 2003); Norwegiancattle in Norway (Klungland et al., 2001;Olsen et al., 2002); and Swedish Red andWhite (SRB) (Holmberg and Andersson-Eklund, 2004). Daughter design analyseshave been performed for Israeli Holsteins(Mosig et al., 2001; Ron et al., 2004). Moststudies have considered the five economicmilk production traits: milk, fat and proteinproduction, and fat and protein concentration,although a number of studies havealso considered somatic cell score (SCS),female fertility, herd life, calving traits,health traits, temperament and conformationtraits. The SCS is a log base 2 functionof the concentration of somatic cells, andhas been shown to be a useful indicator ofudder health. Results are summarized inTable 1.Results for milk, fat and protein production,fat and protein concentration, andSCS from most of the studies listed aboveare summarized at www.vetsci.usyd.edu.au/reprogen/QTL_Map/. Results fromthese traits, and many others includingmeat production, are summarized at http://bovineqtl.tamu.edu. Significant effects werefound on all 29 autosomes, but most effectswere found only in single studies and havenot been repeated. Khatkar et al. (2004)performed a meta-analysis, combining datafrom most of these studies, and foundsignificant across-study effects on chromosomes1, 3, 6, 9, 10, 14 and 20.Methods of incorporatinginformation from genetic markersin genetic evaluation systemsHeritabilities of most economic traits indairy cattle are low to moderate. Geneticevaluation of dairy cattle is complicatedby confounding between genetic and

Chapter 12 – Marker-assisted selection in dairy cattle 211Table 1Summary of dairy cattle genome scansExperimentaldesignBreed Country Traits analysed ReferencesGranddaughter Ayrshire Finland Milk production 1 Vilkki et al., 1997; de Koning et al.,2001; Viitala et al., 2003SCS 2 , mastitis, other treatments Schulman et al., 2004Jersey New Zealand Conformation Spelman, Garrick and vanArendonk, 1999Holstein Canadian Milk production Plante et al., 2001France Milk production Boichard et al., 2003Germany Milk production Thomsen et al., 2001Functional Kuhn et al., 2003Conformation, temperament, Hiendleder et al., 2003milking speedNetherlands conformation, SCS, fertility, Schrooten et al., 2000calving, milking speed,gestation, birth weight,temperamentNew Zealand Conformation Spelman, Garrick and vanArendonk, 1999USA Milk production Ashwell et al., 1998b; Ashwell andTassell 1999; Ashwell et al., 1997,2004; Ashwell, Van Tassell andSonstegard, 2001;Georges et al.,1995; Heyen et al., 1999; Zhang etal., 1998SCS Ashwell et al., 1996, 1997, 1998b;Ashwell and Van Tassell, 1999;Heyen et al., 1999Herdlife Heyen et al., 1999Conformation Ashwell et al., 1998a, 1998b;Ashwell and Van Tassell, 1999Fertility Ashwell et al., 2004Montbeliarde France Milk production Boichard et al., 2003Normande France Milk production Boichard et al., 2003Norwegian Norway Milk production Olsen et al., 2002SCS, mastitis Klungland et al., 2001Swedish Sweden SCS, mastitis, other diseases Holmberg and Andersson-Eklund,2004Daughter Holstein Israel Milk production, SCS, fertility, Ron et al., 2004herdlife% protein Mosig et al., 20011Milk, fat, and protein production, and fat and protein concentration.2Somatic cell concentrationenvironmental factors. Cows are scatteredover many different herds with differentmanagement levels, and distribution of siresacross herds is not random or orthogonal.Furthermore, cows generally producemultiple lactations that are correlated. Inorder to account for the limited heritability,and co-variances among relatives, geneticeffects are generally assumed to berandom, while most environmental effectsare assumed to be fixed. Thus, geneticevaluation is performed by the mixedmodel using best linear unbiased prediction(BLUP) methodology (Henderson, 1984).Beginning in the late 1980s, the modelof choice for genetic evaluation for milkproduction traits was the individual animalmodel, in which a genetic effect is computed

212Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishfor each animal, including animals that didnot have production records (Westall andvan Vleck, 1987). Genetic evaluations forthese animals are derived via the numeratorrelationship matrix, which is included in themodel. In addition, a “permanent environmental”effect is computed for each animalwith records to account for similaritiesamong multiple records of the same cowthat are not due to additive genetic effects.As noted previously, analysis of QTLeffects has generally been based on analysisof genetic evaluations or DYD, which arethe adjusted means of the daughter recordsof a bull but which, unlike genetic evaluations,are not regressed. However, thestatistical properties of DYD are not wellunderstood, and QTL effects derived fromanalysis of DYD are still biased (Israeland Weller, 1998). Theoretically, it shouldbe possible to derive unbiased QTL estimatesif these effects are incorporated intoa genetic evaluation scheme based on analysisof the actual records, such as the animalmodel. In practice, the inclusion of QTLeffects into genetic evaluation models iscomplicated by three main factors:• actual QTL location is unknown, andthere is only partial linkage betweengenetic markers and QTL;• linkage phase between genetic markersand QTL differs among individuals, andis generally unknown;• only a small fraction of the population isgenotyped.An analysis including only genotypedindividuals is not a viable option as it willgenerally not be possible to derive accuratefixed effects, such as herd-year-seasons,from this sample.Fernando and Grossman (1989) proposedmodifying the individual animalmodel described above to a “gametic”model that assumes the two QTL allelesof each individual are random effects sampledfrom a distribution with a knownvariance. They developed a method to estimatebreeding values for all individualsin a population, including QTL effectsvia linkage to genetic markers, providedthat all animals are genotyped and theheritability and recombination frequencybetween the QTL and the genetic markerare known. This model is suitable for anypopulation structure and can also incorporatenon-linked polygenic effects and other“nuisance” effects such as herd or block.The basic model assumes only a singlerecord per individual, but can be adaptedreadily to a situation of multiple recordsper animal. This method is also denoted“marker-assisted BLUP” or “MA-BLUP”.Each individual with unknown ancestorsis assumed to have two unique allelesfor the QTL, which are “sampled” from aninfinite population of alleles. For animalsthat are not genotyped, the probability ofreceiving either allele from either parent willbe equal. However, if both the parent andprogeny are genotyped for a linked geneticmarker, then the probability of receiving aspecific parental allele for a QTL linked tothe genetic marker will be a function of theprogeny marker genotype and recombinationfrequency. Based on these probabilities,Fernando and Grossman (1989) demonstratedhow a variance-co-variance matrixcould be constructed for the QTL gameticeffects. They further described a simplealgorithm to invert this matrix analogousto Henderson's method for invertingthe numerator relationship matrix. Thismethod has been extended to handle multiplemarkers and traits (Goddard, 1992).Cantet and Smith (1991) demonstrated thatthe number of equations could be significantlyreduced by analysis of the reducedanimal model.

Chapter 12 – Marker-assisted selection in dairy cattle 213The disadvantages of this model are thatit assumes that both recombination frequencyand the variance due to the QTL areknown a priori. Studies on simulated datahave demonstrated that although restrictedmaximum likelihood methodology can beused to estimate these parameters, theyare completely confounded for a singlemarker locus (van Arendonk et al., 1994).Methods to estimate the variance contributedby QTL with multiple markers weredeveloped by Grignola, Hoeschele and Tier(1996). Furthermore, as each individualwith unknown parents is assumed to havetwo unique alleles, the prediction errorvariances of the effects for any individualwill be quite large and, therefore, not veryinformative. Finally, the assumption of anormal distribution of possible QTL alleleeffects may not be realistic.Israel and Weller (1998) proposed analternative method that assumes that onlytwo QTL alleles are segregating in thepopulation, and that either a daughter orgranddaughter design has been applied todetermine QTL genotypes of the familyancestors. The QTL effect is then includedin the complete animal model analysis asa fixed effect. For individuals that arenot genotyped, probabilities of receivingeither allele are included as regressionconstants. These probabilities can be readilycomputed for the entire population usingthe segregation analysis method of Kerrand Kinghorn (1996). Israel and Weller(1998) assumed complete linkage betweenthe QTL and a single marker. Israel andWeller (2002) extended the method to QTLanalysis based on flanking marker, usingthe method of Whittaker, Thompsom andVisscher (1996) to estimate QTL effectsand location from the regression estimatesof flanking markers. This method has beentested extensively on simulated populations,and was able to derive unbiased estimatesof QTL effect and location. It has alsobeen applied to actual data from the IsraeliHolstein population for a segregatingQTL on chromosome 14 that affectedmilk production traits (Weller et al., 2003).However, in this case the QTL effectwas underestimated. Further research isrequired to determine the reason for thisdiscrepancy.Methods for QTL detection andMAS in developing countriesAs noted previously, dairy cattle breedingin tropical and subtropical countries is generallybased on crossbreeding between highproduction breeds adapted to temperateclimates, and tropical strains which areadapted to the local environment, includingresistance to local diseases. In other animalspecies, synthetic strains have been producedby selecting those individuals thatretain the positive characteristics fromboth strains. For example, the Assaf sheepbreed was produced from a cross betweenthe Middle East Awassi breed and theEast Friesian breed (www.sheep101.info/breedsA.html). In dairy cattle, the problemof an appropriate strategy for future generationshas not been adequately solved,for reasons considered previously. If theeconomically important genes were identified,then the time and effort required forproduction of the desired synthetic strainscould be reduced.Visscher, Haley and Thompson (1996)considered the situation in which therecipient strain is an outbred populationin an ongoing selection programme, andthe introgressed genes are QTL. Markersflanking the QTL will be required in orderto select backcross progeny that received thedonor QTL allele. As there will be uncertaintywith respect to the QTL location,

214Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishthe flanking markers must be sufficientlyclose to the QTL so that it will be possibleto determine with relative certaintythat the QTL is in fact located betweenthe flanking markers. Although markerassistedintrogression does decrease thenumber of generations required to obtainfixation of the desired allele, it increasestwo key cost elements. First, with traditionalintrogression, half of the progenywill carry the donor allele for the introgressedgene, and all of these can be usedas parents in the next generation. However,if only a small fraction of the progeny isselected based on genetic markers, thenmany more individuals must be producedeach generation. Second, genotyping costsfor a large number of markers at each generationwill also be significant.Crosses between cattle breeds can alsobe used for QTL detection and they havebeen used in developing countries. In mostplant species, the parental lines are completelyinbred, and there will be completeLD in the F 2 or backcross generation.However, cattle are outbreeders and incrosses between breeds there will thereforeonly be partial LD between segregatingQTL and linked genetic markers. Song,Soller and Genizi (1999) proposed the fullsibintercross line (FSIL) design for QTLdetection and mapping for crosses betweenstrains of outcrossing species. They assumedthat the two parental strains differ in allelicfrequencies, but were not at fixation foralternative QTL alleles.For given statistical power, the FSILdesign requires only slightly more individualsthan an F 2 design derived from aninbred line cross, but six- to ten-fold fewerthan a half-sib or full-sib design. In addition,as the population is maintained bycontinued intercrossing, DNA samples andphenotypic information can be accumulatedacross generations. Continued intercrossingin future generations also leads to mapexpansion, and thus to increased mappingaccuracy in the later generations. AnFSIL can therefore be used for fine mappingof QTL and this is considered belowin detail.Although these methods have not as yetbeen applied to detect QTL related to milkproduction, they have been applied to QTLfor disease resistance. Trypanosomosis(sleeping sickness) is a major constrainton livestock productivity in sub-SaharanAfrica. Hanotte et al. (2003) mappedQTL affecting trypanotolerance in a crossbetween the “tolerant” N’Dama breed andthe susceptible Boran breed. Putative QTLaffecting 16 traits associated with diseasesusceptibility were mapped tentatively to18 autosomes. Excluding chromosomeswith ambiguous effects, the allele associatedwith resistance was derived from theN’Dama strain for nine QTL and from theBoran strain for five QTL. These results areconsistent with many plant crossbreedingexperiments in which the strain with overallphenotypic inferiority for the quantitativetrait nevertheless harbours QTL alleles thatare superior to the alleles present in thephenotypically superior strain (e.g. Weller,Soller and Brody, 1988).From QTL to QTN – theoryAs noted by Darvasi and Soller (1997),with a saturated genetic map, the resolvingpower for QTL will be a function of theexperimental design, number of individualsgenotyped and QTL effect. Weller andSoller (2004) computed that the 95 percentconfidence interval (CI) in percent recombinationfor half-sib designs, including thedaughter and granddaughter designs, was3073/d 2 N, where d is the QTL substitutioneffect in units of the standard deviation, and

Chapter 12 – Marker-assisted selection in dairy cattle 215N is the sample size. In the case of a granddaughterdesign, the units for the standarddeviation will be either units of the bulls’DYD or genetic evaluations. For example,if d is 0.5 and N is 400, the CI will be 31 percentrecombination, or approximately 35cM. Thus, except for the largest QTL, CIswill generally include several tens of cM.Considering that each cattle cM includes ~8genes and one million bp, detection of theactual polymorphism responsible for theobserved QTL effects (the quantitative traitnucleotide, QTN) appears at first glance tobe a “mission impossible”.Various strategies have been proposedto reduce the CI based on multiplecrosses, but most are not applicable todairy cattle (e.g. Darvasi, 1998). Meuwissenand Goddard (2000) proposed that CI forQTL location could be reduced to individualcM by application of LD mapping.If a QTL polymorphism is due to a relativelyrecent mutation or to a relativelyrecent introduction from another population,then it should be possible to detectpopulation-wide LD between the QTLand closely linked genetic markers. Thecloser the marker to the QTL, the greaterwill be the extent of LD. They developeda method to estimate QTL location and CIbased on LD between a QTL and a seriesof closely linked markers. The CI can befurther reduced by combining linkage andLD mapping (Meuwissen et al., 2002), andby a multitrait analysis (Meuwissen andGoddard, 2004). However, unless the QTLeffect is very large, the CI will still extendover several cM.In order to determine the actual generesponsible for the QTL, most studies haveused the “candidate gene” approach, i.e.to determine a likely candidate among thegenes within the CI, based on known genefunction, or specific gene expression in theorgan of interest. Examples are given inthe following section. However, even if apolymorphism is detected in the candidategene and the polymorphism has a majorLD effect on the QTL, how does one provethat this polymorphism is not merely inLD with the actual QTN?Mackay (2001) proposed two alternativesfor proof positive that a candidatepolymorphism is in fact the QTN, namely,co-segregation of intragenic recombinantgenotypes in a candidate gene with theQTL phenotype, and functional complementationwhere the trait phenotype is“rescued” in a transgenic organism. Neitherof these is applicable to QTL in dairy cattle.In this case, Mackay (2001) postulated thatthe only option to achieve the standard ofrigorous proof for identification of a geneunderlying a QTL in commercial animalpopulations is to collect “multiple piecesof evidence, no single one of which is convincing,but which together consistentlypoint to a candidate gene”. Evidence can beprovided by concordance of polymorphismwith deduced QTL genotype, quantitativedifferences of gene expression in physiologicallyrelevant organs, SNP capableof encoding a non-conservative aminoacid change, protein differences in cowswith contrasting genotypes for the QTN,orthologous QTL in other species (genesthat are derived from a common ancestralgene) and alteration of gene proteinin bovine cell lines by “short interferingRNA” (siRNA) technology. (The siRNAmolecules bind with proteins to form a unitcalled the “RNA-induced silencing complex”that suppresses the expression of thegene to which it corresponds in the viralgenome, silencing the gene from which thesiRNA is derived.)For dairy cattle, to date, the most compellingevidence is “concordance”, i.e. that

216Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishthe deduced QTL genotypes of a sample ofindividuals correspond completely to theirgenotypes for the putative QTN. All individualsheterozygous for the QTL shouldbe heterozygous for the putative QTN,with the same QTN allele associated withthe same QTL allele in all individuals, andall individuals homozygous for the QTLshould also be homozygous for the QTN.Theoretically, the sample of individualsanalysed should be large enough to rejectstatistically the hypothesis that concordancewas obtained by chance. However,in dairy cattle, the only individuals forwhich QTL genotype can be derived withany level of reliability are sires that havebeen analysed by either a daughter orgranddaughter design, and the numberof these individuals will always be limited.Furthermore, there is at present noaccepted theory to compute concordanceprobabilities by chance, considering thatany polymorphism very close to the QTNwill display significant LD. Several studieshave addressed the problem (Cohen-Zinderet al., 2005; Schnabel et al., 2005). The casefor identification of the QTN is clearlymore compelling if concordance is obtainedin two different populations.From QTL to QTN – resultsTo date, the QTN has been determinedin two cases in dairy cattle, on BTA 6 andBTA 14. In both cases the QTL chieflyaffected fat and protein concentration andthe QTL effect was large enough that theconfidence interval for QTL location was

Chapter 12 – Marker-assisted selection in dairy cattle 217ferentially expressed in the mammary glandduring lactation, as compared with the liver.Furthermore, anti-sense SPP1 transgenicmice displayed abnormal mammary glanddifferentiation and milk secretion (Nemiret al., 2000).Schnabel et al. (2005) found concordancebased on four heterozygous and fourhomozygous sires for the United StatesHolstein population, as determined by agranddaughter design, while Cohen-Zinderet al. (2005) found concordance for threeheterozygous and 15 homozygous siresfrom both the United States and IsraeliHolstein populations. Cohen-Zinder et al.(2005) also analysed the site proposed bySchnabel et al. (2005), and found that thissite was hyper-variable in that at leastfour single nucleotide changes were foundwithin the 20 bp region centred on thepoly-A sequence. Eight of nine Israelisires analysed by the daughter design wereheterozygous for at least one of thesepolymorphisms.Many studies have found a QTLaffecting all five milk production traits andSCS near the middle of BTA 20. Blott et al.(2003) claimed that a mis-sense mutationin the bovine growth hormone receptorwas responsible for the QTL affectingmilk yield and composition on BTA 20,but did not find concordance for the bullsheterozygous for the QTL. Thus, this polymorphismmay be responsible for onlypart of the observed effect on BTA 20, ormay be a physiologically neutral mutationin LD with the QTN.For both the QTL on BTA 6 and 14,the polymorphisms analysed apparently donot account for the entire effect observed inthese chromosomal regions (Bennewitz etal., 2004a; Kuhn et al., 2004; Cohen-Zinderet al., 2005). The effect associated with themis-sense mutation in ABCG2 explains theentire effect observed on milk yield and fatand protein concentration, but does notexplain the effects associated with fat andprotein yield. It is likely that in the nearfuture additional QTN will be resolved.As noted, the meta-analysis (Khatkar et al.,2004) found significant effects on BTA 1, 3,9 and 10, in addition to the effects describedon BTA 6, 14 and 20.Methods and theory for MAS indairy cattleConsidering the long generation interval,the high value of each individual, the verylimited female fertility and the fact thatnearly all economic traits are expressedonly in females, it would seem that dairycattle should be a nearly ideal species forapplication of MAS. However, most theoreticalstudies have been rather pessimisticwith respect to the expected gains that canbe obtained by MAS. As noted by Weller(2001), MAS can potentially increase annualgenetic gain by increasing the accuracy ofevaluation, increasing the selection intensityand decreasing the generation interval.The following dairy cattle breedingschemes that incorporate MAS have beenproposed:• a standard PT system, with informationfrom genetic markers being used toincrease the accuracy of sire evaluationsin addition to phenotypic informationfrom daughter records (Meuwissen andvan Arendonk, 1992);• a MOET nucleus breeding scheme inwhich marker information is used toselect sires for service in the MOETpopulation, in addition to phenotypicinformation on half-sisters (Meuwissenand van Arendonk, 1992);• PT schemes in which information ongenetic markers is used to preselect youngsires for entrance into the PT (Kashi,

218Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishHallerman and Soller, 1990; Mackinnonand Georges, 1998);• selection of bull sires without a PT, basedon half-sib records and genetic markers(Spelman, Garrick and van Arendonk,1999);• selection of sires in a half-sib scheme,based on half-sib records and geneticmarkers (Spelman, Garrick and vanArendonk, 1999);• use of genetic markers to reduce errorsin parentage determination (Israel andWeller, 2000).Meuwissen and van Arendonk (1992)found that inclusion of marker informationto increase the accuracy of sire evaluationsincreased the rate of genetic gain byonly 5 percent when the markers explained25 percent of the genetic variance. Thisresult is not surprising considering that theaccuracy of sire evaluations based on a PTof 50 to 100 daughters is already quite high.In “open” and “closed” nucleus breedingschemes, rates of genetic gain were increasedby 26 and 22 percent, respectively. Theadvantage of MAS in this case is greater,because young sires are not progeny tested,and their reliabilities based only on half-sibinformation are much lower.Mackinnon and Georges (1998) proposed“top-down” and “bottom-up” strategies toapply the third scheme listed above, preselectionof young sires prior to PT. In the“top-down” strategy, QTL genotypes aredetermined for the elite sires used as bullsires by a granddaughter design. If a densemarker map is available, it will then be possibleto determine which QTL allele is passedto each son. Elite bulls from among thesesons are then selected as bull sires for thenext generation. If the original sire was heterozygousfor a QTL, it can be determinedwhich of his sons received the favourableallele. Sons of these sires are then genotypedand selected based on whether they receivedthe favourable grandpaternal QTL alleles.It is assumed that the dams of the candidatesires are also genotyped, and that these cowswill be progeny of the sires evaluated by agranddaughter design. Thus, grandpaternalalleles inherited via the candidates’ dams canalso be traced. A disadvantage of this schemeis that only the grandpaternal alleles are followed.Some of the sons of the original siresthat were evaluated by a granddaughterdesign will also have received the favourableQTL allele from their dams, but not via thegenotyped grandsires. However, young sireswill be selected based only on the grandpaternalhaplotypes.In the “bottom-up” scheme, QTL genotypesof elite sires are determined bya daughter design. These sires are thenused as bull sires. The candidate bullsare then pre-selected for those QTL heterozygousin their sires, based on whichpaternal haplotype they received. As theQTL phase is evaluated on the sires of thebull calves (the candidates for selection),no selection pressure is “wasted” as inthe “top-down” scheme. In addition, thisdesign can be applied to a much smallerpopulation, because only several hundreddaughters are required to evaluate each bullsire. On the negative side, more daughtersthan sons must be genotyped to determineQTL genotype. Mackinnon and Georges(1998) assumed that in either scheme it willnot be necessary to increase mean generationinterval above that of a traditional PTprogramme, although this will probablynot be the case (Weller, 2001).Kashi, Hallerman and Soller (1990),Mackinnon and Georges (1998), and Israeland Weller (2004) all addressed the problemthat QTL determination will be subjectto error. Deciding that a specific sire ishomozygous for the QTL when in fact

Chapter 12 – Marker-assisted selection in dairy cattle 219the sire is heterozygous will be denotedthe “type I” error. Deciding that the QTLis heterozygous in a specific sire, whilethe sire is in reality homozygous will bedenoted the “type II” error. In the firstcase, segregating QTL will be missed while,in the second case, selection for the positiveQTL allele will be applied to no advantage.All three studies found that geneticgains will be maximized with a relativelylarge proportion of type I errors, between5 and 20 percent. This is due to the factthat as type I error increases, type II errordecreases, and more real effects will bedetected and applied in selection. A thirdtype of error is theoretically possible, i.e.determining correctly that the ancestor isheterozygous for the QTL, but incorrectdetermination of QTL phase relative tothe genetic markers. However, Israel andWeller (2004) showed by simulation thatthis error never occurred even when thetype I error rate was set at 20 percent.Spelman, Garrick and van Arendonk(1999) considered three different breedingschemes by deterministic simulation:• a standard PT with the inclusion of QTLdata;• the same scheme except that young bullswithout PT could also be used as bullsires based on QTL information;• a scheme in which young sires could beused as both bull sires and cow sires inthe general population, based on QTLinformation.It was assumed that only bulls weregenotyped but that, once genotyped, theinformation on QTL genotype and effectwas known without error. It was then possibleto conduct a completely deterministicanalysis. They varied the fraction of thegenetic variance controlled by known QTLfrom zero to 100 percent. Even withoutMAS, a slight gain was obtained by allowingyoung sires to be used as bull sires, and agenetic gain of 9 percent was obtained ifyoung sires with superior evaluations werealso used directly as both sires of sires andin general service. As noted previously, thegenetic gain was limited where MAS wasused only to increase the accuracy of youngbull evaluations for a standard PT schemebecause the accuracy of the bull evaluationswas already high. Thus, even if all thegenetic variance was accounted for by QTL,the genetic gain was less than 25 percent.However, if young sires are selected forgeneral service based on known QTL, therate of genetic progress can be doubled. Themaximum rate of genetic gain that can beobtained in the third scheme, the “all bulls”scheme, was 2.2 times the rate of geneticgain in a standard PT. Theoretically, withhalf of the genetic variance due to knownQTL, the rate of genetic gain obtainedwas greater than that possible with nucleusbreeding schemes.The final scheme, with use of geneticmarkers to reduce parentage errors, is themost certain to produce gains, as it doesnot rely on QTL genotype determination,which may be erroneous. Weller et al.(2004) genotyped 6 040 Israeli Holsteincows from 181 Kibbutz herds for 104microsatellites. The frequency of rejectedpaternity was 11.7 percent, and most errorswere due to inseminator mistakes. Mostadvanced breeding schemes already usegenetic markers to confirm parentage ofyoung sires. Israel and Weller (2002) foundby simulations that if the parentage of bulldams and the test daughters of young siresare also verified, genetic gain increasedby 4.3 percent compared with a breedingprogramme with 10 percent incorrectpaternity. This scheme is economicallyjustified if genotyping costs per individualare no more than US$15.

220Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishcurrent status of MAS in dairycattleTwo ongoing MAS programmes in dairycattle have been reported to date, inFrench and German Holsteins (Boichardet al., 2002, 2006; Bennewitz et al., 2004b).Currently in the German programme,markers on three chromosomes are used.The MA-BLUP evaluations (Fernandoand Grossman, 1989) are computed at theVIT-computing centre in Verden, and aredistributed to Holstein breeders who canuse these evaluations for selection of bulldams and preselection of sires for progenytesting. The MA-BLUP algorithm onlyincludes equations for bulls and bull dams,and the dependent variable is the bull’sDYD (Bennewitz et al., 2003b). Linkageequilibrium throughout the population isassumed. To close the gap between thegrandsire families analysed in the Germangranddaughter design and the current generationof bulls, 3 600 bulls were genotypedin 2002. As then, about 800 bulls have beenevaluated each year (N. Reinsch, personalcommunication). Only bulls and bull damsare genotyped as tissue samples are alreadycollected for paternity testing. Thus additionalcosts due to MAS are low and evena very modest genetic gain can be economicallyjustified. This scheme is similar to the“top-down” scheme of Mackinnon andGeorges (1998) in that evaluation of thesons is used to determine which grandsiresare heterozygous for the QTL and theirlinkage phase. This information is thenused to select grandsons based on whichhaplotype was passed from their sires. Itdiffers from the scheme of Mackinnon andGeorges (1998) in that the grandsons arepreselected for PT based on MA-BLUPevaluations, which include general pedigreeinformation in addition to genotypes.The French MAS programme includeselements of both the “top-down” and“bottom-up” MAS designs. Similar to theGerman programme, genetic evaluationsincluding marker information were computedby a variant of MA-BLUP, andonly genotyped animals and non-genotypedconnecting ancestors were includedin the algorithm. Genotyped females werecharacterized by their average performancebased on pre-corrected records (with theappropriate weight), whereas males werecharacterized by twice the yield deviationof their non-genotyped daughters. Twelvechromosomal segments, ranging in lengthfrom 5 to 30 cM, are analysed. Regionswith putative QTL affecting milk productionor composition are located on BTA 3,6, 7, 14, 19, 20 and 26; segments affectingmastitis resistance are located on BTA 10,15 and 21; and chromosomal segmentsaffecting fertility are located on BTA 1, 7and 21. Each region was found to affect oneto four traits and on average three regionswith segregating QTL were found for eachtrait. Each region is monitored by two tofour evenly spaced microsatellites, and eachanimal included in the MAS programme isgenotyped for at least 43 markers. Sires anddams of candidates for selection, all maleAI ancestors, up to 60 AI uncles of candidates,and sampling daughters of bull siresand their dams are genotyped. The numberof genotyped animals was 8 000 in 2001 and50 000 in 2006. An additional 10 000 animalsare genotyped per year, with equalproportions of candidates for selection andhistorical animals.Future prospective for MAS indairy cattleAlthough the first large experiment inQTL detection in dairy cattle was publishedin 1961 by Neimann-Sørensen andRobertson, in 1985 it still looked as if

Chapter 12 – Marker-assisted selection in dairy cattle 221MAS was a long way off for commercialanimal populations as there were very fewknown genetic markers and methodologywas rudimentary. In the last 20 years therehave been huge advances in both DNAtechnology and statistical methodology,and it can now be stated with near certaintythat the technology is available to detectand map accurately segregating QTL indairy cattle. Furthermore, although manyeffects reported in the literature are “falsepositives”, there is a wealth of evidence thatseveral QTL are in fact real as a number ofeffects have been repeated across numerousexperiments, and the actual QTN havebeen identified for at least two QTL.The main limitation at this point todetecting and mapping more QTL is thesample sizes available, especially the numberof progeny tested bulls per family. To mapQTL of smaller magnitude accurately, itwill be necessary to combine data acrossexperiments (e.g. Khatkar et al., 2004) orsignificantly increase sample sizes. This canonly be done by genotyping cows, eventhough power per individual genotypedwill be lower.The fact that only two countries haveactually started MAS programmes highlightsthe current limitations to practicalapplication of MAS. To date, very few segregatingQTL with economic impact havebeen identified in commercial dairy cattlepopulations. Of the two QTNs that havebeen detected, each has disadvantages withrespect to application in MAS. The alleleof DGAT1 that increases fat productionand decreases water content in the milk,both desirable, also decreases protein yield,which is undesirable (Weller et al., 2003).The allele of ABCG2 that decreases milkproduction and increases protein percentis clearly the favourable allele in nearly allcurrent selection indices, but this allele isalready at a very high frequency in all majordairy cattle populations (Ron et al., 2006).In addition to the limitation of definitivelyidentified QTL with economic value,suitable software for genetic evaluationincluding QTL effects is also a limitingfactor. At present, those countries that areapplying MAS are using two-step procedures,i.e. a preliminary analysis to computegenetic evaluations based only on pedigreeand phenotypic data, and then a secondanalysis in which the genetic evaluationsare “adjusted” for QTL effects. Ideally asingle algorithm should be used to derivegenetic evaluations for the entire populationincluding the effects of known QTL.AcknowledgementsI thank M. Ron, E. Seroussi and M. Ashwellfor their input.ReferencesAshwell, M.S. & Van Tassell, C.P. 1999. Detection of putative loci affecting milk, health, and typetraits in a US Holstein population using 70 microsatellite markers in a genome scan. J. Dairy Sci.82: 2497–2502.Ashwell, M.S., Van Tassell, C.P. & Sonstegard, T.S. 2001. A genome scan to identify quantitativetrait loci affecting economically important traits in a US Holstein population. J. Dairy Sci. 84:2535–2542.Ashwell, M.S., Rexroad, C.E., Miller, R.H. & VanRaden, P.M. 1996. Mapping economic trait loci forsomatic cell score in Holstein cattle using microsatellite markers and selective genotyping. Anim.Genet. 27: 235–242.

222Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishAshwell, M.S., Rexroad, C.E., Miller, R.H., VanRaden, P.M. & Da, Y. 1997. Detection of lociaffecting milk production and health traits in an elite US Holstein population using microsatellitemarkers. Anim. Genet. 28: 216–222.Ashwell, M.S., Da, Y., VanRaden, P.M., Rexroad, C.E. & Miller, R.H. 1998a. Detection of putativeloci affecting conformational type traits in an elite population of United States Holsteins using microsatellitemarkers. J. Dairy Sci. 81: 1120–1125.Ashwell, M.S., Da, Y., Van Tassell, C.P., VanRaden, P.M., Miller, R.H. & Rexroad, C.E. 1998b.Detection of putative loci affecting milk production and composition, health, and type traits in aUnited States Holstein population. J. Dairy Sci. 81: 3309–3314.Ashwell, M.S., Heyen, D.W., Sonstegard, T.S., Van Tassell, C.P., DA, Y., VanRaden, P.M., Ron,M., Weller, J.I. & Lewin, H.A. 2004. Detection of quantitative trait loci affecting milk production,health, and reproductive traits in Holstein cattle. J. Dairy Sci. 87: 468–475.Bennewitz, J., Reinsch, N. & Kalm, E. 2003. Comparison of several bootstrap methods for biasreduction of QTL effect estimates. J. Anim. Breed. Genet. 120: 403–416.Bennewitz, J., Reinsch, N., Grohs, C., Leveziel, H., Malafosse, A., Thomsen, H., Xu, N.Y., Looft,C., Kuhn, C., Brockmann, G.A., Schwerin, M., Weimann, C., Hiendleder, S., Erhardt, G.,Medjugorac, I., Russ, I., Forster., M., Brenig, B., Reinhardt, F., Reents, R., Averdunk, G.,Blumel, J., Boichard, D. & Kalm, E. 2003a. Combined analysis of data from two granddaughterdesigns: A simple strategy for QTL confirmation and increasing experimental power in dairy cattle.Genet. Sel. Evol. 35: 319–38.Bennewitz, J., Reinsch, N., Thomsen, H., Szyda, J., Reinhart, F., Kuhn, C., Schwerin, M., Erhardt,G., Weimann, C. & Kalm, E. 2003b. Marker assisted selection in German Holstein dairy cattlebreeding: outline of the program and marker assisted breeding value estimation. Ann. Meet. Europ.Ass. Anim. Prod. 54th Session G1.9. Rome, Italy.Bennewitz, J., Reinsch, N., Paul, S., Looft, C., Kaupe, B., Weimann, C., Erhardt, G., Thaller, G.,Kuhn, C., Schwerin, M., Thomsen, H., Reinhardt, F., Reents, R. & Kalm, E. 2004a. The DGAT1K232A mutation is not solely responsible for the milk production quantitative trait locus on thebovine chromosome 14. J. Dairy Sci. 87: 431–442.Bennewitz, J., Reinsch, N., Reinhardt, F., Liu, Z. & Kalm, E. 2004b. Top down preselection usingmarker assisted estimates of breeding values in dairy cattle. J. Anim. Breed. Genet. 121: 307–318.Blott, S., Kim, J., Moisio, S., Schmidt-Kuntzel, A., Cornet, A., Berzi, P., Cambisano, N., Ford, C.,Grisart, B., Johnson, D., Karim, L., Simon, P., Snell, R., Spelman, R., Wong, J., Vilkki, J., Georges,M., Farnir, F. & Coppieters, W. 2003. Molecular dissection of a quantitative trait locus: A phenylalanine-to-tyrosinesubstitution in the transmembrane domain of the bovine growth hormonereceptor is associated with a major effect on milk yield and composition. Genetics 163: 253–266.Boichard, D., Fritz, S., Rossignol, M.N., Boscher, M.Y., Malafosse, A. & Colleau, J.J. 2002.Implementation of marker-assisted selection in French dairy cattle. Proc. 7th World Cong. Genet.Appl. Livest. Prod. 33: 19–22. Montpellier, France.Boichard, D., Fritz, S., Rossignol, M.N., Guillaume, F., Colleau, J.J. & Druet, T. 2006.Implementation of marker-assisted selection: practical lessons from dairy cattle. Proc. 8th WorldCong. Genet. Appl. Livest. Prod. 22: 11. 13–18 August 2006, Belo Horizonte, MG, Brazil.Boichard, D., Grohs, C., Bourgeois, F., Cerqueira, F., Faugeras, R., Neau, A., Rupp, R., Amigues,Y., Boscher, M.Y. & Leveziel, H. 2003. Detection of genes influencing economic traits in threeFrench dairy cattle breeds. Genet. Sel. Evol. 35: 77–101.

Chapter 12 – Marker-assisted selection in dairy cattle 223Brascamp, E.W., van Arendonk J.A.M. & Groen A.F. 1993. Economic appraisal of the utilization ofgenetic markers in dairy cattle breeding. J. Dairy Sci. 76: 1204–1213.Cantet, R.J.C. & Smith, C. 1991. Reduced animal model for marker assisted selection using bestlinear unbiased prediction. Genet. Sel. Evol. 23: 221–233.Cohen-Zinder, M., Seroussi, E., Larkin, D.M., Loor, J.J., Everts-van der Wind, A., Lee, J.H.,Drackley, J.K., Mark R. Band, M.R., Hernandez, A.G., Shani, M., Lewin, H.A., Weller, J.I. &Ron, M. 2005. Identification of a missense mutation in the bovine ABCG2 gene with a major effecton the QTL on chromosome 6 affecting milk yield and composition in Holstein cattle. GenomeResearch 15: 936–944.Coppieters, W., Kvasz, A., Arranz, J.J., Grisart, B., Riquet, J., Farnir, F. & Georges, M. 1999. Thegreat-grand-daughter design: a simple strategy to increase the power of a grand-daughter design forQTL mapping. Genetical Research 74: 189–199.Cunningham, E.P. 1989. Formulation of breeding plans for dairy and dual-purpose cattle. Brazil. J.Genet. 12: 81.Darvasi, A. 1998. Experimental strategies for the genetic dissection of complex traits in animalmodels. Nat. Genet. 18: 19–24.Darvasi, A. & Soller, M. 1997. A simple method to calculate resolving power and confidence intervalof QTL map location. Behavior. Genet. 27: 125–132.Darvasi, A., Vinreb, A., Minke, V., Weller, J.I. & Soller, M. 1993. Detecting marker-QTL linkageand estimating QTL gene effect and map location using a saturated genetic map. Genetics 134:943–951.de Koning, D.J., Schulman, N.F., Elo, K., Moisio, S., Kinos, R., Vilkki, J. & Maki-Tanila, A. 2001.Mapping of multiple quantitative trait loci by simple regression in half-sib designs. J. Anim. Sci.79: 616–622.de Vaccaro, L. 1990. Survival of European dairy breeds and their crosses with Zebus in the tropics.Anim. Breed. Abstr. 58: 475–494.Fernando R.L. & Grossman, M. 1989. Marker assisted selection using best linear unbiased prediction.Genet. Sel. Evol. 21: 467–477.FAO. 1987. Crossbreeding Bos indicus and Bos taurus for milk production in the tropics, by E.P.Cunningham & O. Syrstad. Anim. Prod. Health Paper No. 68. Rome.Farnir, F., Coppieters, W., Arranz, J.J., Berzi, P., Cambisano, N., Grisart, B., Karim, L., Marcq.,F., Moreau, L., Mni, M., Nezer, C., Simon, P., Vanmanshoven, P., Wagenaar, D. & Georges, M.2000. Extensive genome-wide linkage disequilibrium in cattle. Genome Research 10: 220–227.Georges, M., Nielsen, D., Mackinnon, M., Mishra, A., Okimoto, R., Pasquino, A.T., Sargent, L.S.,Sorensen, A., Steele, M.R., Zhao, X., Womack, J.E. & Hoeschele, I. 1995. Mapping quantitative traitloci controlling milk production in dairy cattle by exploiting progeny testing. Genetics 139: 907–920.Goddard, M.E. 1992. A mixed model for analyses of data on multiple genetic markers. Theor. Appl.Genet. 83: 878–886.Grignola, F.E., Hoeschele, I. & Tier, B. 1996. Mapping quantitative trait loci in outcross populationsvia residual maximum likelihood. I. Methodology. Genet. Sel. Evol. 28: 479–490.Grisart, B., Coppieters, W., Farnir, F., Karim, L., Ford, C. Berzi, P., Cambisano, N., Mni, M., Reid,S., Simon, P., Spelman, R., Georges, M. & Snell, R. 2002. Positional candidate cloning of a QTLin dairy cattle: Identification of a missense mutation in the bovine DGAT1 gene with major effecton milk yield and composition. Genome Research 12: 222–231.

224Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishHaley, C.S. & Knott, S.A. 1992. A simple regression method for mapping quantitative trait loci inline crosses using flanking markers. Heredity 69: 315–324.Hanotte, O., Ronin, Y., Agaba, M., Nilsson, P., Gelhaus, A., Horstmann, R., Sugimoto, Y., Kemp,S., Gibson, J., Korol, A., Soller, M. & Teale, A. 2003. Mapping of quantitative trait loci controllingtrypanotolerance in a cross of tolerant West African N’Dama and susceptible East African Borancattle. Proc. Nat. Acad. Sci. USA 100: 7443–7448.Henderson, C.R. 1984. Applications of linear models in animal breeding. University of Guelph,Ontario, ON, Canada.Heyen, D.W., Weller J.I., Ron, M., Band, M., Beever, J.E., Feldmesser, E., Da, Y., Wiggans, G.R.,VanRaden, P.M. & Lewin, H.A. 1999. A genome scan for QTL influencing milk production andhealth traits in dairy cattle. Physiol. Genom. 1: 165–175.Hiendleder, S., Thomsen, H., Reinsch, N., Bennewitz, J., Leyhe-Horn, B., Looft, C., Xu, N.,Medjugorac, I., Russ, I., Kuhn, C., Brockmann, G.A., Blumel, J., Brenig, B., Reinhardt, F.,Reents, R., Averdunk, G., Schwerin, M., Forster, M., Kalm, E. & Erhardt, G. 2003. Mapping ofQTL for body conformation and behavior in cattle. J. Heredity 94: 496–506.Hill, W.G. 1971. Investment appraisal for national breeding programmes. Anim. Prodn. 13: 37–50.Holmberg, M. & Andersson-Eklund, L. 2004. Quantitative trait loci affecting health traits inSwedish dairy cattle. J. Dairy Sci. 87: 2653–2659.International Human Genome Sequencing Consortium. 2004. Finishing the euchromatic sequenceof the human genome. Nature 431:931–945.Israel, C. & Weller, J.I. 1998. Estimation of candidate gene effects in dairy cattle populations. J. DairySci. 81: 1653–1662.Israel, C. & Weller, J.I. 2000. Effect of misidentification on genetic gain and estimation of breedingvalue in dairy cattle populations. J. Dairy Sci. 83: 181–187.Israel, C. & Weller, J.I. 2002. Estimation of quantitative trait loci effects in dairy cattle populations.J. Dairy Sci. 85: 1285–1297.Israel, C. & Weller, J.I. 2004. Effect of type I error threshold on marker-assisted selection in dairycattle. Livest. Prod. Sci. 85: 189–199.Kashi, Y., Hallerman, E. & Soller, M. 1990. Marker-assisted selection of candidate bulls for progenytesting programmes. Anim. Prod. 51: 63–74.Kerr, R. J. & Kinghorn, B.P. 1996. An efficient algorithm for segregating analysis in large populations.J. Anim. Breed. Genet. 113: 457–469.Khatkar, M.S., Thomson, P.C., Tammen, I. & Raadsma, H.W. 2004. Quantitative trait loci mappingin dairy cattle: review and meta-analysis. Genet. Sel. Evol. 36: 163–190.Klungland, H., Sabry, A., Heringstad, B., Olsen, H.G., Gomez-Raya, L., Vage, D.I., Olsaker, I.,Odegard, J., Klemetsdal, G., Schulman, N., Vilkki, J., Ruane, J., Aasland, M., Ronningen, K. &Lien, S. 2001. Quantitative trait loci affecting clinical mastitis and somatic cell count in dairy cattle.Mamm. Genome 12: 837–842.Knott, S.A., Elsen J.M. & Haley, C.S. 1996. Methods for multiple-marker mapping of quantitativetrait loci in half-sib populations. Theor. Appl. Genet. 93: 71–80.Kosgey, I. S., Kahi, K.A. & van Arendonk J.A.M. 2005. Evaluation of closed adult nucleus multipleovulation and embryo transfer and conventional progeny testing breeding schemes for milk productionin tropical crossbred cattle. J. Dairy Sci. 88: 1582–1594.

Chapter 12 – Marker-assisted selection in dairy cattle 225Kuhn, C., Bennewitz, J., Reinsch, N., Thomsen, H., Looft, C., Brockmann, G.A., Schwerin, M.,Weimann, C., Hiendieder, S., Erhardt, G., Medjugorac, I., Forster, M., Brenig, B., Reinhardt, F.,Reents, R., Russ, I., Averdunk, G., Blumel, J. & Kalm, E. 2003. Quantitative trait loci mapping offunctional traits in the German Holstein cattle population. J. Dairy Sci. 86: 360–368.Kuhn, C., Thaller, G., Winter, A., Bininda-Emonds, O.R.P., Kaupe, B., Erhardt, G., Bennewitz,J., Schwerin, M. & Fries, R. 2004. Evidence for multiple alleles at the DGAT1 locus betterexplains a quantitative tip trait locus with major effect on milk fat content in cattle. Genetics 167:1873–1881.Lander, E.S. & Botstein, D. 1989. Mapping Mendelian factors underlying quantitative traits usingRFLP linkage maps. Genetics 121: 185–199.Mackay, T.F. 2001. The genetic architecture of quantitative traits. Ann. Rev. Genet. 35: 303–339.Mackinnon, M.J. & Georges, M.A.J. 1998. Marker-assisted preselection of young dairy sires prior toprogeny-testing. Livest. Prod. Sci. 54: 229–250.Mackinnon, M.J. & Weller, J.I. 1995. Methodology and accuracy of estimation of quantitative traitloci parameters in a half-sib design using maximum likelihood. Genetics 141: 755–770.Madalena, F.E. 1993. A simple scheme to utilize heterosis in tropical dairy cattle. World Anim. Rev.74/75: 17–25.Martinez, O. & Curnow, R.N. 1992. Estimating the locations and the sizes of the effects of quantitativetrait loci using flanking markers. Theor. Appl. Genet. 85: 480–488.Meuwissen, T.H.E. & Goddard, M.E. 2000. Fine mapping of quantitative trait loci using linkage disequilibriawith closely linked marker loci. Genetics 155: 421–430.Meuwissen, T.H.E. & Goddard, M.E. 2004. Mapping multiple QTL using linkage disequilibrium andlinkage analysis information and multitrait data. Genet. Sel. Evol 36: 261–279.Meuwissen, T.H.E. & van Arendonk, J.A.M. 1992. Potential improvements in rate of genetic gainfrom marker assisted selection in dairy cattle breeding schemes. J. Dairy Sci. 75: 1651–1659.Meuwissen, T.H.E., Karlsen, A., Lien, S., Olsaker, I. & Goddard, M.E. 2002. Fine mapping of aquantitative trait locus for twinning rate using combined linkage and linkage disequilibrium mapping.Genetics 16: 373–379.Mosig, M.O., Lipkin, E., Khutoreskaya, G., Tchourzyna, E., Soller, M. & Friedmann, A. 2001. Awhole genome scan for quantitative trait loci affecting milk protein percentage in Israeli-Holsteincattle, by means of selective milk DNA pooling in a daughter design, using an adjusted false discoveryrate criterion. Genetics 157: 1683–1698.Nadesalingam, J., Plante, Y. & Gibson, J.P. 2001. Detection of QTL for milk production on chromosomes1 and 6 of Holstein cattle. Mammal. Genome 12: 27–31.Neimann-Sørensen, A. & Robertson, A. 1961. The association between blood groups and severalproduction characters in three Danish cattle breeds. Acta. Agr. Scand. 11: 163–196.Nemir, M., Bhattacharyya, D., Li, X., Singh, K., Mukherjee, A.B. & Mukherjee, B.B. 2000. Targetedinhibition of osteopontin expression in the mammary gland causes abnormal morphogenesis andlactation deficiency. J. Biol. Chem. 275: 969–976.Nicholas, F.W. & Smith C. 1983. Increased rates of genetic change in dairy cattle by embryo transferand splitting. Anim Prod. 36: 341–353.Olsen, H.G., Gomez-Raya, L., Vage, D.I., Olsaker, I., Klungland, H., Svendsen, M., Adnoy, T.,Sabry, A., Klemetsdal, G., Schulman, N., Kramer, W., Thaller, G., Ronningen, K. & Lien, S.

226Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fish2002. A genome scan for quantitative trait loci affecting milk production in Norwegian dairy cattle.J. Dairy Sci. 85: 3124–3130.Plante, Y., Gibson, J.P., Nadesalingam, J., Mehrabani-Yeganeh, H., Lefebvre, S., Vandervoort,G. & Jansen, G.B. 2001. Detection of quantitative trait loci affecting milk production traits on 10chromosomes in Holstein cattle. J. Dairy Sci. 84: 1516–1524.Powell, R.L. & Norman, H.D. 1988. Accuracy of cow indexes according to repeatability, evaluation,herd yield, and registry status. J. Dairy Sci. 71: 2232–2240.Ron, M., Heyen, D.W., Weller, J.I., Band, M., Feldmesser, E., Pasternak, H., Da, Y., Wiggans, G.R.,Vanraden, P.M., Ezra, E. & Lewin, H.A. 1998. Detection and analysis of a locus affecting milkconcentration in the US and Israeli dairy cattle populations. Proc. 6th World Cong. Genet. Appl.Livest. Prod. 26: 422–425. Armidale, Australia.Ron, M., Kliger D., Feldmesser E., Seroussi E., Ezra E. & Weller J. I. 2001. Multiple quantitativetrait locus analysis of bovine chromosome 6 in the Israeli Holstein population by a daughter design.Genetics 159:727–735.Ron, M., Cohen-Zinder, M., Peter, C., Weller, J.I. & Erhardt, G. 2006. ABCG2 polymorphism inBos indicus and Bos taurus cattle breeds. J. Dairy Sci. 89: 4921–4923.Ron, M., Feldmesser, E., Golik, M., Tager-Cohen, I., Kliger, D., Reiss, V., Domochovsky, R., Alus,O., Seroussi, E., Ezra, E. & Weller, J.I. 2004. A complete genome scan of the Israeli Holstein populationfor quantitative trait loci by a daughter design. J. Dairy Sci. 87: 476–490.Schnabel, R.D., Kim, J.J., Ashwell, M.S., Sonstegard, T.S., Van Tassell, C.P., Connor, E.E. &Taylor, J.F. 2005. Fine-mapping milk production quantitative trait loci on BTA6: Analysis of thebovine osteopontin gene. Proc. Nat. Acad. Sci. USA 102: 6896–6901.Schrooten, C., Bovenhuis, H., Coppieters, W. & van Arendonk, J.A.M. 2000. Whole genome scanto detect quantitative trait loci for conformation and functional traits in dairy cattle. J. Dairy Sci.83: 795–806.Schulman, N.F., Viitala, S.M., de Koning, D.J., Virta, J., Maki-Tanila, A. & Vilkki, J.H. 2004.Quantitative trait loci for health traits in Finnish Ayrshire cattle. J. Dairy Sci. 87: 443–449.Smith, C. 1988. Genetic improvement of livestock, using nucleus breeding units. World Anim. Rev.65: 2–10.Song, J-Z., Soller, M. & Genizi, A. 1999. The full-sib intercross line (FSIL): a QTL mapping designfor outcrossing species. Genet. Research 73: 61–73.Spelman, R.J., Garrick, D.J. & van Arendonk, J.A.M. 1999. Utilization of genetic variation bymarker assisted selection in commercial diary cattle populations. Livest. Prod. Sci. 59: 51–60.Spelman, R.J., Coppieters, W., Karim, L., van Arendonk, J.A.M. & Bovenhuis, H. 1996.Quantitative trait loci analysis for five milk production traits on chromosome six in the DutchHolstein-Friesian population. Genetics 144: 1799–1808.Spelman, R.J., Huisman, A.E., Singireddy, S.R., Coppieters, W., Arranz, J., Georges, M. &Garrick, D.J. 1999. Quantitative trait loci analysis on 17 nonproduction traits in the New Zealanddairy population. J. Dairy Sci. 82: 2514–2516.Thomsen, H., Reinsch, N., Xu, N., Looft, C., Grupe, S., Kuhn, C., Brockmann, G.A., Schwerin,M., Leyhe-Horn, B., Hiendleder, S., Erhardt, G., Medjugorac, I., Russ, I., Forster, M., Brenig,B., Reinhardt, Kalm, E. 2001. Comparison of estimated breeding values, daughter yield deviationsand deregressed proofs within a whole genome scan for QTL. J. Anim. Breed. Genet. 118:357–370.

Chapter 12 – Marker-assisted selection in dairy cattle 227van Arendonk, J.A.M., Bovehuis, H., van der Beek, S. & Groen, A.F. 1994. Detection and exploitationof markers linked to quantitative traits in farm animals. Proc. 5th World Cong. Genet. Appl.Livest. Prod. 21:193–200. Guelph, Canada.VanRaden, P.M. & Wiggans, G.R. 1991. Derivation, calculation and use of national animal modelinformation. J. Dairy Sci. 74: 2737–2746.Velmala, R.J., Vilkki, H.J., Elo, K.T., de Koning, D.J. & Maki-Tanila, A.V. 1999. A search for quantitativetrait loci for milk production traits on chromosome 6 in Finnish Ayrshire cattle. Anim.Genet. 30: 136–143.Viitala, S.M., Schulman, N.F., de Koning, D.J., Elo, K., Kinos, R., Virta, A., Virta, J., Maki-Tanila,A. & Vilkki, H.J. 2003. Quantitative trait loci affecting milk production traits in Finnish Ayrshiredairy cattle. J. Dairy Sci. 86: 1828–1836.Vilkki, H.J., de Koning, D.J., Elo, K.T., Velmala, R. & Maki-Tanila, A. 1997. Multiple marker mappingof quantitative trait loci of Finnish dairy cattle by regression, J. Dairy Sci. 80: 198–204.Visscher, P.M., Haley, C.S. & Thompson, R. 1996. Marker-assisted introgression in backcrossbreeding programs. Genetics 144: 1923–1932.Visscher, P.M., Thompson, R. & Haley, C.S. 1996. Confidence intervals in QTL mapping by bootstrapping.Genetics 143: 1013–1020.Weller, J.I. 1986. Maximum likelihood techniques for the mapping and analysis of quantitative traitloci with the aid of genetic markers. Biometrics 42: 627–640.Weller, J.I. 1994. Economic aspects of animal breeding. London, Chapman & Hall. 244 pp.Weller, J.I. 2001. Quantitative trait loci analysis in animals. London, CABI Publishing. 287 pp.Weller, J.I. & Soller, M. 2004. An analytical formula to estimate confidence interval of QTL locationwith a saturated genetic map as a function of experimental design. Theor. Appl. Genet. 109:1224–1229.Weller, J. I., Kashi, Y. & Soller, M. 1990. Power of “daughter” and “granddaughter” designs for geneticmapping of quantitative traits in dairy cattle using genetic markers. J. Dairy Sci. 73: 2525–2537.Weller, J.I., Shlezinger, M. & Ron, M. 2005. Correcting for bias in estimation of quantitative trait locieffects. Genet. Sel. Evol. 37: 501–522.Weller, J.I., Soller, M. & Brody, T. 1988. Linkage analysis of quantitative traits in an interspecificcross of tomato (L. esculentum x L. pimpinellifolium) by means of genetic markers. Genetics 118:329–339.Weller, J.I., Song, J.-Z., Heyen, D.W., Lewin, H.A. & Ron M. 1998. A new approach to the problemof multiple comparisons in the genetic dissection of complex traits. Genetics 150: 1699–1706.Weller, J.I., Weller, H., Kliger, D. & Ron, M. 2002. Estimation of quantitative trait locus allele frequencyvia a modified granddaughter design. Genetics 162: 841–849.Weller, J.I., Golik, M., Seroussi, E., Ezra, E. & Ron, M. 2003. Population-wide analysis of a QTLaffecting milk-fat production in the Israeli Holstein population. J. Dairy Sci. 86: 2219–2227.Weller, J.I., Feldmesser, E., Golik, M., Tager-Cohen, I., Domochovsky, R., Alus, O., Ezra, E. &Ron, M. 2004. Factors affecting incorrect paternity assignment in the Israeli Holstein population.J. Dairy Sci. 87: 2627–2640.Westell, R.A. & van Vleck, L.D. 1987. Simultaneous genetic evaluation of sires and cows for a largepopulation of dairy cattle. J. Dairy Sci. 70: 1106–1017.Whittaker, J.C., Thompsom, R. & Visscher, P.M. 1996. On the mapping of QTL by regression ofphenotype on marker-type. Heredity 77: 23–32.

228Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishWinter, A., Kramer, W., Werner, F.A.O., Kollers, S., Kata, S., Durstewitz, G., Buitkamp, J.,Womack, J.E., Thaller, G. & Fries, R. 2002. Association of a lysine-232/alanine polymorphism in abovine gene encoding acyl-CoA:diacylglycerol acyltransferase (DGAT1) with variation at a quantitativetrait locus for milk fat content. Proc. Nat. Acad. Sci. USA 99: 9300–9305.Zhang, Q., Boichard, D., Hoeschele, I., Ernst, C., Eggen, A., Murkve, B., Pfister-Genskow, M.,Witte, L.A., Grignola, F.E., Uimari, P., Thaller, G. & Bishop, M.D. 1998. Mapping quantitativetrait loci for milk production and health of dairy cattle in a large outbred pedigree. Genetics 149:1959–1973.

Chapter 13Marker-assisted selectionin sheep and goatsJulius H.J. van der Werf

230Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishSummarySheep and goats are often kept in low input production systems, often at subsistence levels.In such systems, the uptake of effective commercial breeding programmes is limited, letalone the uptake of more advanced technologies such as those needed for marker-assistedselection (MAS). However, effective breeding programmes exist in a number of countries,the largest ones in Australia and New Zealand aiming for genetic improvement of meat andwool characteristics as well as disease resistance and fecundity. Advances have been madein sheep gene mapping with the marker map consisting of more than 1 200 microsatellites,and a virtual genome sequence together with a very dense single nucleotide polymorphism(SNP) map are expected within a year. Significant research efforts into quantitative traitloci (QTL) are under way and a number of commercial sheep gene tests have alreadybecome available, mainly for single gene effects but some for muscularity and diseaseresistance. Gene mapping in goats is much less advanced with mainly some activity indairy goats. Integration of genotypic information into commercial genetic evaluation andoptimal selection strategies is a challenge that deserves more development.

Chapter 13 – Marker-assisted selection in sheep and goats 231IntroductionThe benefits of marker-assisted selection(MAS) to sheep and goat breedingprogrammes depend on a number ofconditions that are relevant for mostbreeding programmes across species. Theseconditions include the existence of a genotypetest predicting phenotypic differences,the economic value of these differencesand the value of the genotypic informationwithin the breeding programme. The valueof genetic information will depend heavilyon the socio-economic context of thebreeding programme and the productionsystem. In a technical sense, the value ofthis information is basically driven by theincrease in selection accuracy resulting fromknowledge of genotypes, which in turn willdiffer between animals from different ageclasses. In particular, the relative increase inselection accuracy of the youngest selectioncandidates will be critical to the valueof MAS. However, technical argumentsabout increased selection accuracy are oflittle value if these selection criteria arepoorly developed or accepted within theproduction system.The application of new technologies suchas MAS in animal breeding programmestherefore depends not only on a number oftechnical aspects associated with increasedrates of genetic improvement, but also onthe commercial structures of the industry.For example, the uptake of MAS in breedingprogrammes depends on the willingness ofbreeders to invest in genotypic information,and their ability to turn this intoknowledge that helps them improve theircommercial breeding activities. A basicunderstanding of breeding programmecharacteristics, the possible role of geneticinformation within these programmes, andthe commercial relationships among thedifferent players are needed to assess thevalue and predict the application of MASin breeding programmes. These commercialrelationships are distinctly differentin sheep and goat breeding programmesfrom those in the more intensive animalindustries, and the application of MAS willtherefore be different. For example, 96 percentof the world goat population is keptby smallholders in developing countries,and genetic improvement programmes arerare (Olivier et al., 2005).The purpose of this chapter is to describethe use of MAS in breeding programmesfor sheep and goats and the likely rate ofuptake of this technology in these species. Itbegins by characterizing such programmesand describing and comparing existing programmes.MAS is most useful for traits thatcannot be improved easily by phenotypicselection, either because they are difficultto measure on young animals (before sexualreproduction), or because of low heritability.Therefore, breeding objectives arediscussed in general terms and the traits thatare particularly suitable for MAS are identified.Based on some general well-knownadvantages of MAS, its possible role withinbreeding programmes can be predicted andexamples of these are provided. Examplesof marked genes are then described andan overview given of the status of “genediscovery” and gene mapping projects insheep and goats. The chapter concludesby describing cases of using this informationin actual breeding programmes. Somegene tests are based on actual functionalmutations, many of which do not affectquantitative traits that are generally targetedin breeding programmes. Although,the term “MAS” should be replaced insome cases by “genotype assisted selection”(GAS), the term MAS is used looselyto refer to all selection based on genotypicinformation. It will become clear that

232Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishcurrently most real applications of MASfor sheep and goat breeding are based onresearch projects and therefore subsidized.However, the first commercial applicationsare now also emerging. The main requirementsfor a successful commercial andlong-term application of MAS in sheep andgoat breeding are discussed and illustratedbased on examples.MAS applications are often illustrated orsimulated for pure breeding programmes.However, MAS could be particularlyuseful in crossbreeding programmes wheredesirable genotypes in unfavourable backgroundsare introgressed into productivelocal breeds with overall better breedingvalues. The opposite is also possible, wheredisease resistance genes of local breedsare specifically targeted in upgrading programmeswith imported stock with higherproductivity being crossed to local breeds.Crossbreeding and introgression programmesare discussed and, as sheep andgoat production is relatively predominantin developing countries, particular attentionis given to breeding programmes forlow to medium input production systems.Characteristics of sheep andgoat breeding programmesBreeding structuresBreeding programmes for sheep and goatsgenerally operate within an industry that isbased on low levels of resource inputs, i.e.low levels of feeding and low labour costson a per animal basis. Goat production takesplace largely in developing countries whereselective breeding based on performancerecording is often absent. A more substantialproportion of sheep production is foundin developed countries such as Australia,France, New Zealand, South Africa and theUnited Kingdom. These systems are alsopredominantly pastoral-based and extensivein nature. An FAO working group report(Hoste, 2002; Olivier et al., 2005) made thefollowing distinction between productionsystems and the opportunities within themfor breeding programmes: 1) subsistencebasedproduction, among the world’spoorest, with limited market developmentand limited inputs and scope for geneticimprovement; 2) market-based production,with better developed markets targetingurban populations, higher input levels andmore specialized production systems, withscope for genetic improvement dependingon cost of inputs and also on skills andinformation literacy of breeders andproducers; and 3) high-input production,with further specialization, emphasis onincreased land and labour efficiency, andmuch more concern for food quality, foodsafety, animal welfare and the environment.Most of the world’s goat production aswell as many of the sheep systems wouldfall into the first category, whereas sheepproduction in developed countries wouldmainly fall into the second category, withsome of these working towards the thirdcategory.Sheep and goat breeding programmes arecharacterized by a flat breeding structure,meaning that compared with intensivelivestock industries many operationsparticipate in genetic improvement, therebyforming a wide base for the nucleus breedingsector. Reproductive levels of breedinganimals, especially males, are relatively lowcompared with other species. In such asystem, the multiplication factor, i.e. thenumber of commercial expressions resultingfrom investments in improved genotypes inthe breeding nucleus, is relatively low. Thismakes it more difficult to introduce newtechnologies and justify large investments inimproving individual animals. However, likeother breeding programmes, there remains

Chapter 13 – Marker-assisted selection in sheep and goats 233a significant return on overall investment ingenetic improvement. Also, in some moreadvanced sheep breeding programmes,the use of artificial insemination (AI) andacross-flock evaluation has boosted the useof high profile rams and raised the value ofindividual breeding animals.The main investment in breeding programmesis for performance recording.The extent of trait measurement is oftenquite closely aligned with the intensity ofthe production system. Input levels forsheep production vary, depending on breedtype and market. In Australia, for example,there is a significant difference betweenwool producing Merino sheep that arekept extensively in harsh environments,and more intensive lamb production systemsthat are found in higher rainfall areasor on irrigated land. The proportion ofbreeding flocks for which objective traitand pedigree measurements are undertakenis relatively much higher in the Australianterminal sire breeds.Selection takes place within the breedingstuds. AI is common in the stud breedingsector, enabling the genetic linkage of flocks.There are breeder groups with organizedprogeny testing of young sires across flockprogrammes. In Australia, a national geneticevaluation system known as “Lambplan”has driven genetic evaluation for terminalsires and maternal breeds across flocks formore than a decade. Breeders as well as rambuyers are increasingly basing their ramassessment on estimated breeding value(EBV) or dollar index value. Such a systemgives breeders incentives to invest in traitmeasurement and to create genetic linksbetween their flocks, otherwise it wouldbe difficult for a ram to rise to the top ofthe across-flock EBV list. Hence, there isincreasingly an exchange of genetic materialbetween flocks, mainly through the useof AI. Obviously, such a breeding structurewould be more conducive to breedersinvesting in gene marker technology.By contrast, the Australian Merinoindustry has had a much lower proportionof breeders taking up trait and pedigreerecording. The industry is more traditionaland selection is most often based on visualassessment. While this might be due partlyto the sector being more extensive, AIhas been commonly used in the Merinostud sector and top Merino rams havealways been sold for high prices. Therefore,the extensive nature of the industry doesnot fully explain the lack of investmentin performance recording. The traditionalnature of the industry that has hamperedthe uptake of quantitative genetic principlesis also the result of socio-economic factors,with wool producers being traditionallya prominent and relatively wealthy socialclass. The lamb industry has long beenthe wool person’s “poor brother”, but thislack of status has accelerated innovationwith the introduction of new approachessuch as formal recording and across-flockevaluation. Hence, economic as well associal and cultural reasons may explainwhy sheep breeding programmes havedifferent levels of sophistication in termsof recording, genetic evaluation and acrossflockselection.Breeding programmes and traitstargetedMeat sheepLarge-scale genetic evaluation programmesfor sheep are found in Australia, France,New Zealand, South Africa and the UnitedKingdom. In all of these, performancerecording for meat traits is well advanced,with not only weight traits measured, butalso traits related to carcass quality such asbody fat and muscle (based on ultrasound

234Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishscanning and in some cases computer tomography[CT] scanning), disease (mainlyresistance to internal parasites) and reproduction.The national evaluation systemin Australia (“Lambplan”) now has about120 000 new animals from about 450 flocksrecorded each year for terminal sire breedsand maternal breeds (A. Ball, personal communication).Performance recording takesplace only at the stud level, which in a senseis a dispersed nucleus, and a large proportionof the genetic basis of the commercial populationstems from these recorded flocks.The proportion of pedigree recorded individualsis high at the stud level, allowingbest linear unbiased prediction (BLUP) ofEBV. In New Zealand, a similar programmeexists (“Sheep Improvement Limited”[SIL]), in which pedigree and performancerecords are registered with genetic serviceproviders and the information “retailed”back to the breeders. SIL enters more than250 000 new animals per year from some750 recorded flocks, all pedigree recorded,and has a database of more than 5 millionanimal records. Across-flock EBVs areestimated for a proportion of these. In theUnited Kingdom, about 50 000 breedingewes and their lamb records are recordedevery year from 37 different breeds, andindices have been developed for terminalsand maternal (“hill”) breeds (Conington etal., 2004). Across-flock genetic evaluationprogrammes for meat sheep breeds existalso on smaller scales in France, Norwayand South Africa.Most breeding programmes for meatsheep focus on weight traits, and ultrasoundscanning is commonly used for fatand muscle traits. Reproduction traits arerecorded as numbers of lambs born andweaned. Selection for resistance to internalparasites can be based on faecal worm eggcounts (WECs) associated with naturalchallenge in the field, e.g. in Australia andNew Zealand, and this has been shown tobe reasonably heritable in Merino sheep(e.g. Khusro et al., 2004). EBVs for WECare produced for an increasing number offlocks in Australia and New Zealand.The traits that would most obviouslybenefit from MAS in meat sheep wouldbe traits related to carcass and carcassquality, reproduction and disease resistance.Ultrasound measurements are currentlyused to predict carcass fat and muscling.However, genetic correlations with traitsmeasured on carcass are only moderate(Safari, Fogarty and Gilmour, 2005) andspecific meat quality attributes such astenderness and colour might not be wellcaptured by current measurement. Carcasstraits are prime targets for MAS as theycannot be measured on breeding animalsand progeny or sib testing would be neededas an alternative. Reproduction traits aswell as maternal behaviour and ewe survivalare also good MAS targets as they aresex limited and are only expressed after thefirst round of reproduction. Disease resistancetraits are generally hard to measureunder uniform conditions and would alsogreatly benefit from MAS.Wool sheepBreeding for and recording of wool traits islimited to a few countries. The largest acrossflockscheme is found in Australia (mainlyfor the Merino breed), and smaller geneticevaluation schemes are run in New Zealand,South Africa and South America (Merinoand Corriedale). In Australia, the proportionof breeders participating in formalrecording and genetic evaluation is smallerfor wool than for meat sheep. However, theMerino industry is very large, constitutingthe vast majority of the Australian flockthat consists of about 100 million sheep. By

Chapter 13 – Marker-assisted selection in sheep and goats 235the end of 2005, a new single system for anational across-flock genetic evaluation ofMerinos had been introduced in Australia,combining data from previously separateschemes. The number of animals performancerecorded per year is growing rapidly,with about 100 000 new animals now beingentered annually.Wool production efficiency is mainlydetermined by fleece weight and woolquality. Wool quality traits are mainly fibrediameter and staple strength, and these areeconomically much more important for finewools. Staple strength is more expensive tomeasure, but has a high correlation with thecoefficient of variation of fibre diameter,which is therefore a good predictor. Wooltraits have generally high levels of heritability,especially fleece weight and fibrediameter.Reproductive rate in wool sheep hasbeen hard to select for as pedigree recordinghas been limited and the heritability islow. Moreover, genetic improvement ofreproductive rate has been less importantfor wool production because of thepositive net economic benefit of wool producingbreeding females. However, with anincreasing meat/wool price ratio, the situationis changing and reproductive rate iscurrently becoming more important. Also,meat attributes of Merino sheep are nowreceiving increased attention, includingmeasurements of body weight at differentages, fat depth and eye muscle depth (ultrasoundscanned).In pure wool production systems, MASwould be expected to have limited benefitfor wool production traits because of theirhigh heritability and the ability to measurethe traits before the age of first selection.MAS for reproductive traits and motheringability would be more beneficial because oflow heritability and sex-limited recording.Parasite resistance is becoming a trait ofgreater economic importance due to thedevelopment of resistance to all the majorclasses of anthelmintics used and the lackof new anthelmintic classes being developed.Host resistance to internal parasitesis particularly poor in the Merino breed.The trait can be selected for using fieldrecords of WEC. EBVs are being producedfor this trait and genetic progress is beingachieved. However, the procedure is laboriousand there is also some concern aboutuniformity of measurement and trait definition,as well as the existence of differentspecies of parasites in different regions.Various studies have looked at genotypex environment interactions for parasiteresistance and, although some interactionexists, relatively high correlations (~0.8)were found between breeding values in differentenvironments, when environmentswere defined either through worm type(McEwan et al., 1997) or by high and lowflock averages for WEC (Pollot and Greeff,2004). In any case, many of these traitattributes make parasite resistance a goodtarget for MAS. Identifying QTL for parasiteresistance might also shed more lighton the biology of immunity, and possiblyhelp to find other modes of improvement.Feed efficiency, and particularly maternalefficiency, are important determinants ofpastoral production systems (Ferrell andJenkins, 1984) and genetic improvementwould benefit from MAS because of thecost of their measurement. However, feedavailability and feed costs are quite variablewithin and between years, and the abilityof sheep to cope with harsh environmentsand periods of drought is perceived byindustry as being of greater importance.Hardiness and ewe survival are not welldefined characteristics and are not normallymeasured in breeding programmes.

236Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishEwe fitness and adaptation are often usedas the main argument for the existenceof genotype x environment interaction inwool production, inhibiting the exchangeof genetic material among regions. Carrick(2005) found moderate to high genetic correlationsbetween wool production traitsin flock groups differentiated by their phenotypicmeans for a range of productiontraits. Discovering QTL for fitness andsurvival traits in different environmentswould be useful, but these are unlikely tobe found unless the traits themselves areclearly defined and measured.Dairy sheepDairy sheep are predominantly found inthe Mediterranean region with both milkand meat production being economicallyrelevant traits to farmers. A great varietyof breeds are being targeted in selectionprogrammes for the improvement of milkyield and milk composition but the importanceof functional traits such as uddercharacteristics and mastitis susceptibility isincreasing (Barillet, 1997; Barillet, Arranzand Carta, 2005). Genetic improvement fordairy traits, being sex-limited and measuredafter the first offspring are born, wouldparticularly benefit from MAS.GoatsMost goat farming systems focus on meatproduction (about 80 percent), with moreemphasis in developed countries on dairygoat production (Olivier et al., 2005) andfibre production (cashmere, mohair). Indairy goat breeding, the most developedbreeding programmes are found in Franceand are based on a strong goat cheesemarket. Based on AI and milk recording,Caprigene France runs selection schemesfor the Saanen and Alpine breeds, with300 000 goats in 2 500 herds being recordedfor milk traits. Dairy goat production isalso recorded on smaller scales in Italy,Norway and Spain, with no more than afew thousand animals recorded in othercountries (Montaldo and Manfredi, 2002).The main traits in dairy goat production aremilk yield and protein and fat content ofmilk. Being sex-limited and measured onlyafter first production of progeny, thesetraits would benefit from MAS.Goat meat production is widely spreadthroughout the developing world but thereare few breeding programmes of any significance.Genetic evaluation for Boer goatsand other meat breeds is taking place inAustralia and South Africa with weaningweight usually being the main trait measured.Ultrasound measurement of fat andmuscle traits is less common in goats, whilereproductive traits have had less attention,possibly because of their low heritabilityand multiparous nature. There are fewstudies concerning resistance to internalparasites in goats (Olayemi et al., 2002),but these seem to indicate that faecal WECscould be a similar selection trait as in sheep.However, the trait is hard to measure andthere is no systematic recording and evaluationin breeding programmes.Development of sheep and goatgenome mapsSeveral key publications have reportedprogress on the linkage map of the sheepgenome based on an international mappingflock developed in New Zealand (Crawfordet al., 1995; Maddox et al., 2001). The latestsheep linkage map (version 4.3) comprises1 256 gene markers mapped to unique locations(Maddox, 2004) and most genomicregions are well covered with a maximumgap of 20 cM. However, there are quitea number of markers of low quality, so atypical genome scan would leave a number

Chapter 13 – Marker-assisted selection in sheep and goats 237of gaps. Most of the markers are microsatellites.The total number of sheep locilisted in the ARKdb database (http://iowa.thearkdb.org) contains more than 2 000markers, but many of these are not onthe linkage map. The development of thesheep genome map runs somewhat behinddevelopments for other livestock speciesbecause of substantially lower investments.Nevertheless, at the DNA level where thesequence can be aligned, there is a ~90 percenthomology with the cattle sequence andthrough gene coding regions ~96 percent,and the sequencing of the cattle genomewill therefore greatly enhance the developmentof the genome map in sheep. There isgenerally good agreement between sheepand cattle maps, with 598 mainly anonymouscommon microsatellite loci, i.e. genemarkers can be linked to a comparativemap. Based on sequence information inother mammals (mainly cattle) and sheepGeneBank sequences, comparative mappingcan be used to construct a predictedsheep map. This can be accessed fromthe Australian Gene Mapping Web site(Maddox, 2005a). The number of singlenucleotide polymorphism (SNP) markersin sheep is still very low, but with the cattlesequence known and with an internationalcollaborative sheep bacterial artificial chromosome(BAC)-end sequencing projectunder way, it is expected that there will bea large number (~16 000) of SNPs availablefor sheep towards the end of 2006. Thiswill form a set of markers that would allowhigh-density genome-wide scans.The goat map is more sparse thanthe sheep map and contains about halfthe number of markers known in sheep:731 loci with 271 genes and 423 microsatellites(http://locus.jouy.inra.fr/). The lastpub-lished linkage map for goats containsonly 307 markers (Schibler et al., 1998),with coverage of the whole goat genomebeing far from complete. Although thesparsity of the sheep map makes it difficultto develop a good homology between themaps, about two-thirds of the mapped goatmarkers can also be linked to the sheep map(Maddox, 2005b).QTL and gene mappingAn excellent overview of mapping experimentsin sheep can be found on the AustralianGene Mapping Web site (Maddox, 2005a),including references to identified QTL andgenes. Successfully identified genes andQTL are related mainly to fecundity, diseaseresistance and meat quality.FecundityTwo genetic mutations have been reportedfor fecundity: the Booroola mutation: FecBon chromosome 6 (Wilson et al., 2001;Mulsant et al., 2001; Souza et al., 2001) andthe Inverdale gene: FecX on the X chromosome(Galloway et al., 2000). The Booroolagene has a substantial additive effect onovulation rate with each copy increasingthis by about 1.5 eggs (i.e. scanned foetuses).The additional allelic effect of theBooroola mutation on litter size is about0.8 to 0.9 lambs (Davis et al., 1982; Piperand Bindon, 1982; Gootwine et al., 2003)whereas a second copy of the mutation hasa slightly smaller effect (0.4–0.6 lambs). Theeffect on number of lambs weaned is somewhatlower. The effect of the Booroolagene is often perceived as too large and thesurvival of twin and triplet lambs decreasessubstantially in extensive and harsh conditions,typical for many sheep flocks.For example, in the Australian Merinoindustry, the Booroola mutation is notseen as a desirable characteristic. However,the Booroola gene has been introduced inmany sheep populations around the world.

238Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishThe Booroola mutation possibly originatesfrom the Indian Garole (Davis et al., 2002)and, interestingly, the gene effect appearedto be smaller (0.6 lambs born alive) inan Indian introgression programme withDeccani sheep (Nimbkar, Pardeshi andGhalsasi, 2005). This increase in litter sizeappears to be easily managed in shepherdflocks. A smaller effect would be moredesirable for extensive production systems.It is not clear whether the reduced geneeffect arises from a modification due toenvironmental effects or the genetic background.As the reproductive rate is a trait ofhigh economic value, and due to the availabilityof a test for the actual gene mutation,Booroola remains a very interesting genefor MAS and marker-assisted introgression(MAI) programmes.The Inverdale gene has been mappedto the X chromosome and has an effectof about 0.6 lambs per ewe lambing.However, the homozygous ewe is infertile.As carrier rams as well as non-carrier ewesneed to be maintained in a crossbreedingsystem, using this gene in the industry ismore complex. However, the 100 percentaccurate test has made the use of this genemore manageable.A number of other major genes forfecundity have been described by Davis(2005), but the molecular basis of theseeffects has not been formally described.DiseaseInternal parasites are the main cause ofeconomic losses due to health problemsin sheep and goat production systems.Although there is significant research underway to detect and map QTL for host resistanceto internal parasites, there have notyet been any major breakthroughs in termsof detected polymorphisms in functionalgenes. Few QTL have been reported forresistance to internal parasites (see reviewby Dominik, 2005) but not all results arereported in the literature. A major geneeffect for resistance to Haemonchus contortuswas found based on segregationanalysis (Meszaros et al., 1999) but this hasnot been confirmed based on gene markers.The problem of finding distinct QTL forresistance to internal parasites may be dueto the complexity of the underlying biologicalmechanism as well as the difficultyof finding well-defined phenotypes thatmeasure resistance.Transmissible spongiform encephalopathy(TSE) is a prion disease like scrapieand is characterized by the accumulationof a modified form of a protein knownas PrP. The PrP gene has been associatedwith variation in scrapie susceptibility insheep (Moreno et al., 2002), mice (Morenoet al., 2003), and goats (Acin et al., 2003).The gene only explains a proportion of theoverall variation for increased resistance toscrapie. Commercial gene tests are availablefor the PrP gene mutation.A causative mutation has been foundfor the Spider Lamb Syndrome. This is arelatively rare recessive skeletal disorderwith the responsible mutation beingassigned to chromosome 6 (Cockett et al.,1999). A commercial test is available forthis syndrome.A gene test based on the DQA2 gene thatresides on the MHC complex (Hickford etal., 2004) and predicts susceptibility tofoot rot has been developed at LincolnUniversity in New Zealand. Thirty-onedifferent alleles have been identified forDQA2 and a gene marker test rating hasbeen developed based on a susceptibilityscore of the two alleles of a genotype. Thereis a clear association between the test ratingand the relative risk of contracting foot rot.A gene marker test has been available since

Chapter 13 – Marker-assisted selection in sheep and goats 2392001 and has been used extensively (over40 000 tests).The β-3 adrenergic receptor gene hasbeen sequenced (Forrest and Hickford,2000) and eight different alleles have beenfound. This allelic variation is significantlyassociated with increased risk of coldrelatedmortality of lambs.Meat traitsThe first causal mutation found for meattraits in sheep is the callipyge gene causingmuscular hypertrophy. The gene has beenmapped to chromosome 18 and the causativemutation has been identified. However,the trait is expressed in a rather complexmanner, termed polar over-dominance;only lambs that inherit the callipyge mutationfrom their father but not their motherdevelop the trait. Several interacting genesare involved and the complete molecularbasis of callipyge phenotypes has not yetbeen fully resolved (Freking et al., 2002;Cockett et al., 1996, 2005).The Carwell gene somewhat resemblesthe callipyge gene, as it has been mappedto the same genomic region (distal endof chromosome 18) and it also affectsmuscling (McLaren et al., 2001). However,the overall phenotypic effect is not exactlythe same in that the Carwell gene affectsonly the longissimus dorsi and unlike thecallipyge gene it has not been associatedwith a decreased tenderness if the meatis aged appropriately and neither doesit seem to be affected by the parent oforigin (Jopson et al., 2001). The functionalmutation of the Carwell gene, also knownas the rib-eye muscling (REM) gene, hasnot yet been found but close markers inlinkage disequilibrium with the putativegene are being developed in Australia,New Zealand and the United Kingdom. Acommercial gene test termed “LoinMax”was introduced towards the end of 2005 byOvita in New Zealand.A number of gene detection projectshave resulted in significant QTL for muscle,fat and other carcass traits, but not all ofthese have been published, confirmed orfine mapped. A number of studies havereported on QTL for meat traits in sheep(Broad et al., 2000; Walling et al., 2004;Johnson et al., 2005; McRae et al., 2005)and there are probably some unpublishedQTL being further developed. Some ofthese sheep QTL are based on related cattlegenes, e.g. the myostatin gene for doublemuscling (Grobet et al., 1997) and thethyroglobulin gene affecting intramuscularfat (Barendse et al., 2004).Wool traitsIn a recent paper, Purvis and Franklin (2005)reviewed QTL for wool production traitsand wool quality. Although wool traits canbe measured easily and have high heritability,these authors suggested that researchinto certain wool production genes was stilljustified, for example, to break antagonisticcorrelations (between fleece weight andfibre diameter) or to target specific woolquality traits important for the processingof the product.A few Mendelian (single locus)characteristics have been described forwool. There is a known mutation of thehalo hair gene (HH1) causing extremehairiness. This has been found in the NewZealand Romney breed and several lineshave been developed for the production of“carpet wool” using this specific mutation,.A recessive gene for hairlessness (hr) hasbeen described by Finocchiaro et al. (2003).Several QTL for wool traits have beenpublished (see Purvis and Franklin, 2005for an overview), but few of these havebeen confirmed. On the other hand, it is

240Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishprobable that a number of QTL identifiedhave not been published. It is likely thatsome of these wool QTL will be confirmedand available for gene testing over thenext few years. Polymorphisms associatedwith candidate genes for the wool proteinskeratin and sulphur have been described(Rogers, Hickford and Bickerstaffe, 1994;McLaren et al., 1997) and seem to beassociated significantly with fibre diameterand staple strength.The genetic regulation of some formsof pigmented wool fibres has oftenbeen associated with the Agouti gene(chromosome 13) but this has proven tobe a complex pattern of inheritance withseveral mutations seemingly involved(Smith et al., 2002). More specifically, thereappear to be two Agouti loci and at leasttwo different polymorphisms (deletions).Currently, a genetic test for self colouredblack wool is not yet available. Otherpigmented phenotypes such as badger faceand piebald also have a distinct Mendelianinheritance pattern (Sponenberg, 1997) butthe molecular basis of these phenotypicvariations has not been found.Dairy traitsResearch in dairy sheep has mainly focusedon milk protein polymorphisms, in particularαs1-casein and β-lactoglobulin, butresults have been inconclusive, unlike thosein goats. Together with unfavourable allelefrequencies, these results make it unlikelythat these polymorphisms will be veryuseful in a MAS programme. Further QTLmapping work is under way, focusing onproduction and functional traits (Barillet,Arranz and Carta, 2005).OtherThe Horns gene has been found in sheepas described by Montgomery et al. (1996),allowing improved selection efficiency forpolled sheep.GoatsTwo goat genes have been well studied.Substantial mapping work has been dedicatedtowards finding a gene associated withPolled Intersex Syndrome (PIS), and theactual mutation for PIS has been described(Pailhoux et al., 2005). Furthermore, theeffects of the αs1-casein gene on milk solidscontent (protein, fat, casein, casein/proteinratio) have been described in Frenchdairy goat breeds (Barbieri et al., 1995) andthe molecular basis has been unravelled(Yahyaoui et al., 2003).Examples of sheep and goat MASbreeding programmesThere is little formal literature aboutactual applications of MAS in breedingprogrammes for any livestock species, letalone for sheep and goats. In fact, genetesting and MAS in sheep and goats haveonly very recently been introduced, andtherefore the information compiled inthis section is based mainly on informationobtained from communication withcolleagues in a number of countries (seeAcknowledgements).There are currently two types of MASprogrammes. One is the use of genemarkers in selection programmes withinresearch projects. Usually the genotypingis subsidized and the purpose of the projectis to create additional data for confirmatorystudies of the QTL effect, or simply toobtain “proof of concept” where predictionsbased on simulation and modelling arebeing verified based on real data. In theother type of application, commercialgene testing is used. This is the scenariorequired for long-term and sustaineduse of the technology, but there are few

Chapter 13 – Marker-assisted selection in sheep and goats 241breeding programmes where commercialapplications are viable. The basic conditionis that ram breeders and ram buyers areprepared to pay for genetic informationarising from genetic testing. This is morelikely to happen in places where acrossflockgenetic evaluations already exist,combined with objective trait measurementand trait valuation in the form of indices.However, not all genetic information canbe translated into dollar index terms andgenetic testing is often valued beyond theexisting index framework.Experimental sheep MASThe purpose of “experimental MAS” programmesis to demonstrate that geneticchanges can be achieved based on genotypeselection and thereby to encourage uptakeof MAS by commercial breeders. Usually,the programmes are also designed eitherto estimate QTL effects more clearly, orto confirm earlier experimental results inindustry flocks. Examples of such MASprogrammes are:• selection of sheep against susceptibilityto scrapie, being conducted in France andthe United Kingdom;• the MAS Applied to Commercial Sheep(MASACS) Programme in the UnitedKingdom, coordinated by OswaldMatika from the Roslin Institute. Theresearch team in this programme collaborateswith commercial breeders. Threegene marker tests for muscling are beingtrialled in the first year and it is envisagedthat a test for parasite resistance will beintroduced in 2006. The three QTL aretermed “Texel muscling” (chromosome18), “Suffolk muscling” (chromosome 1)and “Charollais muscling” (chromosome1) as described by McRae et al. (2005),and the tests will be applied within therespective breeds.Commercial sheep MASCommercial gene testing in sheep is limitedmostly to service providers in NewZealand, mainly Ovita and the Universityof Lincoln, whereas it is absent in goats.Details about gene tests can be found onthe Australian Gene Mapping Web site(Maddox, 2005). Gene tests currently availableare:• Foot rot, a gene test commercialized bythe University of Lincoln;• Inverdale gene, through Ovita;• Booroola gene, through Genomnz;• Scrapie, (PrP gene), available throughmany companies (see Maddox, 2005);• Carwell gene, available through Ovita asLoinmax;• Texel Muscling gene (Chrom 2), availablethrough Ovita as MyoMax.None of these tests is currently integratedwith formal genetic evaluationsystems. Rather, gene test results and indexvalues based on polygenic quantitativetraits will have to be used separately, andholistic approaches are needed to deviseselection rules. The gene tests for reproductivetraits are not straightforward to use,while the Inverdale gene is only useful ina heterozygous state and requires specificcrossing programmes. The Booroola inheritancemodel is more straightforward butthe effect is too large for most managementsystems found in Australia.It should also be noted that most ofthe commercial gene tests are for traitsthat are not captured by formal EBVs, andcannot be incorporated easily into existingEBVs, e.g. gene tests for disease traits suchas scrapie and foot rot. In principle, thetests for muscle traits could be part of theEBV calculation, but from a ram marketingperspective it might be more useful to exploitthe genotype information obtained moreexplicitly. Furthermore, the proportion of

242Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishbreeding animals genotyped will be smallin relation to the total number of animalsevaluated based on phenotype, making anintegration of genotypic information withthe full evaluation procedure less sensibleat this stage. Finally, service providersoffering the genotyping results are oftendifferent from the service providers of EBVwhich inhibits a full integration of geneticinformation to the breeder.In New Zealand, a significant and rapidlyexpanding part of the performancerecording sheep industry already usesDNA parentage and the above-mentionedtests are now often provided as part of thatsystem. At this point in time, DNA fractionalparentage (Dodds, Tate and Sise,2005) is included within the SIL system,but MAS EBVs for some of the abovementionedtests (Inverdale, LoinMax) areonly carried out on a stand alone basis, inthe case of LoinMax since 1997.Goat MASA GAS programme is operational for thealpha-S1 casein gene for dairy goats inFrance (Manfredi, 2003). The gene is associatedwith protein content and protein yield.In this programme, young bucks are preselectedwithin families based on genotype.The programme is run by a cooperative AIcentre (Capri-IA) and, although started upwith government funding, it is now almostrunning on a fully commercial basis.MAS in developing countriesMost breeding programmes in developingcountries, if existing at all, are small-scalewith modest objectives. Usually, the challengeis to foster the flow of information(measurement and evaluation) as well as theflow of genes (dissemination of improvedstock). These processes are often inhibitedby infrastructural, logistical and socioeconomicfactors. Clearly, gene markertechnology will not be the first priorityin many of these programmes. However,where gene tests exist for clearly definedcharacters with substantial economic benefit,gene markers and MAS could be verybeneficial. Introgression of disease resistancegenes into productive breeds could beof great value, but few of these examplesexist in sheep and goats.A good example of a clear gene effectsuccessfully implemented in a MAI programmeis found in India (Nimbkar et al.,2005). The Booroola gene is being introgressedhere from the small Garole breedinto the local Deccani breed that is suitablefor meat production but has a limitedreproductive performance. The Booroolagene has tremendous economic effects inthis production system, raising the weaningrate by nearly 50 percent. The breedingprogramme is undertaken by a researchinstitute, but there are clear strategies andactivities to ensure that the improved stockfinds its way to shepherd flocks. Evaluationof the results in these shepherd flocks is anexplicit part of the project, and initial resultslook very promising. Therefore, MAS andMAI should not be ruled out for breedingprogrammes in developing countries, butshould be assessed based on the merit ofeach case. However, implementation ofgene marker technology will only workwithin the framework of a sound existingbreeding programme, ensuring the prerequisitethat genetic information is valued andthat the gene marker accounts for substantialeconomic merit.ConclusionSheep and goat breeding programmes existin low- to medium-input agricultural systemswhere there are many independentbreeding units and where trait recording

Chapter 13 – Marker-assisted selection in sheep and goats 243and genetic evaluation are provided byexternal service agents. This situation is differentfrom that in poultry and pigs and tosome extent dairy, and more similar to thatin beef cattle, in the sense that the businessunits that invest in genetic information arenot the same as those providing geneticevaluation, and EBVs are available in thepublic domain. Also, genotypic informationis an explicit part of the marketing ofgenetic material. The result is that genotypicinformation is more likely to be usedoutside the usual EBV system, with thechance of being overvalued once the investmentis made. There is a place for MASand MAI based on genetic tests for clearlydemonstrated phenotypic effects with economicbenefit, for example for disease,fecundity and meat quality.The number of detected and confirmedQTL is low for sheep and goats and genemapping is less advanced than in otherlivestock species. There is significant investmentand progress being made in markerdevelopment and gene discovery, but itwill take some years before large amountsof genetic information become available atlittle cost, e.g. in the form of SNP chips.Until then, genotypic information will provideadditional selection criteria, makingoptimal selection a greater challenge.Ultimately, the additional value of genemarkers will be greatest in breeding programmesthat already use intensive pedigreeand performance recording, and it will helpto shift selection pressure towards traitsthat are hard to improve based on phenotypic(BLUP) selection (i.e. traits suchas fertility, disease resistance and carcassquality). It is not essential that genetictests are based on functional mutations,as gene markers can have predictive valuedue to being in linkage disequilibrium withfunctional genes. In breeding programmeswithout extensive recording, it is moreimportant to rely on direct markers, butthis will only be valuable in practice ifgenes have very large economic effects.The same holds for genetic tests for distinctMendelian traits, but the overall valueof these traits in breeding programmes islimited. In less-developed breeding programmes,investments in pedigree andperformance recording will most likely bemore profitable than investments in genetechnology.Application of MAS or MAI in manysheep and goat breeding programmes indeveloping countries is not a priority, butopportunities exist, conditional on having aclearly visible phenotypic effect and a programmebased on well-defined objectivesand performance based selection.AcknowledgementsThe author would like to thank the followingcolleagues for providing informationabout breeding and MAS programmes:Richard Apps, Alex Ball, Stephen Bishop,Didier Boichard, Joanne Conington, SchalkCloete, Oswald Matika, Eduardo Manfrediand John McEwan. Information from JillMaddox, her Web site and her commentson the marker maps were very helpful, andRobert Banks made useful comments andimprovements to the manuscript.ReferencesAcin, C., Martin-Burriel, I., Monleon, E., Rodellar, C., Badiola, J. & Zaragoza, P. 2003.Characterization of the caprine PrP-gene study of new polymorphisms and relationship with theresistance/susceptibility to the scrapie disease. Proc. Internat. Workshop on Major Genes in sheepand Goats. 8–11 December 2003. CD-ROM Communication No. 2-31. Toulouse, France.

244Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishBarbieri, M.E., Manfredi, E., Elsen, J.M., Ricordeau, G., Bouillon, J., Grosclaude, F., Mahe, M.F. &Bibe, B. 1995. Effects of the alpha-s1-casein locus on dairy performances and genetic parameters ofAlpine goats. Genet. Sel. Evol. 27: 437–450.Barendse, W., Bunch, R., Thomas, M., Armitage, S., Baud, S. & Donaldson, N. 2004. The TG5 thyroglobulingene test for a marbling quantitative trait loci evaluated in feedlot cattle. Austr. J. Exper.Agric. 44: 669 – 674.Barillet, F. 1997. Genetics of milk production. In L. Piper & A. Ruvinsky, eds. The genetics of sheep,pp 539–564. Wallingford, UK, CABI.Barillet, F., Arranz, J.J. & Carta, A. 2005. Mapping quantitative trait loci for milk production andgenetic polymorphisms of milk proteins in dairy sheep. Genet. Sel. Evol. 37 (Suppl. 1): S109–S123.Broad, T.E., Glass, B.C., Greer, G.J., Robertson, T.M., Bain, W.E., Lord, E.A. & McEwan, J.C.2000. Search for a locus near to myostatin that increases muscling in Texel sheep in New Zealand.Proc. New Zealand Soc. Anim. Prod. 60: 110–112.Carrick, M. 2005. Genotype by environment interaction in wool sheep. Report SHGEN.036. NorthSydney, NSW, Australia, Meat & Livestock Australia.Cockett, N.E., Jackson, S.P., Shay, T.L., Farnir, F., Berghmans, S., Snowder, G.D., Nielsen, D.M. &Georges, M. 1996. Polar overdominance at the ovine callipyge locus. Science 273: 236–238.Cockett, N.E, Shay, T.L., Beever, J.E., Nielsen, D., Albretsen, J., Georges, M., Peterson, K.,Stephens, A., Vernon, W., Timofeevskaia, O., South, S., Mork, J., Maciulis, A. & Bunch, T.D.1999. Localization of the locus causing Spider Lamb Syndrome to the distal end of ovine chromosome6. Mamm. Genome 10: 35–38.Cockett, N.E., Smith, M.A., Bidwell, C.A., Segers, K., Hadfield, T.L., Snowder, G., Georges, M. &Charlier, C. 2005. The callipyge mutation and other genes that affect muscle hypertrophy in sheep.Genet. Sel. Evol. 37 (Suppl. 1): S65–S81.Conington, J., Bishop, S.C., Waterhouse, A. & Simm, G. 2004. A bioeconomic approach to deriveeconomic values for pasture-based sheep genetic improvement programmes. J. Anim. Sci. 82:1290–1304.Crawford, A.M., Dodds, K.G., Pierson, C.A., Ede, A.J., Montgomery, G.W., Garmonsway, H.G.,Beattie, A.E., Davies, K., Maddox, J.F., Kappes, S.W., Stone, R.T., Nguyen, T.C., Penty, J.M.,Lord, E.A., Broom, J.E., Buitkamp, J., Schwaiger, W., Epplen, J.T. Matthew, P., Matthews, M.E.,Hulme, D.J., Beh, K.J., McGraw, R.A. & Beattie, C.W. 1995. An autosomal genetic linkage mapof the sheep genome. Genetics 140: 703–724.Davis, G.H. 2005. Major genes affecting ovulation rate in sheep. Genet. Sel. Evol. 37 (Suppl. 1):S11–S23.Davis, G.H., Montgomery, G.W., Allison, A.J., Kelly R.W. & Bray, A.R. 1982. Segregation of amajor gene influencing fecundity in progeny of Booroola sheep. New Zealand J. Agric. Res. 25:525–529.Davis, G.H., Galloway, S.M., Ross, I.K., Gregan, S.M., Ward, J., Nimbkar, B.V., Ghalsasi, P.M.,Nimbkar, C., Gray, G.D., Subandriyo, I.I., Tiesnamurti, B., Martyniuk, E., Eythorsdottir, E.,Mulsant, P., Lecerf, F., Hanrahan, J.P., Bradford, G.E. & Wilson, T. 2002. DNA tests in prolificsheep from eight countries provide new evidence on origin of the Booroola (FecB) mutation. Biol.Reprod. 66: 1869–1874.Dodds, K.G., Tate, M.L. & Sise, J.A. 2005. Genetic evaluation using parentage information fromgenetic markers. J. Anim Sci. 83: 2271–2279.

Chapter 13 – Marker-assisted selection in sheep and goats 245Dominik, S. 2005. Quantitative trait loci for internal nematode resistance in sheep: a review. Genet.Sel. Evol. 37 (Suppl. 1): S83–S96.Ferrell, C.L. & Jenkins, T.G. 1984. Energy utilisation by mature, non-pregnant, non-lactating cowsof different types. J. Anim. Sci.58: 234–243.Finocchiaro, R., Portolano, B., Damiani, G., Caroli, A., Budelli, E., Bolla, P. & Pagnacco, G. 2003.The hairless (hr) gene is involved in the congenital hypotrichosis of Valle del Belice sheep. Genet.Sel. Evol. 35 (Suppl 1): S147–S156.Forrest, R.H. & Hickford, J.G.H. 2000. Rapid communication: nucleotide sequences of the bovine,caprine and ovine β3-adrenergic receptor genes. J. Anim. Sci. 78: 1397–1398.Freking, B.A., Murphy, S.K., Wylie, A.A., Rhodes, S.J., Keele, J.W., Leymaster, K.A., Jirtle, R.L.& Smith, T.P. 2002. Identification of the single base change causing the callipyge muscle hypertrophyphenotype, the only known example of polar overdominance in mammals. Genome Res.12: 1496–1506.Galloway, S.M., McNatty, K.P., Cambridge, L.M., Laitinen, M.P.E., Juengel, J.L., Jokiranta, T.S.,McLaren, R.J., Luiro, K., Dodds, K.G., Montgomery, G.W., Beattie, A.E., Davis, G.H. & Ritvos,O. 2000. Mutations in an oocyte derived growth factor gene (BMP15) causes increased ovulationrate and infertility in a dosage sensitive manner. Nature Genet. 25: 279–283.Gootwine, E., Rozov, A., Bor, A. & Richer, S. 2003. Effect of the FecB (Booroola) gene on reproductiveand productive traits in the Assaf breed. Proc. Internat. Workshop on Major Genes and QTL inSheep and Goats. 8–11 December 2003. CD-ROM Communication No. . 2-12. Toulouse, France.Grobet, L., Royo Martin, L.J., Poncelet, D., Pirottin, D., Brouwers, B., Riquet, J., Schoeberlein, A.,Dunner, S., Menissier, F., Massabanda, J., Fries, R., Hanset, R. & Georges, M. 1997. A deletionin the myostatin gene causes double-muscling in cattle. Nature Genet. 17: 71–74.Hickford, J.G.H., Zhou, H., Slow, S. & Fang, Q. 2004. Diversity of the ovine DQA2 gene. 2004 J.Anim. Sci. 82: 1553–1563.Hoste, C.H. 2002. Research and development challenges for animal breeding in developing countries.Proc. 7th Wrld. Congr. Genetics Appl. Livest. Prodn. 19–23 August 2002. CD-ROMCommunication No. PS-02. Montpellier, France.Johnson, P.L., McEwan, J.C., Dodds, K.G., Purchas, R.W. & Blair, H.T. 2005. A directed search inthe region of GDF8 for quantitative trait loci affecting carcass traits in Texel sheep. J. Anim. Sci.83: 1988–2000.Jopson, N.B., Nicoll, G.B., Stevenson-Barry, J.M., Duncan, S., Greer, G.J., Bain, W.E., Gerard,E.M., Glass, B.C., Broad, T.E. & McEwan, J.C. 2001. Mode of inheritance and effects on meatquality of the rib-eye muscling (REM) QTL in sheep. Proc. Assoc. Adv. Anim. Breed. Genet. 14:111–114.Khusro, M., van der Werf, J.H.J., Brown, D. & Ball, A. 2004. Across flock (co)variance componentsfor faecal worm egg count, live weight, and fleece traits for Australian merinos. Livest. Prod. Sci.91: 35–43.Maddox, J.F. 2004. Enhancing the sheep genome map. Proc. 29th Int. Conf. Anim. Genet. p. 86. 11–16 September 2004. Tokyo, Japan, International Society for Animal Genetics and Japanese Societyof Animal Breeding and Genetics (available at www.isag.org.uk/ISAG/all/ISAG_Tokyo_2004_Proceedings.pdf).Maddox, J.F. 2005a. Australian Gene Mapping Web site (http://rubens.its.unimelb.edu.au/~jillm/jill.htm).

246Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishMaddox, J.F. 2005b. A presentation of the differences between the sheep and goat maps. Genet. Sel.Evol. 37 (Suppl. 1): S1–S10.Maddox, J.F., Davies, K.P., Crawford, A.M., Hulme, D.J., Vaiman, D., Cribiu, E.P., Freking,B.A., Beh, K.J., Cockett, N.E., Kang, N., Riffkin, C.D., Drinkwater, R., Moore, S.S., Dodds,K.G., Lumsden, J.M., van Stijn, T.C., Phua, S.H., Adelson, D.L., Burkin, H.R., Broom, J.E.,Buitkamp, J., Cambridge, L., Cushwa, W.T., Gerard, E., Galloway, S.M., Harrison, B., Hawken,R.J., Hiendleder, S., Henry, H.M., Medrano, J.F., Paterson, K.A., Phua, S.H., Schibler, L., Stone,R.T. & van Hest, B. 2001. An enhanced linkage map of the sheep genome comprising more than1 000 loci. Genome Res.11: 1 275–1 289.Manfredi, E. 2003. The tale of goat αs1-casein. Proc. 3rd Internat. Workshop on Major Genes andQTL in Sheep and Goats.CD-ROM Communication No. 2-31. Toulouse, France.McEwan, J.C., Dodds, K.G., Greer, G.J., Bain, W.E., Wright, C.S., Green, R.S. & Watson, T.G.1997. Genotype rankings for host resistance to internal parasites for sheep grazed in contrastingenvironments. Proc. Assoc. Adv. Anim. Breed. Genet. 12: 40–43.McLaren, R.J., Broad, T.E., McEwan, J.C., Jopson, N.B., Robertson, T.R., Glass, B.C., Gerard,E.M., Greer, G.J., Bain W.E. & Nicoll, G.B. 2001. Identification of positional candidates for theCarwell locus for rib-eye muscling in sheep. Proc. Plant and Animal Genome IX, W46. 13–17January 2001. San Diego, CA, USA (available at www.intl-pag.org/9/abstracts/W17_03.html).McLaren, R., Rogers, G., Davies, K., Maddox, J. & Montgomery, G. 1997. Linkage mapping ofwool keratin and keratin associated protein genes in sheep. Mamm. Genome 8: 938–940.McRae, A.F., Bishop, S.C., Walling, G.A., Wilson, A.D. & Visscher, P.M. 2005. Mapping of multiplequantitative trait loci for growth and carcass traits in a complex commercial sheep pedigree. Anim.Sci. 80: 135–141.Meszaros, S.A., Henshall, J.M., Burgess, S.K., Gray, G.D. & Tier, B. 1999. Detection of a quantitativetrait locus associated with a reduction of faecal egg count in Merino sheep. Proc. Assoc. Adv.Anim. Breed. Genet. 13: 211–218.Montaldo, H.H. & Manfredi, E. 2002. Organization of selection programs for dairy goats. Proc. 7thWrld. Cong. Genet. Appl. Livest. Prodn. Communication No. 01-35. Montpellier, France.Montgomery, G.W., Henry, H.M., Dodds, K.G., Beattie, A.E., Wuliji, T. & Crawford, A.M. 1996.Mapping the Horns (Ho) locus in sheep: a further locus controlling horn development in domesticanimals. J. Hered. 87: 358–63.Moreno, C.R., Lantier, I., Lantier, F., Andreoletti, O., Vaiman, D., Sarradin, P., Echeynne, F.,Cribiu, E.P., Cosseddu, G. & Elsen, J-M. 2002. Proc. 7th Wrld. Cong. Genet. Appl. Livest. Prodn.CD-ROM Communication No. 13-21. Montpellier, France.Moreno, C.R., Lantier, F., Lantier, I., Sarradin, P. & Elsen, J-M. 2003. Detection of new quantitativetrait loci for susceptibility to transmissible spongiform encephalopathies in mice. Genetics 165:2085–2091.Mulsant, P., Lecerf,, F., Fabre, S., Schibler, L., Monget, P., Lanneluc, I., Pisselet, C., Riquet, J.,Monniaux, D., Callebaut, I., Cribiu, E., Thimonier, J., Teyssier, J., Bodin, L., Cognie, Y., Chitour,N. & Elsen, J.M. 2001. Mutation in bone morphogenetic protein receptor-1B is associated withincreased ovulation rate in Booroola Merino ewes. Proc. Nat. Acad. Sci. USA 98: 5104–5109.Nimbkar, C., Pardeshi, V. & Ghalsasi, P. 2005. Evaluation of the utility of the FecB gene to improvethe productivity of Deccani sheep in Maharashtra, India. In H.P.S, Makkar & G.J. Viljoen, eds.

Chapter 13 – Marker-assisted selection in sheep and goats 247Applications of gene-based technologies for improving animal production and health in developingcountries, pp. 145–154. IAEA/FAO Internat. Symp., Netherlands, Springer.Olayemi, M.E., Walkden-Brown, S.W., van der Werf, J.H.J. & Le Jambre, L.F. 2002. Genetic variabilityof resistance indicators for gastrointestinal nematode infection in Angora and Cashmeregoats. Proc. 7th Wrld. Cong. Genet. Appl. Livest. Prodn. CD-ROM Communication No. 13-30,Montpellier, France.Olivier, J.J., Cloete, S.W.P., Schoeman S.J. & Muller, C.J.C. 2005. Performance testing and recordingin meat and dairy goats. Small Rumin. Res. 60: 83–93.Pailhoux, E., Vigier, B., Schibler, L., Cribiu, E.P., Cotinot, C. & Vaiman, D. 2005. Positional cloningof the PIS mutation in goats and its impact on understanding mammalian sex-differentiation.Genet. Sel. Evol. 37 (Suppl 1): S55–S64.Piper, L.R. & Bindon, B.M. 1982. Genetic segregation for fecundity in Booroola Merino sheep. InProc. Wrld. Cong. Sheep and Beef Cattle Breeding, Vol. 1, pp. 395–400. Palmerston North, NewZealand, Dunmore Press.Pollot, G.E. & Greeff, J.C. 2004. Genotype x environment interactions and genetic parameters forfecal egg count and production traits of Merino sheep. J. Anim. Sci. 82: 2840–2851.Purvis, I.W. & Franklin, I.R. 2005. Major genes and QTL influencing wool production and quality:a review. Genet. Sel. & Evol. 37 (Suppl. 1): S97–S107.Rogers, G.R., Hickford, J.G.H. & Bickerstaffe, R. 1994. Polymorphisms in two genes for B2 sulfurproteins of wool. Anim. Genet. 25: 407–415.Safari, E., Fogarty, N.M. & Gilmour, A.R. 2005. A review of genetic parameter estimates for wool,growth, meat and reproduction traits in sheep. Livest. Prodn. Sci. 92: 271–289.Schibler, L., Vaiman, D., Oustry, A., Giraud-Delville, C. & Cribiu, E.P. 1998. Comparative genemapping a fine scale survey of chromosome rearrangements between ruminants and humans.Genome Res. 8: 901–915.Smith, M.A., Shay, T.L., Beever, J.E., Notter, D.R. & Cockett, N.E. 2002. Identification of anAgouti-like locus in sheep. Anim. Genet. 33: 383–385.Souza, C., MacDougall, J., Campbell, C., McNeilly, B.K. & Baird, D.T. 2001. The Booroola (FecB)phenotype is associated with a mutation in the bone morphogenetic receptor type 1B (BMPR1B)gene. J. Endocrin. 169: R1–R6.Sponenberg, D.P. 1997. Genetics of colour and hair texture. In L.R. Piper & A.Ruvinsky, eds. Thegenetics of sheep, pp. 51–86. Wallingford, UK, CABI.Walling, G.A., Visscher, P.M., Wilson, A.D., McTeir, B.L., Simmand, G. & Bishop, S.C. 2004.Mapping of quantitative trait loci for growth and carcass traits in commercial sheep populations. J.Anim. Sci. 82: 2234–2245.Wilson, T., Wu, X., Juengel, J., Ross, I., Lumsden, J., Lord, E., Dodds, K., Walling, G., McEwan,J., O’Connell, A., McNatty, K. & Montgomery, G. 2001. Highly prolific Booroola sheep have amutation in the intracellular kinase domain of bone morphogenetic protein IB receptor (ALK-6)that is expressed in both oocytes and granulosa cells. Biology of Reprod. 64: 1225–1235.Yahyaoui, M.H., Angiolillo, A., Pilla, F., Sanchez, A., & Folch, J.M. 2003. Characterization andgenotyping of the caprine kappa-casein variants. J. Dairy Sci. 86: 2715–2720.

Section IVMarker-assisted selectionin forestry – case studies

Chapter 14Marker-assisted selectionin EucalyptusDario Grattapaglia

252Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishSummaryPlanted Eucalyptus occupies globally more than 18 million hectares and has become themost widely planted hardwood tree in the world, supplying high-quality woody biomassfor several industrial applications. In this chapter an overview is presented on the statusand perspectives of marker-assisted selection (MAS) in species of Eucalyptus. After anintroduction to the main features of modern eucalypt breeding and clonal forestry, someapplications of molecular markers in support to operational breeding are presented. Byreviewing the status of quantitative trait locus (QTL) mapping in Eucalyptus, the challengesand some realistic prospects for the application of MAS to improve relevant traitsare outlined. With the expected availability of more powerful genomic tools, including adraft of the Eucalyptus genome, the main challenges in implementing MAS will be in phenotypingtrees accurately, analysing the overwhelming amount of genomic data availableand translating this into truly useful molecular tools for breeding.

Chapter 14 – Marker-assisted selection in Eucalyptus 253IntroductionPlanted Eucalyptus forests occupy globallymore than 18 million hectares and havebecome the most widely planted hardwoodforest tree in the world (FAO, 2001).Eucalyptus tree species used in productionforestry are long-lived, evergreen speciesbelonging to the angiosperm familyMyrtaceae (Ladiges, Udovicic and Nelson,2003). They are native to Australia andadjacent islands where they occur naturallyfrom sea level to the alpine tree line, fromhigh rainfall to semi-arid zones, and fromthe tropics to latitudes as high as 43° south(Eldridge et al., 1993; Ladiges, Udovicicand Nelson, 2003). Fast growth rates and awide range of adaptability have contributedto the great interest that Eucalyptus speciesreceive in many countries outside theirnative range. Besides the fast growth thatallows for shorter rotations, many speciesdisplay wood properties that make themvery suitable for fuel and charcoal production,pulp and paper manufacture as well assawn wood. While E. globulus is the premierspecies for temperate zone plantationsin Australia, Chile, Portugal and Spain, elitehybrid clones involving E. grandis and E.urophylla are used extensively by the pulpand paper industry in tropical regions orBrazil, China, the Democratic Republic ofthe Congo and South Africa because oftheir wood quality, rapid growth, cankerdisease resistance and high volumetricyield.Planted Eucalyptus stands supply ina rational and efficient way, high-qualitywoody raw material that would otherwisecome from native tropical forests. In thedecades to come, the expansion of these“fibre farms” will likely be limited by thegrowth of crop plantations and by publicopinion pressure. Increased productivityof forests and refinements in the qualityof wood products by selective breedingwill become of increasing strategic importanceto the forest industry. Moleculartools based on the direct identificationof useful variation at the DNA level areexpected to provide new opportunities forthe genetic manipulation of growth, formand especially wood properties of plantedtrees by marker-assisted selection (MAS)approaches.Almost fifteen years have passed sincethe first experiments in molecular breedingof forest trees. The development of linkagemaps and quantitative trait loci (QTL)information in trees was greatly acceleratedby the advent of more accessible DNAmarker techniques, new concepts in linkagemapping and novel strategies for advancedgeneration tree breeding. From the outset,many expectations were generated for fastand accurate methods for early markerbasedselection in trees. Significant progresshas been made and the knowledge gatheredled to some short-term opportunitiesfor the incorporation of genomic analysisin tree genetics and breeding. However, italso became clear that several challengesremained before more refined and higherimpact applications could be implemented.In this chapter, an overview is presentedon the status of MAS in species ofEucalyptus. The term MAS is used in latusensu, i.e. encompassing the several moleculartechniques and approaches that offerpotential to contribute to eucalypt breeding.Some recent reviews have detailed severalaspects of Eucalyptus genome researchincluding gene discovery, candidate genemapping, functional genomics and physicalmapping (Moran et al., 2002; Grattapaglia,2004; Poke et al., 2005; Shepherd and Jones,2005; Myburg et al., 2006). The focusof this chapter is a more applied one,attempting to link the realities of current

254Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fisheucalypt breeding practice and the moleculartools available or in development.To set the stage for a realistic appraisal ofMAS for Eucalyptus, a brief introductionis presented of the main features of moderneucalypt breeding and clonal forestry inorder to provide a better understandingof the challenges and opportunities thatlie ahead for cost-efficient molecularbreeding. Following this section, some currentlow technological input applicationsof molecular markers in support of operationalbreeding are presented, such as thequantification of genetic diversity and relationships,the analysis of mating patternsand paternity in seed orchards and fingerprintingfor quality assurance and qualitycontrol of clonal propagation. Within theframework of MAS for trait advancement,after reviewing the status of QTL mappingin Eucalyptus, the challenges and somerealistic prospects for the application ofMAS to improve relevant traits are outlined.Finally, with the expected availabilityof a draft of the whole Eucalyptus genomewithin the next years, a succinct summaryis presented on the prospects of advancinggenomic approaches for gene identificationand subsequent application of MAS.Eucalyptus breeding and plantationforestryEucalyptus domesticationEucalypts spread rapidly around the worldfollowing their discovery by Europeansin the late eighteenth century (Eldridgeet al., 1993). They were introduced intocountries such as Brazil, Chile, France,India, Portugal and South Africa in thefirst quarter of the 1800s (Doughty, 2000)and rapidly adopted in forest plantationsas their fast growth and good adaptabilitybecame known. During the nineteenthand twentieth centuries, large quantitiesof seeds were collected and distributeddirectly from Australia through a numberof seed collection expeditions carried outboth by government organizations andprivate forestry companies throughout theworld.Eucalyptus species have a mixed matingsystem, but are predominantly outcrossersand animal pollinated. High levels of outcrossingare maintained by protandry andvarious incomplete pre- and post-zygoticbarriers to self-fertilization including strongselection against the products of inbreeding(Pryor, 1976). Although the major eucalyptsubgenera do not hybridize in nature,hybridization among species within thesame subgenus has been detected, oftenmaking separation of species difficult (Pryorand Johnson, 1971). Hybridization becomesmore frequent in exotic conditions outsidethe natural species range. In fact, this propertyhas been widely exploited by eucalyptbreeders who take advantage of the naturallyoccurring genetic variation for growthand wood properties among species (deAssis, 2000). Several artificial hybrid combinationshave been produced, althoughhybrid inviability tends to increase withincreasing taxonomic distance between theparents (Griffin, Burgess and Wolf, 1988;Potts and Dungey, 2004).In several countries the continued plantationfrom local seed sources gave riseto landraces adapted to the specific environmentof the country (Eldridge et al.,1993). Seed collections from such localexotic plantings of multiple species becamecommon and where plantings occurred, F 1hybrids were derived (Potts and Dungey,2004). While several of these F 1 hybrids performedwell, especially when deployed asclones, seed collection from hybrid standsoften resulted in plantations that performedpoorly in subsequent generations and were

Chapter 14 – Marker-assisted selection in Eucalyptus 255extremely variable. A textbook case isthe Rio Claro hybrid swarm in Brazil(Campinhos and Ikemori, 1977; Brune andZobel, 1981), a eucalypt arboretum whereNavarro de Andrade, the “father of eucalypts”in Brazil first introduced and planteda collection of 144 different Eucalyptusspecies between 1904 and 1909. Several ofthese species hybridized once the naturalbarriers to introgression were removed inthe exotic habitat so that seeds collectedfrom these stands were largely interspecifichybrids. Large commercial plantationswere established in Brazil with seeds fromthis arboretum following fiscal incentivesfor reforestation granted by the governmentstarting in 1966. Although some of theresulting forests were on average economicallyinferior, in these very variable standssome outstanding trees for growth, formand disease resistance derived from chanceevents of recombination were found. Theadvent of operational cloning techniquesat the beginning of the 1980s allowed capturingthe superiority of such hybrids thatare still used today in some of the mostproductive eucalypt clonal plantations inthe world.The history of eucalypt breeding, whichis short when compared with crop species,was detailed by Eldridge et al. (1993) andmore recently reviewed and updated byPotts (2004). Some of the earliest breedingwas undertaken by French foresters inMorocco in 1954–55 (Eldridge et al., 1993).The advent of industrially-oriented eucalyptstands in the 1960s led to a more formalapproach to breeding with, for example, theestablishment of the Florida E. grandisbreeding programme in 1961 (Franklin,1986), E. globulus breeding in Portugal in1965–66 (Potts et al., 2004) and large provenancetests of E. camaldulensis in manycountries (Eldridge et al., 1993). However, amajor breakthrough in eucalypt plantationtechnology occurred in the 1970s with theestablishment of the first commercial standsof selected clones derived from hardwoodcuttings in the Democratic Republic of theCongo (Martin and Quillet, 1974) followedby Aracruz in Brazil (Campinhos andIkemori, 1977). At the same time, in manytropical countries such as Brazil and SouthAfrica, efforts were intensified to establishextensive provenance/progeny trialsof species such as E. urophylla, E. grandisand some others that belonged to the samesubgenus Symphyomyrtus (Eldridge et al.,1993). These trials were established fromopen pollinated seed lots collected fromselected trees in the wild and constitutedthe base populations for subsequent selectivebreeding in many countries. This initialeffort, which was carried out typically bygovernment forestry research institutions,was followed during the 1980s by moreintensive collections by private organizations,targeting elite provenances identifiedin earlier collections as being more adaptedfor species such as E. grandis, E. tereticornisand E. viminalis (Eldridge et al., 1993).Eucalyptus breeding and plantationforestryEucalyptus plantation forestry species arewell known for their fast growth, straightform, valuable wood properties, wideadaptability to soils and climates, andease of management through coppicing(Eldridge et al., 1993; Potts, 2004). Theyare now planted in more than 90 countrieswhere the various species are grownfor products as diverse as sawn timber,poles, firewood, pulp, charcoal, essentialoils, honey and tannin as well as for shadeand shelter (Doughty, 2000). They are animportant source of fuel and building materialin rural communities in countries such

256Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishas China, Ethiopia, India, Peru and VietNam. However, it is the increasing globaldemand for short fibre pulp that has driventhe massive expansion of eucalypt plantationsand accompanying breeding practicesthroughout the world during the twentiethcentury (Turnbull, 1999). Their highfibre content relative to other wood components,coupled with the uniformity offibres relative to other angiosperm species,has led to high demand for eucalypt pulpfor coated and uncoated free-sheet paper,bleach board, sanitary products (fluff pulp),and to a lesser extent for top liners on cardboardboxes, corrugating medium, and asa filler in long fibre conifer products suchas newsprint and containerboard (Kellison,2001). In the last ten years, the developmentof new wood drying and sawingtechnologies has also increased interest inusing plantation eucalypts for sawn wood,veneer, medium density fibreboard and asextenders in plastic and moulded timber(Kellison, 2001).An FAO report estimated a total of17.9 million ha of planted Eucalyptus worldwidewith India as the largest planter withover 8 million ha followed by Brazil with3 million (FAO, 2001). The majority ofplantations consists of only a few eucalyptspecies and hybrids. The most importantare E. grandis, E. globulus, E. urophyllaand E. camaldulensis, which together withtheir hybrids account for about 80 percentof the plantation area, followed byE. nitens, E. saligna, E. deglupta, E. pilularis,Corymbia citriodora and E. teriticornis(Eldridge et al., 1993; Waugh, 2004). Marketfavourites for pulpwood are E. grandis, E.urophylla and their hybrids in tropical andsubtropical regions and E. globulus in temperateregions.Although eucalypt breeding is currentlya very dynamic and technically advancedoperation carried out mainly by severalprivate companies, eucalypts should beseen as still in their domestication infancywhen compared with crop species, withmost breeding programmes only one ortwo generations removed from the wild.However, with the combination of amplegenetic variation both at the intra and interspecificlevels and the ability to clone elitegenotypes, eucalypts have quickly becomeamong the most advanced genetic materialin forestry. Breeding of eucalypts has movedfaster in countries such as Brazil, Chile,Portugal and South Africa that adoptedEucalyptus for industrial plantation forestry.Most eucalypt breeding programmesworldwide are focused on geneticallyimproving trees for industrial pulpwoodproduction (Borralho, 2001; Kanowski andBorralho, 2004). The target traits of mostbreeding programmes include volumetricgrowth per hectare, wood density and pulpyield (Borralho, Cotterill and Kanowski,1993). Traits such as pest and disease resistanceand adaptability to abiotic stressessuch as frost, drought or wind are usuallysecondary targets that become importantwhen they have an impact on one or moreof the main traits. Following the standardconcepts in tree breeding, large geneticgains have been obtained in the early stagesof eucalypt domestication, simply throughspecies and provenance selection followedby individual selection and establishmentof clonal or seedling seed orchards orclonal propagation of elite selections fordirect deployment (Eldridge et al., 1993;Kanowski and Borralho, 2004; Potts, 2004).Subsequent population improvement hasalso demonstrated significant genetic gainthrough recurrent selection in an open-pollinatedbreeding population coupled withopen or controlled pollinated populationsof the most elite selections or specialized

Chapter 14 – Marker-assisted selection in Eucalyptus 257Figure 1Basic breeding scheme used in Eucalyptus breeding involving recurrent selection cycles fromwhich elite parent trees are derived for the establishment of seed orchards, as well as individualtrees to be tested and later recommended as clonesNative forest standsSelection of best trees ,possibly unrelatedProvenance trialsPROGENY TRIALSRECOMBINATIONMating of selected trees andgeneration of progeniesSELECTIONPhenotypic evaluation formain target traitsSELECTED POPULATIONVegetative material or seeds collectedClonal tests of selected treesEstablishment of seedorchardsFinal selection and operationalclone recommendationImproved seeds forcommercial forestsbreeds (Potts, 2004). For species that areeasily propagated vegetatively, such as E.grandis, E. urophylla and several of theirhybrids, clonally propagated breeding populationshave enhanced gains by allowingthe capture of additive and non-additivegenetic effects (Figure 1).Clonal forestry of EucalyptusAfter more than 25 years following theintroduction of clonal forestry of Eucalyptus(Campinhos, 1980; Brandão, Campinhosand Ikemori, 1984), this forest productionsystem is now perfectly integrated into thestrategies and plans of advanced generationbreeding programmes. Clonal propagationand hybrid breeding have constituted anextremely powerful combination of toolsfor the rapid improvement of the qualityof wood and wood products. While thefirst hybrid clones were selected based onlarge-scale screening of high-yielding spontaneoushybrids resistant to diseases (such asthe eucalypt canker), today clones are beingderived increasingly from deliberate interspecifichybrid production strategies (Figure2). Eucalypt hybrids, involving two or morespecies deployed as clones, currently makeup a significant proportion of eucalypt plantationforestry, particularly in the tropicsand subtropics. In a recent survey of clonalforestry in Brazil, for example, consideringall the large and medium-sized companies,the area planted with clones corresponded

258Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishFigure 2One-stop pollination of Eucalyptus(A) Elite parent trees, kept as grafts in indoor insect-proof orchards, are induced to flower with growth regulators inapproximately 12 to 15 months. (B) Flowers to be used are still closed; open protandric flowers are discarded. (C) Flowers arecut open before anthesis with a nail cutter. (D) Pollen from the other parent is deposited directly at the base of the style andno bag protection is needed as the greenhouse is kept free of insects. (Photographs courtesy of Teotônio F. de Assis)to more than 1 008 000 ha, involving 362 differentclones at a rate of 2 to 40 clones percompany, and a range of 10 to 34 000 ha perclone (mean 4 150 ha). The annual introductionof new clonal plantations to supportexpansion of forest-based industrial productionis in the order of 238 000 ha, witha mean of 1 820 ha per clone (de Assis,Rezende and Aguiar, 2005).An important paradigm shift in eucalyptbreeding for pulp and paper began inthe 1990s with the increasing realization

Chapter 14 – Marker-assisted selection in Eucalyptus 259that the actual “pulp factory” is the tree.Particularly in vertically integrated pulpproduction systems, as highly productiveclonal forests with over 40 m 3 /ha/yr becamethe standard (Binkley and Stape, 2004), thefocus shifted quickly from volume growthto wood quality with the objective of improvingpulp yield per hectare by reducingwood specific consumption (WSC), i.e. theamount of wood in cubic metres necessaryto produce one tonne of pulp. Trees thatyield more cellulose generate savings all theway from tree harvesting, transportation,chipping and pulping while mitigating theneed for an accelerated expansion of theforest land base.Clonal forestry of E. grandis xE. urophylla selected clones in the 1980swas able to reduce WSC from 4.9 to4.0 m 3 /tonne of pulp (Ikemori, Penchel andBertolucci, 1994). However, it is now wellknown by breeders that E. globulus has thebest combination of wood properties forpulp and paper among the commerciallyplanted Eucalyptus species, resulting in ahigh pulp yield requiring approximately25 percent less wood to produce the sametonne of cellulose. While only 3.0 m 3 ofE. globulus wood are required per tonneof pulp, 4 m 3 are needed from selected E.grandis. E. globulus has a very adequatewood density in the range of 550 kg/m 3 ,the longest fibre length and the largest contentof holocellulose and pentosans of anyother intensively planted species (Sanchez,2002). E. globulus, however, is much moredemanding on soil fertility, is not adaptedto tropical temperatures, is slower growingand more difficult to propagate clonallythan E. grandis. In the last ten years, basedon the very successful pioneering experiencesin Brazil led by Teotônio de Assis,several breeding programmes in tropicalcountries have started an intensive effortto introgress superior E. globulus pulptraits into the tropical and subtropical highyielding genetic backgrounds of E. grandisand E. urophylla. Given the very highgenetic diversity that segregates in suchcrosses together with the technical possibilityof practising intensive within-familyselection and clonal propagation, this efforthas resulted in the development of exceptionaltrees that combine superior growthand adaptability to tropical conditions,higher pulp yielding wood and easy propagationusing minicutting/hydroponicstechnology (de Assis, 2000, 2001; Figure 3).A new wave of clonal forestry is thereforestarting that will most likely result inanother significant jump in the quality ofEucalyptus forests.It is therefore in the context of a highlyspecialized industrially-oriented breedingprogramme that fully exploits the powerof hybrid breeding and clonal forestry thatone needs to discuss the prospects of MASin Eucalyptus. Understanding the fundamentaldifferences between E. grandis andE. globulus at the molecular level to exploitbetter the natural allelic variation that existsin the genus has been the starting point.Marker-assisted managementof genetic variation in breedingpopulationsThe use of genome information for thepractice of directional selection of superiorgenotypes still represents a challenge thatdepends on further and more refined experimentalwork (see below). Nevertheless,molecular markers can be used immediatelyto solve several questions related to themanagement and identification of geneticvariation in breeding and production populations.These applications can be usefulessentially to any breeding programmeindependently of its stage of develop-

260Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishFigure 3Selection and clonal propagation of elite trees by the minicutting technology(A) Elite trees are selected, juvenile sprouts are induced by partial bark stripping while keeping the tree alive. (B) Operationalmother plants in hydroponic sand beds from where apical minicuttings are harvested for propagation. (C) Minicuttings arerooted without any use of growth regulators in controlled environment greenhouses. (D) High productivity clonal foreststands that reach over 60 m 3 /ha/year.ment. Although isozyme markers wereinitially used for these purposes (Moranand Bell, 1983), DNA polymorphismsprovide an enhanced level of resolutionboth at the locus level with much higherexpected heterozygosity values and at thegenome level with greater coverage. DNAmarkers provide a powerful tool to quantifyexisting levels of genetic variation inbreeding and production populations offorest trees. Molecular markers can be usedto estimate the extent of genetic divergencebetween individuals selected to composesuch populations and resolve several issuesof individual identity even at high levels ofrelatedness, including varietal protectionand the verification of alleged parentage inopen pollinated breeding systems. Someoperational applications of molecularmarkers for management of genetic variationin Eucalyptus are outlined below.Identification of elite clonesThe correct identification of clones iscurrently the most common application ofmolecular markers in Eucalyptus operational

Chapter 14 – Marker-assisted selection in Eucalyptus 261breeding and production forestry. Thisapplication is nowadays routinely used byseveral forest companies in Australia, Brazil,Portugal, South Africa and Spain. Qualitycontrol and quality assurance of largescaleclonal plantation operations becomecrucial aspects in forestry, especially invertically integrated production systemswhere the pulp mill plans on the availabilityof clones with specific wood propertiesat specific times. Given the scale of suchoperations that frequently have to feedplantation programmes of several thousandhectares per year, (i.e. several millionseedlings), mislabellings can seriously affectthe expected production. Correct clonalidentity has also important implicationsin several breeding procedures such asseed orchard management or controlledpollination programmes affecting theexpected gains of breeding cycles.Several technologies are available todayto resolve questions of clonal identity inEucalyptus. Dominant markers such asrandom amplified polymorphic DNA(RAPD) or amplified fragment lengthpolymorphism (AFLP) have been used forclonal fingerprinting of eucalypts (Keil andGriffin, 1994; Nesbitt et al., 1997; Costae Silva and Grattapaglia, 1997). Dominantmarkers are, however, very limited intheir ability to establish conclusively theidentity of two redundant individual treesdue to artefact polymorphisms. Dominantmarkers can be used to establish thattwo individuals are not the same, but thestatement that two individuals are identicalis usually only approximate and no formaltest statistics can be attached to thisassertion. The high degree of multi-allelismand the very clear and simple co-dominantMendelian inheritance of microsatellitesprovide an extremely powerful system forthe unique identification of individualsfor fingerprinting purposes and parentagetesting particularly when the individuals areexpected to be related. Kirst et al. (2005a)demonstrated the high resolving powerof this class of markers in Eucalyptus. Abreeding population of 192 individuals of E.grandis was genotyped with a set of six highlypolymorphic microsatellites. The numberof alleles detected ranged from 6 to 33 withan average of 19.8±9.2 and the expectedheterozygosity averaged 0.86±0.11. Usingthree loci all 192 genotypes could be readilydiscriminated. The combined probabilityof identity (i.e. the probability of twoindividuals having the same multilocusgenotype) considering all six loci was less thanone in 2 000 million. Similarity coefficientsestimated from microsatellite data weremuch smaller, thus more discriminative,than those usually obtained in similarstudies with RAPD and AFLP markers.In common with human forensic DNAanalysis, the standard method for clonalidentification in eucalypts today is basedon multiplexed, multicolour fluorescentanalysis of microsatellite markers sized inan automatic sequencer. The identity ofsamples is declared based on a maximumlikelihood ratio where the likelihood ofobserving those genetic data conditionalon the hypothesis of the two samples beingderived from the same clone is comparedwith the alternative hypothesis, i.e. thatthe two samples are derived from differentclones. Furthermore, the repeatabilityand precision of multilocus genotypedetermination allows correct comparisonsacross laboratories and at different times.Varietal protectionFollowing publication of the varietalprotection law in Brazil, specific instructionsfor protecting Eucalyptus cloneswere published in 2002 by the Ministry

262Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishof Agriculture of Brazil based on a set ofvalidated morphological descriptors. Tothe best of the author’s knowledge, this isthe only country today that has formalizedsuch descriptors, which include 36 morphologicaltraits of leaves, flowers, barkand fruit as well as wood density. Althoughthese descriptors generally satisfy the basicrequirements of stability and low environmentalinfluence, they are still difficult toevaluate, especially those related to maturetraits in flowers and fruits. Furthermore,it is common that clones are related bycommon ancestry making their discriminationeven more difficult. The high powerof discrimination coupled with the generalacceptance of DNA technology by eucalyptbreeders in Brazil resulted in the inclusionof molecular markers as additionaldescriptors (Grattapaglia et al., 2003). Theinclusion of DNA markers represented aremarkable advance that Brazil made in theinternational landscape of varietal protectionof forest trees. Currently all requestsfor clonal protection are accompanied bya multilocus DNA profile (DNA fingerprint)of 15 to 20 microsatellite markersthat were recommended based on severalaspects such as robustness, polymorphicinformation content and general availabilityin the public domain. The perspective forthe following years points to an increasednumber of applications for clone protectionby forest companies in view of theoutstanding value of elite eucalypt clonesfor the maintenance of competitivenessof the forestry-based industry. It can beexpected that DNA markers will add asignificant power of resolution for distinctness,uniformity and stability (DUS) testsin varietal protection of eucalypt clones,especially when closely related individualsare under scrutiny in legal disputes overclonal property.Characterization of breedingpopulationsBreeding populations can be characterizedby quantifying the levels and organizationof genetic variation within and betweenbreeding groups, sublines and progenies.These data can immediately be used toimprove the structure of breeding populations,infuse new material and decideon selection, enrichment or eliminationof germplasm entries. In the incompletepedigree systems frequently used in eucalypts,marker-based systems have been usedto monitor the levels of random geneticvariation throughout the different cyclesof a breeding programme thus allowingmuch greater flexibility and control overthe rate of reduction of genetic variability.For example, RAPD markers were successfullyused to characterize the wide range ofgenetic variation in a germplasm bank of E.globulus and thereby assist in the designingof further seed collections (Nesbitt et al.,1995). Gaiotto and Grattapaglia (1997) estimatedthe distribution of genetic variabilitywithin and between open pollinated familiesof a long-term breeding population of E.urophylla, and proposed a selection strategywithin and between families for incompletepedigreed populations based on theincorporation of genetic diversity measures.Marcucci-Poltri et al. (2003) used AFLPand microsatellite markers to obtain quantitativeestimates of genetic diversity in a E.dunnii breeding population selected for fitnessto subtropical and cold environments.Molecular data were used to design a clonalseed orchard using the nine most divergentpairs of genotypes, thereby retaining95.2 percent of the total number of allelesfrom the 140 polymorphic AFLP loci andthe four microsatellite loci analysed. In a subsequentstudy, Zelener et al. (2005) selectedE. dunni trees using trait selection index

Chapter 14 – Marker-assisted selection in Eucalyptus 263and genetic diversity measures estimatedfrom AFLP and microsatellites. Genetic differentiationestimates consistently showedlow differentiation among provenances andgreat differentiation among families suggestingthat orchard design should be basedon individual or family selection rather thanon provenance selection.Mating and deployment designs basedon genetic distanceGiven the wide genetic diversity and multiplesources of available germplasm foreucalypt breeding, choices typically haveto be made as to which elite parents shouldbe mated. Some selection based on the individual’sown performance or on pedigreeinformation is used before including it ina mating design. Any means of predictingtree performance deserves attention. One ofthe “holy grails” of molecular breeders hasbeen the ability to predict progeny performanceaccurately based on distance estimatesamong parents from genetic marker data.Vaillancourt et al. (1995a) used genetic distancesbased on RAPD markers to predictheterosis in E. globulus progenies. Theability of genetic distance to predict heterosiswas significant but accounted for lessthan 5 percent of the variation in specificcombing ability. Baril et al. (1997) used thestructure of RAPD genetic diversity withinand between E. grandis and E. urophylla towork out prediction equations for the treetrunk volume of individual hybrids at 38months. Surprisingly, this study showedthat a genetic distance based on RAPDmarkers with similar frequencies in the twospecies successfully predicted the value ofa cross. Through this model, the distancecalculated between species explained thegeneral combining ability and the specificcombining ability of volume growth witha global coefficient of determination of81.6 percent. RAPD markers were usedto recommend more divergent crosses in areciprocal recurrent selection programmefor hybrid breeding in Brazil (Ribeiro,Bertolucci and Grattapaglia, 1997). A setwith the 20 most and 20 least divergentcrosses between populations was recommended.Matings between more divergentindividuals will potentially allow segregationto be maximized in the resultingprogenies and transgressive segregants tobe recovered and used as clones.RAPD data were used to quantifyrelatedness among elite eucalypt clonesfor deployment purposes. As the historyof selective breeding in eucalypts is veryrecent, little, if any, pedigree informationis typically available. Furthermore, clonalplantations of Eucalyptus generally involveonly a few superior genotypes of unknownorigin. Costa e Silva and Grattapaglia(1997) used RAPD markers to quantify thegenetic relatedness among a group of 15elite clones. Comparative similarity analysesshowed that there was significantlymore genomic variation in the group ofclones than both within and between unrelatedhalf-sib families from a single species.Data on genetic similarity among cloneswere also used to propose a deploymentstrategy in a “genetic mosaic”, i.e. avoidingplanting more genetically related clones sideby side in contiguous forest blocks. Thisproposed strategy was based on the premisethat related clones share a common originand ancestry, have been subject to similarevolutionary selective pressures, and thereforeshare common susceptibility/tolerancealleles at pest and pathogen defence loci.Mating system and paternity inbreeding populationsOpen pollinated breeding by controllingexclusively the maternal progenitor and

264Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishFigure 4Retrospective selection strategy of elite parents based on paternity testing of progenyindividuals displaying superior performance using microsatellite markersSeed orchardSeed harvest with maternal controlCommercial plantation by half- sib familyblocks or half-sib progeny trial?????????????ES76EG62EG65???????Identification of top trees for specificphenotypic traits of interestRetrospective selection of parents with highspecific combining ability (SCA) to be usedin controlled crosses and/or to cull fromseed orchard parents of low SCAPaternity testing of selected trees usingmicrosatellite markers to identifyprecisely theirpollen parentshalf-sib progeny testing is still commonpractice in some eucalypt breeding programmes.It is a low cost option thatallows good estimation of the breedingvalue of maternal parents. Nevertheless, alarge amount of genetic variation is usuallyencountered within half-sib families andselection intensity within families is limitedby the number of individuals usuallydeployed in a progeny test. Knowledge ofoutcrossing versus selfing rates is essentialfor maintaining adequate levels of geneticvariability for continuous gains. A numberof studies have shown that eucalypts arepreferentially outcrossed both in naturalpopulations as well as seed orchards.Isozyme markers were originally used(Moran, Bell and Griffin, 1989), but othertypes of markers now provide a muchhigher level of resolution. Outcrossing ratein an open pollinated breeding populationof E. urophylla was estimated at 93 percentusing RAPD markers, indicating predominantoutcrossing and maintenance ofadequate genetic variability within families(Gaiotto, Bramucci and Grattapaglia, 1997).A complex pattern of mating was describedin a E. regnans seed orchard in Australiawhere gene dispersal was influenced bycrop fecundity and orchard position ofmother trees with approximately 50 percentof effective pollen gametes comingfrom males more than 40 metres away frommother trees (Burczyk et al., 2002). In a

Chapter 14 – Marker-assisted selection in Eucalyptus 265detailed mating system study in a E. grandisorchard in Madagascar, the outcrossing ratewas found to be 96.7 percent but a pollinationrate from outside the seed orchardof 39.2 percent was estimated based on sixmicrosatellite markers (Chaix et al., 2003).The ability to determine paternity preciselyusing DNA markers was recentlyproposed as a short-term breeding tacticfor Eucalyptus. The conventional wayto drive modifications in old forest treeseed orchards is to establish progeny trialsinvolving each parent tree and then evaluateits contribution to the performanceof the progeny by estimating its generaland specific combining ability (GCA andSCA). Grattapaglia, Ribeiro and Rezende(2004) successfully applied an alternativeretrospective parent selection tactic basedon paternity testing of superior offspring.After identifying seed mixtures, selfedindividuals and offspring sired by pollenparents outside the orchard, one particularpollen parent was found to have siredsignificantly more high-yielding progenytrees. Based on these results, low reproductivesuccess parents were culled fromthe orchard and management procedureswere adopted to minimize external pollencontamination. A significant difference(p < 0.01) in mean annual increment wasobserved between forest stands producedwith seed from the orchard before andafter selection of parents and revitalizationof the orchard. An average realizedgain of 24.3 percent in volume growth wasobtained from the selection of parents asmeasured in forest stands at age two to fouryears. The marker-assisted tree breedingtactic efficiently identified top parents in aseed orchard and resulted in an improvedseed variety. It should be applicable for rapidlyimproving the quality of output fromseed orchards especially when the breederis faced with an emergency demand forimproved seeds (Figure 4).Molecular breedingMolecular markers and maps forEucalyptusIn the last ten years a number of studieshave reported genetic maps for Eucalyptusbuilt from combinations of several hundredRAPD, AFLP or RFLP markers(Grattapaglia and Sederoff, 1994; Verhaegenand Plomion, 1996; Marques et al., 1998;Myburg et al., 2003), together with RFLP,isozymes, EST, genes and some microsatellites(e.g. Byrne et al., 1995; Gion et al.,2000; Bundock, Hayden and Vaillancourt,2000; Thamarus et al., 2002; Brondani etal., 2002). In contrast to crop species wheremapping populations are designed basedon contrasting inbred lines, map constructionin eucalypts has relied on availablepedigrees drawn from operational breedingprogrammes. These pedigrees generallyinvolve only the highly heterozygous parentsand their F 1 progeny, either full-sibsof half-sibs. Genetic mapping has thereforebeen carried out using a pseudo-testcrossstrategy, analysing dominant markerspresent in one parent and absent in theother (Grattapaglia and Sederoff, 1994).Maps are therefore individual-specific andcannot be aligned or integrated as suchunless other markers common to both mapsare also used. Consequently, although somegenome maps of eucalypts have been constructed,the use of the linkage informationtends to remain restricted to the pedigreeemployed as the mapping population,limiting the interexperimental sharing oflinkage mapping and QTL data generated.It is now well accepted that true advancementsin QTL validation across pedigrees forthe eventual practice of MAS in Eucalyptus,will strongly depend on the availability of

266Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishhigher-throughput, higher polymorphismtyping systems such as microsatellites,organized in dense genetic maps (Brondaniet al., 1998; Thamarus et al., 2002). Only 137autosomal microsatellite markers have beenpublished to date for species of Eucalyptus,including 67 from E. globulus, E. nitens,E. sieberi and E. leucoxyon (Byrne et al.,1996; Steane et al., 2001; Glaubitz, Emebiriand Moran, 2001; www.ffp.csiro.au/tigr/molecular/eucmsps.html; Ottewell et al.,2005) and 70 from E. grandis and E. urophylla(Brondani et al., 1998; Brondani,Brondani and Grattapaglia, 2002). Recentlya set of 35 chloroplast DNA microsatelliteswas developed based on the full cp-DNAsequence of E. globulus (Steane, Jones andVaillancourt, 2005). Microsatellite transferabilityacross species of the subgenusSymphyomyrtus, which includes all themost widely planted species, varies between80 and 100 percent depending on the sectionto which they belong. It still remainsaround 50 to 60 percent for species of differentsubgenera such as Idiogenes andMonocalyptus and goes down to 25 percentfor the related genus Corymbia (Kirstet al., 1997). Microsatellite comparativemapping data have also shown that genomehomology across species of the same subgenusSymphyomyrtus is very high, notonly in terms of microsatellite flankingsequence conservation, but also markerorder along linkage maps (Marques et al.,2002). Although some tens of microsatelliteshave been mapped on existing RAPDand AFLP framework maps (Brondani,Brondani and Grattapaglia, 2002; Marqueset al., 2002; Thamarus et al., 2002), thegenus Eucalyptus still lacks a more comprehensivegenetic map widely useful formolecular breeding practice. To fill this gap,a novel set of 230 new microsatellites hasrecently been developed and a consensusmap assembled covering at least 90 percentof the recombining genome of Eucalyptus.This map has 234 mapped loci on 11 linkagegroups, an observed length of 1 568 cM anda mean distance between markers of 8.4cM (Brondani et al., 2006). This representsan important step forward for Eucalyptuscomparative genomics, opening stimulatingperspectives for evolutionary studies andmolecular breeding applications. The generalizeduse of an increasingly larger setof interspecific transferable markers andconsensus mapping information will allowfaster and more detailed investigations ofQTL synteny among species, validationof QTL and expression-QTL across variablegenetic backgrounds, and positioningof a growing number of candidate genesco-localized with QTL, to be tested inassociation mapping experiments.QTL mapping in EucalyptusFollowing the construction of linkagemaps, several groups have reported theidentification of genomic regions that havea significant effect on the expression of economicallyimportant traits in Eucalyptus.QTL mapping experiments have, withoutexception, found a few major effect QTLfor all traits considered in spite of the limitedexperimental precision, the lack ofpre-designed pedigree to maximize phenotypicsegregation, and the relatively smallsegregating populations evaluated. This canbe explained by the undomesticated natureand wide genetic heterogeneity of eucalyptsadded to the fact that most QTL mappingexperiments were carried out in interspecificpopulations thus taking advantage ofcontrasting gene pools. QTL for juveniletraits such as seedling height, leaf area andseedling frost tolerance have been mapped(Vaillancourt et al., 1995b; Byrne et al.,1997a, b), while traits related to vegetative

Chapter 14 – Marker-assisted selection in Eucalyptus 267propagation ability such as adventitiousrooting, stump sprouting and in vitro shootmultiplication have also been detected(Grattapaglia, Bertolucci and Sederoff,1995; Marques et al., 1999), as has a majorQTL for early flowering (Missiaggia,Piacezzi and Grattapaglia, 2005). In addition,QTL for insect resistance and essentialoil traits were mapped (Shepherd, Chaparroand Teasdale, 1999) and recently a majorQTL for Puccinia psidii rust resistance withquasi Mendelian inheritance was found andmapped in E. grandis (Junghans et al., 2003).Major QTL were also found for rotationage traits such as volume growth, woodspecific gravity, bark thickness and stemform (Grattapaglia et al., 1996; Verhaegenet al., 1997; Thamarus et al., 2004; Kirst etal., 2004, 2005b).Although QTL of relatively large effectshave been detected for growth traits, whenit comes to potential application in MASthe best opportunities for QTL mappingare those related to specialized wood propertiesthat have a direct impact on industrialprocesses. These traits are usually difficult tomeasure both because they require destructivewhole stem sampling and because theyare traits that are expressed late. Myburg(2001) demonstrated the application of indirect,high-throughput phenotyping of woodquality traits in Eucalyptus by near infraredreflectance spectroscopy (NIRS) for QTLmapping in a hybrid E. grandis x E. globulusbackcross population. Approximately300 individuals that had been previouslygenotyped with AFLP markers were analysedby NIRS, and predictions made forpulp yield, alkali consumption, basic density,fibre length and coarseness, and severalwood chemical properties (lignin, celluloseand extractives). A variety of molecularmarker classes and pedigree types wereused in these experiments. QTL weredetected in F 1 , inbred or outbred F 2 andhalf-sib families with or without clonalreplicates. Also looking at wood qualitytraits, Thamarus et al. (2004) used novelhigh-throughput and traditional methodsto quantify wood density, fibre length, pulpyield and microfibril angle (MFA) in twofull-sib families of E. globulus that shareda common parent. Pulp yield and cellulosecontent were determined by NIRS, andMFA was quantified by SilviScan. Exceptfor fibre length, QTL for all traits couldbe detected in both populations, includingthree QTL in common genetic regions onboth crosses for wood density, one for pulpyield and one for MFA. The proportionof phenotypic variation explained by theQTL identified in both crosses ranged from3.2 to 15.8 percent.Recently QTL analysis of transcriptlevels of lignin-related genes showed thattheir mRNA abundance is regulated bytwo genetic loci co-localized with QTL forgrowth, suggesting that the same genomicregions are regulating growth, lignin contentand composition (Kirst et al., 2004).In a subsequent study, Kirst et al. (2005b)showed that one identified expression QTLexplained up to 70 percent of the transcriptlevel variation for over 800 genesand that hotspots with co-localized expressionQTL were identified on single treeAFLP typically containing genes associatedwith specific metabolic and regulatorypathways, suggesting coordinated geneticregulation. The correlation of gene expressionprofiles in segregating progeny can alsoextend knowledge about genes involved inthese pathways. Complementary DNAsrepresenting previously uncharacterized orhypothetical genes, whose transcript levelsare strongly correlated with those of geneswith known functions, may be associatedwith the same pathway or biological process.

268Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishSimilarly, new functions can tentatively beassigned to previously characterized genesthat had not been described in the contextof revealing pleiotropic action of thesegenes. However, a major limitation in thistype of study in Eucalyptus is the lack ofa completed genome sequence because,without this, the relative locations of largenumbers of genes and their expressionQTL cannot be determined. This informationis required to assess whether thegenetic control of gene expression variationis in the cis- or trans-form for each gene. Inthe context of MAS in Eucalyptus, a betterunderstanding of such “master expressionQTL” that apparently control cascades ofgene expression of important biochemicalpathways may be very promising targetsfor detailed characterization in associationmapping experiments to uncover relevantpolymorphisms to be used in molecularbreeding practice.In summary, although the number ofreports detecting QTL in Eucalyptus hasgrown and these have become increasinglysophisticated, the large majority of mappedQTL have been localized on RAPD orAFLP maps. Consequently, it is impossibleto compare positions of QTL for the sameor correlated traits, seriously limiting thelong-term value of such mapping for MAS.Exceptions are QTL studies where transferablemarkers such as a few microsatellites(Marques et al., 2002; Thamarus et al.,2004) or candidate genes (Gion et al., 2000;Thamarus et al., 2004) were also mapped sothat it is at least possible to make a roughpreliminary comparison of QTL locationsat the linkage group level. Especially in thegenus Eucalyptus where breeders worldwidetake advantage of interspecific geneticvariation for wood properties and diseaseresistance through hybridization, therecent availability of a robust, genus-widegenetic map with highly transferable microsatellitemarkers (Brondani et al., 2006)should stimulate improved genomic undertakingsincluding QTL validation acrosspedigrees, co-localization of QTL andcandidate genes for guiding associationmapping experiments, positional cloning ofQTL and eventually MAS.MAS in EucalyptusTwenty years have passed since the firstdemonstrations that QTL for major effectscould be mapped with molecular markers(Stuber et al., 1980; Paterson et al., 1988;Lander and Botstein, 1989), and severalreviews have described the potential benefitsand caveats of MAS in the plant geneticsliterature (e.g. Tanksley, 1993; Beavis, 1998;Young, 1999; Mauricio, 2001; Dekkers andHospital, 2002). Yet, large-scale operationalMAS is still largely restricted to very fewcrops and for very specific applications.Maize is probably the best example, wherethe financial returns on hybrid seed developmentcoupled with the ability to controlgermplasm fully, has prompted large-scaleinvestments in MAS by the private sectorbased on high-throughput single nucleotidepolymorphism (SNP) genotyping platforms.Based on a detailed understandingof the molecular architecture of quantitativetraits, current applications include yieldoriented advanced backcross QTL (AB-QTL) systems as well as accelerated lineconversion following trait introgression bymarker-assisted backcrossing (MABC). InEucalyptus and forest tree breeding in general,the application of molecular markersfor directional selection is still an unfulfilledpromise. This is largely due to: the recentdomestication of tree crops and hence thewide genetic heterogeneity of breedingpopulations; the inability to develop inbredlines at least on a short-term basis to allow a

Chapter 14 – Marker-assisted selection in Eucalyptus 269Figure 5Classification of three different types of marker-trait associations relevant to Eucalyptus MAS(see text for details)QTLLE markers – Linkage equilibriumEx. microsatellite markers flanking a QTL mapped in a highLD pedigree - Centimorgan resolution, ~ 10 5 a 10 7 bpQTL or candidate geneResolutionLD markers – Linkage disequilibriumEx. SNPs strongly associated with the QTL or candidategene - Subcentimorgan resolution ~ 10 2 a 10 4 bpGene and exact polymorphism (QTN) identifiedDirect markersEx. causal SNPs ( QTNs ) of quantitative variationMaximum resolutionand identification of exact allelemore precise understanding of genetic architectureof quantitative traits; the absence ofsimply inherited traits that could be immediatelyand more easily targeted; and finallyto the very limited number of scientistsactually working on forest trees.If applying MAS in other intensivelystudied crops besides maize is already asignificant challenge; this challenge is evenmore difficult and complex for Eucalyptusand forest trees in general as it pre-supposes:(i) the manipulation of polygenic traits withvariable heritabilities in breeding populationswith a heterogeneous genetic baseand in linkage equilibrium; (ii) its incorporationin breeding schemes that involvealtering the frequencies of favourable allelesthrough recurrent selection in large populations;and (iii) dealing with age x agetrait correlations, and late expressing phenotypes(Grattapaglia, 2000). In applyingMAS for forest trees, more will likelybe learned from experiences in livestock(Dekkers, 2004; Chapter 10) than fromannual crop plants, with the added advantagethat gains can be quickly realized bylarge-scale cloning of selected individuals.In this context, the categorization of threedifferent levels of marker-trait associationdescribed by Dekkers (2004) are relevant totrees: (a) direct markers, i.e. loci that codefor the functional mutation; (b) linkage disequilibrium(LD) markers: loci that are inpopulation-wide LD with the functionalmutation; (c) linkage equilibrium (LE)markers: loci that are in population-wideLE with the functional mutation in outbredpopulations (Figure 5). In forest trees,besides the recent encouraging discoveryof an LD marker for MFA in Eucalyptus(Thumma et al., 2005), only LE markertraitassociations have been described. LEmarkers have been readily detected ona genome-wide basis by analysing largefull-sib families with sparse marker mapsallowing the detection of most QTL ofmoderate to large effects. For the othertwo types of marker-trait association, it isonly now that the first association mappingexperiments are being started to uncover

270Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishFigure 6Rapid decay of intragenic linkage disequilibrium estimated as r 2 in the cinnamyl-alcoholdehydrogenase (cad) gene in two different Eucalyptus species10.90.80.70.60.50.40.30.20.1E. urophyllar 2 000 500 1000 1500 2000 2500 3000Distance in basepairs10.9E. grandis0.80.7r 20.60.50.40.30.20.10 500 1000 1500 2000 2500Distance in basepairsSource: Faria et al., 2006.LD markers, i.e. polymorphisms that aresufficiently close to the functional mutation(Neale and Savolainen, 2004). The challenge,however, is considerable, as LD inoutcrossing forest trees such as pines decaysvery rapidly, in general within 1 500 to 2 000bp (Neale and Savolainen, 2004), and similarbehaviour has been seen in the few Eucalyptusgenes analysed to date with significant LDextending for only a few hundred base pairs(Thumma et al., 2005; Kirst, Marques andSederoff, 2005; Faria et al., 2006) (Figure 6).Genome-wide association studies for LDmarker-trait discovery in tress will requirevery high SNP marker densities that are currentlystill impracticable (but see below), sothat the only alternative left is a candidategene approach. Finally, direct markers (i.e.polymorphisms that code for the functionalmutations) would be the most valuable and

Chapter 14 – Marker-assisted selection in Eucalyptus 271directly applicable in breeding. However,they are the most difficult to detect becausecausality is very difficult to prove unlessvery high penetrance Mendelian inheritancesare tackled.Prospects for using MAS in EucalyptusEucalyptus breeding programmes varybroadly according to several aspectsincluding the target species or hybrid,the possibility of deploying clones andthe amount of resources available to thebreeder. However, from the standpoint ofintegrating MAS, a reasonable premise isthat this will only be a justifiable optionwhen the breeding programme has alreadyreached a relatively high level of sophistication,fully exploiting all the accessiblebreeding and propagation tools. Advancedbreeding programmes that aim at elite cloneselection involve a significant amount oftime and effort being devoted to clonaltesting before effective recommendationscan be made concerning new clones foroperational plantations. Small sublinebreeding for hybrid performance combinedwith clonal propagation of selected individualsis being used increasingly for extractingnew elite clones (Potts, 2004). The recombinationstep of a breeding cycle involvesthe generation of several segregating progeniesfrom selected parents derived fromrecurrent selection programmes for generalcombining ability, or reciprocal recurrentselection programmes for hybrid combiningability. This latter strategy has been adoptedin tropical countries where the two reciprocalpopulations are actually two differentspecies such as E. grandis and E. urophylla.Controlled crosses that were oncean important obstacle for implementingpedigreed selection methods are now usedroutinely after the relatively recent advancesmade in controlled pollination methods forEucalyptus (Harbard, Griffin and Espejo,1999; de Assis, Warburton and Harwood,2005) (see Figure 2). Progeny trials, togetherwith expanded single family plots wherelarger numbers of full-sibs per family aredeployed, are used to allow very intensivewithin-family selection based on all theavailable information, both at the familyand individual level using BLUP-basedselection indices. This selection is generallycarried out at half-rotation age based ongrowth performance and on a preliminaryassessment of wood specific gravity usingindirect non-destructive techniques suchas pilodyn penetration and/or NIRS andRaman spectroscopy (Schimleck, Michelland Vinden, 1996). Vegetative propagulesare then rescued from selected trees eitherby coppicing, sequential grafting or in vitrotechniques, multiplied and then used forthe establishment of clonal tests.This breeding scheme generates largeamounts of linkage disequilibrium byhybridization and substantial amounts ofnon-additive genetic variation can be capturedby vegetative propagation. These arefavourable conditions for MAS in foresttrees (Strauss, Lande and Namkoong,1992). Favourable alleles at QTL segregatingwithin-families could be efficientlytagged with microsatellite markers in linkageequilibrium with the actual functional polymorphismsand used for marker-assistedwithin-family selection for superior individuals.QTL linked markers could be usedto carry out early selection thus reducing thetime necessary to carry out the first selectionespecially for traits related to woodproperties, and at the same time reducingthe number of trees to be selected, propagatedand advanced all the way to clonaltrials (Figure 7). Therefore, in the contextof molecular breeding, given their relativelyshort rotations and the possibility of

272Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishFigure 7Proposed scheme for early marker-assisted selection of plus trees to be used as clonesMAS STAGEXF 1F 1Mating between plus hybrid trees tomaximize segregation for several traitsin the outbred F 2 Deployment of a large number (> 1000) of F 2progeny individuals toQTL MAPPING STAGEQTL mapping for wood properties traitssuch as lignin, fibreand wood densityembra_622*embra_28*embra_94*embra_627*EG_62*embra_372*embra_32*embra_31*embra_233*embra_50*embra_8*embra_328*embra_324*embra_173*embra_51*embra_25*embra_106*embra_248*embra_258*embra_105*embra_646*embra_175*ES_157*embra_277*embra_362*embra_189embra_115EG_94embra_227embra_122*EG_98*embra_125*embra_114embra_361F 2embra_189*EG_94*embra_227*embra_350*embra_321*EG-98*embra_125*embra_629*EN_14*embra_109embra_197embra_57EN_1embra_53EG_67embra_103embra_624ES_140embra_18embra_131embra_210.1*embra_210embra_204*embra_217*embra_365*embra_701*embra_19embra_393embra_66*embra_78*embra_137*ES_54*embra_645*QTL mapping informationto be used in MASF 2maximize probabilityof generatinga recombinant individual with asuperior multipleQTL allele contentGenotyping > 1000 seedlings witha small (~ 6 to 10) set of flankingmarkers for a targetednumber of QTL for wood qualitytraitsEarly MAS. Selection intensity is increased by MAS for late expressing traitsbut number of trees commonly deployed in progeny (~ 100) test iskept the same, thus allowing large variation to select for othertraits such as volume growth, form and branching habitQTL mapping is carried out and flanking markers in linkage disequilibrium with favourable alleles at major effect QTLalleles are identified. These markers are then used for early within-family selection in expanded progeny sizes at veryearly age to select for late expression traits such as wood density and lignin content.deploying clones to capture non-additivegenetic variation, it is reasonable to state thateucalypt is the forest tree crop for whichMAS has the best prospects of application.Quantitative theory as well as commonsense suggest that MAS in forest treesshould help, particularly in situations wheretrait heritability is low and selection occursat the level of the individual tree. However,implementing MAS for such traits is a challengingtask as extremely precise QTLmapping information is required and thiscan only be derived from experiments withlarge progeny sizes (in the order of severalhundred individuals), clonal replicatesfor increased precision, representative andmultiple genetic backgrounds and environments.To date, mapped QTL in foresttrees still do not fall into this descriptionalthough improved experiments are underway (Grattapaglia, 2004). Most experimentshave mapped QTL for traits thatdisplay intermediate to high heritabilityand probably did not tag the top alleles thatexist in the breeding populations as only avery limited sample of crosses were conducted.Furthermore, given the relativelysmall progeny sizes used for QTL detection(around 100 to 200 individuals), theestimated magnitude of the effects werelargely overestimated following the wellknown “Beavis effect” (Beavis, 1998).

Chapter 14 – Marker-assisted selection in Eucalyptus 273It is frequently stated that MAS for treeswould be most useful for volume growth asthis is a universal trait of interest and typicallyof low heritability at the individualtree level. However, in tropical conditions,it is most likely this will not be the targettrait of first choice for MAS. Broad senseheritability at the clone mean level, whichis the typical selection unit, is frequentlyabove 0.8, allowing an almost perfectranking and selection of clones even atvery early ages (less than two years) undertropical conditions (Rezende, Bertolucciand Ramalho, 1994). Molecular markersfor volume growth in these conditionswill hardly make a significant contributionto increasing gain per unit time. Thecost of scoring molecular markers dictatesthat the most likely application of MAS inEucalyptus will be for traits that providesignificant added value to the final productsuch as branching habit (for solid wood)and wood chemical traits, or allow clonaldeployment such as adventitious rooting orsomatic embryogenesis response. Within allpossible quality traits, the option would befor those that display medium to high heritabilitiesbut where phenotype assessment isdifficult, expensive or requires waiting untilthe tree reaches maturity. Wood qualitytraits typically require the tree to startaccumulating late wood and involve relativelylengthy procedures for phenotypicevaluation in the laboratory. These kindsof traits could be interesting targets forMAS in Eucalyptus, given that the costsof genotyping are sufficiently competitiveand precision is high when comparedwith direct phenotype measurements. It isimportant to point out, however, that withthe recent developments of fast samplingand indirect wood chemistry measurementsbased on NIRS (Schimleck, Michell andVinden, 1996), the potential gain will onlybe realized on the basis of the time savingsprovided by very early selection. Selectedindividuals could be recombined more rapidlyfollowing flower induction (Griffin etal., 1993) to produce the next generation,potentially increasing the genetic gain perunit time.MAS for multiple traits will face manyof the same difficulties faced by conventionalmultiple trait selection. Very largeprogeny sizes would have to be deployed tohave a reasonable probability of recoveringgenotypes with a combination of favourablealleles at many QTL for many traits.When using MAS, priorities will have to beestablished not only for traits but also forspecific QTL. This will require a very goodunderstanding of the relative magnitude ofeach QTL, potential QTL x backgroundinteractions and pleiotropic effects of QTL.Linkage mapping, however, will allow thebreeder to understand the basis of negativecorrelation between traits and possibly tobreak unwanted linkages by selecting specificrecombinant genotypes.Once the challenging issues relatedto the discovery of robust marker-traitassociations, either within family (LEmarkers) or at the population level (LD ordirect markers), are dealt with, a realisticstrategy for the implementation of MAS inEucalyptus might be to tackle only a fewmajor QTL for a quality trait of significantadded value. Theoretically, when the totalproportion of the additive genetic varianceexplained by the marker loci exceeds theheritability of the character, selection on thebasis of the markers alone is more efficientthan selection on the individual phenotype.Such a goal might be achieved for a specifictrait with just a few QTL alleles responsiblefor large effects. On the other hand, if nomajor gene is detected in an experiment ofreasonable size, it might be wiser to dismiss

274Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishMAS for that particular trait. Estimates ofheritability for a trait might be useful togive an initial clue. Intuitively, the probabilityof major genes existing for traitsof low heritability is lower than for traitsof high heritability. However, this shouldnot be taken as a measure to discard possibleQTL mapping experiments as, evenwith low heritabilities, traits might still displaymajor QTL, and MAS would have thegreatest impact particularly in such cases.Conclusions and perspectivesThe successful application of molecularbreeding in Eucalyptus will dependheavily on first demonstrating and validatingthe clear-cut association between aDNA polymorphism and a quantitativelyinherited phenotypic trait. In highly heterogeneouseucalypts, while conventionalQTL mapping can reveal useful markersto be exploited in within-family selectionpractices, only a more direct LD mappingapproach can uncover populationwide applicable marker-trait associations.Such studies based on candidate genes havebegun and the first candidate gene associationfor MFA was detected. However, thisassociation explains only a small proportion(3.4 percent) of the variation to bereally exciting news to breeders (Thummaet al., 2005). One of the key issues whenembarking on an association mapping experimentis the selection of candidate genes.Maximizing the probability of choosing theproper genes requires levels of knowledgeof biochemistry, physiology and developmentthat are generally not yet availableeven for well defined phenotypes and/orknown metabolic pathways.Following the path taken in humangenetics, co-localization of candidate genesand QTL for relevant traits on linkagemaps together with integrative expression-QTL mapping (Kirst et al., 2004) could be apowerful way forward, although choosingthe correct candidate depends heavily onthe precision of the QTL localization. Atthe moment, there are two possibilities forcircumventing the dilemma of choosingcandidate genes correctly. The first is microarray-basedgenotyping with ultra-densearrays of short (25 nt) oligonucleotides(Borevitz et al., 2003; Hazen and Kay, 2003;West et al., 2006) that would allow sufficientthroughput for association geneticanalysis of thousands of genes at a time.Such an array format could later turn out tobe a useful instrument for MAS once validatedmarker-trait associations have beenestablished. The second would be to haveaccess to a whole genome sequence so thatcandidate genes in a fine mapping intervaldelimited by markers flanking a QTL withcentimorgan resolution could be mined,reannotated and then analysed in associationmapping experiments.A draft genome of E. camaldulensis is currentlybeing sequenced at the Kazusa DNAResearch Institute in Japan (T. Hibino, personalcommunication), and the possibilityexists that a fully public 4X draft of theE. grandis genome will be sequenced bythe Joint Genome Institute of the UnitedStates Department of Energy within thenext years (J. Tuskan, personal communication)following a proposal recentlysubmitted by an international group ofEucalyptus geneticists (www.ieugc.up.ac.za/DOE%20proposal%20-%20final%20-%2026%20July%202006.pdf) who recentlyformed the International EucalyptusGenome Network (EUCAGEN) (www.ieugc.up.ac.za; Myburg, 2004). Such publiccollaborative efforts should contributegreatly to the advancement of Eucalyptusgenetics, genomics and molecular breedingby bringing together existing private data-

Chapter 14 – Marker-assisted selection in Eucalyptus 275bases and genomic resources and therebyexpanding the value of such genomesequences. As such genome projectsadvance and new and more powerful analyticaltools become accessible, the truechallenge to dissecting the complexity ofeconomically-important traits in Eucalyptusand implementing MAS will depend to alarge extent on our ability to phenotypetrees accurately, analyse the overwhelmingamount of genomic data available and translatethis into truly useful molecular tools forbreeding. MAS should be considered on acase-by-case basis and without overstatingthe gains to be expected until hard experimentaldata are accumulated on the actualgains made from its application withinindustrial forests beyond those which canbe attained by comparable investment inconventional phenotypic selection.AcknowledgmentsI am very thankful to the Ministry ofScience and Technology of Brazil and theparticipating forestry companies in theGenolyptus project for the continuedcompetitive grant support for research andthe Brazilian National Research Council(CNPq) for its support through a researchfellowship. I am also very grateful to allmy undergraduate and graduate studentsand research collaborators from severaluniversities in Brazil and around the worldand to breeders from private companiesfor discussions and scientific inputs. I wishto make a special acknowledgment to mygood friend Teotônio Francisco de Assis,an icon of Eucalyptus breeding worldwide,for having introduced me to this wonderfulworld of Eucalyptus genetics and breeding.ReferencesBaril, C.P., Verhaegen, D., Vigneron, P., Bouvet, J.M. & Kremer, A. 1997. Structure of the specificcombining ability between two species of Eucalyptus. I. RAPD data. Theor. Appl. Genet. 94:796–803.Beavis, W.D. 1998. QTL analyses: power, precision, and accuracy. In H.A. Paterson, ed. Moleculardissection of complex traits, pp. 145–162. Boca Raton, FL, USA, CRC.Binkley, D. & Stape, J.L. 2004 Sustainable management of eucalypt plantations in a changingworld. In M. Tomé, ed. Proc. IUFRO Conf. Eucalyptus in a Changing World. CD-ROM. Aveiro,Portugal, Instituto Investigação de Floresta e Papel (RAIZ). Aveiro, Portugal,Borevitz, J.O., Liang, D., Plouffe, D., Chang, H.S., Zhu, T., Weigel, D., Berry, C.C., Winzeler, E.& Chory, J. 2003. Large-scale identification of single-feature polymorphisms in complex genomes.Genome Res. 13: 513–523.Borralho, N.M.G. 2001. The purpose of breeding is breeding for a purpose. In S. Barros, ed. Proc.IUFRO Internat. Symp. Developing the Eucalypt of the Future. CD-ROM. Valdivia, Chile,INFOR.Borralho, N.M.G., Cotterill, P.P. & Kanowski, P.J. 1993. Breeding objectives for pulp production ofEucalyptus globulus under different industrial cost structures. Can. J. For. Res. 23: 648–656.Brandão, L.G., Campinhos, E. & Ikemori, Y.K. 1984. Brazil’s new forest soars to success. Pulp Pap.Int. 26: 38–40.Brondani, R.P.V., Brondani, C. & Grattapaglia, D. 2002. Towards a genus-wide reference linkagemap for Eucalyptus based exclusively on highly informative microsatellite markers. Mol. Genet.Genomics 267: 338–347.

276Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishBrondani, R.P.V., Brondani, C., Tarchini, R. & Grattapaglia, D. 1998. Development, characterizationand mapping of microsatellite markers in Eucalyptus grandis and E. urophylla. Theor. Appl.Genet. 97: 816–827.Brondani, R.P.V., Williams, E.R., Brondani, C. & Grattapaglia, D. 2006. A microsatellite-basedconsensus linkage map for species of Eucalyptus and a novel set of 230 microsatellite markers forthe genus. BMC Plant Biology 6: 20 (available at www.biomedcentral.com/1471-2229/6/20).Brune, A. & Zobel, B.J. 1981. Genetic base populations, gene pools and breeding populations forEucalyptus in Brazil. Silvae Genetica 30: 146–149.Bundock, P.C., Hayden, M. & Vaillancourt, R.E. 2000. Linkage maps of Eucalyptus globulus usingRAPD and microsatellite markers. Silvae Genet. 49: 223–232.Burczyk, J., Adams, W.T., Moran, G.F. & Griffin, A.R. 2002. Complex patterns of mating revealedin a Eucalyptus regnans seed orchard using allozyme markers and the neighbourhood model. Mol.Ecol. 11: 2379–91.Byrne, M., Murrell, J.C., Allen, B. & Moran, G.F. 1995. An integrated genetic linkage map forEucalyptus using RFLP, RAPD and isozyme markers. Theor. Appl. Genet. 91: 869–875.Byrne, M., Marquezgarcia, M.I., Uren, T., Smith, D.S. & Moran, G.F. 1996 Conservation andgenetic diversity of microsatellite loci in the genus Eucalyptus. Aust. J. Bot. 44: 331–341.Byrne, M., Murrell, J.C., Owen, J.V., Kriedemann, P., Williams, E.R. & Moran, G.F. 1997a.Identification and mode of action of quantitative trait loci affecting seedling height and leaf area inEucalyptus nitens. Theor. Appl. Genet. 94: 674–681.Byrne, M., Murrell, J.C., Owen, J.V., Williams, E.R. & Moran, G.F. 1997b. Mapping of quantitativetrait loci influencing frost tolerance in Eucalyptus nitens. Theor. Appl. Genet. 95: 975–979.Campinhos, E. 1980. More wood of better quality through intensive silviculture with rapid growthimproved Brazilian Eucalyptus. Tappi 63: 145–147.Campinhos, E. & Ikemori, Y.K. 1977. Tree improvement program of Eucalyptus spp.: preliminaryresults. In 3 rd Wrld. Consult. Forest Tree Breeding, pp. 717–738. Canberra, CSIRO.Chaix, G., Gerber, S., Razafimaharo, V., Vigneron, P., Verhaegen, D. & Hamon, S. 2003. Geneflow estimation with microsatellites in a Malagasy seed orchard of Eucalyptus grandis. Theor. Appl.Genet. 107: 705–712.Costa e Silva, C. & Grattapaglia, D. 1997. RAPD relatedness of elite clones, applications in breedingand operational clonal forestry, Proc. Internat. IUFRO Conf. Eucalyptus Genet. & Silvicult., pp.161–166. Salvador, Brazil.de Assis, T.F. 2000. Production and use of Eucalyptus hybrids for industrial purposes. In D.G.,Nikles, ed. Proc. QFRI/CRC Workshop on Hybrid Breeding & Genetics of Forest Trees, pp. 63–74.Brisbane, Australia, Department of Primary Industries.de Assis, T.F. 2001. The evolution of technology for cloning Eucalyptus in a large scale. Proc. IUFROInternat. Symp. Developing the Eucalypt of the Future. CD-ROM. Valdivia, Chile, INFOR.de Assis, T.F., Rezende, G.D.S.P. & Aguiar, A.M. 2005. Current status of breeding and deploymentfor clonal forestry with tropical eucalypt hybrids in Brazil. Proc. XXII IUFRO World Congress.Forests in the balance: linking tradition and technology. International Forestry Review 7(5): 61.Brisbane, Australia.de Assis, T.F., Warburton, P. & Harwood, C. 2005. Artificially induced protogyny: an advance in thecontrolled pollination of Eucalyptus. Austr. Forestry 68: 27–33.

Chapter 14 – Marker-assisted selection in Eucalyptus 277Dekkers, J.C.M. 2004. Commercial application of marker-and gene-assisted selection in livestock:strategies and lessons. J. Anim Sci. 82 (E supp.): 313–328.Dekkers, J.C.M. & Hospital, F. 2002. The use of molecular genetics in the improvement of agriculturalpopulations. Nature Revs. Genet. 3: 22–32.Doughty, R.W. 2000. The Eucalyptus: a natural and commercial history of the gum tree. Baltimore,MD, USA, and London, The Johns Hopkins University Press.Eldridge, K., Davidson, J., Harwood, C. & van Wyk, G. 1993. Eucalypt domestication and breeding.Oxford, UK, Clarendon Press.FAO. 2001. Global forest resources assessment 2000: main report. FAO Forestry Paper No. 140. Rome(also available at www.fao.org/forestry/fo/fra/main/index.jsp).Faria, D.A., Alves, T.P.M., Pereira, R.W. & Grattapaglia, D. 2006. Frequência de SNPs e extensão dodesequilíbrio de ligação ao longo dos genes CCR e CAD em E. grandis, E. globulus e E. urophylla.Proc. 52 nd Brazilian Genet. Congr., Abstract GP251. Foz do Iguaçu.Franklin, E.C. 1986. Estimation of genetic parameters through four generations of selection inEucalyptus grandis. Proc. IUFRO Joint Mtg. Working Parties on Breeding Theory, Progeny Testingand Seed Orchards, pp. 12–17. Williamsburg, VA, USA.Gaiotto, F.A. & Grattapaglia, D. 1997. Estimation of genetic variability in a breeding populationof Eucalyptus urophylla using AFLP (amplified fragment length polymorphism) markers. Proc.Internat. IUFRO Conf. Eucalyptus Genet. & Silvicult., pp. 46–52. Salvador, Brazil, IUFROGaiotto, F.A., Bramucci, M. & Grattapaglia, D. 1997. Estimation of outcrossing rate in a breedingpopulation of Eucalyptus urophylla s.t. Blake with dominant RAPD and AFLP markers. Theor.Appl. Genet. 95: 842–849.Gion, J-M., Rech, P., Grima-Pettenati, J., Verhaegen, D. & Plomion, C. 2000. Mapping candidategenes in Eucalyptus with emphasis on lignification genes. Molec. Breeding 6: 441–449.Glaubitz, J.C., Emebiri, L.C. & Moran, G.F. 2001. Dinucleotide microsatellites from Eucalyptussieberi: inheritance, diversity, and improved scoring of single-base differences. Genome 44:1014–1045.Grattapaglia, D. 2000. Molecular breeding of Eucalyptus - State of the art, operational applicationsand technical challenges, In S.M. Jain & S.C. Minocha, eds. Molecular biology of woody plants, pp.451–474. Dordrecht, Netherlands, Kluwer Academic Publishers.Grattapaglia, D. 2004. Integrating genomics into Eucalyptus breeding. Gen. Mol. Res. 3: 369–379.Grattapaglia, D. & Sederoff, R. 1994. Genetic linkage maps of Eucalyptus grandis and Eucalyptusurophylla using a pseudo-testcross: mapping strategy and RAPD markers. Genetics 137:1121–1137.Grattapaglia, D., Bertolucci, F.L.G. & Sederoff, R. 1995. Genetic mapping of QTLs controllingvegetative propagation in Eucalyptus grandis and E. urophylla using a pseudo-testcross mappingstrategy and RAPD markers. Theor. Appl. Genet. 90: 933–947.Grattapaglia, D., Ribeiro, V.J. & Rezende, G.D. 2004. Retrospective selection of elite parent treesusing paternity testing with microsatellite markers: an alternative short term breeding tactic forEucalyptus. Theor. Appl. Genet. 109: 192–199.Grattapaglia, D., Bertolucci, F.L.G., Penchel, R. & Sederoff, R. 1996. Genetic mapping of quantitativetrait loci controlling growth and wood quality traits in Eucalyptus grandis using a maternalhalf-sib family and RAPD markers. Genetics 144: 1205–1214.

278Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishGrattapaglia, D., Pimenta, D., Campinhos, E.N., Rezende, G.D.S. & de Assis, T.F. 2003. Marcadoresmoleculares na proteção varietal de Eucalyptus. Proc. 8 th Brazilian Forestry Congr., pp. 1–13. SãoPaulo, Brazilian Society of Silviculture.Griffin, A.R., Burgess, I.P. & Wolf, L. 1988. Patterns of natural and manipulated hybridisation in thegenus Eucalyptus L’Hérit: a review. Austr. J. Bot. 36(1): 41–66.Griffin, A.R., Whiteman, P., Rudge, T., Burgess, I.P. & Moncur, M. 1993. Effect of paclobutrazolon flower-bud production and vegetative growth in two species of Eucalyptus. Can. J. For. Res.23: 640–647.Harbard, J.L., Griffin, A.R. & Espejo, J. 1999. Mass controlled pollination of Eucalyptus globulus: apractical reality. Can. J. For. Res. 29: 1457–1463.Hazen, S.P. & Kay, S.A. 2003. Gene arrays are not just for measuring gene expression. Trends in PlantSci. 8: 413–416.Ikemori, Y.K., Penchel, R.M. & Bertolucci, F.L.G. 1994. Integrating biotechnology into Eucalyptusbreeding. In Proc. Internat. Symp. Wood Biotech., pp. 79–84. Tokyo, Japan Wood Research Societyand Nippon Paper Industries.Junghans, D.T., Alfenas, A.C., Brommonschenkel, S.H., Oda, S., Mello, E.J. & Grattapaglia, D.2003. Resistance to rust (Puccinia psidii Winter) in Eucalyptus: mode of inheritance and mapping ofa major gene with RAPD markers. Theor. Appl. Genet. 108: 175–80.Kanowski, P.J. & Borralho, N.M.G. 2004. Economics of tree improvement. In J.A. Youngquist, ed.Encyclopedia of forest science, pp. 1561–1568. Oxford, UK, Elsevier Science.Keil, M. & Griffin, A.R. 1994. Use of random amplified polymorphic DNA (RAPD) markers in thediscrimination and verification of genotypes in Eucalyptus. Theor. Appl. Genet. 89: 442–450.Kellison, R.C. 2001. Present and future uses of eucalypts wood in the world. In S. Barros, ed. Proc.IUFRO Internat. Symp. Developing the Eucalypt of the Future. CD-ROM. Valdivia, Chile,INFOR.Kirst, M., Marques, C.M. & Sederoff, R. 2005. Nucleotide diversity and linkage disequilibrium inthree Eucalyptus globulus genes. IUFRO Tree Biotech. Conf., Section 5, poster 28. 6–11 November2005. Pretoria, South Africa.Kirst, M., Brondani, R.P.V., Brondani, C. & Grattapaglia, D. 1997. Screening of designed primerpairs for recovery of microsatellite markers and their transferability among species of Eucalyptus.Proc. IUFRO Conf. Eucalyptus Genet. & Silvicult., pp. 167–171. Salvador, Brazil, Embrapa.Kirst, M., Myburg, A.A., De Leon, J.P., Kirst, M.E., Scott, J. & Sederoff, R. 2004. Coordinatedgenetic regulation of growth and lignin revealed by quantitative trait locus analysis of cDNAmicroarray data in an interspecific backcross of Eucalyptus. Plant Physiol. 135: 2368–2378.Kirst, M., Cordeiro, C.M., Rezende, G.D. & Grattapaglia, D. 2005a. Power of microsatellitemarkers for fingerprinting and parentage analysis in Eucalyptus grandis breeding populations. J.Hered. 96: 161–166.Kirst, M., Basten, C.J., Myburg, A., Zeng, Z-B. & Sederoff, R. 2005b. Genetic architecture of transcriptlevel variation in differentiating xylem of Eucalyptus hybrids. Genetics 169: 2295–2303.Ladiges, P.Y., Udovicic, F. & Nelson, G. 2003. Australian biogeographical connections and the phylogenyof large genera in the plant family Myrtaceae. J. Biogeogr. 30: 989–998.Lander, E.S. & Botstein, D. 1989. Mapping mendelian factors underlying quantitative traits usingRFLP linkage maps. Genetics 121: 185–199.

Chapter 14 – Marker-assisted selection in Eucalyptus 279Marcucci-Poltri, S.N., Zelener, N., Rodriguez Traverso, J., Gelid, P. & Hopp, H. 2003. Selection ofa seed orchard of Eucalyptus dunnii based on genetic diversity criteria calculated using molecularmarkers. Tree Physiol. 23: 625–32.Marques, C.M., Araujo, J.A., Ferreira, J.G., Whetten, R., O’Malley, D.M., Liu, B.H. & Sederoff,R. 1998. AFLP genetic maps of Eucalyptus globulus and E. tereticornis. Theor. Appl. Genet. 96:727–737.Marques, C.M., Vasquez-Kool, J., Carocha, V.J., Ferreira, J.G., O’Malley, D.M., Liu, B.H. &Sederoff, R. 1999. Genetic dissection of vegetative propagation traits in Eucalyptus tereticornis andE. globulus. Theor Appl Genet 99: 936–946.Marques, C.M., Brondani, R.P.V., Grattapaglia, D. & Sederoff, R. 2002. Conservation and syntenyof SSR loci and QTLs for vegetative propagation in four Eucalyptus species. Theor. Appl. Genet.105: 474–478.Martin, B. & Quillet, J. 1974. The propogation by cuttings of forest trees in the Congo. Bois et Forêtsdes Tropiques 155: 15–33.Mauricio, R. 2001. Mapping quantitative trait loci in plants: uses and caveats for evolutionary biology.Nature Revs. Genet. 2: 370–381.Missiaggia, A.A., Piacezzi, A.L. & Grattapaglia, D. 2005. Genetic mapping of Eef1, a major effectQTL for early flowering in Eucalyptus grandis. Tree Genet. & Genomes 1: 79–84.Moran, G.F. & Bell, J.C. 1983. Eucalyptus. In S.D. Tanksley & T.J. Orton, eds. Isozymes in plantgenetics and breeding, pp. 423–441. Amsterdam, Elsevier.Moran, G., Bell, J.C. & Griffin, A.R. 1989. Reduction in levels of inbreeding in a seed orchard ofEucalyptus regnans F. Muell, compared with natural populations. Silvae Genet. 38: 32–36.Moran, G.F., Thamarus, K.A., Raymond, C.A., Qiu, D., Uren, T. & Southerton, S.G. 2002.Genomics of Eucalyptus wood traits. Ann. For. Sci. 59: 645–650.Myburg, A.A. 2001. Genetic architecture of hybrid fitness and wood quality traits in a wide interspecificcross of Eucalyptus tree species. North Carolina State University, Raleigh, NC (PhD thesis;available at www.lib.ncsu.edu/theses/available/etd-20010723-175234).Myburg, A.A. 2004. The International Eucalyptus Genome Consortium (IEuGC): Opportunities andResources for Collaborative Genome Research in Eucalyptus. In B. Li, & S. McKeand, eds. Proc.IUFRO Conf. Forest Genet. & Tree Breeding in the Age of Genomics: Progress and Future, pp.154–155. Charleston, SC, USA.Myburg, A.A., Griffin, A.R., Sederoff, R.R. & Whetten, R.W. 2003. Comparative genetic linkagemaps of Eucalyptus grandis, Eucalyptus globulus and their F1 hybrid based on a double pseudobackcrossmapping approach. Theor. Appl. Genet. 107: 1028–42.Myburg, A.A., Potts, B., Marques, C.M., Kirst, M., Gion, J.M., Grattapaglia, D. & Grima-Pettenati, J. 2006. Eucalyptus. In C. Kole, ed. Genome mapping & molecular breeding in plantsvolume 7: forest trees. Heidelberg, Berlin, New York and Tokyo, Springer (in press).Neale, D.B. & Savolainen, O. 2004. Association genetics of complex traits in conifers. Trends PlantSci. 9: 325–330.Nesbitt, K.A., Potts, B.M., Vaillancourt, R.E., West, A.K. & Reid, J.B. 1995. Partitioning and distributionof RAPD variation in a forest tree species, Eucalyptus globulus (Myrtaceae). Heredity 74:628–637.Nesbitt, K.A., Potts, B.M., Vaillancourt, R.E. & Reid, J.B. 1997. Fingerprinting and pedigree analysisin Eualyptus globulus using RAPDs. Silvae Genet. 46: 6–11.

280Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishOttewell, K.M., Donnellan, S.C., Moran, G.F. & Paton, D.C. 2005. Multiplexed microsatellitemarkers for the genetic analysis of Eucalyptus leucoxylon, Myrtaceae and their utility for ecologicaland breeding studies in other Eucalyptus species. J. Hered. 96: 445–451.Paterson, A.H., Lander, E.S., Hewitt, J.D., Peterson, S., Lincoln, S.E. & Tanksley, S.D. 1988.Resolution of quantitative traits into Mendelian factors by using a complete linkage map of restrictionfragment length polymorphisms. Nature 335: 721–726.Poke, F.S., Vaillancourt, R.E., Potts, B.M. & Reid, J.B. 2005. Genomic research in Eucalyptus.Genetica 125: 79–101.Potts, B.M. 2004. Genetic improvement of eucalypts, pp 1480–1490. In J. Burley, J. Evans, & J.A.Youngquist, eds. Encyclopedia of forest science. Oxford, UK, Elsevier Science.Potts, B.M. & Dungey, H.S. 2004. Hybridisation of Eucalyptus: key issues for breeders and geneticists.New Forests 27: 115–138.Potts, B.M., Vaillancourt, R.E., Jordan, G.J., Dutkowski, G.W., Costa e Silva, J., McKinnon, G.E.,Steane, D.A., Volker, P.W., Lopez, G.A., Apiolaza, L.A., Li, Y., Marques, C. & Borralho, N.M.G.2004. Exploration of the Eucalyptus globulus gene pool. In M. Tomé ed. Eucalyptus in a changingworld, pp. 46–61. Aveiro, Portugal, Instituto Investigação de Floresta e Papel (RAIZ).Pryor, L.D. 1976. The biology of eucalypts. London, Edward Arnold.Pryor, L.D. & Johnson, L.A.S. 1971. A classification of the eucalypts. Canberra, Australian NationalUniversity Press.Rezende, G.D.S.P., Bertolucci, F.L.G. & Ramalho, M.A.P. 1994. Eficiência da seleção precoce narecomendação de clones de eucalipto avaliados no norte do Espírito Santo e sul da Bahia. Cerne1: 45–50.Ribeiro, V.J., Bertolucci, F.L.G. & Grattapaglia, D. 1997. RAPD marker - guided matings in a reciprocalrecurrent selection program of Eucalyptus. Proc. Internat. IUFRO Conf. Eucalyptus Genet.& Silvicult., pp. 156–160. Salvador, Brazil.Sanchez, J.R. 2002. Panel de tecnologia, pasta y papel. In Graficas Diumaró, ed. Socioeconomia, patologia,tecnologia y sostenibilidade del eucalipto, pp. 187–192. Brazil, Universidad de Vigo.Schimleck, L.R., Michell, A.J. & Vinden, P. 1996. Eucalypt wood classification by NIR spectroscopyand principal components analysis. Appita J. 49: 319–324.Shepherd, M., Chaparro, J.X. & Teasdale, R. 1999. Genetic mapping of monoterpene compositionin an interspecific eucalypt hybrid. Theor. Appl. Genet. 99: 1207–1215.Shepherd, M. & Jones, M. 2005. Molecular markers in tree improvement: Characterisation and usein Eucalyptus,. In H. Lörz & G. Wenzel, eds. Molecular marker systems in plant breeding and cropimprovement, pp 399–409. Heidelberg, Germany, Springer-Verlag.Steane, D.A., Jones, R.C. & Vaillancourt, R.E. 2005. A set of chloroplast microsatellite primers forEucalyptus, Myrtaceae. Mol. Ecol. Notes 5: 538–541.Steane, D.A., Vaillancourt, R.E., Russell, J., Powell, W., Marshall, D. & Potts, B.M. 2001.Development and characterization microsatellite loci in Eucalyptus globulus (Myrtacea). SilvaeGenet. 50: 89.Strauss, S.H., Lande, R. & Namkoong, G. 1992. Obstacles to molecular-marker-aided selection inforest trees. Can. J. For. Res. 22: 1050–1061.Stuber, C.W., Moll, R.H., Goodman, M.M., Schaffer, H.E. & Weir, B.S. 1980. Allozyme frequencychanges associated with selection for increased grain yield in maize. Genetics 95: 225–236.Tanksley, S.D. 1993. Mapping polygenes. Ann. Rev. Genet. 27: 205–233.

Chapter 14 – Marker-assisted selection in Eucalyptus 281Thamarus, K.A., Groom, K., Murrell, J., Byrne, M. & Moran, G. 2002. A genetic linkage map forEucalyptus globulus with candidate loci for wood, fibre and floral traits. Theor. Appl. Genet. 104:379–387.Thamarus, K.A., Groom, K., Bradley, A., Raymond, C.A., Schimleck, L.R., Williams, E.R. &Moran, G.F. 2004. Identification of quantitative trait loci for wood and fibre properties in two fullsibpedigrees of Eucalyptus globulus. Theor. Appl. Genet. 109: 856–864.Thumma, R.F., Nolan, M.F., Evans, R. & Moran, G.F. 2005. Polymorphisms in cinnamoyl CoAreductase (CCR) are associated with variation in microfibril angle in Eucalyptus spp. Genetics 171:1257–1265.Turnbull, J.W. 1999. Eucalypt plantations. New Forests 17: 37–52.Vaillancourt, R.E., Potts, B.M., Watson, M., Volker, P.W., Hodge, G.R., Reid, J.B. & West, A.K.1995a. Prediction of heterosis using RAPD generated distances between Eucalyptus globulus trees.In Proc. CRCTHF-IUFRO Conf. Eucalypt Plantations: Improving Fibre Yield and Quality, pp.455–456. 19–24 February 1995, Hobart, Australia(available at www.forestry.crc.org.au/iufro95.htm).Vaillancourt, R.E., Potts, B.M., Manson, A., Eldridge, T. & Reid, J.B. 1995b Using RAPD’sto detect QTLs in an interspecific F 2 hybrid of Eucalyptus. In Proc. CRCTHF-IUFRO Conf.Eucalypt Plantations: Improving Fibre Yield and Quality, pp. 430–433. 19–24 February 1995,Hobart, Australia (available at www.forestry.crc.org.au/iufro95.htm).Verhaegen, D. & Plomion, C. 1996. Genetic mapping in Eucalyptus urophylla and E grandis. RAPDmarkers. Genome 39: 1051–1061.Verhaegen, D., Plomion, C., Gion, J.M., Poitel, M., Costa, P. & Kremer, A. 1997. Quantitative traitdissection analysis in Eucalyptus using RAPD markers. 1. Detection of QTL in interspecific hybridprogeny, stability of QTL expression across different ages. Theor. Appl.Genet. 95: 597–608.Waugh, G. 2004. Growing Eucalyptus globulus for high-quality sawn products. In M. Tomé, ed.Eucalyptus in a changing world, pp. 79–84. Aveiro, Portugal, Instituto Investigação de Floresta ePapel (RAIZ).West, M.A., van Leeuwen, H., Kozik, A., Kliebenstein, D.J., Doerge, R.W., St Clair, D.A. &Michelmore, R.W. 2006. High-density haplotyping with microarray-based expression and singlefeature polymorphism markers in Arabidopsis. Genome Res. 16: 787–795.Young, N.D. 1999. A cautiously optimistic view for marker-assisted selection. Mol. Breed. 5:505–510.Zelener, N., Poltri, S.N., Bartoloni, N., Lopez, C.R. & Hopp, H.E. 2005. Selection strategy for aseedling seed orchard design based on trait selection index and genomic analysis by molecularmarkers: a case study for Eucalyptus dunnii. Tree Physiol. 25: 1457–1467.

Chapter 15Marker-assisted selectionin forestry speciesPenny Butcher and Simon Southerton

284Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishSummaryThe primary goal of tree breeding is to increase the quantity and quality of wood productsfrom plantations. Major gains have been achieved using recurrent selection in geneticallydiverse breeding populations to capture additive variation. However, the long generationtimes of trees, together with poor juvenile-mature trait correlations, have promotedinterest in marker-assisted selection (MAS) to accelerate breeding through early selection.MAS relies on identifying DNA markers, which explain a high proportion of variationin phenotypic traits. Genetic linkage maps have been developed for most commercialtree species and these can be used to locate chromosomal regions where DNA markersco-segregate with quantitative traits (quantitative trait loci, QTL). MAS based on QTLis most likely to be used for within-family selection in a limited number of elite familiesthat can be clonally propagated. Limitations of the approach include the low resolutionof marker-trait associations, the small proportion of phenotypic variation explained byQTL and the low success rate in validating QTL in different genetic backgrounds andenvironments. This has led to a change in research focus towards association mapping toidentify variation in the DNA sequence of genes directly controlling phenotypic variation(gene-assisted selection, GAS). The main advantages of GAS are the high resolution ofmarker-trait associations and the ability to transfer markers across families and evenspecies. Association studies are being used to examine the adaptive significance of variationin genes controlling wood formation and quality, pathogen resistance, cold tolerance anddrought tolerance. Single nucleotide polymorphisms (SNPs) in these gene sequencesthat are significantly associated with trait variation can then be used for early selection.Markers for SNPs can be transferred among individuals regardless of pedigree or familyrelationship, increasing opportunities for their application in tree breeding programmes indeveloping as well as developed countries. Significant reductions in genotyping costs andimproved efficiencies in gene discovery will further enhance these opportunities.

Chapter 15 – Marker-assisted selection in forestry species 285IntroductionTree breeding offers a unique set ofchallenges associated with long generationtimes, outcrossing breeding systems and arelatively short history of genetic improvement.Breeding populations are often onlyone or two generations from the wild state.This has the advantage over crop breedingof providing vast stores of genetic variationthat can be utilized in tree improvement.Tree breeding programmes have generallyrelied on testing and selecting large numbersof genotypes derived from multiple geneticbackgrounds, the maintenance of highgenetic diversity in production forests,and sexual propagation and capture ofadditive genetic variation through recurrentselection (Strauss, Lande and Namkoong,1992). Inbreeding depression and longgeneration intervals have precluded the useof inbred lines, although research into theirdevelopment continues (Wu, Abarquezand Matheson, 2004). The greatest use ofinterspecific hybrids in operational treebreeding has been with introduced species;for example, Pinus elliottii x P. caribaea inAustralia (Nikles, 1996), hybrid eucalyptsin South Africa, Brazil and the Congo(Eldridge et al., 1993), Acacia mangiumx A. auriculiformis in Viet Nam (Kha,Hai and Vinh, 1998) and hybrid poplarsin temperate regions. These programmesoften rely on clonal propagation fordeployment.The goal of commercial tree breeding isto increase the quantity and quality of woodproducts from plantations. Productionof industrial timber was estimated at2.8 thousand million cubic metres in 2004and has been increasing at an average annualrate of 2.4 percent since 1998 (FAOSTAT)with much of the recent increase beingdue to rapid economic growth in China.Consumption of fuelwood is increasing ata similar rate (Carson, Walter and Carson,2004). Rising demand together withrestrictions on the supply of timber fromnative forests mean that increases in forestproductivity will be required. To date,increased production has been achievedby expanding the area of plantations,particularly in tropical regions wherehigh growth rates can be achieved. Gainshave also been made using conventionalbreeding, but further productivity increasesare required to reduce pressure on nativeforests and limit the increases in land arearequired for plantations. MAS has thepotential to enhance plantation productivityif the relationship between genetic variationin gene sequences and phenotypic variationin traits can be demonstrated.The relatively long generation timesand poor juvenile-mature trait correlationsin forest trees have promoted interest inMAS to accelerate breeding through earlyselection. MAS relies on identifying DNAmarkers which explain a high proportionof additive variation in phenotypic traits.Initially, research focused on the use ofDNA markers in genome-wide linkageanalysis of progeny arrays (Lander andBotstein, 1989). By identifying patternsof co-segregation in complex traits andpolymorphic markers (QTL), these studiesaimed to reveal causative regions of thechromosome or gene that were inheritedintact over a few generations. The QTLapproach can be used for marker-aidedbreeding within families. The low successrate in validating QTL in different geneticbackgrounds and environments (Neale,Sewell and Brown, 2002) led to a changein research focus towards population-levelassociation mapping. This approach seeksto find alleles of genes that affect thephenotype directly (Neale and Savolainen,2004), and relies on the retention of much

286Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishsmaller regions of intact DNA over manygenerations. Candidate genes targeted inthese studies can be identified by genemapping, expressed sequence tag (EST)sequencing, gene expression profiling orfunctional studies (transgenics). If variationcan be found in the sequence of these genesin different phenotypes, it allows MAS tobe used for within- and between-familyselection in forest trees.The application of biotechnology intree improvement research is currentlytaking different paths in developed anddeveloping countries due to contrastingregulatory approval processes for geneticallymodified plants and differences inpublic acceptance of genetically modifiedorganisms (GMOs). There is considerableresistance in developed countries towardstransgenic trees, which has more to do withtheir possible effects on other plants andon the environment than with concernsabout transgenic wood (Sedjo, 2004).Long-term field trials are needed to ensurethe stability of any genetic modificationand the absence of negative impacts ongrowth and resistance to environmentalstresses before they can be incorporatedinto industrial plantations (Strauss et al.,1998; Strauss et al., 2004). Regulationcosts, possible trade restrictions, lack ofpublic acceptance of transgenics and lackof support by major forestry certificationgroups such as the Forest StewardshipCouncil (FSC) are currently barriers to thedevelopment of transgenics (Sedjo, 2004).Consequently, trials of transgenic trees indeveloped countries remain in the researchphase, mostly conducted with young treesgrown under glasshouse conditions (seeWalter and Killerby, 2004 for review). Dueto these problems, some research has shiftedtowards alternative methods of investigatinggene function and incorporating desirablegenes into breeding populations, mainlythrough association mapping.Recent MAS research in forest trees hasbeen greatly assisted by advances in ourunderstanding of tree genomes. The completesequencing of plant genomes such asArabidopsis (Arabidopsis Genome Initiative,2000) and rice (Yu et al., 2002) is improvingour understanding of the number of genesinvolved (25 000–55 000) in the developmentof different organs and the functionof the genes.The Populus genome was the first treegenome to be sequenced with 58 036predicted genes (www.jgi.doe.gov/poplar)and efforts are under way to sequencethe Eucalyptus genome (www.ieugc.up.ac.za), a genus of particular importance incountries with developing economies inAsia and South America. To date, partialcoverage of the E. camaldulensis genome(600 Mb) has been completed by randomshotgun sequencing, through collaborationbetween Oji Paper and the Kasuza DNAResearch Institute in Chiba, Japan (S. PotterEnsis, personal communication). A draftsequence, based on four-fold coverage ofthe genome is expected to be availableby mid-2007 (Poke et al., 2005). Thelarge genome size of conifers is currentlya barrier to whole genome sequencing;however, comprehensive EST sequencingis likely to yield most genes expressed intarget tissues.Genomic resources and tools are nowbeing established for important foresttree species. Rapidly growing numbersof ESTs are publicly available in a rangeof species including Eucalyptus grandis,Pinus radiata, P. taeda, Picea abies, Populustrichocarpa, P. tremula x tremuloides andCryptomeria japonica (see Krutovskii andNeale, 2001 and Strabala, 2004 for reviews)and Avicennia marina (Mehta et al., 2005).

Chapter 15 – Marker-assisted selection in forestry species 287Over 80 000 ESTs have been sequencedfrom pine (http://pinetree.ccgb.umn.edu/),over 130 000 from poplar (http://poppel.fysbot.umu.se/) and over 100 000 fromspruce (www.arborea.ulaval.ca/en/; www.treenomix.ca/).Comprehensive microarrays (Shena etal., 1996) are now being used in manyof these species, allowing transcriptionprofiling of thousands of genes incontrasting phenotypes in a range of tissuesunder different environmental/stress/developmental regimes. Identification ofcandidate genes from expression profilingis based on the assumption that the genesshowing genotype-specific differences intheir level of expression are causing variationin that trait (Morgante and Salamini, 2003).Microarrays are being used to identify genesthat are regulated differentially in individualswith contrasting wood traits in eucalypts(Moran et al., 2002), symbiosis-regulatedgenes in Eucalyptus globulus-Pisolithustinctorius ectomycorrhiza (Voiblet et al.,2001) and genes involved in embryogenesisin pines (van Zyl et al., 2003). Usingmicroarrays, many genes involved in cellwall biosynthesis have been identifiedin loblolly pine (Whetten et al., 2001;Pavy et al., 2005), eucalypts (Paux et al.,2004) and hybrid aspen (Populus tremulax P. tremuloides) (Hertzberg et al., 2001).A combination of proteomics, whichexamines changes in protein expression indifferent tissue and developmental stages,and microarray technology is also beingused to give a more complete picture ofgene function, for example of droughttolerance in Pinus pinaster (Plomion et al.,2004). This discovery work is uncoveringlarge numbers of candidate genes that areexcellent targets for both QTL mappingand association studies aimed at identifyingmarkers for use in MAS.Family-based genetic linkagemapping and QTL analysisGenetic linkage or recombination mappingrelies on finding sufficient polymorphismusing DNA markers in progeny arraysfrom full-sib pedigrees to identifyassociations between linked loci on a chromosome.Genetic linkage maps have beenconstructed for most of the commerciallyimportant forest tree genera (summarizedin Table 1), and updated information ongenetic linkage maps for forest trees isavailable at http://dendrome.ucdavis.edu/index.php. The number and location ofchromosomal regions affecting a trait(QTL) and the magnitude of their effectcan then be investigated by QTL mapping.QTL are identified by a statistical associationbetween variation in a quantitativetrait and segregation of alleles at a markerlocus in a segregating population (mappingpedigree).Most phenotypic traits of interest for treebreeding are characterized by continuousvariation. Such traits are usually influencedby a number of genes with a small effectinteracting with other genes and the environment.The main traits targeted for QTLmapping are wood properties and traitsrelated to adaptation and growth (reviewedby Sewell and Neale, 2000). These includephysical wood properties that affect thestrength of sawn timber (e.g. wood densityand microfibril angle), and properties thataffect paper pulping, e.g. pulp yield, fibrelength and the relative proportion of cellulose,hemicellulose and lignin, generallymeasured as percentage cellulose. In addition,QTL have been identified for diseaseresistance, growth, flowering, vegetativepropagation, frost tolerance and leaf oilcomposition (Table 2).The detection of QTL requires largesample sizes; the lower the heritability

288Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishTable 1Genetic linkage maps constructed for forest trees, markers used and location of mapping pedigreesSpecies Markers 1 Country ReferenceAcacia mangium RFLP, SSR Australia Butcher and Moran, 2000Cryptomeria japonica RFLP, RAPD, isozyme Japan Mukai et al., 1995Eucalyptus camaldulensis RAPD, RFLP, SSR Egypt Agrama, George and Salah, 2002Eucalyptus globulus RAPD, SSR Australia Bundock, Hayden and Vaillancourt, 2000Candidate genes, isozymes, Australia Thamarus et al., 2002ESTP, RFLP, SSREucalyptus grandis x E. AFLP Uruguay Myburg et al., 2003globulusEucalyptus grandis x E.urophylllaRAPDBrazilCongoGrattapaglia and Sederoff, 1994;Verhaegen and Plomion, 1996Eucalyptus nitens Isozyme, RAPD, RFLP Australia Byrne et al., 1995Eucalyptus tereticornis x E. AFLP Portugal Marques et al., 1998globulusEucalyptus urophylla x E. RAPD China Gan et al., 2003tereticornisFagus sylvatica AFLP, RAPD, SSR Italy Scalfi et al., 2004Hevea braziliensis x H. AFLP, isozymes, RFLP, SSR French Lespinasse et al., 1999benthamiana (rubber tree)GuyanaLarix decidua & L. kaempferi AFLP, ISSR, RAPD France Arcade et al., 2000Picea abies RAPD Italy Binelli and Bucci, 1994RAPD Denmark Skov and Wellendorf, 1998AFLP, SAMPL, SSR Italy Paglia, Olivieri and Morgante, 1998Picea glauca ESTP, RAPD, SCAR Canada Gosselin et al., 2002Pinus edulis AFLP USA Travis et al., 1998Pinus elliottii var elliottii RAPD USA Nelson, Nance and Doudrick, 1993Pinus elliottii var. elliottii & AFLP, SSR Australia Shepherd et al., 2003P. caribaea var. hondurensisPinus palustris RAPD USA Kubisiak et al., 1996Pinus pinaster RAPD France Plomion, O’Malley and Durel, 1995AFLP, RAPD, protein France Costa et al., 2000AFLP France Chagne et al., 2002Pinus radiata RFLP, RAPD, SSR Australia Devey et al., 1996Pinus sylvestris RAPD Sweden Yazdani, Yeh and Rimsha, 1995Pinus taeda Isozymes, RAPD, RFLP USA Devey et al., 1994; Sewell, Sherman andNeale, 1999AFLP USA Remington et al., 1999Pinus thunbergii AFLP, RAPD Japan Hayashi et al., 2001PopulusAFLP, candidate genes,isozymes, ISSR, RAPD, RFLP,STS, SSRBelgium,France, USASee review in Cervera et al., 2004;Yin et al., 2004Pseudotsuga menziesii RAPD, RFLP USA Jermstad et al., 1998;Krutovskii et al., 1998Quercus roburIsozyme, minisatellite, RAPD, France Barreneche et al., 1998SCAR, SSR. 5SrDNASalix viminalis AFLP, SSR UK Hanley et al., 2002Salix viminalis x S. schwerinii AFLP, RFLP Sweden Tsarouhas, Gullberg and Lagercrantz,20021AFLP = amplified fragment length polymorphism; ESTP = expressed sequence tagged polymorphism; ISSR = inter-simplesequence repeats; RAPD = random amplified polymorphic DNA; RFLP = restriction fragment length polymorphism; SAMPL= selective amplification of microsatellite polymorphic loci; SCAR = sequence characterized amplified regions; SSR = simplesequence repeat (microsatellite); STS = sequence-tagged sites

Chapter 15 – Marker-assisted selection in forestry species 289Table 2Quantitative trait loci reported for forest tree speciesSpecies Markers 1 QTL ReferenceAcacia mangium RFLP, SSR Disease resistance Butcher, 2004Cryptomeria japonica Isozyme, RAPD, RFLP Juvenile growth, flowering,vegetative propagationYoshimaru et al., 1998RAPD Wood quality Kuramoto et al., 2000Eucalyptus globulus Isozyme, RFLP, SSR Wood density, pulp yield,microfibril angleThamarus et al., 2004Eucalyptus grandis RAPD Growth, wood density Grattapaglia et al., 1996;RAPD Disease resistance Junghaus et al., 2003E.grandis x E. urophylla RAPD Vegetative propagation Grattapaglia, Bertolucci andSederoff, 1995; Marques et al.,1999RAPD Growth, wood density Verhaegen et al., 1997RAPD Leaf oil composition Shepherd, Chaparro and Teasdale,1999Eucalyptus nitens RFLP Growth Byrne et al., 1997aEucalyptus tereticornisx E. globulusRFLP Frost tolerance Byrne et al., 1997bAFLP Vegetative propagation Marques et al., 2002Fagus sylvatica AFLP, RAPD, SSR Leaf traits, growth Scalfi et al., 2004Hevea braziliensis x AFLP, isozyme, RFLP, Disease resistance Lespinasse et al., 2000H. benthamiana SSRPinus palustris x RAPD Juvenile growth Weng et al., 2002P. elliottiiPinus pinaster RAPD Bud set, frost tolerance Hurme et al., 2000AFLP Growth, water use efficiency Brendel et al., 2002Pinus radiata RAPD Growth Emebiri et al., 1997,1998a,bAFLP, RAPD, SSR Wood density Kumar et al., 2000RFLP, SSRGrowth, wood density, disease Devey et al., 2004a,bresistancePinus sylvestris AFLP Growth, cold acclimation Lerceteau, Plomion andAndersson, 2000; Yazdani et al.,2003Pinus taeda Isozymes, RAPD, RFLP Growth Kaya, Sewell and Neale, 1999RFLP Physical wood properties Groover et al., 1994; Sewell etal., 2000RFLP Chemical wood properties Sewell et al., 2002ESTP, RFLP Wood properties Brown et al., 2003PopulusAFLP, ISSR, RAPD, RFLP,SCAR, SSR, STSGrowth, form, leafarchitecture, leaf & budphenology, disease resistance,wood qualitySee review in Cervera et al., 2004Pseudotsuga menziesii RAPD, RFLP Vegetative bud flush Jermstadt et al., 2001aRAPD, RFLP Cold hardiness Jermstadt et al., 2001bRAPD, RFLP QTL x environment Jermstadt et al., 2003Quercus robur AFLP, RAPD, SCAR, SSR Growth, bud burst Saintagne et al., 2004Salix dasyclados xS. viminalisAFLPGrowth, drought tolerance,bud flushTsarouhas, Gullberg andLagercrantz, 2002, 2003;Rönnberg-Wästljung, Glynn andWeih, 20051AFLP = amplified fragment length polymorphism; ESTP = expressed sequence tagged polymorphism; ISSR = inter-simplesequence repeats; RAPD = random amplified polymorphic DNA; RFLP = restriction fragment length polymorphism; SCAR =sequence characterized amplified regions; SSR = simple sequence repeat (microsatellite); STS = sequence-tagged sites

290Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishof the trait and the larger the number ofgenes affecting the trait, the larger thesample size required (see Strauss, Landeand Namkoong, 1992). As shown byBrown et al. (2003), the use of small mappingpopulations of 100–200 segregatingindividuals, typical of most QTL studiesin trees, is likely to cause an upward biasin the estimated phenotypic effect of QTL.Simulation and practical studies have shownthat, in addition to sample size, QTL detectionis affected by genetic background,environment and interactions among QTL.The location of QTL is imprecise as theycan only be mapped to 5–10 cM. This maytranslate into a physical distance of severalmegabases, which may contain severalhundred genes. The effect of a QTL is alsolikely to vary over time in perennial plantswith changing biotic and abiotic factors(Brown et al., 2003). This highlights thenecessity of verifying QTL in differentseasons, environments and genetic backgrounds(Sewell and Neale, 2000). Thechallenges of developing and genotypingthe large progeny arrays required to locateQTL accurately in outbred pedigrees, andof verifying these QTL in different environmentsand ages, are such that MAS hasnot yet been applied in any commercial treebreeding programme.In one of the most intensive studies onapplying MAS to date, and based on datafrom over 1 300 individuals for wood density,4 400 individuals for wood diameterfrom a single pedigree and using selectivegenotyping of the 50 highest and lowestscoring individuals for density and 100of each for diameter, Devey et al. (2004a)were able to validate (in the same pedigree)two out of 13 QTL for diameter and eightout of 27 QTL for wood density in Pinusradiata. The effect of each QTL rangedfrom 0.8 to 3.6 percent of phenotypicvariation, implying that these traits werecontrolled by a large number of genes, eachof small effect. Using a different approach,Brown et al. (2003) used a verificationpopulation of 447 progeny (derived fromre-mating the parents of the QTL pedigree)and an “unrelated population” of 445progeny from the base pedigree to verifyQTL in Pinus taeda. They found about halfthe QTL were detected in multiple seasonsand fewer QTL were common to differentpopulations.An area where QTL mapping may assistbreeders is in breaking linkages betweennegatively correlated traits. For example inE. grandis and E. urophylla, Verhaegen et al.(1997) reported co-location of QTL for thenegatively correlated traits of wood densityand growth. If these traits are controlled bytightly linked genes, markers could be usedto select favourable recombinants.Most markers used in QTL mappinghave been anonymous markers that areunlikely to occur in a gene influencing aquantitative trait. In an attempt to increasethe power of QTL mapping, candidategenes that may control the trait in questionare being used as molecular markers.Candidate genes are typically sourced fromthe tissue of interest (e.g. xylem or leaves)and have either a known function intuitivelyrelated to the trait, or are of interestfrom studies of their expression usingDNA microarrays. Comparative mappingand candidate gene approaches can utilizesuch information to search for homologousgenes in different genomes. Candidategenes have been mapped to QTL for woodquality in E. urophylla and E. grandis (Gionet al., 2000), Pinus taeda (Neale, Sewell andBrown, 2002), and E. globulus (Thamaruset al., 2004). They have also been mappedto QTL for bud set and bud flush inPopulus deltoides (Frewen et al., 2000).

Chapter 15 – Marker-assisted selection in forestry species 291Kirst et al. (2004) measured transcriptabundance in 2 608 genes in the differentiatingxylem of 91 E. grandis backcrossprogeny. QTL analysis of lignin-relatedtranscripts (expressed gene QTL [eQTL])showed that their mRNA abundance isregulated by two genetic loci. Coordinateddown-regulation of genes encoding ligninenzymes was observed in fast growing individuals,indicating that the same genomicregions are regulating growth and the lignincontent and composition in the progeny.Comparative mapping has shown that genecontent and gene order are conserved overlong chromosomal regions among relatedspecies. Comparative maps are thereforelikely to play an important role in enablinginformation on gene location andfunction to be transferred between speciesand genera. However, this will depend onorthologous genetic markers being mappedin each species (Krutovskii and Neale,2001). To date, comparative maps have beenpublished for Populus (Cervera et al., 2001),Pinus (Devey et al., 1999; Krutovsky et al.,2004), Quercus and Castanea (Barrenecheet al., 2004).MAS, based on QTL, is most likely to beused for within-family selection in a limitednumber of elite families that can be propagatedclonally for deployment in large-scaleindustrial plantations. It is most suitablefor traits that are expensive to measure orcan only be detected after plants have beensubjected to a particular stress or pathogen,and that have poor juvenile-mature correlations.Limitations of the approach includethe low resolution of the marker-trait associations,the low proportion of phenotypicvariation explained by QTL (generallyless than 10 percent), and the low successrate in validating QTL in different geneticbackgrounds and environments (Sewell andNeale, 2002). Recombination-based methodologieshave been applied to inbred croplines to positionally clone genes of interestin QTL regions (Salvi et al., 2002); however,the use of this technique in forest treesis not practicable due to their outcrossingbreeding system.Population-based associationstudiesLimitations of the QTL approach have ledto a change in research focus towards identifyingvariations in the DNA sequence ofgenes directly controlling phenotypic variation,known by some as GAS. One of themain advantages of association genetics isthe high resolution of marker-trait associations.As natural populations are usedin association studies, recombinations thataccumulate over many generations of thepopulation break any long range associationsbetween marker and trait leaving shortstretches of the genome associated with thetrait. If alleles or SNPs can be found that arestrongly associated with superior phenotypes,they can be used for selection acrossbreeding populations. This methodology isbetter suited to tree breeding programmes,which aim to maintain a broad genetic base,i.e. programmes with a large number offamilies. In contrast, the QTL approach isused for within-family selection. Spuriousassociations may be observed in associationstudies where there is undetectedgenetic structure in the breeding populationthat invalidates standard statisticaltests. Strategies for dealing with populationstratification have been developed to avoidthese problems (Pritchard et al., 2000; Wuand Zeng, 2001).In the first association study publishedin forest trees, Thumma et al. (2005)identified 25 common SNP markers inthe lignin biosynthesis gene CCR fromEucalyptus nitens. Using single-marker

292Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishand haplotype analyses in 290 trees froma natural population, they observed twohaplotypes that were significantly associatedwith microfibril angle, a major determinantof timber strength. These results wereconfirmed in a full-sib family in E. nitensand in the related species E. globulus. In apowerful demonstration of the resolutionof association genetics, Thumma et al.(2005) detected an alternatively-splicedvariant of the CCR gene from the region ofthe significant haplotype, thereby revealingthe probable molecular basis of the traitvariation.Association mapping is a particularlyuseful approach when genes are availablethat are likely to be functionally relevantto the trait of interest. Homologues ofgenes characterized in model species suchas Arabidopsis, maize or rice, and poplarare excellent targets for association studiesin forest species. In most cases, putativeorthologues can be identified by comparingESTs to gene sequences in public databases.In some cases, gene function may bedetermined by modulating the expressionof selected genes using sense and antisensemodification to up- and down-regulategene expression, or intron RNA hairpinconstructs to silence genes (Smith et al.,2000). However, one of the advantages ofassociation studies is the capacity to studya large number of genes simultaneouslywithout the need for transformation (Peterand Neale, 2004).There is considerable interest in understandingthe genes controlling wood fibrecell wall development in forest trees asfibre microstructure is a major determinantof the commercial value of wood. Forexample, the angle of orientation of cellulosemicrofibrils (MFA) in fibre cell wallsis known to affect timber strength andstiffness as well as fibre collapsibility, animportant determinant of tensile strength inpaper. Knowledge of cell wall biosynthesiswould also assist in understanding andmanipulating the development of abnormalwood, e.g. tension/compression wood (seePaux et al., 2005; Pavy et al., 2005), whichis known to have an impact on woodstability, sawing patterns and pulpability.Wood is primarily composed of secondaryxylem, and its properties are the product ofsequential developmental processes fromcambial cell division and expansion, to secondarywall formation and lignification.Genes expressed during xylogenesis determinethe physical and chemical propertiesof wood. Important genes are now beingidentified that control the synthesis of themajor constituents of the cell wall: cellulose,hemicellulose and lignin. Genes forcellulose synthesis (CesA) have been clonedin aspen (Joshi, Wu and Chiang, 1999; Wu,Joshi and Chiang, 2000), poplar (Djerbiet al., 2005) and loblolly pine (Nairn andHaselkorn, 2005). Characterizing the CesAgene in aspen revealed strong similaritywith a secondary cell wall protein in cotton,indicating that they serve similar functionsin the two evolutionarily divergent genera.Transformation of cellulose synthase genesin aspen (Joshi, 2004) should further elucidategene function. Each of the threeloblolly CesA genes is orthologous to oneof the three angiosperm secondary cell wallCesAs, suggesting functional conservationbetween angiosperms and gymnosperms.A search of the poplar genome revealed18 distinct CesA gene sequences in Populustrichocarpa (Djerbi et al., 2005). The CesAgenes belong to a superfamily of CesA-like(Csl) genes, which includes a very largenumber of glycosyltransferases that arelikely to be involved in the synthesis ofthe numerous non-cellulosic polysaccharidesin plants (Liepman, Wilkerson and

Chapter 15 – Marker-assisted selection in forestry species 293Keegstra, 2005).Lignin biosynthesis is well understoodat the molecular level in plants (reviewedby Boerjan, Ralph and Baucher, 2003 andPeter and Neale, 2004) and is of particularinterest in forest trees as removal of ligninfor paper-making has major economic andenvironmental costs. In some cases, geneticmodification of lignin structure has beenshown to improve delignification in pulpand paper-making (Jouanin and Goujon,2004), and down regulation of ligninpathway enzymes has also been shown toincrease cellulose content (Hu et al., 1999).Gymnosperms and angiosperms share acommon set of enzymes that are responsiblefor the formation of guaiacyl lignin,while angiosperms have evolved at least twoenzymes that catalyse the production ofsyringyl lignin. Association studies are nowbeing carried out in loblolly to examine theadaptive significance of sequence variationin monolignol biosynthetic genes (Peterand Neale, 2004) and other genes controllingwood properties (Brown et al.,2004). Similar research (Table 3) aimed atidentifying genes controlling wood formationis being undertaken in Douglas fir(Krutovsky et al., 2005), maritime pine (Potet al., 2004), radiata pine (S.G. Southertonand G.F. Moran, personal communication),spruce (MacKay et al., 2005) and eucalypts(Moran et al., 2002).The availability of genes linked to arange of other traits in model plants opensup new areas of investigation in associationgenetics. For example, associationstudies are in progress to identify genescontrolling pathogen resistance (Ersoz etal., 2004; MacKay et al., 2005), droughttolerance (Ersoz et al., 2004), cold tolerance(Krutovsky et al., 2005) and budset (Paoli and Morgante, 2005) (Table 3).Further opportunities exist for associationstudies aimed at identifying SNPs linked toimportant traits. Flowering is particularlywell understood at the molecular level (Zikand Irish, 2003), and increasing numbersof genes controlling flowering have beencloned in angiosperm tree species includingeucalypts (Southerton et al., 1998; Watsonand Brill, 2004), silver birch (Elo et al.,2001), poplar (Rottmann et al., 2000) andgymnosperm tree species including spruce(Tandre et al., 1995; Rutledge et al.,1998)and pines (Mouradov et al., 1998, 1999).Another important technologicaladvance that is making large-scale associationstudies possible is the recentdevelopment of rapid, high-throughputTable 3Association studies in progress for forest tree speciesSpecies Trait ReferenceEucalyptus nitens Wood properties Moran et al., 2002; Thumma et al., 2005Populus Wood properties MacKay et al., 2005Disease resistance MacKay et al., 2005Picea glauca Wood properties MacKay et al., 2005Disease resistance MacKay et al., 2005Picea abies Bud set Paoli and Morgante, 2005Pseudotsuga menziesii Wood properties Krutovsky et al., 2005;Cold hardiness Krutovsky et al., 2005;Pinus radiata Wood properties Southerton and Moran unpub. dataPinus taeda Wood properties Peter and Neale, 2004; Brown et al., 2004Disease resistance Ersoz et al., 2004Drought tolerance Ersoz et al., 2004Pinus pinaster Wood properties Pot et al., 2004

294Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishgenotyping techniques that have drasticallyreduced the cost of genotyping SNPs inassociation populations (www.illumina.com/products/prod_snp.ilmn). It is nowfeasible to genotype SNPs in hundreds ofgenes potentially associated with a trait.QTL mapping remains largely a researchtool to improve our understanding of thenumber, distribution and mode of action ofgenes controlling quantitative traits. QTLcan also play a role in GAS as a vehicle forvalidating significant SNP correlations identifiedin association populations (Thummaet al., 2005). In the near future, associationstudies promise to yield numerous SNPmarkers that could be used in breedingprogrammes for early selection of superioralleles associated with a wide range of traits.As the efficiency of techniques for microarrayanalysis, SNP discovery, genotypingand other molecular procedures improvefurther, the opportunities to incorporatemolecular technologies into breeding programmesfor forest trees will increase.use of MAS to enhance breedingprogrammes in developingcountriesThe adoption of molecular techniquesvaries widely, not only between developedand developing countries but also amongthe less developed economies (Chaix andMonteuuis, 2004). Countries such as China,India, Indonesia, Malaysia, Thailand andViet Nam have established molecular laboratoriesfor genotyping. Molecular markersare being used routinely to assess the levelof genetic diversity in breeding programmesand to monitor any changes following selection(Butcher, 2003; Marcucci Poltri et al.,2003). They are also being used to estimatelevels of contamination and inbreeding inopen-pollinated seed orchards (Chaix etal., 2003; Harwood et al., 2004), to validateintra- and interspecies crosses and todetermine error rates in clonal propagationor trial establishment (see, for example,Bell et al., 2004). This has identified relativelyhigh error rates in several breedingprogrammes, affecting calculations of heritability,breeding value and genetic gain.Genetic linkage maps have been publishedfor eucalypts in China (Gan et al., 2003)and Brazil is prominent in eucalypt mappingand genomics (Grattapaglia, Chapter14). EST libraries have been developed formangroves in India as a first step towardscharacterizing genes associated withsalinity tolerance (Mehta et al., 2005), whileDNA markers have been used for backwardselection to identify superior maleparents in eucalypt seed orchards in Brazil(Grattapaglia et al., 2004). This approachhas some potential for accelerating thebreeding cycle in open-pollinated breedingprogrammes, particularly with species thatare difficult to hand pollinate (Butcher,Moran and DeCroocq, 1998). The applicationof QTL-MAS in developing countriesremains limited, exceptions being selectionof coconut parents for breeding (FAO,2003) and identification of QTL for rubbertree improvement (Lespinasse et al., 2000).More widespread application may dependon economic considerations. Reports onthe financial viability of MAS differ, withJohnson, Wheeler and Strauss (2000) indicatingthat large areas would need to beplanted with MAS-improved germplasm tojustify initial investment, while Wilcox etal. (2001) suggest significant financial gainsare possible even when selection is basedon DNA markers linked to a few loci eachof relatively small effect. Association mappinghas the potential for more widespreadapplication in developing countries due,in part, to the ability to transfer markersamong individuals, regardless of pedigree

Chapter 15 – Marker-assisted selection in forestry species 295or family relationships. The possibility oftransferring SNP markers among specieshas already been demonstrated in eucalypts(Thumma et al., 2005).Forest trees, including many species indeveloping countries, are near their wildstate, and significant improvement can stillbe made quite rapidly based on selectionamong existing genotypes (FAO, 2003).Association studies are ideally suited toexploiting variation in natural populationsand do not rely on the existence of extensivepedigrees from controlled crosses.Suitable populations should include a largenumber of unrelated individuals of thesame age growing on the one site. It hasbeen estimated that 500 individuals wouldbe necessary to detect association betweena quantitative nucleotide responsiblefor 5 percent or more of the phenotypicvariance (Long and Langley, 1999). Thedevelopment of such populations wouldprovide a good foundation for future GASresearch in developing countries. While themarkers developed using this approach arelikely to be more easily transferred betweenbreeding programmes, the application ofGAS would require the subsequent developmentof advanced breeding programmeswhere the selection of superior alleles couldtake place. However, publicly fundedforestry research is suboptimal in manydeveloping countries and development prioritiesdo not necessarily include geneticimprovement programmes (FAO, 2003).The major costs of GAS are associatedwith identifying candidate genes potentiallylinked to the relevant traits, and discoveringSNPs. In some cases, public databases maycontain large numbers of genes from thetarget or closely related species but, if not,there would be additional costs associatedwith EST sequencing of genes fromthe relevant tissue (i.e. xylem genes forwood traits). These additional costs may beoffset partially by EST sequencing clonesfrom mixed cDNA libraries derived from anumber of unrelated trees from the associationpopulation.Previously, the cost of genotyping SNPswas prohibitive, but this has fallen dramaticallyin recent years as high-throughputtechnologies have been developed for thehuman HapMap project (InternationalHapMap Consortium, 2003). The IlluminaBeadstation technology (www.illumina.com) is particularly suited to smaller-scalegenotyping projects such as those beingundertaken in forest trees. Cost is certainlya limitation in many developing countriesincluding much of Africa and someSouth American countries; however, inmost Asian countries and countries suchas Brazil where molecular genetic laboratoriesare already well established, costswould not be prohibitive. The full benefitsof GAS would require development of efficientclonal propagation and deploymentsystems before it was applied routinely.Less stringent regulatory approval processesand greater public acceptance ofgenetically modified plants have allowedBrazil and China to take a lead role in commercializingtransgenic tree technology.China is the only country to announce thecommercial release of transgenics (poplar)with 300–500 hectares being planted in2002 (Wang, 2004). Regulatory approvalfor the release of Bacillus thuriengensis(Bt) insect resistant eucalypts in Brazil ispending (Sedjo, 2004). Given the difficultyof carrying out long-term transgenic fieldtrials with long rotation conifers, transgenicapproaches are likely to be restrictedto modification of high-value traits such aswood fibre properties in short rotation speciesgrown on a large scale. Conventionalbreeding, using either open or controlled

296Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishpollination in seed orchards, will continueto be the most important mechanism fordeveloping new genotypes for increasedgenetic gain (Carson, Walter and Carson,2004).AcknowledgementsWe thank Simon Potter and Reddy Thumma(ENSISjv) for constructive comments onthe manuscript.ReferencesAgrama, H.A., George, T.L. & Salah, S.F. 2002. Construction of genome map for Eucalyptuscamaldulensis Dehn. Silv. Genet. 51: 201–206.Arabidopsis Genome Initiative. 2000. Analysis of the genome sequence of the flowering plantArabidopsis thaliana. Nature 408: 796–815.Arcade, A., Anselin, F., Faivre Rampant, P., Lesage, M.C., Paques, L.E. & Prat, D. 2000. Applicationof AFLP, RAPD and ISSR markers to genetic mapping of European and Japanese larch. Theor.Appl. Genet. 100: 299–307.Barreneche, T., Bodénès, C., Lexer, C., Trontin, J.F., Fluch, S., Streiff, R., Plomion, C., Roussel, G.,Steinkellner, H., Burg, K., Favre, J.M., Glössl, J. & Kremer, A. 1998. A genetic linkage map ofQuercus robur L. (pedunculate oak) based on RAPD, SCAR, microsatellite, minisatellite, isozyme,and 5S rDNA markers. Theor. Appl. Genet. 97: 1090–1103.Barreneche, T., Casosoli, M., Russell, K., Akkak, A., Meddour, H., Plomion, C., Villai, F. & Kremer,A. 2004. Comparative mapping between Quercus and Castanea using simple-sequence repeats(SSRs). Theor. Appl. Genet. 108: 558–566.Bell, J.C., Powell, M., Devey, M.E. & Moran, G.F. 2004. DNA profiling, pedigree lineage analysis andmonitoring in the Australian breeding program of radiata pine. Silv. Genet. 53: 130–134.Binelli, G. & Buchi, G. 1994. A genetic linkage map of Picea abies based on RAPD markers, as a toolin population genetics. Theor. Appl. Genet. 88: 283–288.Boerjan, W., Ralph, J. & Baucher, M. 2003. Lignin biosynthesis. Ann. Rev. Plant Biol. 54: 519–46.Brendel, O., Pot, D., Plomion, C., Rozenberg, P. & Guehl, J.M. 2002. Genetic parameters and QTLanalysis of d13C and ring width in maritime pine. Plant Cell Environ. 25: 945–953.Brown, G.R., Bassoni, D.L., Gill, G.P., Fontana, J.R., Wheeler, N.C., McGraw, R.A., Davis, M.F.,Sewell, M.M., Tuskan, G.A. & Neale, D.B. 2003. Identification of quantitative trait loci influencingwood property traits in loblolly pine (Pinus taeda L.) III. QTL verification and candidate genemapping. Genetics 164: 1537–1546.Brown, G.R., Gill, G.P., Kuntz, R.J., Beal, J.A., Nelson, D., Wheeler, N.C., Penttila, B., Roers, J.& Neale, D.B. 2004. Associations of candidate gene single nucleotide polymorphisms with woodproperty phenotypes in loblolly pine, Abstract W98. In Plant & Animal Genomes XII Conference,San Diego, CA, USA.Bundock, P.C., Hayden, M. & Vaillancourt, R. 2000. Linkage maps of Eucalyptus globulus usingRAPD and microsatellite markers. Silv. Genet. 49: 223–232.Butcher, P.A. 2003. Molecular breeding of tropical trees. In A. Rimbawanto, & M. Susanto, eds. Proc.Internat. Conf. on Advances in Genetic Improvement of Tropical Tree Species, pp. 59–66. Centrefor Forest Biotechnology and Tree Improvement, Yogyakarta, Indonesia.Butcher, P.A. 2004. Genetic mapping in acacias. In S. Kumar & M. Fladung, eds. Molecular geneticsand breeding of forest trees, pp. 411–427. New York, USA, Haworth Press.

Chapter 15 – Marker-assisted selection in forestry species 297Butcher, P.A. & Moran, G.F. 2000. Genetic linkage mapping in Acacia mangium 2. Development ofan integrated map from two outbred pedigrees using RFLP and microsatellite loci. Theor. Appl.Genet. 101: 594–605.Butcher, P.A., Moran, G.F. & DeCroocq, S. 1998. Use of molecular markers in domestication andbreeding programs for acacias. In J.W.Turnbull, H.R. Crompton, & K. Pinyopusarek, eds. Recentdevelopments in acacia planting, pp. 211–215. ACIAR Proc. No. 82, Canberra. ACIAR.Byrne, M., Murrell, J.C., Allen, B. & Moran, G.F. 1995. An integrated genetic linkage map for eucalyptsusing RFLP, RAPD and isozyme markers. Theor. Appl. Genet. 91: 869–875.Byrne, M., Murrell, J.C., Owen, J.V., Kriedemann, P., Williams, E.R. & Moran, G.F. 1997a.Identification and mode of action of quantitative trait loci affecting seedling height and leaf area inEucalyptus nitens. Theor. Appl. Genet. 94: 674–681.Byrne, M., Murrell, J.C., Owen, J.V., Williams, E.R. & Moran, G.F. 1997b Mapping of quantitativetrait loci influencing frost tolerance in Eucalyptus nitens. Theor. Appl. Genet. 95: 975–979.Carson, M., Walter, C. & Carson, S. 2004. The future of forest biotechnology. In R. Kellison, S.McCord, & K.M.A Gartland, eds. Forest biotechnology in Latin America, pp. 13–38. Proceedings,Workshop Biotecnolegia Forestal, Biotechnology Forum, Conceptión, Chile (available at www.forestbiotech.org/pdffiles/ChlePDFfinal31Jan2005.pdf).Cervera, M.T., Sewell, M.M., Rampant, P.-F., Storme, V. & Boejan, W. 2004. Genome mapping inPopulus. In S. Kumar, & M. Fladung, eds. Molecular genetics and breeding of forest trees, pp. 387–410. New York, USA, Haworth Press.Cervera, M.T., Stormel, V., Ivens, B., Gusmão, J., Liu, B.H., Hostyn, V., Van Slyckenc, J., VanMontagu, M. & Boerjan, W. 2001. Dense genetic linkage maps of three Populus species (Populusdeltoides, P. nigra and P. trichocarpa based on AFLP and microsatellite markers. Genetics 158:787–809.Chagne, D., Lalanne, C., Madur, D., Kumar, S., Frigerio, J.-M., Krier, C., Decroocq, S., Savoure,Al, Bou-Dagher-Karrat, M., Bertocchi, E., Brach, J. & Plomion, C. 2002. A high density linkagemap for maritime pine based on AFLPs. Ann. For. Sci. 59: 627–636.Chaix, G., Gerber, S., Razafimaharo, V., Vigneron, P., Verhaegen, D. & Hamon, S. 2003. Gene flowestimation in a Malagasy seed orchard of Eucalyptus grandis. Theor. Appl. Genet. 107: 705–712.Chaix, G. & Monteuuis, O. 2004. Biotechnology in the forestry sector. In Preliminary review ofbiotechnology in forestry including genetic modification. FAO Forest Genetic Resources WorkingPaper FGR/59E. Rome (available at www.fao.org/docrep/008/ae574e/ae574e00.htm).Costa, P., Pot, D., Dubos, C., Frigerio, J.M., Pionneau, C., Bodenes, C., Bertocchi, E., Cervera, M.-T., Remington, D.L. & Plomion, C. 2000. A genetic map of Maritime pine based on AFLP, RAPDand protein markers. Theor. Appl. Genet. 100: 39–48.Devey, M.E., Fiddler, T.A, Liu, B.-H., Knapp, J. & Neale, D.B. 1994. An RFLP linkage map forloblolly pine based on a three-generation pedigree. Theor. Appl. Genet. 88: 273–278.Devey, M.E., Bell, J.C., Smith, D.N., Neale, D.B. & Moran, G.F. 1996. A genetic linkage mapfor Pinus radiata based on RFLP, RAPD and microsatellite markers. Theor. Appl. Genet. 92:673–679.Devey, M.E., Sewell, M.M., Uren, T.L. & Neale, D. 1999. Comparative mapping in loblolly andradiata pine using RFLP and microsatellite markers. Theor. Appl. Genet. 99: 656–662.

298Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishDevey, M.E., Carson, S.D., Nolan, M.F., Matheson, A.C., Te Riini, C. & Hohepa, J. 2004a. QTLassociation for density and diameter in Pinus radiata and the potential for marker-aided selection.Theor. Appl. Genet. 108: 516–524.Devey, M.E., Groom, K.A., Nolan, M.F., Bell, J.C., Dudzinski, M.J., Old, K.M., Matheson,A.C. & Moran, G.F. 2004b. Detection and verification of quantitative trait loci for resistance toDothistroma needle blight in Pinus radiata. Theor. Appl. Genet. 108: 1056–1063.Djerbi, S., Lindskog, M., Arvestad, L., Sterky, F. & Teeri, T.T. 2005. The genome sequence of blackcottonwood (Populus trichocarpa) reveals 18 conserved cellulose synthase (CesA) genes. Planta221: 739–746.Eldridge, K.G., Davidson, J., Harwood, C.E. & van Wyk, G. 1993. Eucalypt Domestication andBreeding. Oxford, Clarendon Press.Elo, A., Lemmetyinen, J., Turunen, M.-L., Tikka, T. & Sopanen, T. 2001. Three MADS-box genessimilar to APETALA 1 and FRUITFULL from silver birch (Betula pendula). Physiol. Plant 112:95–103.Emebiri, L.C., Devey, M.E., Matheson, A.C. & Slee, M.U. 1997. Linkage of RAPD markers toNESTUR, a stem growth index in radiata pine seedlings. Theor. Appl. Genet. 95: 119–124.Emebiri, L.C., Devey, M.E., Matheson, A.C. & Slee, M.U. 1998a. Age-related changes in the expressionof QTLs for growth in radiata pine seedlings. Theor. Appl. Genet. 97: 1053–1061.Emebiri, L.C., Devey, M.E., Matheson, A.C. & Slee, M.U. 1998b. Interval mapping of quantitativetrait loci affecting NESTUR, a stem growth efficiency index of radiata pine seedlings. Theor. Appl.Genet. 97: 1062–1068.Ersoz, E.S., Gonzalez-Martinez, S., Gill, G., Brown G, Morse, A., Davis, J., White, T. & Neale,D.B. 2004. SNP discovery in candidate genes for disease resistance (pitch canker and fusiformrust) and drought tolerance in loblolly pine, Abstract W95. In Plant & Animal Genomes XIIConference, San Diego, CA, USA.FAO. FAOSTAT interactive database (available at http://faostat.fao.org/default.aspx).FAO. 2003. Molecular marker-assisted selection as a potential tool for genetic improvement of crops,forest trees, livestock and fish in developing countries. Summary Document to Conference 10 (17November to 14 December 2003) of the FAO Biotechnology Forum (available at www.fao.org/biotech/logs/C10/summary.htm).Frewen, B.E., Chen, T.H.H., Howe, G.T., Davis, J., Rohde, A., Boerjan, W. & Bradshaw, H.D.Jr.2000. Quantitative trait loci and candidate gene mapping of bud set and bud flush in Populus.Genetics 154: 837–845.Gan, S., Shi, J., Li, M., Wu, K., Wu, J. & Bai, J. 2003. Moderate density molecular maps of Eucalyptusurophylla S.T. Blake and E. tereticornis Smith genomes based on RAPD markers. Genetica 118:59–67.Gion, J.-M., Rech, P., Grima-Pettenati, J., Verhaegen, D. & Plomion, C. 2000. Mapping candidategenes in Eucalyptus with emphasis on lignification genes. Mol. Breed. 6: 441–449.Gosselin, I., Zhou, Y., Bousquet, J. & Isabel, N. 2002. Megagametophyte-derived linkage maps ofwhite spruce (Picea glauca) based on RAPD, SCAR and ESTP markers. Theor. Appl. Genet. 104:987–997.Grattapaglia, D. & Sederoff, R.R. 1994. Genetic linkage maps of Eucalyptus grandis andEucalyptus urophylla using a pseudo-testcross mapping strategy and RAPD markers. Genetics 137:1121–1137.

Chapter 15 – Marker-assisted selection in forestry species 299Grattapaglia, D., Bertolucci, L.G. & Sederoff, R.R. 1995. Genetic mapping of QTLs controllingvegetative propagation in Eucalyptus grandis and E. urophylla using a pseudo-testcross strategy andRAPD markers. Theor. Appl. Genet. 90: 933–947.Grattapaglia, D., Bertolucci, L.G., Penchel, R. & Sederoff, R.R. 1996. Genetic mapping of quantitativetrait loci controlling growth and wood quality traits in Eucalyptus grandis using a maternalhalf-sib family and RAPD markers. Genetics 144: 1205–1214.Grattapaglia, D., Ribeiro, V.J. & Rezende, G.D.S.P. 2004. Retrospective selection of elite parenttrees using paternity testing with microsatellite markers: an alternative short term breeding tacticfor Eucalyptus. Theor. Appl. Genet. 109: 192–199.Groover, A., Devey, M., Fiddler, T., Lee, J., Megraw, R., Mitchel-Olds, T., Sherman, B., Vujcic, S.,Williams, C. & Neale, D. 1994. Identification of quantitative trait loci influencing wood specificgravity in an outbred pedigree of loblolly pine. Genetics 138: 1293–1300.Hanley, S., Barker, J.H.A., van Oijen, J.W., Aldam, C., Harris, S.L., Ahman, I., Larsson, S. &Karp, A. 2002. A genetic linkage map of willow (Salix viminalis) based on AFLP and microsatellitemarkers. Theor. Appl. Genet. 105: 1087–1096.Harwood, C.E., Ha Huy Thinh, Tran Ho Quang, Butcher, P.A. & Williams, E.R. 2004. The effectof inbreeding on early growth of Acacia mangium in Vietnam. Silv. Genet. 53: 65–69Hayashi, E., Kondo, T., Terada, K., Kuramoto, N. Goto, Y., Okamura, M. & Kawasaki, H. 2001.Linkage map of Japanese black pine based on AFLP and RAPD markers including markers linkedto resistance against the pine needle gall midge. Theor. Appl. Genet. 102: 871–875.Hertzberg, M., Aspeborg, H., Schrader, J., Andersson, A., Erlandsson, R., Blomqvist, K., Bhalerao,R., Uhlén, M., Teeri, T.T., Lundeberg, J., Sundberg, B., Nilsson, P. & Sandberg, G. 2001. A transcriptionalroadmap to wood formation. Proc. Nat Acad. Sci. USA 98: 14732–14737.Hu, W.-J., Harding, S.A., Lung, J., Popko, J.L., Ralph, J., Stokke, D.D., Tsai, C.-J. & Chiang,V.L. 1999. Repression of lignin biosynthesis promotes cellulose accumulation and growth. NatureBiotech. 17: 808–812.Hurme, P., Sillanpaa, M.J., Arjas, E., Repo, T. & Savolainen, O. 2000. Genetic basis of climatic adaptationin Scots pine by Bayesian quantitative trait locus analysis. Genetics 156: 1309–1322.International HapMap Consortium. 2003. The International HapMap Project. Nature 426:789–796.Jermstad, K.D., Bassoni, D.L., Jech, K.S., Ritchie, G.A., Wheeler, N.C. & Neale, D.B. 2003.Mapping of quantitative trait loci controlling adaptive traits in coastal Douglas fir. III. QTL byenvironment interactions. Genetics 165: 1489–1506.Jermstad, K.D., Bassoni, D.L., Jech, K.S., Wheeler, N.C. & Neale, D.B. 2001a. Mapping of quantitativetrait loci controlling adaptive traits in coastal Douglas fir. I. Timing of vegetative bud flush.Theor. Appl. Genet. 102: 1142–1151.Jermstad, K.D., Bassoni, D.L., Wheeler, N.C., Anekonda, T.S., Aitken, S.N., Adams, W.T. &Neale, D.B. 2001b. Mapping of quantitative trait loci controlling adaptive traits in coastal Douglasfir. II. Spring and fall cold hardiness. Theor. Appl. Genet. 102: 1152–1158.Jermstad, K.D., Bassoni, D.L., Wheeler, N.C. & Neale, D.B. 1998. A sex-averaged linkage map incoastal Douglas-fir (Pseudotsuga menziesii [Mirb.] Franco) based on RFLP and RAPD markers.Theor. Appl. Genet. 97: 762–770.Johnson, G.R., Wheeler, N.C. & Strauss, S.H. 2000. Financial feasibility of marker-aided selectionin Douglas fir. Can. J. For. Res. 30: 1942–1952.

300Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishJoshi, C.P. 2004. Molecular genetics of cellulose biosynthesis in trees. In S. Kumar & M. Fladung,eds. Molecular genetics and breeding of forest trees, pp. 141–165. New York, USA, HaworthPress.Joshi, C.P., Wu, L. & Chiang, V.L. 1999. A novel xylem specific and tension stress responsive cellulosesynthase gene from quaking aspen, Abstract p.19. In M. Campbell, ed. Forest Biotechnology99. Oxford, UK, Oxford University Press.Jouanin, L. & Goujon, T. 2004. Tuning lignin metabolism through genetic engineering in trees. InS. Kumar & M. Fladung, eds. Molecular genetics and breeding of forest trees, pp. 167–191. NewYork, USA, Haworth Press.Junghaus, D.T., Alfenas, A.C., Brommonschenkel, S.H., Oda,S., Mello, E.J. & Grattapaglia, D.2003. Resistance to rust (Puccinia psidii Winter) in Eucalyptus: mode of inheritance and mapping ofa major gene with RAPD markers. Theor. Appl. Genet. 108: 175–180.Kaya, Z., Sewell, M.M. & Neale, D.B. 1999. Identification of quantitative trait loci influencing annualheight- and diameter-increment growth in loblolly pine. Theor. Appl. Genet. 98: 586–592.Kha, L.D., Hai, N.D. & Vinh, H.Q. 1998. Clonal tests and propagation options for naturalhybrids between Acacia manguim and A. auriculiformis. In J.W. Turnbull H.R. Crompton & K.Pinyopusarek, eds. Recent developments in acacia planting, pp. 203–210. ACIAR Proc. No. 82,Canberra, ACIAR.Kirst, M., Myburg, A.A., De León, J.P.G., Kirst, M.E., Scott, J. & Sederoff, R. 2004. Coordinatedgenetic regulation of growth and lignin revealed by quantitative trait locus analysis of cDNAmicroarray data in an interspecific backcross of Eucalyptus. Plant Physiol. 135: 2368–2378.Krutovskii, K.V. & Neale, D.B. 2001. Forest genomics for conserving adaptive genetic diversity.FAO Forest Genetic Resources Working Paper FGR/3 (E). Rome (available at www.fao.org/DOCREP/003/X6884E/X6884E00.HTM).Krutovskii, K.V., Vollmer, S.S., Sorensen, F.C., Adams, W.T., Knapp, S.J. & Strauss, S.H. 1998.RAPD genome maps of Douglas fir. J. Hered. 89: 197–205.Krutovsky, K.V., Troggio, M., Brown, G., Jermstad, K.D. & Neale, D.B. 2004. Comparative mappingin the Pinaceae. Genetics 168: 447–461.Krutovsky, K.V., Pande, P., Jermstad, K.D., Howe, G.T., St. Clair, J.B., Wheeler, N.C. & Neale, D.B.2005. Nucleotide diversity and linkage disequilibrium in adaptive trait related candidate genes inDouglas fir, Abstract W110. In Plant & Animal Genomes XII Conference, San Diego, CA, USA.Kubisiak T.L., Nelson, C.D., Nance, W.L. & Stine, M. 1996. Comparison of RAPD linkage mapsconstructed for a single longleaf pine from both haploid and diploid mapping populations. For.Genet. 3: 203–211.Kumar, S., Spelman, R.J., Garrick, D.J., Richardson, T.E., Lausberg, M. & Wilcox, P.L. 2000.Multiple marker mapping of wood density loci in an outbred pedigree of radiata pine. Theor. Appl.Genet. 100: 926–933.Kuramoto, N., Kondo, T., Fujisawa, Y., Nakata, R., Hayashi, E. & Goto, Y. 2000. Detection ofquantitative trait loci for wood strength in Cryptomeria japonica. Can. J. For. Res. 30: 1525–1533.Lander, E.S. & Botstein, D. 1989. Mapping Mendelian factors underlying quantitative traits usingRFLP linkage maps. Genetics 121: 185–199.Lerceteau, E., Plomion, C. & Andersson, B. 2000. AFLP mapping and detection of quantitative traitloci (QTLs) for economically important traits in Pinus sylvestris: a preliminary study. Mol. Breed.6: 451–458.

Chapter 15 – Marker-assisted selection in forestry species 301Lespinasse, D., Grivet, L., Troispoux, V., Rodier-Goud, M., Pinard, F. & Seguin, M. 2000.Identification of QTLs involved in resistance to South American leaf blight (Microcyclus ulei) inthe rubber tree. Theor. Appl. Genet. 100: 975–984.Lespinasse, D., Rodier-Goud, M., Grivet, L., Leconte, A., Legnate, H. & Seguin, M. 1999. A saturatedlinkage map of rubber tree (Hevea spp.) based on RFLP, AFLP, microsatellite and isozymemarkers. Theor. Appl. Genet. 100: 127–138.Liepman, A.H., Wilkerson, C.G. & Keegstra, K. 2005. Expression of cellulose synthase-like (Csl)genes in insect cells reveals that CslA family members encode mannan synthases. Proc. Nat. Acad.Sci. USA 102: 2221–2226.Long, A.D. & Langley, C.H. 1999. The power of association studies to detect the contribution ofcandidate genetic loci to variation in complex traits. Gen. Res. 9: 720–731.MacKay, J., Bousquet, J., Cooke, J., Isabel, N., Pavy, N., Regan, S., Retzel, E. & Séguin, A. 2005.Functional genomic investigations of wood formation and defense response in trees, and beyond,Abstract W101. In Plant & Animal Genomes XII Conference, San Diego, CA, USA.Marcucci Poltri, S.N., Zelena, N., Rodriguez Traverso, J., Gelid, P. & Hopp, H.E. 2003. Selection ofa seed orchard of Eucalyptus dunnii based on genetic diversity criteria calculated using molecularmarkers. Tree Physiol. 23: 625–632.Marques, C.M., Araújo, J.A., Ferreira, J.G., Whetten, R., O’Malley, D.M., Liu, B.-H. & Sederoff, R.1998. AFLP genetic maps of Eucalyptus globulus and E. tereticornis. Theor. Appl. Genet. 96: 727–737.Marques, C.M., Vasquez-Kool, J., Carocha, V.J., Ferreira, J.G., O’Malley, D.M., Liu, B.-H. &Sederoff, R. 1999. Genetic dissection of vegetative propagation traits in Eucalyptus tereticornis andE. globulus. Theor. Appl. Genet. 99: 936–946.Marques, C.M., Brondani, R.P.V., Grattapaglia, D. & Sederoff, R. 2002. Conservation and syntenyof SSR loci and QTLs for vegetative propagation in four Eucalyptus species. Theor. Appl. Genet.105: 474–478.Mehta, P., Sivaprakash, K., Parani, M., Venkataraman, G. & Parida, A.K. 2005. Generation andanalysis of expressed sequence tags from the salt tolerant mangrove species Avicennia marina(Forsk.) Vierh. Theor. Appl. Genet. 110: 416–424.Moran, G.F., Thamarus, K.A., Raymond, C.A., Qiu, D., Uren, T. & Southerton, S. 2002. Genomicsof Eucalyptus wood traits. Ann. For. Sci. 59: 645–650.Morgante, M. & Salamini, F. 2003. From plant genomics to breeding practice. Curr. Opin. Biotech.14: 214–219.Mouradov, A., Glassick, T., Hamdorf, B., Murphy, L., Fowler, B., Marla, S. & Teasdale, R.D. 1998.NEEDLY, a Pinus radiata ortholog of FLORICAULA/LEAFY genes, expressed in both reproductiveand vegetative meristems. Proc. Nat. Acad. Sci. USA 95: 6537–6542.Mouradov, A., Hamdorf, B., Teasdale, R.D., Kim, J.T. Winter, K.U. & Theißen, G.A. 1999. DEF/GLO-like MADS-Box gene from a gymnosperm: Pinus radiata contains an ortholog of angiospermB class floral homeotic genes. Dev. Genet. 25: 245–252.Mukai, Y., Suyama, Y., Tsumura, Y., Kawahara, T., Yoshimaru, H., Kondo, T., Tomaru, N.,Kuramoto, N. & Murai, M. 1995. A linkage map for sugi (Cryptomeria japonica) based on RFLP,RAPD and isozyme loci. Theor. Appl. Genet. 90: 835–840.Myburg, A.A., Griffin, A.R., Sederoff, R.R. & Whetten, R.W. 2003. Comparative genetic linkagemaps of Eucalyptus grandis, Eucalyptus globulus and their F1 hybrid based on a double pseudobackcrossmapping approach. Theor. Appl. Genet. 107: 1028–1042.

302Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishNairn, C.J. & Haselkorn, T. 2005. Three loblolly pine CesA genes expressed in developing xylem areorthologous to secondary cell wall CesA genes of angiosperms. New Phytologist 166: 907–915.Neale, D.B. & Savolainen, O. 2004. Association genetics of complex traits in conifers. Trends PlantSci. 9: 325–330.Neale, D.B., Sewell, M.M. & Brown, G.R. 2002. Molecular dissection of the quantitative inheritanceof wood property traits in loblolly pine. Ann. For. Sci. 59: 595–605.Nelson, C.D., Nance, W.L. & Doudrick, R.L. 1993. A partial genetic linkage map of Slash pine (Pinuselliottii Englem var. elliottii) based on random amplified polymorphic DNAs. Theor. Appl. Genet.87: 145–151.Nikles, D.G. 1996. The first 50 years of the evolution of forest tree improvement in Queensland. InM.J. Dieters, A.C. Matheson, D.G. Nikles, C.E. Harwood, & S.M. Walker, eds. Tree improvementfor sustainable tropical forestry, pp. 51–64. Proc. QFRI-IUFRO Conference, Caloundra,Queensland, Australia.Paglia, G.P., Olivieri, A.M. & Morgante, M. 1998. Towards second-generation STS (sequence-taggedsites) linkage maps in conifers: a genetic map of Norway spruce (Picea abies K.). Mol. Gen. Genet.258: 466–478.Paoli, E.D. & Morgante, M. 2005. Nucleotide diversity and linkage disequilibrium in a set of candidategenes for time of bud set in Picea abies, Abstract W106. In Plant & Animal Genomes XIIConference, San Diego, CA, USA.Paux, E., Tamasloukht, M., Ladouce, N., Sivadon, P. & Grima-Pettenati, J. 2004. Identification ofgenes preferentially expressed during wood formation in Eucalyptus. Plant Mol. Biol. 55: 263–280.Paux, E., Carocha, V., Marques, C., Mendes de Sousa, A., Borralho, N., Sivadon, P. & Grima-Pettenati, J. 2005. Transcript profiling of Eucalyptus xylem genes during tension wood formation.New Phytol. 167: 89–100.Pavy, N., Laroche, J., Bousquet, J. & MacKay, J. 2005. Large-scale statistical analysis of secondaryxylem ESTs in pine. Plant Mol. Biol. 57: 203–224.Peter, G. & Neale, D. 2004. Molecular basis for the evolution of xylem lignification. Curr. Opin.Plant Biol. 7: 737–742.Plomion, C., O’Malley, D.M. & Durel, C.E. 1995. Genomic analysis in maritime pine (Pinus pinaster).Comparison of two RAPD maps using selfed and open-pollinated seeds of the sameindividual. Theor. Appl. Genet. 90: 1028–1034.Plomion, C., Bahrman, N., Costa, P., Dubos, C., Frigerio, J.-M., Lalanne, C., Madur, D., Pionneau,C. & Gerber, S. 2004. Proteomics for genetic and physiological studies in forest trees: applicationin maritime pine. In S. Kumar & M. Fladung, eds. Molecular genetics and breeding of forest trees,pp. 53–79. New York, USA, Haworth Press.Poke, F.S., Vaillancourt, R.E., Potts, B.M. & Reid, J.B. 2005. Genomic research in Eucalyptus.Genetica 125: 79–101.Pot, D., Eveno, E., Garnier-Géré, P. & Plomion, C. 2004. Association studies for wood properties inmaritime pine, Abstract W88. In Plant & Animal Genomes XII Conference, San Diego, CA, USA.Pritchard, J.K., Stephens, M., Rosenberg, N.A. & Donnelly, P. 2000. Association mapping in structuredpopulations. Am. J. Hum. Genet. 67: 170–181.Remington, D.L., Whetten, R.W., Liu, B.-H. & O’Malley, D.M. 1999. Construction of an AFLPgenetic map with nearly complete coverage in Pinus taeda. Theor. Appl. Genet. 98: 1279–1292.

Chapter 15 – Marker-assisted selection in forestry species 303Rönnberg-Wästljung, A.C., Glynn, C. & Weih, M. 2005. QTL analysis of drought tolerance andgrowth for a Salix dasyclados x Salix viminalis hybrid in contrasting water regimes. Theor. Appl.Genet. 110: 537–549.Rottmann, W.H., Meilan, R., Sheppard, L.A. Brunner, A.M., Skinner, J.S., Ma, C., Cheng,S., Jouanin, L., Pilate, G. & Strauss, S.H. 2000. Diverse effects of overexpression of LEAFYand PTFL, a poplar (Populus) homolog of LEAFY/FLORICAULA, in transgenic poplar andArabidopsis. Plant J. 22: 235–245.Rutledge, R., Regan, S., Nicolas, O., Fobert, P., Côté, C., Bosnich, W., Kauffeldt, C., Sunohara,G., Séguin, A. & Stewart, D. 1998. Characterization of an AGAMOUS homologue from theconifer black spruce (Picea mariana) that produces floral homeotic conversions when expressed inArabidopsis. Plant J. 15: 625–634.Saintagne, C.-S., Bodenes, C., Barreneche, T., Bertocchi, E., Plomion, C. & Kremer, A. 2004.Detection of quantitative trait loci controlling bud burst and height growth in Quercus robur L.Theor. Appl. Genet. 109: 1648–1659.Salvi, S., Tuberosa, R., Chiapparrino, E., Maccaferri, M., Veillet, S., van Beuningen, L. Isaac, P.,Edwards, K. & Phillips, R.L. 2002. Towards positional cloning of Vgt1, a QTL controlling thetransition from the vegetative to the reproductive phase in maize. Plant Mol. Biol. 48: 601–613.Scalfi, M., Troggio, M., Piovani, P., Leonardi, S., Magnaschi, G., Vendramin, G.G. & Menozzi, P.2004. A RAPD, AFLP and SSR linkage map, and QTL analysis in European beech (Fagus sylvaticaL.). Theor. Appl. Genet. 108: 433–441.Sedjo, R.A. 2004. Transgenic trees and trade: problems on the horizon? Resources 155: 8–13.Sewell, M.M. & Neale, D.B. 2000. Mapping quantitative traits in forest trees. In S.M. Jain & S.C.Minocha, eds. Molecular biology of woody plants, Vol. 1, pp. 407–423. Dordrecht, Netherlands,Kluwer.Sewell, M.M., Sherman, B.K. & Neale, D.B. 1999. A consensus map for loblolly pine (Pinus taedaL.). I. Construction and integration of individual linkage maps from two outbred three-generationpedigrees. Genetics 151: 321–330.Sewell, M.M., Bassoni, D.L., Megraw, R.A., Wheeler, N.C. & Neale, D.B. 2000. Identification ofQTLs influencing wood property traits in loblolly pine (Pinus taeda L.) I. Physical wood properties.Theor. Appl. Genet. 101: 1273–1281.Sewell, M.M., Davis, M.S., Tuskan, G.A., Wheeler, N.C., Elam, C.C., Bassoni, D.L. & Neale, D.B.2002. Identification of QTLs influencing wood property traits in loblolly pine (Pinus taeda L.) II.Chemical wood properties. Theor. Appl. Genet. 104: 214–222.Shena, M., Shalon, D., Davis, R.W. & Brown, P.O. 1996. Quantitative monitoring of gene expressionpatterns with a complementary DNA microarray. Science 270: 467–470.Shepherd, M., Chaparro, J.X. & Teasdale, R. 1999. Genetic mapping of monoterpene compositionin an interspecific eucalypt hybrid. Theor. Appl. Genet. 99: 1207–1215.Shepherd, M., Cross, M., Dieters, M.J. & Henry, R. 2003. Genetic maps for Pinus elliottii var. elliottiiand P. caribaea var. hondurensis using AFLP and microsatellite markers. Theor. Appl. Genet.106: 1409–1419.Skov, E. & Wellendorf, H. 1998. A partial linkage map of Picea abies clone V6470 based on recombinationof RAPD-markers in haploid megagametophytes. Silv. Genet. 47: 273–282.Smith, N.A., Singh, S.P., Wang, M.B., Stoutjesdijk, P.A., Green, A.G. & Waterhouse, P.M. 2000.Total silencing by intron-spliced hairpin RNAs. Nature 407: 319–320.

304Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishSoutherton, S.G., Strauss, S.H., Olive, M.R., Harcourt, R.L., Decroocq, V., Zhu, X., Llewellyn,D.J., Peacock, W.J. & Dennis, E.S. 1998. Eucalyptus has a functional equivalent of the Arabidopsisfloral meristem identity gene LEAFY. Plant Mol. Biol. 37: 897–910.Strabala, T.J. 2004. Expressed sequence tag databases from forestry tree species. In S. Kumar & M. Fladung,eds. Molecular genetics and breeding of forest trees, pp. 19–51. New York, USA, Haworth Press.Strauss, S.H., Lande, R. & Namkoong, G. 1992. Limitations of molecular-marker-aided selection inforest tree breeding. Can. J. For. Res. 22: 1050–1061.Strauss, S.H., Brunner, A.M., DiFazio, S.P. & James, R.R. 1998. Environmental effects of geneticallyengineered woody biomass crops. Biomass & Bioenergy 14: 403–414.Strauss, S.H., Brunner, A.M., Busov, V.B., Caiping, M.A. & Meilon R. 2004. Ten lessons from 15years of transgenic Populus research. Forestry 77: 455–465.Tandre, K., Albert, V.A., Sundas, A. & Engstrom, P. 1995. Conifer homologues to genes that controlfloral development in angiosperms. Plant Mol. Biol. 27: 69–78.Thamarus, K.A., Groom, K., Murrell, J., Byrne, M. & Moran G.F. 2002. A genetic linkage map forEucalyptus globulus with candidate loci for wood, fibre and floral traits. Theor. Appl. Genet. 104:379–387.Thamarus, K.A., Groom, K., Bradley, A., Raymond, C.A., Shimleck, L.R., Williams, E.R. &Moran G.F. 2004. Identification of quantitative trait loci for wood and fibre properties in two fullsibpedigrees of Eucalyptus globulus. Theor. Appl. Genet. 109: 856–864.Thumma, B.R., Nolan, M.F., Evans R. & Moran G.F. 2005. Polymorphisms in cinnamoyl CoAreductase (CCR) are associated with variation in microfibril angle in Eucalyptus spp. Genetics 171:1257–1265.Travis, S.E., Ritland, K., Whitham, T.G. & Keim, P. 1998. A genetic linkage map of pinyon pine (Pinusedulis) based on amplified fragment length polymorphisms. Theor. Appl. Genet. 97: 871–880.Tsarouhas, V., Gullberg, U. & Lagercrantz, U. 2002. An AFLP and RFLP linkage map and quantitativetrait locus (QTL) analysis of growth traits in Salix. Theor. Appl. Genet. 105: 277–288.Tsarouhas, V., Gullberg, U. & Lagercrantz, U. 2003. Mapping of quantitative trait loci controllingtiming of bud flush in Salix. Hereditas 138: 172–178.van Zyl, L., Bozhkov, P.V., Clapham, D.H., Sederoff, R.R. & von Arnold, S. 2003. Up, down andup again is a signal global gene expression pattern at the beginning of gymnosperm embryogenesis.Gene Expr. Patterns 3: 83–91.Verhaegen, D. & Plomion, C. 1996. Genetic mapping in Eucalyptus urophylla and Eucalyptus grandisusing RAPD markers. Genome 39: 1051–1061.Verhaegen, D., Plomion, C., Gion, J.-M., Poitel, M., Costa, P. & Kremer, A. 1997. Quantitative traitdissection analysis in Eucalyptus using RAPD markers: 1.Detection of QTL in interspecific hybridprogeny, stability of QTL expression across different ages. Theor. Appl. Genet, 95: 597–608.Voiblet, C., Duplessis, S., Encelot, N. & Martin, F. 2001. Identification of symbiosis-regulated genesin Eucalyptus globulus – Pisolithus tinctorius ectormycorrhiza by differential hybridization ofarrayed cDNAs. Plant J. 25: 181–191.Walter, C. & Killerby, S. 2004. A global study on the state of forest tree genetic modification.In Preliminary review of biotechnology in forestry including genetic modification. FAO ForestGenetic Resources Working Paper FGR/59E. Rome (available at www.fao.org/docrep/008/ae574e/ae574e00.htm).

Chapter 15 – Marker-assisted selection in forestry species 305Wang, H. 2004. The state of genetically modified trees in China. In Preliminary review of biotechnologyin forestry, including genetic modification, pp. 96–105. FAO Forest Genetic ResourcesWorking Paper FGR/59E. Rome (available at www.fao.org/docrep/008/ae574e/ae574e00.htm).Watson, J.M. & Brill, E.M. 2004. Eucalyptus grandis has at least two functional SOC 1-like floralactivator genes. Funct. Plant Biol., 31: 225–234.Weng, C., Kubisiak, T.L., Nelson, C.D. & Stine, M. 2002. Mapping quantitative trait loci controllingearly growth in a (longleaf pine x slash pine) x slash pine BC1 family. Theor. Appl. Genet. 104:852–859.Whetten, R.W., Sun, Y.H., Zhang, Y. & Sederoff, R. 2001. Functional genomics and cell wall biosynthesisin loblolly pine. Plant Mol. Biol. 47: 275–291.Wilcox, P.L., Carson, S.D., Richardson, T.E., Ball, R.D., Horgan, G.P. & Carter, P. 2001. Benefitcostanalysis of DNA marker-based selection in progenies of Pinus radiata seed orchard parents.Can. J. For. Res. 31: 2213–2224.Wu, R. & Zeng, Z-B. 2001. Joint linkage and linkage disequilibrium mapping in natural populations.Genetics 157: 899–909.Wu, H.X., Abarquez, A. & Matheson, A.C. 2004. Inbreeding in Pinus radiata V. The effects ofinbreeding on fecundity. Silv. Genet. 53: 80–87.Wu, L., Joshi, C.P. & Chiang, V.L. 2000. A xylem specific cellulose synthase gene from aspen(Populus tremuloides) is responsive to mechanical stress. Plant. J. 22: 495–502.Yazdani, R., Nilsson, J.E., Plomion, C. & Mathur, G. 2003. Marker trait association for cold acclimationand growth rhythm in Pinus sylvestris. Scand. J. For. Res. 18: 29–38.Yazdani, R., Yeh, F.C. & Rimsha, J. 1995. Genomic mapping of Pinus sylvestris (L.) using randomamplified polymorphic markers. For. Genet. 2: 109–116.Yin, T.M., DiFazio, S.P., Gunter, L.E., Riemenshneider, D. & Tuskan, G.A. 2004. Large scale heterospecificsegregation distortion in Populus revealed by a dense genetic linkage map. Theor. Appl.Genet. 109: 451–463.Yoshimaru, H., Ohba, K., Tsurumi, K., Tomaru, N., Murai, M., Mukai, Y., Suyama, Y., Tsumura,Y., Kawahara, T. & Sakamaki, Y. 1998. Detection of quantitative trait loci for juvenile growth,flower bearing and rooting ability based on a linkage map of sugi (Cryptomeria japonica). Theor.Appl. Genet. 97: 45–50.Yu, J. Hu, S., Wang, J., Wong, G.K-S., Li, S., Liu, B., Deng, Y., Dai, L., Zhou, Y., Zhang, X., Cao,M., Liu, J., Sun, J., Tang, J., Chen, Y., Huang, X., Lin, W., Ye, C., Tong, C., Cong, L., Geng, J.,Han, Y., Li, L., Li, W., Hu, G., Huang, X, Li, W., Li, J., Liu, Z., Li, L., Liu, J., Qi, O., Liu, J., Li,L., Li, T., Wang, X., Lu, H., Wu, T., Zhu, M., Ni, P., Han, H., Dong, W., Ren, X., Feng, X., Cui,P., Li, X., Wang, H., Xu, X., Zhai, W., Xu, Z., Zhang, J., He, S., Zhang, J., Xu, J., Zhang, K.,Zheng, X., Dong, J., Zeng, W., Tao, L., Ye, J., Tan, J., Ren, X., Chen, X., He, J., Liu, D., Tian, W.,Tian, C., Xia, H., Bao, O., Li, G., Gao, H., Cao, T., Wang, J., Zhao, W., Li, P., Chen, W., Wang,X., Zhang, Y., Hu, J., Wang, J., Liu, S., Yang, J., Zhang, G., Xiong, Y., Li, Z., Mao, L., Zhou,C.,Zhu, Z., Chen, R., Hao, B., Zheng, W., Chen, S., Guo, W., Li, G., Liu, S., Tao, M., Wang, J., Zhu,L., Yuan, L. & Yang, H. 2002. A draft sequence of the rice genome (Oryza sativa L. ssp. indica).Science 296: 79–92.Zik, M. & Irish, V.F. 2003. Flower development: initiation, differentiation, and diversification. Annu.Rev. Cell Dev. Biol. 19: 119–140.

Section VMarker-assisted selectionin fish – case studies

Chapter 16Possibilities for marker-assistedselection in aquaculturebreeding schemesAnna K. Sonesson

310Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishSummaryFAO estimates that there are around 200 species in aquaculture. However, only a fewspecies have ongoing selective breeding programmes. Marker-assisted selection (MAS) isnot used in any aquaculture breeding scheme today. The aim of this chapter, therefore, isto review briefly the current status of aquaculture breeding schemes and to evaluate thepossibilities for MAS of aquaculture species. Genetic marker maps have been publishedfor some species in culture. The marker density of these maps is, in general, rather lowand the maps are composed of many amplified fragment length polymorphism (AFLP)markers anchored to few microsatellites. Some quantitative trait loci (QTL) have beenidentified for economically important traits, but they are not yet mapped at a high density.Computer simulations of within-family MAS schemes show a very high increase ingenetic gain compared with conventional family-based breeding schemes, mainly due tothe large family sizes that are typical for aquaculture breeding schemes. The use of geneticmarkers to identify individuals and their implications for breeding schemes with controlof inbreeding are discussed.

Chapter 16 – Possibilities for marker-assisted selection in aquaculture breeding schemes 311IntroductionAquaculture is an expanding industry,with a total global value of US$61 billion(FAO, 2003). FAO estimates that thereare around 200 species in culture, ofwhich carps and oysters have the largestworldwide production. However, only afew species have ongoing selective breedingprogrammes.MAS is not used in any aquaculturebreeding scheme today. The aim of thischapter, therefore, is to review briefly thecurrent status of aquaculture breedingschemes and to evaluate the possibilities forMAS of aquaculture species.Traits of breeding interestGrowth rate is the most important trait formost aquaculture species under selection. Itis recorded on the selection candidates, andcan easily be improved using mass selection.Sexual maturation is a trait that leadsto reduced growth, reduced feed conversionefficiency and reduced fillet qualityin several aquaculture species. Therefore,selection against early maturation is oftenperformed, i.e. the individuals that becomesexually mature before market size are discardedas selection candidates. In tilapia,late maturity is desirable because of excessivespawning that results in overcrowdingof ponds and reduced size of the fish.For many other traits, however, accuratemeasurement techniques for live individualsare inadequate. Hence, selectionmust be based on information from otherfamily members, e.g. on siblings. MASwould be especially valuable for traits thatare difficult and/or costly to measure onthe selection candidate or for traits thatare measured late in life or at slaughter(Lande and Thompson, 1990; Poompuangand Hallerman, 1997). Examples of theseimportant traits are:• Disease resistance. Challenge tests existfor viral (e.g. white spot syndrome inshrimps and infectious pancreatic necrosisin most marine fishes) and bacterial(e.g. furunculosis and vibriosis) diseases,as well as for parasites (e.g. sea lice).When challenge tests are used in breedingprogrammes, however, survivingindividuals cannot enter the breedingnucleus because of the risk that theywill introduce the disease to the nucleus.Therefore, these individuals cannot beselection candidates and only their sibs,who have no records for these traits, arecandidates.• Fillet quality traits. To this group of traitsbelong colour, texture, gaping, differentfat-related traits (e.g. fat percentage anddistribution) and dressing percentage.Accurate measurements of these traits areavailable only for slaughtered individuals.Techniques for measuring fillet colour onlive fish are under development.• Feed conversion efficiency is a trait thatcan be measured practically only at thefamily level at a young age in the breedingnucleus, but not at the individual levelor in grow-out operations. The valueof such early records is rather limitedbecause of the unknown correlation withfeed conversion efficiency at a later age.Feed intake is, in general, a difficult traitto measure in aquaculture species due tounequal feed intake over days. No activeselection programme for aquaculturespecies selects directly for feed conversionefficiency; rather, indirect selectionis practised by selecting for growth.• Salinity and low temperature toleranceare two traits of interest for tilapia breedingprogrammes. Today, tilapias are producedin freshwater in tropical and subtropicalareas. The purpose of selectingfor salinity and temperature tolerance

312Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishis to develop fish that could reproduceand grow in areas of higher salinity andlower temperature, i.e. to increase theproduction area for tilapias. Althoughthese traits could be measured early inthe life of the fish and therefore couldbe selected, sexually mature fish mayrespond differently to the temperatureand salinity conditions.Gjedrem and Olesen (2005) provide amore complete list of aquaculturally relevanttraits and their heritabilities andcorrelations.Structure of breeding schemesMost aquaculture species are currently bredin mono- or polyculture systems (i.e. withone or several species reared together) usingmass selection for growth rate. Only about30 family-based breeding programmesworldwide utilize sib information in theestimation of breeding values (B. Gjerde,personal communication). The main partof a family-based breeding programme isa closed breeding nucleus, with trait informationfrom sibs coming from test stations.Breeding programmes for species with limitedreproductive ability (e.g. salmonids asopposed to several highly fecund marinespecies such as Atlantic cod) have a multiplierunit, where genetic material fromthe nucleus is used to produce eggs or fryfor the grow-out producers. The limitingfactor for the breeding nucleus is often thenumber of tanks, where the fry of a certainfull-sib family are kept until individuals arelarge enough to be physically tagged. Thenumber of offspring per full-sib family islarge, such that a very high selection intensitycan be achieved. Generally, each male ismated to two females in order that the tankeffect can be estimated separately from theadditive genetic effects.The high intensity of selection withinthe nucleus can easily result in high rates ofinbreeding. Introduction of unrelated wildstock is often practised to reduce the rates ofinbreeding. However, introduction of wildstock also leads to reduced genetic gain,and should generally be avoided in ongoingbreeding schemes. Optimum contributionis an approach that maximizes genetic gainwhile restricting the rates of inbreeding forschemes with discrete (Meuwissen, 1997;Grundy, Villanueva and Woolliams, 1998)or overlapping (Meuwissen and Sonesson,1998; Grundy, Villanueva and Woolliams,2000) generation structures or for traitswith a polygenic effect and the effect ofa known single gene (Meuwissen andSonesson, 2004). The key determinationis the number of matings (full-sib families)per selected individual. One practicalconstraint in some marine species is thatmatings are volitional (natural mating in,for example, a tank) and thus depend on theavailability of individuals ready to spawn ata certain moment. Hence, for these species,the number of matings per male or femaleis restricted for each spawning. The use offrozen milt makes the use of males moreflexible. Milt from many species includingsalmonids, carp and shrimps can be frozen(Stoss, 1983), but the practical use of cryopreservedsperm in aquaculture breedingprogrammes has not been fully utilized.Genetic marker mapsA genetic marker map is an ordered listingof the genes or molecular markers occurringalong the length of the chromosomesin the genome. Distances between genesor markers are estimated in terms of howfrequently recombination occurs betweenthem. Genetic marker maps are available forsome aquaculture species (Table 1). Most ofthese genetic maps are constructed usingamplified fragment length polymorphism

Chapter 16 – Possibilities for marker-assisted selection in aquaculture breeding schemes 313(AFLP) markers (Vos et al., 1995), whichare generally anchored to a smallercollection of microsatellites. The markerdensity of the maps is currently rather low,and the markers are spread unevenly overthe genome, which may explain why thenumber of linkage groups found in, forexample, channel catfish (Waldbieser et al.,2001) or white shrimp (Pérez et al., 2004)does not correspond to the number ofchromosomes. In rainbow trout, tetraploidyhas been found for 20 chromosome arms(Sakamoto et al., 2000). Recombinationrates can differ between males and females,and hence marker map lengths can differconsiderably between males and females.In salmonids, females have the higherrecombination rate, which implies that mostinformation comes from the females whenconstructing the marker map. The ratiobetween female and male recombinationrates was 3.25:1.00 for rainbow trout(Sakamoto et al., 2000), 1.69:1.00 for Arcticchar (Woram et al., 2004) and 8.25:1.00for Atlantic salmon (Moen et al., 2004a).However, in other aquaculture species,males have the higher recombination rate.For example, the ratio between male andfemale recombination rates was 7.4:1.00 inJapanese flounder (Coimbra et al., 2003).There are also differences in recombinationrate over the length of the chromosomes inmales, i.e. recombination rate was higher intelomeric regions than in proximal regionsin rainbow trout (Sakamoto et al., 2000).Mapping of QTLQTL are loci whose variability underliesvariation in expression of a quantitativecharacter (Geldermann, 1975). Detectionof QTL would help in understanding thegenetic architecture of the trait, i.e. thenumbers and relative effects of genes thatdetermine expression of a trait. A small,Table 1Aquaculture species for which there are geneticmarker mapsSpeciesReferenceScallop Li et al. (2005)Wang et al. (2004)Pacific oyster Hubert and Hedgecock (2004)Eastern oyster Yu and Guo (2003)White shrimp Pérez et al. (2004)Kuruma prawn Li et al. (2003)Black tiger shrimp Wilson et al. (2002)Kuruma prawn Moore et al. (1999)Atlantic salmon Moen et al. (2004a)Gilbey et al. (2004)Arctic char Woram et al. (2004)Rainbow trout Nichols et al. (2003)Sakamoto et al. (2000)Young et al. (1998)Salmonids May and Johnson (1990)Common carp Sun and Liang (2004)European sea bass Chistiakov et al. (2005)Channel catfish Waldbieser et al. (2001)Liu et al. (2003)Tilapia Lee et al. (2005)McConnell et al. (2000)Agresti et al. (2000)Kocher et al. (1998)Japanese flounder Coimbra et al. (2003)but growing, number of QTL for importanttraits have been identified in farmedaquatic species (Table 2). In tilapias, QTLhave been identified for cold tolerance(Cnaani et al., 2003; Moen et al., 2004b).QTL for upper temperature tolerance(Jackson et al., 1998; Danzmann, Jacksonand Ferguson, 1999; Perry, Ferguson andDanzmann, 2003; Somorjai, Danzmannand Ferguson 2003), and for resistance todifferent disease traits have been foundin salmonids, e.g. for infectious hematopoieticnecrosis virus (Rodriguez et al.,2005), infectious pancreatic necrosis virus(Ozaki et al., 2001), Ceratomyxa shasta(Nichols, Bartholomew and Thorgaard,2003) and infectious salmon anemia (Moenet al., 2004c, 2006). QTL for general diseaseresistance and immune response have beenfound in tilapias (Cnaani et al., 2004) and

314Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishTable 2Known marker-QTL linkages in aquaculture speciesTraitReferenceSalmonidsSpawning time Leder, Danzmann and Ferguson (2006)Early development Martinez et al. (2005)Pyloric caeca number Zimmerman et al. (2005)Natural killer cell-like activity Zimmerman et al. (2004)Hematopoetic necrosis resistance Rodriguez et al. (2005)Development rate Sundin et al. (2005)Infectious salmon anemia resistance Moen et al. (2004c, 2006)Ceratomyxa shasta resistance Nichols, Bartholomew and Thorgaard (2003)Infectious pancreatic necrosis resistance Ozaki et al. (2001)Infectious hematopoietic necrosis resistance Khoo et al. (2004)Body weight and condition factor Reid et al. (2005)Spawning date and body weight O’Malley et al. (2003)Growth and maturation Martyniuk et al. (2003)Temperature tolerance Somorjai, Danzmann and Ferguson (2003)Meristic traits Nichols, Wheeler and Thorgaard (2004)Embryonic development Robison et al. (2001)Albinism Nakamura et al. (2001)Development rate Nichols et al. (2000)Spawning time Sakamoto et al. (1999)Upper temperature tolerance, size Perry et al. (2001, 2003)Upper temperature tolerance Danzmann, Jackson and Ferguson (1999)Upper temperature tolerance Jackson et al. (1998)TilapiaCold toleranceMoen et al. (2004b)Cold tolerance and fish size Cnaani et al. (2003)Stress and immune response Cnaani et al. (2004)Colour pattern Streelman, Albertson and Kocher (2003)Early survival Palti et al. (2002)Sex determination Lee, Penman and Kocher (2003); Lee, Hulata and Kocher (2004)salmonids (Zimmerman et al., 2004). Thedata used for quantifying disease resistanceand temperature tolerance traits are oftenbased on challenge tests, for which modelsaccounting for non-normality of data andspecial algorithms that take account ofcensored data (survival models) are usedin combination with the QTL mappingmethods (e.g. Moen et al., 2006). In salmonids,QTL have been found related tobody weight and size (Martyniuk et al.,2003; O’Malley et al., 2003; Reid et al.,2005), for colouration pattern (Streelman,Albertson and Kocher, 2003) and for oneform of albinism (Nakamura et al., 2001).Zimmerman et al. (2005) found QTL forpyloric caeca number, a trait related tofeed conversion efficiency. Epistasis hasbeen found for upper temperature toleranceand body length in rainbow trout(Danzmann, Jackson and Ferguson, 1999;Perry, Ferguson and Danzmann, 2003); theepistasis depended on the genetic background,which would result in reducedeffectiveness of MAS (Danzmann, Jacksonand Ferguson, 1999).The genetic diversity of wild and culturedpopulations, high fecundity, and thepossibility of interspecific hybridization andreproductive manipulation of aquaculturespecies can be exploited in QTL mappingstudies. These features have resulted in a

Chapter 16 – Possibilities for marker-assisted selection in aquaculture breeding schemes 315wide diversity of experimental populationsbeing used in such studies. Double haploids(see below) have been used for QTLmapping in salmonids (by androgenesis;Robison et al., 2001; Zimmerman et al.,2005) and tilapias (by gynogenesis; Paltiet al., 2002). Backcross populations havebeen set up where strains with large phenotypicdifferences in the trait of interestare crossed, e.g. for temperature tolerancein rainbow trout (Danzmann, Jackson andFerguson, 1999). The strains can come fromone species, but crosses between speciesalso have been used (Streelman, Albertsonand Kocher, 2003 for tilapia; Rodriguez etal., 2005 for salmonids). Finally, familiesderived from a breeding nucleus have beenused for QTL mapping. (e.g. Moen et al.,2006 for Atlantic salmon).Most QTL mapping is based on markerassociation studies (e.g. Sakamoto et al.,1999) or on a marker association studyfollowed by interval mapping (Moen etal., 2006). Combined linkage/linkage disequilibriummethods (e.g. Meuwissen etal., 2002) have high precision for mappingQTL in outbred populations, but have notbeen used for QTL mapping in aquaculturespecies. This lack could be explained by thesparcity of genetic marker maps for the speciesunder study and by many of the studieshaving used special crosses as mentionedabove, where linkage is the main source ofinformation for mapping the QTL.Various reproductive manipulationsmay be applied to aquaculture species.One interesting reproductive manipulationtechnique for outbred populations formarker and QTL mapping is gyno- andandrogenetic double haploids (Chourrout,1984). In gynogenesis, the sperm’s chromosomesare inactivated by irradiation andfollowing fertilization, diploidy is restoredby applying a temperature or hydrostaticpressure shock. The result is an individualthat is double haploid with only the female’schromosomes. Depending on when diploidyis restored, two types of doublehaploid individuals can be produced. Ifan early shock is applied, extrusion ofthe second polar body is suppressed and,when the two maternal chromosome setsfuse, some heterozygosity is retained. Ifa late shock is applied, the first mitoticcleavage of the zygote is suppressed and,when the two maternal chromosome setsfuse, the resulting individual is virtuallyhomozygous. In androgenesis, the egg isirradiated and, after “fertilization”, the eggis shocked to suppress the first mitosis(Parsons and Thorgaard, 1984). The resultis an individual that is a double haploidwith only the male’s chromosomes and thatis virtually homozygous.The power of an experiment to detectQTL depends on the effect of QTL alleles,the recombination rate among the markerand QTL loci, and the sample size of themapping population. The effect of QTLgenotypes is higher for double haploid thanfor full- or half-sib family designs in anoutbred population because the QTL genotypesoccur only in a homozygous form indouble haploids (i.e. in their most extremeform). The relative advantage of a populationof mitotic double haploids, where thetwo chromosome sets are copies of eachother, is largest when the QTL has a smalleffect (Martinez, Hill and Knott, 2002). Inmeiotic haploid individuals, the two chromosomesets in an individual have beenrecombined. The power of QTL detectionin these meiotic double haploid populationsis therefore expected to resemble thatof selfed populations (Martinez, Hill andKnott, 2002). Double haploids have beenused for genetic marker and QTL mapping,as noted above.

316Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishOn the other hand, the extremely largefull-sib family size that is possible foraquaculture species may make use of specialreproductive techniques for markerand QTL mapping studies unnecessary.For example, in Atlantic cod, both malesand females produce millions of gametes atspawning. Also, Atlantic cod and halibutare examples of repeat spawners, in contrastto salmonids that normally die after a singlespawning. Repeated spawning makes themating structure more flexible, e.g. a certainpairing can be repeated or an individualmay be used in multiple pairings. Anotherdisadvantage of using double haploids isthat they are fully inbred individuals andmay therefore express the trait of interestdifferently than non-inbred animals.An example of this is the environmentalvariance for wing length in Drosophilamelanogaster, which has been shown tobe larger for inbred individuals than noninbredindividuals (Falconer and Mackay,1996). Traits with dominant expressionmay be expressed differently because ofinbreeding depression.MAS schemesAfter QTL detection experiments, breederswill have knowledge of marker-QTL linkagesand an estimate of the respective effectsof QTL alleles on the trait in the population.This knowledge may be appliedusing MAS of spawners for producing thenext generation. Generally, QTL detectionwould be carried out in one experimentand MAS in another (Poompuang andHallerman, 1997). For within-family MASschemes, the phase of marker and QTLalleles needs to be established for all familieson which selection will be practised.Two within-family MAS schemes have beenwell studied for livestock populations. Thefirst is a three-generation scheme, which issuitable for breeding schemes with progenytesting (Kashi, Hallerman and Solomon,1990). Progeny tests are not, however, currentlyperformed upon aquaculture species.The second is a two-generation scheme(Mackinnon and Georges, 1998), whichmay be modified for application to fishpopulations.In the two-generation scheme (Figure 1),it is assumed that both sires and dams havegenotypic records for markers linked tothe QTL and that there are two groupsof progeny from these parents. The groupof test progeny has both phenotypic performanceand genotypic records, while thegroup of selection candidates only hasgenotypic records. Phenotypic evaluationoften implies that these individuals cannotbe used later for breeding, perhaps becausethey were used in a challenge test or wereslaughtered to obtain sib information forcarcass traits. The genetic markers of thesire are denoted M1 and M2, and thoseof the dam M3 and M4, with M1 and M3linked with the performance-increasingQTL alleles. With these linkage relationships,M1M3-bearing progeny would beselected while some M1M4- and M2M3-,and no M2M4-bearing progeny would beselected. The proportions of each genotypeselected would vary depending upon selectionintensity.When QTL are mapped densely (upto 5 cM between markers), both linkagedisequilibrium within families and population-widelinkage disequilibrium can beused in the MAS scheme (Smith and Smith,1993; Dekkers and Hospital, 2002).Simulation of two-generation withinfamilyMAS schemesThe attractive feature of MAS is the potentialfor increasing the genetic gain in aselective breeding programme. Lande and

Chapter 16 – Possibilities for marker-assisted selection in aquaculture breeding schemes 317Figure 1Two-generation marker evaluation and selection scheme for aquaculture speciesParental generation: with genotypic record♂ ( M1 M2 ) x ( M3 M4 )♀Progeny generation:Group 1: Group 2:Test- progeny with phenotypicand genotypic records, usedto estimate marker effectsBroodstock candidateprogeny with genotypic recordonlyM1 M3SelectM1 M4 Perhaps select someM2 M3Perhaps select someM2 M4 Select noneIndividuals with marker genotypes shown in bold are selected. The proportion of each genotype selected will vary withselection intensity.Thompson (1990) showed that the efficiencyof MAS relative to conventionalselection alone depends upon the heritabilityof the trait under selection, theproportion of genetic variance associatedwith marker loci and the particular selectionscheme at issue.Stochastic simulations of two-generationwithin-family MAS schemes in a typicalaquaculture breeding programme werecarried out by Sonesson (2006) for selectionfor one trait. Selection was truncatedbased upon best linear unbiased prediction(BLUP) breeding values including informationof the genetic marker (Fernando andGrossman, 1989). The heritability, h 2 , ofthe trait under selection was 0.06 or 0.12,and 20 percent of the genetic variance wasaccounted for by the QTL.Genetic gain was 0.202 for MAS and0.176 for conventional breeding, after selectionin generation 1 (Table 3), i.e. MAS had15 percent higher genetic gain than conventionalbreeding. After selection in generation2, genetic gain was 68 percent higher forMAS than conventional breeding. The performance-increasingQTL allele was thenalmost fixed (i.e. its frequency approached1). The increase in genetic gain is mainlydue to the increased frequency of the positiveQTL allele for the MAS scheme, wherea higher QTL frequency implies more

318Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishTable 3Genetic gain in schemes with heritability (h 2 ) of 0.06 or 0.12Generation number1 2 3 4h 2 = 0.06Conventional 0.176 0.121 0.134 0.117MAS 0.202 0.203 0.135 0.114h 2 = 0.12Conventional 0.337 0.206 0.196 0.191MAS 0.361 0.318 0.206 0.177genetic variance (as long as the frequencyof the positive allele is less than 0.5). Forsituations with a higher h 2 of 0.12, geneticgain was 7 percent higher after selection ingeneration 1 and 54 percent higher afterselection in generation 2, i.e. the superiorityof MAS was somewhat lower for schemeswith the higher heritability.Identification of individualsusing genetic markersIn traditional family-based breeding programmes,individuals from the same full-sibfamilies are reared separately (e.g. in tanks)until they are large enough to be taggedphysically. This mode of rearing is verycostly, and the number of full-sib familiestherefore limits the size of the breedingnucleus. In addition, separate rearing offull-sib families results in common environmental(tank) effects that need to beestimated, which in turn affects other partsof the design and analysis of the breedingprogramme. That is, in the mating design, asire has to be mated to several dams (or viceversa) in order to be able to separate analyticallythese common environmental effectsfrom additive genetic effects. The tank effectis generally higher for newly domesticatedspecies, where feed and other environmentaleffects are not yet standardized.For example, the tank effect was 3–12 percentfor juvenile Atlantic cod (Gjerde etal., 2004), and the nursery pond effect was32 percent for rohu carp (Gjerde et al.,2003) compared with, for example, a tankeffect of 2–6 percent for Atlantic salmonand rainbow trout (Rye and Mao, 1998;Pante et al., 2002). Were it possible to poolprogeny groups into one tank, tank effectswould be eliminated, a smaller number oftanks would be needed per spawner andmore pairs could be spawned.Parentage assignment using molecularmarkers is useful for tracing pedigrees inbreeding programmes, and can be used toidentify parents in breeding schemes whereprogeny groups are pooled. Parentageassignment implies that parents and offspringare all genotyped for a numberof genetic markers that are well spreadover the genome and that the informationon genotypes is used to assign individualprogeny to the correct parental pair. Theparent-offspring relationship is such thateach offspring inherits one allele from eachparent, making it possible to exclude possibleparents when this condition is notfulfilled.Exclusion of parents is the basic methodof assigning parents. The exclusion probabilityper locus (E l ) can be calculatedaccording to the formulae of Hanset (1975)and Dodds et al. (1996). The global exclusionprobability over loci (E g ) is:E g = 1 - Π = (1 -= 1llLE l)

Chapter 16 – Possibilities for marker-assisted selection in aquaculture breeding schemes 319where L is the total number of marker lociscreened. Genotyping errors can result inthe true parents being excluded, becauseuncertainty is not accounted for with thismethod. Even a low error rate reduces thecorrect assignment rate, which increaseswith the number of loci and number ofalleles. SanCristobal and Chevalet (1997)derived a likelihood and a Bayesian-basedmethod that could take account of genotypingerrors. They used the likelihoodmethod on data for 50 parental pairs andtheir offspring and a genotyping errorrate of 2 percent, but with no error rateaccounted for in the likelihood calculation.The correct parentage assignment ratewas only 88 percent using a system of fiveloci with five alleles, and 83 percent using asystem of eight loci with five alleles. Afterincluding a genotyping error rate of 10 -3 inthe same calculations, the correct assignmentrate increased to nearly 100 percent.The number of loci and alleles per locineeded for correct parent assignment wasestimated deterministically and validatedstochastically by Villanueva, Verspoorand Visscher (2002). They showed thatnine loci with five alleles per locus or sixloci with ten alleles would assign 99 percentof offspring to the correct parentsfrom 100 or 400 crosses. Similar resultswere found by Bernatchez and Duchesne(2000). These results agree well with theresults from empirical studies of aquaculturepopulations. For example, Herbingeret al. (1995) assigned 66 percent of theoffspring to correct parental pairs in acomplete factorial cross between ten malesx ten females (i.e. 100 parental pairs) ofrainbow trout using four loci with fourto ten alleles per locus. Perez-Enriquez,Takagi and Taniguchi (1999) reported73 percent correct assignment of parentalpairs using five microsatellites when 7 800parental pairs were possible for a populationof red sea bream. Norris, Bradley andCunningham (2000) assigned over 95 percentof the parental pairs correctly usingeight polymorphic loci with 10–29 allelesper locus in Atlantic salmon when thenumber of possible parental pairs was over12 000. Jackson, Martin-Robichaud andReith (2003) assigned 96–100 percent of theprogeny to the correct parental pair in F 1Atlantic halibut populations using five microsatelliteloci. Castro et al. (2004) assignedover 99 percent of all parental pairs correctlywith six microsatellites for 176 full-sibfamilies of turbot. Vandeputte et al. (2004)assigned 95.3 percent of all parental pairsin a complete factorial cross of 10 damsx 24 sires of common carp using eightmicrosatellites. In addition to assessing parentage,full- and half-sib relationships havealso been estimated using genetic markersin aquaculture stocks, e.g. Atlantic salmon(Norris, Bradley and Cunningham, 2000)and rainbow trout (McDonald, Danzmannand Ferguson, 2004).There are underlying assumptions thataffect experimental power for assigningparents. Some of these assumptions are:• Hardy-Weinberg equilibrium (HWE).Small effective population sizes, nonrandommating and unequal family sizewill lead to deviations from HWE. HWEis not an issue in assigning parental pairsto offspring, but affects the assignmentof offspring genotypes to the parents(Estoup et al., 1998), i.e. some genotypesof parents might be more difficult thanothers from which to assign offspring.If these genotypes are present in largefamilies, parental assignment rate will bereduced relative to what would occur ifthey are present in small families.• Zero mutation rate and no scoring errors.Castro et al. (2004) reported a mutation

320Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishrate of 6.7 x 10 -4 when 13 464 gametes andmarker information from 12 loci werescreened in a turbot population. Mutationsand scoring error have the sameeffect on excluding potential parents, givingrise to incorrect assignments.• Unlinked loci and linkage equilibrium.Linkage and linkage disequilibriumbetween the loci will reduce the effectivenumber of loci used for the parentalassignment. Note that the power to assignparental pairs correctly can thereby differbetween different sets of microsatellites.Estoup et al. (1998) quantified the differencein the power of microsatellite markersets used to assign parents correctly inturbot (eight loci, eight alleles per loci)and rainbow trout (eight loci, four allelesper loci) populations by calculating thefrequency of good and unique parentassignments (f gu ). The more variable setof microsatellites resulted in higher f gu forlarger maximal mating schemes for turbotthan the less variable set of microsatellitesfor rainbow trout. In general, a set of lociwith an equal number of alleles has thehighest exclusion probability (Weir, 1996;Jamieson and Taylor, 1997).Walk-back selection schemesPractical breeding schemes using molecularmarker-based parental assignment havebeen reported. Doyle and Herbinger (1994)proposed carrying out parentage assignmentfor individuals using genetic markers,such that full-sib families would not have tobe kept separately until tagging but, rather,would be held in one large tank. Note thatindividuals that are genotyped also need tobe physically tagged, so that genotypingresults can be traced back to particularindividuals. At the time for selection, fishin the tank were first ranked on their phenotypicvalue, assuming that selection wasfor only one trait that could be measuredon the selection candidates (e.g. weight).Then the individual with the highest phenotypicvalue was selected and genotypedfor family identification. Thereafter, theindividual with the second highest phenotypicvalue was genotyped and selected if itwas not a full- or half-sib of other, alreadyselectedindividuals, such that within-familyselection was performed. This procedurecontinued until the desired number ofbrood stock was selected. This approachof progressing through the performanceranking, genotyping and selecting the bestperformingindividuals within families wastermed “walk-back” selection. Matingssubsequently would be made betweenfamilies, a strategy preferred because itwould keep the rate of inbreeding low(Falconer and Mackay, 1996). Herbingeret al. (1995) reported setting up a rainbowtrout breeding programme using the walkbackselection programme of Doyle andHerbinger (1994) and genetic markers toestimate full/half relationships among thecandidates using a likelihood ratio methodand thereafter performing within-familyselection.Using stochastic simulations, Sonesson(2005) studied a combination of optimumcontribution selection and walk-back selection.Optimum contribution is a selectionmethod that maximizes genetic gain witha restriction on the rate of inbreeding(see earlier in this chapter). Hence, thecombination of optimum contribution andwalk-back selection ensures that the rateof inbreeding is under control, while thegenetic gain is higher than in the withinfamilyselection schemes of Doyle andHerbinger (1994) because selection is bothwithin and between families. In the studyby Sonesson (2005), batches of candidateswere pre-selected from a single tank on

Chapter 16 – Possibilities for marker-assisted selection in aquaculture breeding schemes 321their phenotypic values and the batch sizevaried from 50 to all (1 000, 5 000 or 10 000)candidates. Relatively small batches of fishwere genotyped at any one time to minimizegenotyping costs. Thereafter, BLUPbreeding values (Henderson, 1984) wereestimated and the optimum contributions ofthe candidates calculated using the methodof Meuwissen (1997). If the constraint onthe rate of inbreeding could not be achieved,another batch of fish was genotyped andincluded in the total number of candidates.Results showed that with a batch size of100, 76–92 percent of the genetic gain wasachieved compared with schemes where all1 000, 5 000 or 10 000 fish were genotypedto provide candidates for the optimum contributionselection algorithm. Hence, highgenetic gain was achieved at a fixed rate ofinbreeding with low genotyping costs.The main practical advantage of thesemarker-assisted breeding schemes is thatexpenses associated with separate rearingof full-sib families are not incurred, whichdecreases start-up and operational costs forthe breeding scheme. The most importanttrait at the start of a breeding programmeis probably growth, which can easily bemeasured on the candidate. Use of theoptimum contribution selection algorithmkeeps the rate of inbreeding under control,which is especially important in breedingprogrammes for aquaculture species whereselection intensity can be very high due tothe large family sizes. In combination withBLUP estimated breeding values, whichhave a high within-family correlation suchthat individuals with the highest breedingvalues will tend to come from only a fewfamilies, high rates of inbreeding can result(Sonesson, Gjerde and Meuwissen, 2005).However, there remain unsolved issueswith the combined optimum contributionand walk-back selection method:• Biased BLUP breeding values lead toa reduction in accuracy of selection,because not all selection candidates areincluded in the estimation of breedingvalues.• Low and unequal survival of families maylead to reduced genetic variation and thusincreased rate of inbreeding. However,the optimum contribution selection willcorrect for some of this loss by selectingfrom more families to keep the geneticbase broader than when selecting onlyfor the BLUP estimated breeding values.One option for reducing this problem ofunequal and low survival is to pool anequal number of individuals from eachfamily after the main period of earlymortality is over. In general, it is possibleto use more parents in these programmescompared with conventional familybasedselection programmes, which couldcompensate for some of the loss of familiescontributing to the next generation dueto low and unequal survival.• Multitrait selection is probably the largestpractical problem to solve. One alternativecould be to base the pre-selection ononly one or two traits that are inexpensiveto measure on the candidate. Techniquesfor measuring more traits on live selectioncandidates are steadily evolving (e.g.fat content in Atlantic salmon, Solberg etal., 2003), such that the sib-testing systemmight be unnecessary for these traits inthe future.Introgression schemesMany fish breeding schemes have beenstarted with a relatively narrow geneticbase, selecting for only one or two traits inrelatively few animals. However, becauseall farmed aquatic species still have wildancestors, introgression of genes (i.e. identifiedgenes or QTL) from these wild

322Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishancestors into breeding populations is possible.However, one assumes that all othertraits of the wild fish are unwanted in thebreeding population, such that only theparticular gene of interest should be introgressed,leaving the genome of the breedingpopulation otherwise intact (Hospital andCharcosset, 1997). In white shrimp, forexample, wild stocks have been found tohave higher disease resistance but lowergrowth rates than stocks in culture. In thisexample then, only the genes for diseaseresistance should be introgressed (takingaccount of the possible effect of these geneson growth). The problem of actually implementingmarker-assisted introgression inpopulations is to find the trait of interest inwild populations at a reasonable cost, andthen to identify genes or marked QTL forthe trait to be introgressed (Visscher, Haleyand Thompson, 1992; Koudande et al.,2000). This is a costly and time-consumingprocess, especially for species with longgeneration intervals. Methods for simultaneousQTL mapping and introgressionwould be useful.AcknowledgementsI am grateful to Bjarne Gjerde, EricHallerman and Thomas Moen for theirhelp with the manuscript.ReferencesAgresti, J.J., Seki, S., Cnaani, A., Poompuang, S., Hallerman, E.M., Umiel, N., Hulata, G., Gall,G.A.E. & May, B. 2000. Breeding new strains of tilapia: development of an artificial center of originand linkage map based on AFLP and microsatellite loci. Aquaculture 185: 43–56.Bernatchez, L. & Duchesne, P. 2000. Individual-based genotype analysis in studies of parentage andpopulation assignment: how many loci, how many alleles? Can. J. Fish Aquat. Sci. 57: 1–12.Castro, J., Bouza, C., Presa, P., Pino-Querido, A., Riaza, A., Ferreiro, I., Sánchez, L. & Martínez,P. 2004. Potential sources of error in parentage assessment of turbot (Scophthalamus maximus)using microsatellite loci. Aquaculture 242: 119–135.Chistiakov, D.A., Hellemans, B., Haley, C., Law, A.S., Tsigenopoulos, C.S., Kotulas, G., Bertotto,D., Libertini, A. & Volckaert, F.A.M. 2005. A microsatellite linkage map of the European sea bassDicentrarchus labrax L. Genetics 170: 1821–1826.Chourrout, D. 1984. Pressure-induced retention of second polar body and suppression of firstcleavage in rainbow trout: production of all-triploids, all-tetraploids, and heterozygous andhomozygous gynogenetics. Aquaculture 36:111–126.Cnaani, A., Hallerman, E.M., Ron, M., Weller, J.I., Indelman, M., Kashi, Y., Gall, G.A.E. &Hulata, G. 2003. Detection of a chromosomal region with two quantitative trait loci, affecting coldtolerance and fish size, in an F2 tilapia hybrid. Aquaculture 223: 117–128.Cnaani, A., Zilberman, N., Tinman, S. & Hulata, G. 2004. Genome-scan analysis for quantitativetraits in an F 2 tilapia hybrid. Mol. Gen. Genomics 272: 162–172.Coimbra, M.R.M., Kabayashi, K., Koretsugu, S., Hasegawa, O., Ohara, E., Ozaki, S., Sakamoto,T., Naruse, K. & Okamoto, N. 2003. A genetic map of the Japanese flounder Paralichthys olivaceus.Aquaculture 220: 203–218.Danzmann, R.G., Jackson, T.R. & Ferguson, M.M. 1999. Epistasis in allelic expression at upper temperaturetolerance QTL in rainbow trout. Aquaculture 173: 45–58.Dekkers, J.C.M. & Hospital, F. 2002. The use of molecular genetics in the improvement of agriculturalpopulations. Nature Revs. Genet. 3: 22–32.

Chapter 16 – Possibilities for marker-assisted selection in aquaculture breeding schemes 323Dodds, K.G., Tate, M.L., McEwan, J.C. & Crawford, A.M. 1996. Exclusion probabilities for pedigreetesting farm animals. Theor. Appl. Genet. 92: 966–975.Doyle, R.W. & Herbinger, C.M. 1994. The use of DNA fingerprinting for high-intensity withinfamilyselection in fish breeding. Proc. 5th Wrld. Congr. Genetics Appl. Livest. Prodn. pp. 364–371.Guelph, Ontario. Department of Animal and Poultry Science, University of Guelph.Estoup, A., Gharbi, K., SanCristobal, M., Chevalet, C., Haffray, P. & Guyomard, R. 1998Parentage assignment using microsatellites in turbot (Scophthalmus maximus) and rainbow trout(Oncorhynchus mykiss) hatchery populations. Can. J. Fish. Aquat. Sci. 55: 715–725.Falconer, D.S. & Mackay, T.F.C. 1996. Introduction to quantitative genetics, 4th edition. Harlow, UK,Longman Publishing.FAO. 2003. FAOSTAT statistical database. (available at http://faostat.fao.org).Fernando, R.L. & Grossman, M. 1989. Marker-assisted selection using best linear biased prediction.Genet. Sel. Evol. 21: 467–477.Geldermann, H. 1975. Investigations on inheritance of quantitative characters in animals by genemarkers. I. Methods. Theor. Appl. Genet. 46: 319–330.Gilbey, A., Verspoor, E., McLay, A. & Houlihan, D. 2004. A microsatellite linkage map for Atlanticsalmon (Salmo salar). Anim. Genet. 35: 98–105.Gjedrem, T. & Olesen, I. 2005. Basic statistical parameters. In T. Gjedrem, ed. Selection and breedingprograms in aquaculture, pp. 45–72. Doordrecht, Netherlands, Springer.Gjerde, B., Mahapatra, K.D., Reddy, P.V.G.K., Saha, J.N., Jana, R.K. & Rye, M. 2003. Genetic andphenotypic parameters for growth in rohu (Labeo rohita) in mono- and polyculture productionsystems. In Proc. 8th Internat. Symp. Genet. in Aquaculture, p. 85. Puerto Varas, Chile.Gjerde, B., Terjesen, B.F., Barr, Y., Lein, I. & Thorland, I. 2004. Genetic variation for juvenile growthand survival in Atlantic cod (Gadus morhua). Aquaculture 236: 167–177.Grundy, B., Villanueva, B. & Woolliams, J.A. 1998. Dynamic selection procedures for constrainedinbreeding and their consequences for pedigree development. Genet. Res. 72: 159–168.Grundy B., Villanueva B. & Woolliams J. 2000. Dynamic selection for maximizing response withconstrained inbreeding in schemes with overlapping generations. Anim. Sci. 70 373–382.Hanset, R. 1975. Probabilité d’exclusion de paternité et de monozygotie, probabilité de similtude.Généralisation à N alleles co-dominants. Ann. Med. Vet. 119: 71–80.Henderson, C.R. 1984. Applications of linear models in animal breeding. Guelph, Ontario,Canada,Guelph Univ. Press.Herbinger, C.M., Doyle, R.W., Pitman, E.R., Paquet, D., Mesa, K.A., Morris, D.B., Wright, J.M.& Cook, D. 1995. DNA fingerprint-based analysis of paternal and maternal effects on offspringgrowth and survival in communally reared rainbow trout. Aquaculture 137: 245–256.Hubert, S. & Hedgecock, D. 2004. Linkage maps of microsatellite DNA markers for the Pacificoyster Crassostrea gigas. Genetics 168: 351–362.Hospital, F. & Charcosset, A. 1997. Marker-assisted introgression of quantitative trait loci. Genetics147: 1469–1485.Jackson, T.R., Ferguson, M.M., Danzmann, R.G., Fishback, A.G., Ihssen, P.E., O’Connell, M. &Crease, T.J. 1998. Identification of two QTL influencing upper temperature tolerance in threerainbow trout (Onchorhynchus mykiss) half-sib families. Heredity 80: 143–151.Jackson, T.R., Martin-Robichaud, D.J. & Reith, M.E. 2003. Application of DNA markers to the managementof Atlantic halibut (Hippoglossus hippoglossus) broodstock. Aquaculture 220: 245–259.

324Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishJamieson, A. & Taylor, S.C.S. 1997. Comparisons of three probability formulae for parentage exclusion.Anim. Genet. 28: 397–400.Kashi, Y., Hallerman, E. & Soller, M. 1990. Marker-assisted selection of candidate bulls for progenytesting programs. Anim. Prod. 51: 63–74.Khoo, S.K., Ozaki, A., Nakamura, F., Arakawa, T., Ishimoto, S., Nikolov, R., Sakamoto, T.,Akatsu, T., Mochuzuki, M., Denda, I. & Okamoto, N. 2004. Identification of a novel chromosomalregion associated with infectious hematopoietic necrosis (IHN) resistance in rainbow trout,Oncorhynchus mykiss. Fish Path. 39: 95–101.Kocher, T.D., Lee, W.J., Sobolewska, H., Penman, D. & McAndrew, B. 1998 A genetic linkage mapof a cichlid fish, the tilapia (Oreochromis niloticus). Genetics 148: 1225–1232.Koudandé, O.D., Iraqi, F., Thomson, P.C., Teale, A.J. & van Arendonk, J.A.M. 2000. Strategies tooptimize marker-assisted introgression of multiple linked QTL. Mamm. Genome 11:145–150.Lande, R. & Thompson, R. 1990. Efficiency of marker-assisted selection in the improvement ofquantitative traits. Genetics 124:743–756.Leder, E.H., Danzmann, R.G. & Ferguson, M.M. 2006. The candidate gene, Clock, localizes to astrong spawning time quantitative trait locus region in rainbow trout. J. Heredity 97: 74–80.Lee, B.-Y., Penman, D.J. & Kocher, T.D. 2003. Identification of the sex-determining region in tilapia(Oreochromis niloticus) using bulked segregant analysis. Anim. Genet. 34: 379–383.Lee, B.-Y., Hulata, G. & Kocher, T.D. 2004. Two unlinked loci control the sex of blue tilapia(Oreochromis aureus). Heredity 92: 543–549.Lee, B.-Y., Lee, W.J., Streelman, T., Carleton, K.L., Howe, A.E., Hulata, G., Slettan, A., Stern, J.E.,Terai, Y. & Kocher, T.D. 2005. A second-generation genetic linkage map of tilapia (Oreochromisspp.). Genetics 170: 237–244.Li, Y., Byrne, K., Miggiano, E., Whan, V., Moore, S., Keys, S., Crocos, P., Preston, N. & Lehnert,S. 2003. Genetic mapping of the kuruma prawn (Penaeus japonicus) using AFLP markers.Aquaculture 219: 143–156.Li, L., Xiang, J., Liu, X., Zhang, Y., Dong, B. & Zhang, X. 2005. Construction of AFLP-basedgenetic linkage map for Zhikong scallop, Chlamys farreri Jones et Preston and mapping of sexlinkedmarkers. Aquaculture 245: 63–73.Liu, Z., Karsi, A., Li, P., Cao, D. & Dunham, R. 2003. An AFLP-based genetic linkage map ofchannel catfish (Ictalurus punctatus) constructed by using an interspecific hybrid resource family.Genetics 165: 687–694.Mackinnon, M.J. & Georges, M.A.J. 1998. Marker-assisted preselection of young dairy sires prior toprogeny-testing. Livest. Prod. Sci. 54: 229–250.Martinez, V.A., Hill, W.G. & Knott, S.A. 2002. On the use of double haploids for detecting QTL inoutbred populations. Heredity 88: 423–431.Martinez, V., Thorgaard, G., Robison, B. & Sillanpää, M.J. 2005. An application of Bayesian QTLmapping to early development in double haploid lines of rainbow trout including environmentaleffects. Genet. Res. Camb. 86: 209–221.Martyniuk, C.J., Perry, G.M.L., Mogahadam, H.K., Ferguson, M.M. & Danzmann, R.G. 2003.The genetic architecture of correlations among growth-related traits and male age at maturation inrainbow trout. J. Fish Biol. 63: 746–764.May, B. & Johnson, J.E. 1990. Composite linkage map of salmonid fishes. In S.J. O’Brien, ed.Genetic maps, 5th edition, pp. 4.151–4.159. Cold Spring Harbor, NY, USA, Cold Spring HarborLaboratory Press.

Chapter 16 – Possibilities for marker-assisted selection in aquaculture breeding schemes 325McConnell, S.K.J., Beynon, C., Leamon, J. & Skibinski, D.O.F. 2000. Microsatellite marker-basedgenetic linkage maps of Oreochromis aureus and O. niloticus (Cichlidae): extensive linkage groupsegment homologies revealed. Anim. Genet. 31: 214–216.McDonald, G.J., Danzmann, R.G. & Ferguson, M.M. 2004. Relatedness determination in theabsence of pedigree information in three cultured strains of rainbow trout (Oncorhynchus mykiss).Aquaculture 233: 65–78.Meuwissen, T.H.E. 1997. Maximizing the response of selection with a predefined rate of inbreeding.J. Anim. Sci. 75: 934–940.Meuwissen, T.H.E. & Sonesson, A.K. 1998. Maximizing the response of selection with a predefinedrate of inbreeding – overlapping generations. J. Anim. Sci. 76: 2575–2583.Meuwissen, T.H.E. & Sonesson, A.K. 2004. Genotype-assisted optimum contribution selection tomaximise selection response over a specified time period. Genet. Res. 84: 109–116.Meuwissen, T.H.E., Karlsen, A., Lien, S., Olsaker, I. & Goddard, M.E. 2002. Fine mapping of aquantitative trait locus for twinning rate using combined linkage and linkage disequilibrium mapping.Genetics 161: 373–379.Moen, T., Hoyheim, B., Munck, H. & Gomez-Raya, L. 2004a. A linkage map of Atlantic salmon(Salmo salar) reveals an uncommonly large difference in recombination rate between the sexes.Anim. Genet. 35: 81–92.Moen, T., Agresti, J.J., Cnaani, A., Moses, H., Famula, T.R., Hulata, G., Gall, G.A.E. & May, B.2004b. A genome scan of a four-way tilapia cross supports the existence of a quantitative trait locusfor cold tolerance on linkage group 23. Aquaculture 35: 893–904.Moen, T., Fjalestad, K.T., Munck, H. & Gomez-Raya, L. 2004c. A multistage testing strategy fordetection of quantitative trait loci affecting disease resistance in Atlantic salmon. Genetics 167:851–858.Moen, T., Hayes, B., Sonesson, A.K., Lien, S., Høyheim, B., Munck, H. & Meuwissen, T.H.E. 2006.Mapping of a quantitative trait loci for resistance against infectious salmon anemia in Atlanticsalmon (Salmo salar): comparing survival models with binary models. Genetics. Submitted.Moore, S.S., Whan, V., Davis, G.P., Byrne, K., Hetzel, D.J.S. & Preston, N. 1999. The developmentand application of genetic markers for the kuruma prawn (Penaeus japonicus). Aquaculture 173:19–32.Nakamura, K., Ozaki, A., Akutsu, T., Iwai., K., Sakamoto, T., Toshizaki, G. & Okamoto, N. 2001.Genetic mapping of the dominant albino locus in rainbow trout (Oncorhynchus mykiss). Mol.Genet. Genom. 265: 687–93.Nichols, K.M., Bartholomew, J. & Thorgaard, G.H. 2003. Mapping multiple genetic loci associatedwith Ceratomyxa shasta resistance in Oncorhynchus mykiss. Dis. Aquat. Org. 56: 145–154.Nichols, K.M., Wheeler, P.A. & Thorgaard, G.H. 2004. Quantitative trait loci analyses for meristictraits in Oncorhynchus mykiss. Env. Biol. Fish. 69: 317–331.Nichols, K.M., Robison, B.D., Wheeler, P.A. & Thorgaard, G.H. 2000. Quantitative trait loci (QTL)associated with developmental rate in clonal Oncorhynchus mykiss strains. Aquaculture 209:233.Nichols, K.M., Young, W.P., Danzmann, R.G., Robison, B.D., Rexroad, C., Noakes, M., Phillips,R.B., Bentzen, P., Spies, I., Knudsen, K., Allendorf, F.W., Cunningham, B.M., Brunelli, J.,Zhang, H., Ristow, S., Drew, R., Brown, K.H., Wheeler, P.A. & Thorgaard, G.H. 2003. A consolidatedlinkage map for rainbow trout (Oncorhynchus mykiss). Anim. Genet. 34: 102–115.Norris, A.T., Bradley, D.G. & Cunningham, E.P. 2000. Parentage and relatedness determination infarmed Atlantic salmon (Salmo salar) using microsatellite markers. Aquaculture 182: 73–83.

326Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishO’Malley, K.G., Sakamoto, T., Danzmann, R.G. & Ferguson, M.M. 2003. Quantitative trait loci forspawning date and body weight in rainbow trout: testing for conserved effects across ancestrallyduplicated chromosomes. J. Hered. 94: 273–84.Ozaki, A., Sakamoto, T., Khoo, S., Nakamura, K., Coimbra, M.R.M., Akutsu, T. & Okamoto,N. 2001. Quantitative trait loci (QTLs) associated with resistance/susceptibility to infectious pancreaticnecrosis virus (IPNV) in rainbow trout (Oncorhynchus mykiss). Mol. Genet. Genom. 265:23–31.Palti, Y., Shirak, A., Cnaani, A., Hulata, G., Avtalion, R.R. & Ron, M. 2002. Detection of genes withdeleterious alleles in an inbred line of tilapia (Oreochromis aureus). Aquaculture 206: 151–164.Pante, M.J.R., Gjerde, B., McMillan, I. & Misztal, I. 2002. Estimation of additive and dominancegenetic variances for body weight at harvest in rainbow trout, Oncorhynchus mykiss. Aquaculture204: 383–392.Parsons, J.E. & Thorgaard, G.H. 1984. Induced androgenesis in rainbow trout. J. Exp. Zool. 231:407–412.Pérez, F., Erazo, C., Zhinaula, M., Volckaert, F. & Calderon, J. 2004. A sex-specific linkage map of thewhite shrimp Penaeus (Litopenaeus) vannamei based on AFLP markers. Aquaculture 242: 105–118.Perez-Enriquez, R., Takagi, M. & Taniguchi, N. 1999. Genetic variability and pedigree tracing of ahatchery-reared stock of red sea bream (Pagrus major) used for stock enhancement, based on microsatelliteDNA markers. Aquaculture 173: 413–423.Perry, G.M., Ferguson, M.M. & Danzmann, R.G. 2003. Effects of genetic sex and genomic backgroundon epistasis in rainbow trout (Oncorhynchus mykiss). Genetica 119: 35–50.Perry, G.M., Danzmann, R.G., Ferguson, M.M. & Gibson, J.P. 2001. Quantitative trait loci forupper thermal tolerance in outbred strains of rainbow trout (Oncorhynchus mykiss). Heredity 86:333–341.Poompuang, S. & Hallerman, E.M. 1997. Toward detection of quantitative trait loci and markerassistedselection in fish. Rev. Fisheries Sci. 5: 253–277.Reid, D.P., Szanto, A., Glebe, B., Danzmann, R.G. & Ferguson, M.M. 2005. QTL for body weightand condition factor in Atlantic salmon (Salmo salar): comparative analysis with rainbow trout(Oncorhynchus mykiss) and Arctic charr (Salvelinus alpinus). Heredity 94: 166–172.Robison, B.D., Wheeler, P.A., Sundin, K., Sikka, P. & Thorgaard, G.H. 2001. Composite intervalmapping reveals a major locus influencing embryonic development rate in rainbow trout(Oncorhynchus mykiss). J. Hered. 92: 16–22.Rodriguez, M.F., LaPatra, S., Williams, S., Famula, T. & May, B. 2005. Genetic markers associatedwith resistance to infectious hematopoietic necrosis in rainbow and steelhead trout (Oncorhynchusmykiss) backcrosses. Aquaculture 241: 93–115.Rye, M. & Mao, I.L. 1998. Nonadditive genetic effects and inbreeding depression for body weight inAtlantic salmon (Salmo salar). Livest. Prodn. Sci. 57: 15–22.Sakamoto, T., Danzmann, R.G., Okamoto, N., Ferguson, M.M. & Ihssen, P.E. 1999. Linkageanalysis of quantitative trait loci associated with spawning time in rainbow trout (Oncorhynchusmykiss). Aquaculture 173: 33–43.Sakamoto, T., Danzmann, R.G., Gharbi, K., Howard, P., Ozaki, A., Khoo, S.K., Woram, R.A.,Okamoto, N., Ferguson, M.M., Holm, L.-E., Guyomard, R. & Hoyheim, B. 2000. A microsatellitelinkage map of rainbow trout (Oncorhynchus mykiss) characterized by large sex-specificdifferences in recombination rates. Genetics 155: 1331–1345.

Chapter 16 – Possibilities for marker-assisted selection in aquaculture breeding schemes 327SanCristobal, M. & Chevalet, C. 1997. Error tolerant parent identification from a finite set of individuals.Genet. Res. Camb. 70: 53–62.Smith, C. & Smith, D.B. 1993. The need for close linkages in marker-assisted selection for economicmerit in livestock. Anim. Breed. Abstr. 61: 197–204.Solberg, C., Saugen, E., Swenson, L.P., Bruun, L. & Isaksson, T. 2003. Determination of fat in livefarmed Atlantic salmon using non-invasive NIR techniques. J. Sci. Food Agr. 83: 692–696.Somorjai, I.M., Danzmann, R.G. & Ferguson, M.M. 2003. Distribution of temperature tolerancequantitative trait loci in Arctic charr (Salvelinus alpinus) and inferred homologies in rainbow trout(Oncorhynchus mykiss). Genetics 165: 1443–1456.Sonesson, A.K. 2005. A combination of walk-back and optimum contribution selection for fish – asimulation study. Genet. Sel. Evol. 37: 587–599.Sonesson, A.K. 2006. Marker-assisted selection in breeding schemes for aquaculture species. Genet.Sel. Evol. (in press).Sonesson, A.K., Gjerde, B. & Meuwissen, T.H.E. 2005. Truncation selection for BLUP-EBV andphenotypic values in fish breeding schemes. Aquaculture 243: 61–68.Stoss, J. 1983. Fish gamete preservation and spermatozoan physiology. In W.S. Hoar, D.J. Randall, & E.M.Donaldson, eds. Fish physiology, vol. 9, part B. pp. 305–350. New York, NY, USA, Academic Press.Streelman, J.T., Albertson, R.C. & Kocher, T.D. 2003. Genome mapping of the orange blotch colourpattern in cichlid fishes. Molec. Ecol. 12: 2465–2471.Sun, X. & Liang, L. 2004. A genetic linkage map of common carp (Cyprinus carpio L.) and mappingof a locus associated with cold tolerance. Aquaculture 238: 165–172.Sundin, K., Brown, K.H., Drew, R.E., Nichols, K.M., Wheeler, P.A. & Thorgaard, G.H. 2005.Genetic analysis of development rate QTL in backcrosses of clonal rainbow trout. Aquaculture247: 75–83.Vandeputte, M., Kocour, M., Mauger, S., Dupont-Nivet, M., De Guerry, D., Rodina, M., Gela,D., Vallod, D., Chevassus, B. & Linhart, O. 2004. Heritability estimates for growth-relatedtraits using microsatellite parentage assignment in juvenile common carp (Cyprinus carpio L.).Aquaculture 235: 233–236.Villanueva, B., Verspoor, E. & Visscher, P.M. 2002. Parental assignment in fish using microsatellitegenetic markers with finite numbers of parents and offspring. Anim. Genet. 33: 33–41.Visscher, P.M., Haley, C.S. & Thomson, R. 1992. Marker-assisted introgression in backcross breedingprograms. Genetics 144: 1923–1932.Vos, P., Hogers, R., Bleeker, M., Reijans, M., vanden Lee, T., Hornes, M., Fritters, A., Pot, J.,Paleman, J., Kuiper, M. & Zabeau, M. 1995. AFLP: a new technique for DNA fingerprinting.Nucl. Acids Res. 23: 4407–4414.Waldbieser, G.C., Bosworth, B.G., Nonneman, D.J. & Wolters, W.R. 2001. A microsatellite-basedgenetic linkage map for channel catfish (Ictalurus punctatus). Genetics 158: 727–734.Wang, S., Bao, Z., Pan, J., Zhang, L., Yao, B., Zhan, A., Bi, K. & Zhang, Q. 2004. AFLP linkage mapof an intraspecific cross in Chlamys farreri. J. Shellfish Res. 23: 491–499.Weir, B.S. 1996. Genetic data analysis, 2nd edition. Sunderland, MA, USA., Sinauer Associates, Inc.Wilson, K., Li, Y., Whan, V., Lehnert, S., Byrne, K., Moore, S., Pongsomboon, S., Tassanakajon, A.,Rosenberg, G., Ballment, E., Fayazi, A., Swan, J., Kenway, M. & Benzie, J. 2002. Genetic mappingof the black tiger shrimp (Penaeus monodon) with amplified fragment length polymorphism.Aquaculture 204: 297–309.

328Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishWoram, R.A., McGowan, C., Stout, J.A., Gharbi, K., Ferguson, M.M., Hoyheim, B., Sakamoto,J., Davidson, W., Rexroad, C. & Danzmann, R.G. 2004. A genetic linkage map for Arctic char(Salvelinus alpinus): evidence for higher recombination rates and segregation distortion in hybridversus pure strain mapping parents. Genome 47: 304–315.Young, W.P., Wheeler, P.A., Coryell, V.H., Keim, P. & Thorgaard, G.H. 1998. A detailed map ofrainbow trout produced using double haploids. Genetics 148: 839–850.Yu, Z.N. & Guo, X.M. 2003. Genetic linkage map of the eastern oyster (Crassostrea virginicaGmelin). Biol. Bull. 204: 327–338.Zimmerman, A.M., Evenhuis, J.P., Thorgaard, G.H. & Ristow, S.S. 2004. A single major chromosomalregion controls natural killer cell-like activity in rainbow trout. Immunogen. 55: 825–35.Zimmerman, A.M., Wheeler, P.A., Ristow, S.S. & Thorgaard, G.H. 2005. Composite interval mappingreveals three QTL associated with pyloric caeca number in rainbow trout, Oncorhynchusmykiss. Aquaculture 247: 85–95.

Chapter 17Marker-assisted selection in fishand shellfish breeding schemesVictor Martinez

330Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishSummaryThe main goals of breeding programmes for fish and shellfish are to increase the profitabilityand sustainability of aquaculture. Traditionally, these have been carried outsuccessfully using pedigree information by selecting individuals based on breeding valuespredicted for traits measured on candidates using an “animal model”. This methodologyassumes that phenotypes are explained by a large number of genes with small effectsand random environmental deviations. However, information on individual genes withmedium or large effects cannot be used in this manner. In selective breeding programmesusing pedigree information, molecular markers have been used primarily for parentageassignment when tagging individual fish is difficult and to avoid causing common environmentaleffects from rearing families in separate tanks. The use of these techniques in suchconventional breeding programmes is discussed in detail.Exploiting the great biological diversity of many fish and shellfish species, differentexperimental designs may use either chromosomal manipulations or large family sizesto increase the likelihood of finding the loci affecting quantitative traits, the so-calledQTL, by screening the segregation of molecular markers. Using information on identifiedloci in breeding schemes in aquaculture is expected to be cost-effective compared withtraditional breeding methods only when the accuracy of predicting breeding values israther low, e.g. for traits with low heritability such as disease resistance or carcass quality.One of the problems facing aquaculture is that some of the resources required to locateQTL accurately, such as dense linkage maps, are not yet available for the many species.Recently, however, information from expressed sequence tag (EST) databases has beenused for developing molecular markers such as microsatellites and single nucleotidepolymorphisms (SNPs). Marker-assisted selection (MAS) or genome-wide marker-assistedselection (G-MAS) using linkage disequilibrium within families or across populations arenot widely used in aquaculture, but their application in actual breeding programmes isexpected to be a fertile area of research. This chapter describes how genomic tools can beused jointly with pedigree-based breeding strategies and the potential and real value ofmolecular markers in fish and shellfish breeding schemes.

Chapter 17 – Marker-assisted selection in fish and shellfish breeding schemes 331IntroductionThe main goals of fish and shellfish breedingprogrammes are to increase the profitabilityand sustainability of productionenterprises, while maintaining genetic variabilityin the cultured stock. Traditionally,selective breeding has targeted traits suchas body weight that can be easily improvedusing mass selection. Relatively few studieshave analysed other traits that are economicallyimportant. However, disease resistanceand carcass quality are traits that are difficultto measure on candidates for selection,but have major effects on the productionefficiency and profitability of many speciesin aquaculture.When developing efficient breedingprogrammes, pedigree information isrequired to maximize effective populationsizes and to use information from relativesto increase the accuracy of predictingbreeding values for all traits included inthe breeding objective. In most commercialapplications, pedigree information islacking; therefore, markers can be used toinfer relatedness between individuals, withor without parental information. Severalissues need to be considered on a case-bycasebasis when applying such molecularinformation for increasing the profitabilityof breeding programmes in practice.For traits that are difficult to measureon candidates for selection, prediction ofbreeding value has to rely on measurementson relatives. Under these circumstances,the accuracy of predicted breeding values(and thus, response) is lower than whenrecords are obtained directly on candidatesfor selection. In addition, there is anincreased probability of co-selecting relatives.It is especially for these traits thatmolecular markers that directly affect orare linked to quantitative trait loci (QTL)have been regarded as useful for markerassistedselection (MAS) or gene-assistedselection (GAS) programmes.This chapter begins by discussing thestatus of “conventional” breeding programmes,the challenges involved whenstarting such programmes for new speciesand the possibilities of incorporatingmarker information in “conventional” programmes.An outline is then provided ofthe molecular markers developed for aquaculturespecies and of their use for geneticanalysis. The main features of designs forQTL mapping, including the use of chromosomalmanipulations, are described,followed by a discussion of the prospectsand challenges of GAS or MAS for diseaseor carcass traits. Finally, new genomic toolsare considered briefly.Breeding programmes andresponse to selectionManagement of modern breeding programmesin aquaculture requires usingpedigree information to carry out soundand efficient statistical evaluations (usingbest linear unbiased prediction [BLUP]methodology). This approach enablesbreeders to maximize genetic gain whilelimiting rates of inbreeding to acceptablelevels (Meuwissen, 1997; Toro and Mäki-Tanila, 1999).Most of the genetic improvement infish and shellfish species to date has beenmade through the use of traditional selectivebreeding (reviewed by Hulata, 2001).Well-designed breeding programmes haveshown substantial response to selectionfor body weight, e.g. Atlantic salmon, 10–14 percent (Gjøen and Bentsen, 1997). Inrainbow trout, rates of genetic gain variedfrom 8 percent for indirect selection forbody weight at sea (Kause et al., 2005)to 13 percent for direct selection (Gjerde,1986). The response to selection was about

332Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fish10 percent for body weight in a breedingprogramme for coho salmon (CMG-IFOP)initially funded by FAO (Martinez andHidalgo, unpublished data), and a similarresponse was obtained for this species inthe United States of America (Hershbergeret al., 1990). Estimates for tilapia followlargely the same trend, with a response ofabout 10 percent (Ponzoni et al., 2005). Incommon carp, responses to selection forbody weight were inconsistent between upselectedand down-selected lines, althoughexhaustion of additive genetic variationfor increased growth rate, genotype-byenvironmentinteraction, or competitioneffects could not be ruled out (Moav andWohlfarth, 1976). In oysters, asymmetricalresponse to selection for body weight wasfound (Toro et al., 1995; Ward, English andMcGoldrick, 2000).Although responses to selection havenot been well documented, significant estimatesof genetic parameters have beenobtained for carcass traits (Gjerde andSchaeffer, 1989; Kause et al., 2002; Quinton,McMillan and Glebe, 2005) and diseaseresistance (Gjøen et al., 1997; Henryon etal., 2002, 2005). Rates of genetic gain areexpected to be lower for these traits thanfor body weight because breeding valuepredictions rely solely on measurementsfrom relatives.Several breeding programmes have beeninitiated recently for new aquaculture species,such as mussels, scallops, Artemia andshrimp. The biology of these species posesinteresting avenues for the design of conventionalbreeding programmes, taking intoaccount factors such as self-fertilization,intrafamily competition, cannibalism, lackof methods for physical tagging, and matingpreferences. For example, competition canaffect the expression of quantitative traitsdue to co-variances among members of agroup managed together in a pond or tankand, if not considered properly, this effectcan seriously affect the rates of response toselection (Muir, 2005). However, this effectcan be included explicitly in the model ofanalysis using the co-variance among membersof a group, the so-called “associativeeffects” from other genotypes in the group.The theory of Griffing (1967) for BLUPevaluation was developed in the context oftree breeding, but deserves further investigationin the analysis of fish and shellfishbreeding. This may be especially true forspecies taken recently from the wild orthose that show cannibalistic behaviour.Another recent example is the developmentof scallop breeding programmes.Argopecten purpuratus is a simultaneouslyhermaphroditic species. In the first breedingphase, the scallop liberates sperm, afterwhich the eggs are expelled. To decreasethe level of self-fertilization, it is customaryto use only the last pulses of eggs. Thissystem reduces rates of self-fertilizationto 20 percent (A. Vergara, personal communication),but a residual proportion ofeggs are still already fertilized with spermfrom the same individual. As this processoccurs within the reproductive tract, itis not possible to detect which individualsare selfed or outcrossed, althoughthe rate of residual self-fertilization varieswidely among families and produces biasedestimates of heritability (Martinez and diGiovanni, 2006). Information from molecularmarkers can be of benefit under thesecircumstances (see below).DNA markers used in aquacultureMutations in the genome create geneticvariability (or polymorphism), whichis reflected as allelic diversity of molecularmarkers. While genomic sequencingwould greatly facilitate the development

Chapter 17 – Marker-assisted selection in fish and shellfish breeding schemes 333of molecular markers, the many species inaquaculture would make this a costly task(Liu and Cordes, 2004). Hence, a varietyof approaches have been taken to developgenetic markers for aquaculture species.Dominantly-expressed markers havebeen used extensively in aquaculture studies.Amplified fragment length polymorphism(AFLP) markers (Vos et al., 1995) providea cost-effective alternative for specieswhere DNA sequencing is not under wayor when there are restricted resources forQTL mapping. Dominant AFLP markersare preferred over random amplified polymorphicDNA (RAPD) markers becausethey are more reproducible both in otherlines or populations and in other laboratories(e.g. Nichols et al., 2003), and theycan generate hundreds of markers (a singlepolymerase chain reaction commonly generatesover ten markers). Furthermore,heterozygotes can often be distinguishedfrom homozygotes using the fluorescentband intensity (Piepho and Koch, 2000;Jansen et al., 2001).Microsatellite markers are simplesequence repeats (SSRs) arranged in tandemarrays scattered throughout the genome,both within known genes and in anonymousregions. Microsatellite markers areused increasingly in aquaculture species(reviewed by Liu and Cordes, 2004), dueto their elevated polymorphic informationcontent (PIC), co-dominant mode ofexpression, Mendelian inheritance, abundanceand broad distribution throughoutthe genome (Wright and Bentzen, 1994).Microsatellites are generally Type IImarkers, which are associated with genomicregions that have not been annotatedto known genes (O’Brien, 1991). Othermolecular markers can be distinguished asType I markers, which are linked to genes(of known function). Type I markers aremore desirable because they are generallymore conserved across evolutionarilydistant organisms, enabling comparativegenomics, assessment of genome evolutionand candidate gene analysis.Two procedures are used to generatemicrosatellite markers. The first uses agenomic library enriched with microsatellite-bearingsequences to generate clonesthat bear specific SSRs. These clones arethen sequenced to identify microsatellite-bearingsequences and then to designprimers to amplify the regions with specificSSR. Validation is required to studythe level of polymorphism and the numberof null alleles, and to identify any loci thatare duplicates due to any recent evolutionarygenome duplication event givingrise to multiple copies of loci in the haploidgenome (Coulibaly et al., 2005). This isdone by screening a sample of individualsfrom the target population.Many laboratories have been workingon developing expressed sequence tags(ESTs) derived from complementary DNA(cDNA) libraries for a variety of fish andshellfish species (Panitz et al., 2002; Riseet al., 2004a; Hayes et al., 2004; Rexroad etal., 2005; A. Alcivar-Warren, personal communication).EST sequences can be usedfor marker development in species wherethe full genome is not currently beingsequenced. The cDNA libraries are constructedusing messenger RNA (mRNA)that was expressed in different tissues, suchas kidney and gills. The expressed fragmentsof sequence data are not the full sequence ofa known gene, but what was incorporatedinto a mature mRNA molecule.In addition to the library-basedmethod of marker development previouslydescribed, microsatellites can be developedfrom EST databases or from known genesequences. As it is possible to connect the

334Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishTable 1Recently published linkage maps for various fish and shellfish species used in aquacultureSpeciesAtlanticsalmonRainbowtroutNumber ofmarkersMarkertypeMap lengthFemale/MaleMale Female ReferencecM(Kosambi)cM(Kosambi)473 AFLP 8.26:1 103 901 Moen et al., 2004a54 Microsatellites65 Microsatellites 3.92 np np Gilbey et al., 2004226 Microsatellites - 4 590 Nichols et al., 2003973 AFLP4 Allozymes72 VNTR29 Known genes12 Minisatellites5 RAPDs38 SINE*Oysters 115 Microsatellites 1.31:1 776 1 020 Houbert and Hedgecock, 2004Sea bass 174 Microsatellites 1.6:1 567.4 905.9 Chistiakov et al., 2005Kuruma 195 AFLP 1 780 1 026 Li et al., 2003prawnTilapia 525 Microsatellites 1:1 1 300 Lee et al., 200521 GenesScallops 503 AFLP 1.27:1 2 468 3 130 Wang et al., 2005Common 110 Microsatellites - 4 111 Sun and Liang, 2004carp105 Known genes57 RAPDsJapanese 111 Microsatellites 7.4:1 741.1 670.4 Coimbra et al., 2003flounder352 AFLPChannelcatfish313 Microsatellites 3.18:1 1 958 B Waldbieser et al., 2001B Sex-averaged* Short interspersed elementsnp = not publishedfunction of the transcript of genes (from anEST sequence) with the presence of a microsatellite,these markers are Type I markers(O’Brien, 1991; Serapion et al., 2004; Nget al., 2005). This strategy of developingmicrosatellite markers from known genesand ESTs has been used for common carp(Yue, Ho and Orban, 2004), rainbow trout(Rexroad et al., 2005; Coulibaly et al.,2005) and Atlantic salmon (Ng et al., 2005;Vasemägi, Nilsson and Primmer, 2005).In all these analyses, high levels oftransferability between populations andspecies can be expected if the microsatellitesare included in coding regions. Suchtransferability has been observed e.g.between Atlantic salmon and rainbow trout(Vasemägi et al., 2005; Rexroad et al., 2005),making these markers ideal for analyses ofpopulation genetics and comparative maps.For example, microsatellites derived fromEST sequences have been used to studydivergence of Atlantic salmon populationsin salt, brackish and freshwater habitats(Vasemägi, Nilsson and Primmer, 2005).Bioinformatic tools can be used forpotential discovery of SNPs using DNAsequence alignment “in silico” (Marth et al.,

Chapter 17 – Marker-assisted selection in fish and shellfish breeding schemes 3351999). Although it is possible to use basequality values to discern true allelic variationsfrom sequencing errors, validationis a key step for true positive detection ofSNPs (Marth et al., 1999). This is generallycarried out using a proportion of the SNPsdetected in a sample of individuals from thetarget population. This strategy has beenused recently for SNP detection using ESTsequences from Atlantic salmon (Panitz etal., 2002; Hayes et al., 2004).Linkage mapsA linkage map is an ordered collection ofthe genes and genetic markers occurringalong the lengths of the chromosomes ofa species, with distances between themestimated on the basis of the number ofrecombination events observed in the data.Genetic linkage maps have been publishedfor rainbow trout (Young et al., 1998;Sakamoto et al., 2000; Nichols et al., 2003),channel catfish (Waldbieser et al., 2001),tilapias (Kocher et al., 1998; Lee et al., 2005)and Japanese flounder (Coimbra et al.,2003). References to updated linkage mapsof the major aquaculture species are givenin Table 1. Dense linkage maps includinga relatively large number of markers areunder development.Different patterns of recombinationappear among regions of linkage groups incertain male maps, with markers clusteredin centromeric regions, an extremeexample being Atlantic salmon whererecombination in males is greatly reduced(Moen et al., 2004b). The molecularmechanisms responsible for the differencesin recombination rates between sexes arenot well understood, although studies onmodel organisms such as zebrafish, wheregenomic sequencing is currently underway, may help to clarify this (Singer etal., 2002).Using markers to aidconventional fish and shellfishbreeding programmesMolecular markers may be used in a numberof ways to aid conventional breeding of fishand shellfish species, and some of these aredescribed and exemplified below.Parentage analysisOne of the main constraints facing effectivebreeding programmes for fish and shellfishis that newborn individuals are too small tobe tagged individually. Application of theanimal model approach (i.e. using a statisticalgenetic model to predict individualbreeding values) requires tagging a constantnumber of individuals from each familywith passive integrated transponders (PITtags) when they become sufficiently largeafter a period of individual family rearing.However, this system of early managementcreates common environmental (i.e.tank) effects for full-sib families (Martinez,Neira and Gall, 1999). To address thisissue, mixtures of equal-aged progeny fromdifferent families can be reared communallyto preclude the development of suchfamily-specific environmental effects, andgenetic markers can be used subsequentlyto assign individuals to families after evaluationof individual performance (Doyleand Herbinger, 1994). Thus, the impactof early common environmental effectsis considerably reduced if markers areused for parentage analysis when selectingindividuals for early growth rate traits(Herbinger et al., 1999; Norris, Bradley andCunningham, 2000).The amount of marker data needed toachieve acceptable levels of correct parentageassignment depends on the numberof loci, the number of alleles and the numberof parent-pairs (sires and dams) availablefor reconstructing the pedigree (Jamieson

336Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishFigure 1Predicted probability of assigning the correct sire-dam (parent-pair) for a given number ofparent-pairs (x axis) using different combinations of three microsatellites (L1, L2 and L3)amplified in Oncorhynchus kisutch1L1 (0.985) L2 (0.644) L3 (0.646)L1+L2 (0.995) L1+L3 (0.995) L2+L3 (0.874)L1+L2+L3 (0.998)0.9P (correctly assigning the true parent-pair)0.80.70.60.50.40.30.20.100 50 100 150 200 250 300 350 400Number of parent-pairsProbabilities of exclusion included between parentheses in the legend are obtained using the deterministic methodof Villanueva, Verspoor and Visscher, 2002.Data obtained from Diagnotec SA.and Taylor, 1997; Villanueva, Verspoor andVisscher, 2002). The information from themarker data available for each species canbe studied using exclusion probabilities,which are then used to calculate the probability(PC) of correctly assigning the trueparent-pair (sire and dam) to offspring thatare genotyped (Villanueva, Verspoor andVisscher, 2002).Figure 1 presents the results for threemicrosatellites and combinations of microsatellitesto predict the probabilities ofexclusion and PC. The allelic frequenciesof the three microsatellites were calculatedwith a sample (n=100) from a coho salmon(O. kisutch) farm in southern Chile managedunder commercial conditions. The analysisshowed that the probability of assigningthe true parent-pair depended greatly onthe number of parent pairs available forparentage. Only for an unrealistically smallnumber of ten sires and dams is there ahigh probability of assigning the correctparent-pair to offspring. For a breedingprogramme of 200 or 300 parent-pairs,PC decreased considerably. Therefore, inthis example, more markers are needed foraccurate pedigree reconstruction. Successfulparentage assignment experiments typicallyhave used six to eight microsatellite markers(Herbinger et al., 1995; Garcia de Leon etal., 1998; Norris, Bradley and Cunningham,2000; Castro et al., 2004). In practice, thepresence of genotyping errors, null alleles,realized mutations and non-Mendeliansegregation can seriously affect the efficiencyof parentage assignment (Castro etal., 2004). Parentage assignment in the con-

Chapter 17 – Marker-assisted selection in fish and shellfish breeding schemes 337text of fish breeding is also discussed bySonesson (this volume).For most breeding programmes, physicaltagging will prove efficient both ineconomic and biological terms to achieveacceptable rates of genetic gain, while minimizingrates of inbreeding. Genetic markertechnology can still be costly for routineassignment of parentage, althoughthese costs can be reduced using multiplexpolymerase chain reaction (PCR)technology (Paterson, Piertney and Knox,2004; Taris, Baron and Sharbel, 2005) inwhich more than one marker can be genotypedsimultaneously in a single gel laneor capillary. This is especially the casewhen only DNA markers are used withoutphysical tagging, as individuals must be retypedwhen records for multiple traits areincluded in the selection criteria (Gjerde,Villaneuva and Bentsen, 2002).It is expected that rates of genetic gainfor economic traits will not be affected significantlywhen common environmentaleffects are present. This is because, in manyspecies of cultured salmonids, the commonenvironmental effect decreases considerably,from about 20 percent for alevin weight to5 percent for body weight at harvest, whichis the trait with most impact on profit(Herbinger et al., 1999; Henryon et al.,2002; Kause et al., 2005). Hence, commonenvironmental effects should not decreasethe rates of genetic gain for traits measuredat harvest when physical tagging is used.Furthermore, multistage selection offers thepossibility of first selecting individuals on awithin-family basis directly from tanks (fortraits influenced by common environmentaleffects), and then selecting at a second stagefor traits measured at harvest (Martinez,2006a). This alternative would either maintainrates of gain while decreasing the costsassociated with tagging, or even increaserates of gain, when recording from tanks(within families) can be carried out relativelyinexpensively (Martinez, 2006a).The sample size (i.e. the numbers ofindividuals and markers required forreconstructing the pedigree of a populationaccurately) is a practical issue, asnot all individuals in a population can begenotyped for all markers available. Suchissues arise in species where physical taggingis not possible or not economicallysound, as in nucleus populations withoutelectronic tagging (e.g. when recovering aback-up population for nucleus breedingprogrammes) or when disease challenges(e.g. for infectious pancreatic necrotic virus[IPNV]) are carried out early in the lifecycle (Martinez et al., in preparation). Smallsample sizes, together with sperm competition(Withler and Beacham, 1994), matingpreference (as in Artemia; G. Gajardo, personalcommunication) and other biologicalfactors after fertilization can increase thevariance of family size, thereby decreasingthe effective population size to unsustainablelevels (Brown, Woolliams andMcAndrew, 2005).Another problem arises in practicewhen selection is carried out before genotypingwith markers. In this case, BLUPof breeding values is likely to be biasedbecause not all phenotypic informationis used when predicting breeding values.The magnitude of re-ranking is dependenton the amount of information from afamily within the selected group. In theseinstances, the mixed model equations needto be modified to account for such selecteddata (Morton and Howarth, 2005).Establishing breeding programmesusing molecular informationThe choices made at the founding of abreeding programme have a critical

338Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishbearing on its ultimate success. Criteria forchoosing individuals that will be foundersshould be essentially the same as those usedwhen the selection response is optimizedunder restricted co-ancestry when pedigreeinformation is available (Meuwissen,1997; Toro and Mäki-Tanila, 1999). Thus,it is necessary to avoid matings betweenclose relatives for managing existing quantitativegenetic variation at the start of theprogramme. Experiments with the planktonicmicrocrustacean Daphnia spp. haveshown that neutral genetic variation giveslittle indication of the levels of quantitativegenetic variation available for selection(Pfrender et al., 2000). However, increasingthe population size at the beginning of thebreeding programme will diminish the subsequenteffect of random genetic drift, andtherefore larger founding populations willhave an increased likelihood of showingresponse to selection. Lack of adequatebase populations is the main reason for thelack of selection response observed in somespecies of fish (Gjedrem, 2000).The effective population size (N e )required for setting up a breeding programmedepends on the policy regardingrisk management (Brown, Woolliams andMcAndrew, 2005), but to prevent declinein fitness, some authors have recommendedN e values ranging from 31 to 250, which interms of rates of inbreeding should be lessthan 2 percent (Meuwissen and Woolliams,1994). Due to the large family sizes possiblefor many fish and shellfish species, breedingprogrammes that fail to control the geneticcontributions of parents in every generationare expected to incur relatively high rates ofinbreeding (Meuwissen, 1997). The situationis even more extreme when selection isbased on a complex breeding objective thatincludes information from relatives andmany traits jointly (Martinez, 2006b).Fish within commercial productionpopulations generally are not taggedindividually and pedigree information istherefore lacking. Genetic markers allow theestimation of pairwise relatedness betweenindividuals or sib-ship reconstructioneven with unknown ancestors (Toro andMäki-Tanila, 1999; Thomas and Hill,2000; Toro, Barragán and Óvilo, 2002;Wang, 2004; Fernandez and Toro, 2006).There is a plethora of estimators forcalculating pairwise relatedness (Quellerand Goodnight, 1989; Lynch and Ritland,1999). The efficiency of inferring pairwiserelatedness using markers without parentalinformation is affected by assuming knownallele frequencies in the base populationand unlinked loci in Hardy-Weinbergequilibrium. Furthermore, pair-wisemethods can lead to inconsistent assignationsbetween triplets of individuals because theyuse information from only two individualsat a time (Fernandez and Toro, 2006). Inaddition, it is difficult to set thresholds forclaiming different types of relatedness inthe data (Thomas and Hill, 2000; Norris,Bradley and Cunningham, 2000). On theother hand, sib-ship reconstruction methodsdo not attempt to calculate co-ancestry;rather, they attempt to reconstruct full- orhalf-sib or other family groups (Thomasand Hill, 2000; Emery, Boyle and Noble,2001; Smith, Herbinger and Merry, 2001).Such reconstructions of full- or half-sibfamilies or even other groups of relativesappear robust to lack of knowledge of basepopulation allele frequencies (Thomas andHill, 2000; Fernandez and Toro, 2006).Marker information can be used to inferrelatedness between individuals available ascandidate broodstock to generate the firstgeneration of offspring in the breeding programme,and thereby avoid mating amongclose relatives. This approach uses molec-

Chapter 17 – Marker-assisted selection in fish and shellfish breeding schemes 339ular information to infer the genealogicalpedigree. A simulation was conducted toreconstruct the pedigree of 100 potentialcandidates from ten full-sib families (witha Poisson family size equal to ten using sixequally-frequent microsatellites, withoutparental genotypes (Martinez, 2006c). Theposterior probability of either full [P(FS)]or half-sib [P(HS)] groups was obtainedusing the Bayesian model of Emery, Boyleand Noble (2001). In the simulation results,there was a tendency to overestimate relationships,with posterior probabilities over0.5 when individuals were in fact unrelated.On the other hand, not all true full-sibswere assigned to the correct full-sib familywith the greatest probability, and some truefull-sib family members were reconstructedas half-sibs. On average (among ten replicates),the probability of mating relatedindividuals was significantly smaller wheninformation from molecular markers wasused, compared with what was expectedby chance (4.7 percent versus 18.1 percent,p = 0.002). The practical implication is thatinbreeding in the progeny generation wouldaverage 5 percent when random matingis used and 1 percent when optimizationusing molecular information is used.In practice, to perform mating in thebase population, the relatedness inferredfrom molecular information does not needto be perfectly accurate, but it does requirethat relatedness is not underestimatedgreatly. Among the technical issues thatarise when using marker data are that apair of individuals could be misclassifiedas related when they are in fact unrelated(Type I error) or a pair may be wronglyclassified as unrelated when the pair is infact related (Type II error). Type II erroris of greatest concern as this could result inrelated pairs being mated. This is becausemating of individuals (males and females)as unrelated when in fact they are fullsibs will increase true inbreeding in thepopulation, while misclassification leadingto unrelated individuals being assigned toa full-sib family would not increase theinbreeding in the progeny. The presence ofmutations, null alleles or genotyping errorswill underestimate the true relationships inthe population and eventually increase theprobability of mating true full-sibs (Butler etal., 2004). Recently, Wang (2004) suggesteda method for inferring relationships formarker data with a high error rate andmutation that can be used to address thisissue. It should also be noted that studiesdealing with estimation of heritability orprediction of breeding values with pedigreesreconstructed using molecular markersmay be very inefficient when pedigreesare reconstructed with an increased rate ofType I errors (Mosseau, Ritland and Heath,1998; Thomas, Pemberton and Hill, 2000).Detecting self-fertilization in scallopsIn scallops, a main drawback whenimplementing breeding programmes is theoccurrence of self-fertilization, even whengametes from later spawning pulses are usedfor obtaining family material (Martinezand di Giovanni, 2006), i.e. a mixtureof selfed and outcrossed individuals canbe present even at later stages within asingle family. Bias in estimating geneticparameters is expected due to this residualself-fertilization, which can occur withconsiderable frequency (average 20 percent)within particular families.A simulation study was used to investigateto what extent markers with differentinformation content can be used todiscriminate between selfed and outcrossedindividuals within a family (Figure 2). Theresults showed that microsatellites gavemean values of posterior probabilities greater

340Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fish100%Figure 2Identification of selfed individuals within families of scallopsusing different types of marker data 190%80%70%60%50%40%30%20%10%0%MICRO-5 SNPs-5 AFLP-5 AFLP-15 AFLP-30Vertical bars represent the proportion of the 100 replicates in which the mean posterior probabilities of being selfed(for true selfed individuals) were greater than 0.95.1MICRO-5: five microsatellites with six equally frequent alleles each. SNPs-5: five SNP markers with equal allelefrequencies. AFLP-5, -15 or -30: 5, 15 or 30 AFLP markers. The design of the simulations of self-fertilization in scallops:the amount of self-fertilization was modelled using a truncated normal distribution which best fitted the empiricaldistribution of self-fertilization (Martinez and di Giovanni, 2006). A Bayesian model was used to infer mutuallyexclusive posterior probabilities of being either selfed or outbred (Anderson and Thompson, 2002). It was assumedthat parental information was lacking, with unlinked markers and vague priors. Selfed individuals were regarded ashaving been detected correctly when the posterior probabilities of being selfed were greater than 0.95 (this criterionwas determined empirically for operational reasons).Source: V. Martinez, in preparation.than 0.95 in about 90 percent of the familiessimulated (100 in total). Similar results wereobtained with 30 AFLP markers, but thesepercentages were considerably reduced forsmaller numbers of AFLPs or SNPs.The information from these markerscan be used to cull individuals, to constructa relationship matrix in which all unusualrelationships are incorporated in analysesused for obtaining unbiased estimates ofheritability and genetic correlations, andfor estimating breeding values from realdata sets (Martinez, 2006a).Identifying QTL and majorgenes influencing complexquantitative traitsMolecular biology can greatly help the discoveryof factors influencing the expressionof quantitative traits. There are a number ofways in which this information can be used,the difference between them being the levelof resolution with which these factors canbe mapped. For example, loci with majoreffects on quantitative traits (QTL) aremapped by using markers to track inheritanceof chromosomal regions in familiesor in inbred line crosses using the extentof linkage disequilibrium generated in thepopulation. This approach gives a limitedamount of mapping resolution. Fine mappingrequires information from additionalmarkers and individuals sampled across theoutbred population and, while helping tonarrow the confidence interval of the positionof the QTL, this is only the startingpoint for identifying the polymorphisms inthe performance-determining genes themselves.In practice, identification of genesinfluencing specific traits is achieved using

Chapter 17 – Marker-assisted selection in fish and shellfish breeding schemes 341a combination of genetic mapping (linkageand fine mapping) to localize the QTL toa small region on the chromosome underanalysis, and candidate gene or positionalcloning approaches to identify the geneswithin the QTL region.In some cases, sufficient biochemicalor physiological information is availableto investigate the association between thequantitative expression and the level ofmarker polymorphisms within specificgenes. Nevertheless, this approach requiresa great amount of detailed information inorder to choose which gene explains thegreatest effect and to have sufficient powerto detect the association. This informationis starting to appear in the aquaculture literaturefrom multinational projects suchas the Consortium of Genomic Resourcesfor All Salmonids Project (cGRASP) (Nget al., 2005).QTL mapping in fish using linkage disequilibrium:theoretical and practicalconsiderationsValue of chromosomal manipulationsThe great reproductive flexibility of fishenables different breeding designs to beimplemented relatively easily. Completelyhomozygous fish can be produced inonly one generation using chromosomeset manipulations, without the many generationsof inbreeding needed in othervertebrates. These manipulations enabledoubling of the chromosomal complementof a haploid gamete (Young et al.,1996; Corley-Smith, Lim and Bradhorst,1996). Androgenetic double haploid individualscan be obtained by fertilizing eggsthat were inactivated with gamma radiation,yielding haploid embryos containingonly paternal chromosomes. Alternatively,gynogenetic double haploid individuals canbe obtained by activating the developmentof eggs with ultraviolet-inactivated sperm,yielding haploid embryos containing onlymaternal chromosomes. In each case,diploidy is restored using methods thatsuppress the first mitotic division (Figure 3;Streisinger et al., 1980; Corley-Smith, Limand Bradhorst, 1996; Bijma, van Arendonkand Bovenhuis, 1997; Young et al., 1998).The use of these reproductive manipulationsto provide experimental populationsfor genetic analysis of complex quantitativetraits has been well described (Bongers etal., 1997; Robison, Wheeler and Thorgaard,2001; Tanck et al., 2001).Double haploids from inbred line crossesAfter a second round of uniparental reproduction(Figure 3), a collection of clonal linescan be obtained that collectively is likelyto represent all the genetic variants fromthe base population (Bongers et al., 1997).Crosses of sex-reversed double haploid individualsfrom lines that diverge for the traitsof interest can produce F 1 lines in completelinkage disequilibrium. These F 1 populationscan be used for further experimentationbased on F 2 or backcross designs. Anotherround of androgenesis of F 1 individuals willproduce a population of fully homozygousindividuals. This design will have twice thepower for detecting QTL as the standard F 2design (Martinez, 2003). The standard deviationof QTL position estimates is halved forthe double haploid design. This is due toan increase in the additive genetic variance,which is doubled for the double haploiddesign due to redistribution of the genotypefrequencies in the progeny generation(Falconer and Mackay, 1996).Informative double haploid populationsof this sort have been utilized to performQTL analysis for embryonic developmentrate in rainbow trout (Robison, Wheelerand Sundin, 2001; Martinez et al., 2002;

342Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishFigure 3Chromosomal manipulations in fishOUTBRED MALE POPULATIONRECOMBINANT INBRED PROGENY(DOUBLED HAPLOID)DNA inactivation(ova)INBRED CLONAL (SEX REVERSAL)LateSchockDOUBLED HAPLOIDDoublehaploidHaploidNormaldiploidANDROGENESISIn androgenesis, fertilization is carried out using radiation-inactivated ova, and a late shock is used to suppress first mitosisANDROGENESISand thereby restore diploidy.Adapted from Thorgaard and Allen, 1987.Martinez et al., 2005). At least four QTLof relatively large effect explain about40 percent of the phenotypic variance ofthe mapping population and most of the2.5 standard deviations of the differencebetween the original clonal lines usedto generate the F 1 population (Robison,Wheeler and Thorgaard, 1999). Twolinked QTL were in repulsion phase inthe F 1 population, and were undetectedin the analysis using composite intervalmapping. This result was not surprising asevidence was accumulated among replicatesof lines that were incubated at differenttemperatures (Robison, Wheeler andSundin, 2001), and the Bayesian multipleQTL method incorporated all the availableinformation of environmental co-variatesin the analysis (Martinez et al., 2005).Recently, these double haploid lines havebeen used for mapping QTL related to thenumber of pyloric caeca (Zimmerman et al.,2005) and for confirming QTL influencingdevelopment rate (Sundin et al., 2005).When traits are associated and by takinginto account the correlated structure of thedata, multivariate estimation of QTL effectsis expected to be more powerful than singletrait analysis (Jiang and Zeng, 1995). Also,from a genetic standpoint, joint analysisprovides the means for testing differenthypotheses about the mode by which genesexplained the genetic co-variation (Wu etal., 1999). For example, after hypothesistesting (following Knott and Haley, 2000),a single pleiotropic QTL with oppositeeffects for development rate and lengthbest explained the multivariate data (asdetailed earlier by Martinez et al., 2002b).This finding was also consistent with thenegative correlation estimated with the data(Martinez et al., 2002).Double haploids in outbred populationsMartinez, Hill and Knott (2002) derivedanalytical formulae to predict the power oflinkage analysis for interval mapping underthree different mating designs in outbred

Chapter 17 – Marker-assisted selection in fish and shellfish breeding schemes 343populations: full-sib mating, hierarchicalmating, or double haploid designs. Thisanalysis suggested that the use of doublehaploids appeared to be of benefit whendetecting QTL, particularly when both thevariance of the QTL and of the polygeniceffects was small. Furthermore, given therelatively large size of full-sib families infish, there appeared to be little advantageof hierarchical mating over full-sib matingdesigns for detecting QTL, the optimumfamily size depending on the size of theQTL and the population structure usedfor mapping (Martinez, Hill and Knott,2002). The gain in power of the double haploiddesign comes from the increase in thevariance of the Mendelian sampling termwithin families, which is effectively doubledfor traits that are explained by additiveeffects (Falconer and Mackay, 1996).As experimental settings constrain thetotal number of individuals genotyped,designs aimed at QTL mapping shouldinclude a small number of families of relativelylarge size in order to maximizethe likelihood of detecting the QTL. Thisis because most of the information formapping QTL uses linkage informationthat comes from within-family segregation(Muranty, 1996; Xu and Gessler, 1998).However, increasing power comes at theexpense of reducing the accuracy of estimatingthe additive genetic variance forpolygenic effects. A QTL mapping methodhas been developed for double haploids,which efficiently accommodates all theuncertainties that pertain to outbred populations,such as unknown linkage phasesand differing levels of marker informativeness,using the identical-by-descentvariance component method (see below;Martinez, 2003). Also, it is possible to combinedouble haploids and outbred relativesin the same family. Simulations of differingamounts of marker information and heritabilityfor the QTL were used to comparethe empirical power of the double haploidand full-sib designs. While the power ofthe full-sib design was lower than that fordouble haploids, QTL position estimatesfor double haploids had large confidenceintervals (about 30 cM as compared with 40cM for full-sibs; Martinez, 2003).The double haploid design was usedfor mapping QTL for stress response incommon carp using single marker analysis(Tanck et al., 2001). The authors foundonly suggestive evidence for QTL, which isnot surprising due to limited genome coveragefor markers used in the analysis.Published results have shown that doublehaploid lines are a useful resource for QTLdetection studies. However, double haploidlines are difficult to develop due to theexpression of deleterious recessive alleles(McCune et al., 2002) and the low survivalfollowing shocks applied to restore diploidyto the haploid embryo. As the rate of malerecombination is depressed, the precisionof mapping QTL in androgenetic families islower than that obtained using recombinationevents from females. Another practicalmatter is the labour needed for developinga clonal line, as at least two generations arerequired (Figure 3). This delay can be quiteexpensive and time-consuming for specieswith a long generation interval, such assalmon or trout (two to four years).Aspects of QTL mapping in outbredpopulations of fishInbred line crosses are ideal for mappingQTL because they are expected to be completelyinformative for both markers andQTL, providing that the inbred lines arefixed for alternative alleles. Outbred populationsare not completely informative forboth QTL and markers; thus, experimental

344Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishpower is expected to be lower than that forcrosses between clonal lines. The powerfor detecting the QTL depends on allelefrequencies, the probability of sampling aninformative parent and family size.Factors influencing the power of detectingQTLDue to the large family sizes that can beobtained in many fish species, differentmating designs using full-sib groups can becarried out for outbred populations. Forexample, full factorial designs may be usedin which many males and females are matedto one another, and hierarchical designsmay be applied in which each male is matedwith multiple females, or each female withmultiple males. For a given size of experiment,factorial and hierarchical designs havepotentially a lower probability of samplinga heterozygous parent (because fewer siresand or dams are sampled overall), comparedwith the full-sib design in whicheach family has potentially two informativeparents. For this reason, factorial and hierarchicaldesigns can potentially give lowerpower compared with the simple full-sibdesign (Muranty, 1996; Martinez, Hill andKnott, 2002).The optimum number of full-sib familiessampled in the QTL mapping populationdepends on the intrinsic power of theexperiment (i.e. size of the QTL effect andsize of the population). As expected, largefamily sizes are needed for detecting QTLof small effects (Martinez, Hill and Knott,2002). When the QTL explains 10 percentof the phenotypic variance, the optimumfamily size appears to be 50 individuals perfamily for a reasonably-sized QTL mappingexperiment in outbred populations(Figure 4). Further increases in the numberof individuals per family provide only amodest increase in power. Further, the sameresults used simulation models showingdominance and additive effects under thevariance components method for mappingQTL (Martinez et al., 2006a).Methods of analysisThe method of choice when analysing datafrom outbred populations is the variancecomponent method, in which QTL effectsare included as random effects with a covarianceproportional to the probability thatrelatives (e.g. full-sibs) share alleles identicalby descent conditional on marker data (Xuand Atchley, 1995). This model is similarto the one used more generally for geneticevaluation of candidate fish for selection,but includes the random QTL effect.A considerable proportion of the geneticvariance for growth-related traits in fishpopulations has been explained by dominance(Rye and Mao, 1998; Pante, Gjerdeand McMillan, 2001; Pante et al., 2002).When mapping QTL using the randommodel, it is assumed that only additiveeffects are of importance and thereforeonly matrices of additive relationships conditionalon marker data are fitted in theresidual effect maximum likelihood procedure(George, Visscher and Haley, 2000;Pong-Wong et al., 2001). However, thelarge family sizes in fish enable hypothesesfor different modes of inheritance atthe QTL to be tested using the withinfamilyvariance. While some authors havespeculated that including dominance in themodel will increase the power of detectingQTL (Liu, Jansen and Lin, 2002), others(Martinez, 2003; Martinez, 2006a) haveshown that power to detect QTL was comparablebetween models including or notincluding dominance. This was particularlythe case for the larger family sizes simulatedand it was concluded that for mostscenarios, the additive model was quite

Chapter 17 – Marker-assisted selection in fish and shellfish breeding schemes 345Figure 4Deterministic power calculation (following Martinez, Hill and Knott, 2002) for a QTL explaining10 percent of the phenotypic variance for a variable total population size (x axis) andfamily size (25, 50 and 100)10.90.8Power of detecting QTL0.70.60.50.40.30.20.125 50 1000100 200 300 400 500 600 700 800 900 1 000Population sizerobust for detecting QTL and there waslittle loss of information for detecting QTLwhen dominance is present but not used inthe QTL mapping analysis.QTL mapping in practiceTo date, QTL mapping in fish using outbredpopulations has been carried out mostlywith single marker analysis (microsatellitesand AFLP markers), and using relativelysparse linkage maps when interval mappingis used. In tilapia, the F 2 design and afour-way cross between different species ofOreochromis have been used for detectingQTL affecting cold tolerance and bodyweight (Cnaani et al., 2003; Moen et al.,2004c). In outbred populations of salmonids,QTL that influence body weight havebeen mapped (Reid et al., 2005 and referencestherein).Studies seeking linkage of markers totraits amenable to MAS, such as diseaseresistance, have begun to appear in the literatureover the past few years. For example,QTL for resistance have been mappedfor infectious pancreatic necrosis virus(Ozaki et al., 2001), infectious salmonidanemia (Moen et al., 2004c), infectioushaematopoietic necrosis (Rodriguez et al.,2004; Khoo et al., 2004), and stress andimmune response (Cnaani et al., 2004).Also, Somorjai, Danzmann and Ferguson(2003 and references therein) reported evidenceof QTL for upper thermal tolerancein salmonids with differing effects in differentspecies and genetic backgrounds.From fine mapping to finding genesinfluencing complex traitsWhen the number of meioses in thegenotyped pedigree is not sufficient for thelinkage analysis to obtain a precise positionfor the QTL, there is a wide confidenceinterval around an estimated QTL position.

346Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishFine mapping methods attempt to overcomethis problem by quantifying the gameticphase or linkage disequilibrium (LD)present in an outbred population, i.e. acrossfamilies. This method makes use of thenumber of generations as the appearanceof a mutation and can produce extremelyprecise estimates of the QTL position(Pérez-Enciso et al., 2003). The rationalebehind using LD for mapping QTL is thatwhen the population size is rather small,founders of the population would haveonly a limited number of haplotypes, andwith very tightly linked loci there may notbe sufficient time for recombination tobreak up the association between markersand the mutation affecting the quantitativetrait.LD mapping is carried out by calculatingthe probabilities that haplotypes shared byindividuals are identical by descent from acommon ancestor conditional on markerdata (assuming t generations as the commonancestor and a certain N e ; Meuwissen andGoddard, 2001). The LD in the populationdepends on a number of populationparameters such as the degree of admixtureor stratification in the population andthe actual level of association between thecausal mutation and the polymorphisms.The correct determination of phases andof genotypes at the QTL is required forfine mapping purposes (Meuwissen andGoddard, 2001; Pérez-Enciso, 2003). Forthese reasons, a pure LD analysis is likelyto result in a large number of false positives,i.e. falsely inferring association when thereis no linkage.Methods that incorporate the linkageinformation (within families) and LDjointly are preferred, because the likelihoodof spurious association (i.e. LD withoutlinkage) diminishes, making much betteruse of the whole data set (Meuwissen andGoddard, 2001, 2004; Pérez-Enciso, 2004).All of these methods, however, require agreat deal of genotyping of tightly linkedmarkers such as SNPs, which currently arenot widely available for fine mapping inaquaculture species.Using fine mapping techniques, the confidenceinterval for QTL position can bereduced considerably. However, to developa direct test for a favourable polymorphismrequires use of comparative mappingapproaches with model species, such aszebrafish or fugu, to select the candidategenes that most likely affect the trait ofinterest. Otherwise, enrichment of markersin a specific region of the genome (tonarrow further the most likely position ofthe polymorphism) following sequencingis needed to compare sequences betweenindividuals that show different phenotypesor alternative QTL alleles.Candidate gene analysisIt is tempting to invoke variation at geneswith a known role in the physiology underlyinga complex trait such as growth toexplain phenotypic variability for the trait.These genes can be searched for polymorphisms(e.g. SNPs) and the variants thentested to determine whether they are correlatedwith the expression of the quantitativetrait. This approach requires knowledgeof the biology of the species, biochemicalpathways and gene sequences in order totarget variation at those specific genes. Inaquaculture, most of this information is currentlylacking, but it is expected that moregenes will be incorporated in databases inthe near future. The possibility exists to utilizedata from highly studied model species,such as zebrafish or rainbow trout, in comparativebioinformatic approaches.To date, this strategy has not proven particularlysuccessful for explaining genetic

Chapter 17 – Marker-assisted selection in fish and shellfish breeding schemes 347variation underlying complex (polygenic)traits. This is because although the biologyof the trait and the genes most likelyinvolved in the expression of the phenotypemay be known, in complex traitsmany other genes may be involved in themetabolic pathway that are not obviouscandidates. For example, in aquaculture species,candidate genes have been studied forgrowth-related traits using ten conservedgene sequences known to be related to thegrowth hormone axis (Tao and Boulding,2003). In this study of Arctic charr, only asingle SNP (of ten) from five of ten geneswas found to be associated with growthrate.Another example for disease resistancetraits is the major histocompatibility complex(MHC). The genes of this complexencode highly polymorphic cell surfaceglycoproteins involved in specific immuneresponses and either specific alleles orheterozygotes at this complex were associatedwith resistance and susceptibility toA. salmonicida or infectious haematopoieticnecrosis (IHN) virus (Langefors, Lohm andGrahn, 2001; Lohm et al., 2002; Arkush etal., 2002; Grimholt et al., 2003; Bernatchezand Landry, 2003). Nevertheless, thebackground genome was quite importantfor explaining the difference in resistancebetween individuals within a family(Kjøglum, Grimholt and Larsen, 2005).Microarrays, gene expression andidentification of candidate genes forQTL analysisMicroarray technology (Knudsen, 2002)enables the expression of thousands ofgenes to be studied simultaneously. Untilnow, this information has been used primarilyfor following gene expression intreatment and control experiments in manyfields such as disease exposure and stressresponse. This information can be used todiscover new sets of candidate genes, possiblywith or without functional assignmentthat may be related to the quantitative traitof interest (Walsh and Henderson, 2004).Genes whose expression differs betweentreatments are likely to be trans-actinggenes, i.e. their expression is regulatedby other genes. Therefore, it seems likelythat seeking polymorphisms within thesegenes may not yield information aboutfactors that explain the phenotype, andthere might be problems assigning the correctsignificance threshold (Pérez-Encisoet al., 2003). Further, because many genesare part of metabolic pathways and do notact individually, the expression of a singlegene may be insufficient to explain phenotypicdifferences between individuals. Onlythose genes that directly affect phenotypicexpression (i.e. cis-acting genes) can betreated as candidate genes for subsequentuse in MAS after studying polymorphismsin their sequences. In salmonids, a microarraymade available from the Consortiumfor Genomics Research on all SalmonidsProject (cGRASP) has been used to studygene expression in fish exposed or notexposed to Pisciricketsia salmonis (Rise etal., 2004b), and microarrays in other fishand shellfish species are currently underdevelopment.A gene expression pattern can itself beregarded as a quantitative trait. Here, theinterest is in finding associations betweendifferent patterns of gene expression andmarker loci. This analysis was coined as“genetical genomics” by Jansen and Nap(2001). As is usual in QTL mapping, theanalysis attempted to dissect the transcriptionalregulation of the entire transcriptomeand to identify the effects of individualQTL affecting gene expression (the socalledeQTL; e.g. Hubner et al., 2005). To

348Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishdate, this analysis relies upon the use ofsegregating populations (of known origin)such as recombinant inbred lines (Carlborget al., 2005), and the analysis of outbredpopulations poses greater challenges (Pérez-Enciso, 2004). Still, aquaculture species canprovide sufficient information due to thelarge family sizes needed to unravel complexregulatory gene networks. How allthis information can be included in MASprogrammes is yet unclear.Incorporating molecular markersinto breeding programmes forfish and shellfishGeneral aspects of incorporatingmolecular information in breedingprogrammesThe response to selection ∆G is estimatedas:∆G =iσ H rwhere i = the intensity of selection, r = thecorrelation between the breeding objectiveand the selection criteria (i.e. accuracy), andσ H = the additive genetic standard deviationfor the breeding objective. As the majorimpact of incorporating information frommolecular markers will be on accuracyestimates, improvement of the response toselection will be higher for traits that haverelatively small accuracy than for traits ofrelatively large accuracy. Thus, breedingprogrammes for traits with low heritabilityand relatively few records per trait measuredsuch as carcass and disease resistanceare those most benefiting from incorporatingmarker information (Meuwissen,2003).The relative increase in accuracy dependson the amount of variation explained bymarkers, which in turn depends on thenumber of QTL identified and used in MASor GAS schemes (Lande and Thompson,1990). QTL experiments in other specieshave shown that the effects of markedgenes have a leptokurtic distribution, witha small number of genes having large effectsand polygenes (Hayes and Goddard, 2001),which is likely to be the case in aquaculturespecies (Martinez et al., 2005). Hence, it isexpected that more than a single markedgene will be needed for MAS schemes tobe efficient.Due to the biology of many fish andshellfish species, multistage selectionwill likely prove useful in MAS or GASschemes. Basically, a first stage of selectioncan be applied for traits expressedearly in the life cycle (e.g. body weight),and a second stage of selection will incorporateinformation from relatives plusmarked QTL. Optimization will be neededto determine the intensity of selection thatshould be applied at each stage to maximizeprofit while reducing the cost and labourof keeping individuals until later stages(Martinez et al., 2006b).Health and carcass traits are difficultto select for in fish and shellfish becausephenotypic records are obtained from relativesand not from candidates for selection(Gjoen and Bentsen, 1997). Sib or pedigreeevaluation has many disadvantages inrelation to the amount of genetic progressthat can be realized within a selection programmeusing only pedigree informationto predict breeding values using an animalmodel. First, selection accuracy using sibinformation is lower than when predictingbreeding values based on an individual’sown information (Falconer and Mackay,1996). Second, there is no variation ofestimated breeding value for polygeniceffects. Thus, variation of Mendelian samplingeffects within a family cannot be usedand consequently there may be a limitedscope for constraining rates of inbreeding

Chapter 17 – Marker-assisted selection in fish and shellfish breeding schemes 349to acceptable levels when the number offamilies is relatively low.To date, little has been publishedregarding the economic profits arising fromthe extra genetic gain obtained by MASor GAS schemes in aquaculture or terrestrialspecies. Information of this natureis essential because the additional gainsare dependent on the magnitude of theallelic effects and thus the marginal increaseshould offset the costs of applying thetechnology. This trade-off may be moreimportant when a single marked QTL,rather than multiple marked QTL (andmultiple traits), is targeted by selection.Pleiotropic effects can be important ifthe polymorphisms under MAS or GASalso have negative effects on fitness orother traits of economic importance. Forexample, negative genetic correlations havebeen found for resistance to viral and bacterialdiseases (Gjøen et al., 1997; Henryonet al., 2002, 2005), which may be a problemin practical breeding when the goal is toselect fish resistant to a range of pathogens.For example, in natural and selectedpopulations, MHC polymorphism is likelyto be maintained by frequency-dependentselection (Langefors, Lohm and Grahn,2001; Lohm et al., 2002; Bernatchez andLandry, 2003), suggesting that selectionfavours rare alleles, but works against thesame alleles at high frequency. Therefore,it seems likely that a MAS scheme usingMHC information or QTL in LD with diseaseresistance should focus on maintainingpolymorphism rather than on selecting fora particular combination of alleles.MAS in populations in linkageequilibriumWhen populations are in LE betweenmarkers and QTL, the information usedfor selection purposes is given by theMendelian co-segregation of markers andQTL within each of the full-sib familiesin the population under selection. Inpractical terms, this means that co-ancestryconditional on marker information needsto be computed within a family for agiven segment in the genome. In effect,the segregation of regions that individualsshare as identical-by-descent (“more” or“less” than average) is being traced and,under such circumstances, the accuracy ofpredicting breeding values using markerinformation is mainly dependent on theproportion of the within-family variancedue to the QTL (Ollivier, 1998).The effect of family size on the relativeaccuracy of predicting breeding values(comparing MAS and BLUP) using markerinformation was studied in detail using simulations(Table 2; V. Martinez, unpublisheddata). Compared with the GAS schemespresented below, for LE-MAS to be efficient,large full-sib families are requiredfor predicting breeding values for the QTLaccurately. This is because breeding valueprediction is carried out on a within-familybasis; thus, large families are required toobtain breeding values for predicting QTLeffects with reasonable accuracy. Whenindividuals do not have records for thequantitative trait, the extra accuracy ofMAS was highest for the largest family sizesimulated (50 individuals, 25 with recordsand 25 without records; the difference isequal to 7 percent). The accuracy of predictingbreeding values was very similar inBLUP or MAS for individuals that haverecords for the trait in most of the scenariossimulated, suggesting that MAS is expectedto be of little use under these circumstances(Villanueva, Pong-Wong and Woolliams,2002).The advantage of MAS will comeboth from increased accuracy and from

350Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishTable 2Empirical correlation between predicted breeding values using molecular* and pedigree information(M+BLUP) or pedigree information (BLUP) and true breeding valuesScenarioIndividualswith recordsFamily size (number of families)10 (100) 20 (50) 50 (20)M+BLUPBLUP M+BLUPBLUP M+BLUPI NO 0.47 0.45 0.55 0.52 0.64 0.57YES 0.60 0.60 0.65 0.64 0.70 0.65II NO 0.41 0.41 0.49 0.47 0.56 0.52YES 0.58 0.58 0.62 0.61 0.64 0.63* Molecular information comprises a completely informative marker bracket of 10 cM around a QTL and all individualsgenotyped for the markers. The matrix of identity-by-descent values was calculated using the deterministic method ofMartinez (2003). The estimated values of h 2 using residual effect maximum likelihood for the polygenic and QTL effectswere, on average, 0.13 and 0.09, respectively. The results are presented for different nuclear family sizes (number offamilies, between parentheses) and for candidates with or without phenotypic records. The population size was equal to1 000, where 50 percent (Scenario I) or 25 percent (Scenario II) of the individuals within each full-sib family had records forthe trait.Source: Martinez, unpublished data.BLUPincreasing the realized selection intensityin sustainable breeding schemes withrestricted rates of inbreeding. In sib-testingschemes, candidates without records canonly be selected randomly within familiesbecause an estimate of the Mendelian samplingterms cannot be obtained. Markersprovide an estimate of the QTL effects thatsegregate within a family, and thereforethe realized selection differential (at thesame rates of inbreeding) is expected to begreater than that obtained using standardsib/family testing.All the benefits outlined above come atan expense. MAS using LD within familiesrequires a great deal of genotyping andrecording of phenotypes on relatives, dueto the fact that the linkage phase betweenmarkers and QTL needs to be re-estimatedin each generation. This is because LDbetween markers and the QTL is establishedonly within families in each generation andnot across the population. For this reason,it is not possible to predict breeding valuesfor the QTL using molecular marker datawithout records when exploiting informationfrom a single generation. Therefore,pre-selection using this approach is moredifficult to apply in practice. This meansthat for disease resistance or carcass qualitytraits, challenge (measurement) will have tobe carried out at every generation, in all thefamilies available within the programme, asis always the case for conventional breedingprogrammes.Due to the low resolution when mappingthe QTL, it is likely that inaccurateestimates of position will lead to overoptimisticestimates of rates of geneticgain. In the simulations, it was assumedthat the QTL position was known withinthe interval and the markers surroundingthe QTL were completely informative.Thus, the increase in accuracy presentedin Table 2 represents the upper bounds ofaccuracy estimates.Utilizing direct test of genes in GASschemesThe mean phenotype of the populationfor a quantitative trait can be modifiedby increasing the frequency of favourablealleles of genes influencing the trait.In the literature, greater genetic gain hasbeen predicted for GAS schemes than forMAS schemes (using LE populations) at

Chapter 17 – Marker-assisted selection in fish and shellfish breeding schemes 351Figure 5Relative efficiency of combined GAS (for different family sizes [full-sibs]) and a known QTL,explaining 10 percent of the genetic variance) versus an index using information from full-sibsonly for different values of the overall heritability (h²)80Relative efficiency of combined MAS (full-sibs plus aQTL) versus full-sib information only(percentage)706050403020100.1 0.2 0.3 0.400 10 20 30 40 50Family sizeThe data were obtained from the ratio of square root of the variance of the indices. Selection indexformulae were derived from Lande and Thompson (1990).the same rate of inbreeding (Pong-Wonget al., 2002). This is because the accuracyof predicting QTL effects using markersis always smaller than when the QTLeffects are known, as in GAS schemes. Inreality, it is likely that MAS will be carriedout using information from manymarkers to predict the allelic effects of morethan one QTL simultaneously whereas, inGAS schemes, only a limited number ofpolymorphisms are likely to be available.Therefore, on the whole, MAS schemesmay yield greater genetic response becausea greater proportion of the genetic variationis marked and used. Still, more markergenotyping is required for MAS schemes,which means that the additional proportionof the variance typed should pay for theincrease in the cost of many markers typedsimultaneously.Due to the biology of many speciesin aquaculture, large family sizes can beused in a breeding programme. Followingthe deterministic model of Lande andThompson (1990), Figure 5 describes theeffect of family size and amount of polygenicvariation on the relative efficiency ofaccuracy estimates for an index using differentnumbers of full-sibs measured forthe trait, versus an index also includinginformation on candidates for selection genotypedat loci targeted for GAS schemes (V.Martinez, unpublished results). For a singleQTL explaining 10 percent of the geneticvariance, when the heritability is relativelylarge, family size has a small impact on theaccuracy. On the other hand, when the heritabilityof the trait is small, selection for aknown QTL has a major impact on relativeefficiency, particularly when the family size

352Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishis relatively small. Hence, this approach canbe important for traits that are expensive ordifficult to measure such as carcass quality,disease resistance or antibody response.Given the research efforts carried out atdiverse laboratories worldwide, it is likelythat direct tests will be available in the nearfuture for GAS schemes for different traits.With an increasing amount of data on ESTs,together with a greater understanding ofthe function of known genes in aquaculturespecies and new gene discovery, thereis a possibility of more rapidly identifyingand subsequently using polymorphismsthat are within coding regions. However,the research effort required to develop testsfor polymorphisms explaining allelic effectscannot be underestimated, and the factorsinfluencing the profitability of GAS willinclude:• the amount of variation explained by thetest and the number of tests (genes) availablefor explaining the phenotype;• the frequency of the favourable allele in,and the presence of the direct test (e.g.SNPs), for the target population;• the interaction between the polymorphismand the background genome andpossible pleiotropic effects on fitness;• the trade-off between the marginal returngiven by the additional genetic gainobtained through the non-linear changesin the allele frequency of the favourableallele until fixation;• fixed costs of implementing genotypingand patenting.MAS in populations in LDUsing information from dense markermaps, it is possible to make use of LDbetween the markers and the beneficialmutations influencing the quantitative traitsacross the population. Under this scenario,there are two possible ways to use the LDin MAS programmes i.e. using informationon a single haplotype effect in LD with thebeneficial polymorphism across the population,or predicting the total genetic valueusing genome-wide, dense marker maps(genome-wide marker-assisted selection,or G-MAS) (Lande and Thompson, 1990;Meuwissen, Hayes and Goddard, 2001).The effectiveness of each scenario islargely dependent on the actual magnitudeof the effects associated with the polymorphism,either across the whole genome orat specific genes. It is likely that, in thenear future, high-throughput SNP technologywill make dense marker maps costeffective for selective breeding purposes inaquaculture. Thus, it can be expected thatLD-MAS will be implemented over thewhole genome, basically using markers tounravel the genetic architecture of quantitativetraits. Information from multiple traitsjointly and for multiple genes (and theirinteractions within and between loci) willbe used, rather than first relying on mappingQTL in experimental populations andthen implementing this information in MASprogrammes. A profit analysis includingmultiple traits (e.g. to study undesirablepleiotropic effects on the breeding goal)will be needed on a case-by-case basis todetermine whether the use of a single ormultiple haplotypes simultaneously is mostprofitable and which method of LD-MASbetter suits the population under selection.Specific genes are not being evaluatedwhen LD is used across the population;rather, haplotype effects on the phenotypeare being estimated. As this is done on asingle generation across the whole genome,it would be possible to use these haplotypeeffects for selecting candidates some generationsafter the initial estimation withoutrelying on phenotypes (Meuwissen, Hayesand Goddard, 2001). Recombination will

Chapter 17 – Marker-assisted selection in fish and shellfish breeding schemes 353erode the initial LD and therefore it isexpected that accuracy of estimating thebreeding value of many haplotypes willdecay (Zhang and Smith, 1992), the extentof the erosion being dependent on severalpopulation parameters (Meuwissen,Hayes and Goddard, 2001). In practice, theresponse to selection obtained needs to beverified in each generation; thus, re-estimationcan be used based on a random sampleof individuals from the population.One possible caveat is that by assuminga certain mode of gene action (i.e. onlyadditive effects), there may in fact be a morecomplicated genetic architecture influencingquantitative traits. For example, whenestimating dominance and epistasis with thesame data, more haplotype effects need tobe estimated. Therefore, it is likely that theaccuracy of individual effects will decrease.Another potential complication that ariseswhen the true model involves non-additiveeffects is that assignment of potential matesneeds to be optimized to increase the meanphenotype of the population simultaneouslythrough heterosis arising from combinationof different QTL alleles. In the long term,the frequency of homozygotes that areidentical-by-descent will increase withinthe population as a whole; consequently,methods are required to constrain the ratesof inbreeding to obtain similar changes ofthe population mean across generations.Furthermore, expression of differentcombinations of alleles after selection willrequire re-estimation of between-haplotypeeffects in each generation.ConclusionQTL mapping and MAS are not as welladvanced in aquaculture species as interrestrial plants and animals. However,the merger between genetics and genomicsis expected to be a fertile area of researchin the coming years due to the plethora ofinformation that is currently being gatheredby many laboratories around the world.It is through these research efforts thatvariations affecting complex traits in fishand shellfish species may be detected andused for increasing the usefulness of MASschemes. In the final analysis, however, allthese techniques must be cost-effective ifthey are to be profitable in actual breedingprogrammes.AcknowledgementsI am indebted for comments from Mr EricHallerman who greatly helped clarify themanuscript. This chapter has been fundedpartially by: INNOVA-Corporación deFomento de la Producción (CORFO)(05CT6 PP-10) on “Using functionalgenomics for understanding disease resistancein salmon” from the Governmentof Chile; Vicerrectoria de Investigaciony Creación, Universidad de Chile hatchproject 03/03312; and Fondo de Cienciias yTecnologia (FONDECYT) 1061190.ReferencesAnderson, E.C. & Thompson, E.A. 2002. A model-based method for identifying specieshybrids using multilocus genetic data. Genetics 160: 1217–1229.Arkush, K.D., Giese, A.R., Mendonca, H.L., McBride, A.M., Marty, G.D. & Hedrick, P.W. 2002.Resistance to three pathogens in the endangered winter-run Chinook salmon (Oncorhynchustshawytscha): effects of inbreeding and major histocompatibility complex genotypes. Can. J.Fisheries and Aquatic Sci. 59: 966–975.

354Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishBernatchez, L. & Landry, C. 2003. MHC studies in nonmodel vertebrates: what have we learnedabout natural selection in 15 years? J. Evolut. Biol. 16: 363–377.Bijma, P., van Arendonk, J.A.M. & Bovenhuis, J.H. 1997. Breeding value and variance componentestimation from data containing inbred individuals: application to gynogenetic families in commoncarp.(Cyprinus carpio L.). Genet.145: 1243–1249.Bongers, A.B., Bovenhuis, J.H., van Stokkom, A.C., Wiegertjes, G.F., Zandieh-Doulabi, B.,Komen, J. & Ritcher, C.J.J. 1997. Distribution of genetic variance in gynogenetic or androgeneticfamilies. Aquaculture 153: 225–238.Brown, C.R., Woolliams, J.A. & McAndrew, B.J. 2005. Factors influencing effective population sizein commercial populations of gilthead seabream, Sparus aurata. Aquaculture 247: 219–225.Butler, K., Field, C., Herbinger, C.M. & Smith, B.R. 2004. Accuracy, efficiency and robustnessof four algorithms allowing full sibship reconstruction from DNA marker data. Mol. Ecol. 13:1589–1600.Carlborg, Ö., De Koning, D.J., Manly, K.F., Chesler, E., Williams, R.W. & Haley, C.S. 2005.Methodological aspects of the genetic dissection of gene expression. Bioinformatics 21: 2383–2393.Castro, J., Bouza, C., Presa, P., Pino-Querido, A., Riaza, A., Ferreiro, I., Sanchez, L. & Martinez,P. 2004. Potential sources of error in parentage assessment of turbot (Scophthalmus maximus) usingmicrosatellite loci. Aquaculture 242: 119–135.Christiakov, D.A., Hellemans, B., Haley, C.S., Law, A.S., Tsigenopoulos, C.S., Kotoulas, G.,Bertotto, D., Libertini, A. & Volckaert, F.A.M. 2005. A microsatellite linkage map of theEuropean seabass, Dicentrarchus labrax L. Genetics 170: 1821–1826.Cnaani, A., Hallerman, E.M., Ron, M., Weller, J.I., Indelman, M., Kashi, Y., Gall, G.A.E. &Hulata, G. 2003. Detection of a chromosomal region with two quantitative trait loci, affecting coldtolerance and fish size, in an F 2 tilapia hybrid. Aquaculture 223: 117–128.Cnaani, A., Zilberman, N., Tinman, S., Hulata, G. & Ron, M. 2004 Genome-scan analysis for quantitativetrait loci in an F 2 tilapia hybrid. Mol. Genet. Genom. 272: 62–172.Coimbra, M.R., Kobayashi, M.K., Koretsugu, S., Hasegawa, O., Ohara, E., Ozaki, A., Sakamoto,T., Naruse, K. & Okamoto. N. 2003. A genetic linkage map of the Japanese flounder, Paralichthysolivaceus. Aquaculture 220: 203–218.Corley-Smith, G.E., Lim, C.H.J. & Brandhorst, B.P. 1996. Production of androgenetic zebrafish(Danio rerio). Genetics 142: 1265–1276.Coulibaly, I., Gharbi, K., Danzmann, R., Yao, J. & Rexroad, C.E. 2005. Characterization andcomparison of microsatellites derived from repeat-enriched libraries and expressed sequence tags.Anim. Genet. 36: 309–315.Doyle, R.W. & Herbinger, C.M. 1994. The use of DNA fingerprinting for high intensity, withinfamilyselection in fish breeding. Proc. 5th Wrld. Congr. Genetics Appl. Livestock Prodn., Vol. 19,pp. 364–371. Guelph, Ontario.Emery, W., Boyle, C. & Noble, F. 2001. Assignment of paternity groups without access to parentalgenotypes: multiple mating and development plasticity in squid. Molec. Ecol. 10: 1265–1278.Falconer, D.S. & Mackay, T.F.C. 1996. Introduction to quantitative genetics, 4th edition. Harrow,Essex, U.K, Longman.Fernandez, J. & Toro. M. 2006. A new method to estimate relatedness using molecular markers.Molec. Ecol. (in press).

Chapter 17 – Marker-assisted selection in fish and shellfish breeding schemes 355Garcia de Leon, F.J., Canonne, M., Quillet, E., Bonhomme, F. & Chatain, B. 1998. The applicationof microsatellite markers to breeding programs in the sea bass, Dicentrarchus labrax. Aquaculture159: 303–316.George, A., Visscher, P. & Haley, C.S. 2000. Mapping quantitative trait loci in complex pedigrees: atwo step variance component approach. Genetics 156: 2081–2092.Gilbey, J., Verspoor, E., McLay, A. & Houlihan, D. 2004. A microsatellite linkage map for Atlanticsalmon (Salmo salar). Anim. Genet. 5: 98–105.Gjedrem, T. 2000. Genetic improvement of cold-water fish species. Aquaculture Res. 31: 25–33.Gjerde, B. 1986. Growth and reproduction in fish and shellfish. Aquaculture 57: 37–55.Gjerde, B. & Schaeffer, L.R. 1989. Body traits in rainbow trout: II. Estimates of heritabilities and ofphenotypic and genetic correlations. Aquaculture 80: 25–44.Gjerde, B., Villanueva, B. & Bentsen, H.B. 2002. Opportunities and challenges in designing sustainablefish breeding programs. Proc. 7th Wrld. Congr. Genetics Appl. Livestock Prodn., CD-ROM06-01. Montpellier, France.Gjøen, H. & Bentsen, H. 1997. Past present and future of genetic improvement in salmon aquaculture.ICES J. Marine Sci. 54: 1009–1014.Gjøen, H.M., Refstie, T., Ulla, O. & Gjerde, B. 1997. Genetic correlations between survival ofAtlantic salmon in challenge and field tests. Aquaculture 158: 277–288.Griffing, B. 1967. Selection in reference to biological groups. I. Individual and group selection appliedto populations of unordered groups. Austr. J. Biol. Sci. 20: 127–139.Grimholt, U., Larsen, S., Nordmo, R., Midtlyng, P., Kjoeglum, S., Storset, A., Saebo, S. & Stet, R.J.2003. MHC polymorphism and disease resistance in Atlantic salmon (Salmo salar); facing pathogenswith single expressed major histocompatibility class I and class II loci. Immunogen. 55: 210–219.Hayes, B. & Goddard, M.E. 2001. The distribution of the effects of genes affecting quantitative traitain livestock. Genet. Sel. Evol. 33: 209–229.Hayes, B., Lærdahl, J., Lien, S., Adzhubei, A. & Høyheim, B. 2004. Large scale discovery of singlenucleotide polymorphism (SNP) markers in Atlantic salmon (Salmo salar). AKVAFORSK,Institute of Aquaculture Research (available at www.akvaforsk.no/norsk/HentFil.asp?fid=41).Henryon, M., Jokumsen, A., Berg, P., Lund, I., Pedersen, P.B., Olesen, N.J. & Slierendrecht, W.J.2002. Genetic variation for growth rate, feed conversion efficiency and disease resistance existswithin a farmed population of rainbow trout. Aquaculture 209: 59–76.Henryon, M., Berg, P., Olesen, N.J., Kjaer, T.E., Slierendrecht, W.J., Jokumsen, A. & Lund. I. 2005.Selective breeding provides an approach to increase resistance of rainbow trout (Onchorhynchusmykiss) to the diseases, enteric redmouth disease, rainbow trout fry syndrome and viral haemorrhagicsepticaemia. Aquaculture 250: 621–636.Hershberger, W.K., Myers, J.M., Iwamoto, R.N., Mcauley, W.C. & Saxton, A.M. 1990. Geneticchanges in the growth of coho salmon (Oncorhynchus kisutch) in marine net-pens, produced byten years of selection. Aquaculture 85: 187–197.Herbinger, C.M., Doyle, R.W., Pitman, E.R., Paquet, D., Mesa, K.A., Morris, D.B., Wright, J.M.& Cook, D. 1995. DNA fingerprint-based analysis of paternal and maternal effects of offspringgrowth and survival in communally reared rainbow trout. Aquaculture 137: 245–256.Herbinger, C.M., O’Reilly, P.T., Doyle, R.W., Wright, J.M. & O’Flynn, F. 1999. Early growth performanceof Atlantic salmon full-sib families reared in single family tanks versus in mixed familytanks. Aquaculture 173: 105–116

356Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishHoubert, S. & Hedgecock, D. 2004. Linkage maps of microsatellite DNA markers for the Pacificoyster (Crassostrea gigas) Genetics 168: 351–362.Hubner, N., Wallace, C.A., Zimdahl, H., Petretto, E., Schulz, H., Maciver, F., Mueller, M., Hummel,O., Monti, J., Zidek, V., Musilova, A., Kren, V., Causton, H., Game, L., Born, G., Schmidt. S.,Muller, A., Cook, S.A., Kurtz, T.W., Whittaker, J., Pravenec, M. & Aitman, T.J. 2005. Integratedtranscriptional profiling and linkage analysis for identification of genes underlying disease. NatureGenet. 37: 243–253.Hulata, G. 2001. Genetic manipulations in aquaculture: a review of stock improvement by classicaland modern technologies. Genetica 111: 155–173.Jamieson, A. & Taylor, S.T. 1997. Comparison of three probability formulae for parentage exclusion.Anim. Genet. 28: 397–400.Jansen, R.C., Geerlings, H., Van Oeveren, A.J. & Van Schaik R.C. 2001. A comment on codominantscoring of AFLP markers. Genetics 158: 925–926.Jansen, R.C. & Nap, J.P. 2001. Genetical genomics: the added value from segregation. Trends Genet.17: 388–391.Jiang, C. & Zeng, Z.B. 1995. Multiple trait analysis of genetic mapping for quantitative trait loci.Genetics 140: 1111–1127.Kause, A., Ritola, O., Paananen, T., Ritola, O., Paananen, T., Mäntysaari, E. & Eskelinen, U.2002. Coupling body weight and its composition: a quantitative genetic analysis in rainbow trout.Aquaculture 211: 65–79.Kause, A., Ritola, O., Paananen, T., Wahlroos, H. & Mantysaari, E.A. 2005. Genetic trends ingrowth, sexual maturity and skeletal deformations and rate of inbreeding in a breeding programmefor rainbow trout (Oncorhynchus mykiss). Aquaculture 247: 177–187.Khoo, S.K., Ozaki, A., Nakamura, F., Arakawa, T., Ishimoto, S., Nickolov, R., Sakamoto, T.,Akutsu, T., Mochizuki, M., Denda, I. & Okamoto, N. 2004. Identification of a novel chromosomalregion associated with infectious hematopoietic necrosis (IHN) resistance in rainbow trout.Fish Path. 39: 95–102.Kjøglum, S., Grimholt, U. & Larsen S. 2005. Non-MHC genetic and tank effects influence diseasechallenge tests in Atlantic salmon (Salmo salar). Aquaculture 250: 102–10.Knott, S.A. & Haley, C.S. 2000. Multitrait least squares for quantitative trait loci detection. Genetics156: 899–911.Knudsen, S. 2002. A biologist’s guide to analysis of DNA microarray data. New York, NY, USA, JohnWiley and Sons.Kocher, T. D., Lee, W. J., Sobolewska, H., Penman, D. & McAndrew, B. 1998. A genetic linkage mapof a cichlid fish, the tilapia (Oreochromis niloticus). Genetics 148: 1225–1232.Lande, R. & Thompson, R. 1990. Efficiency of marker-assisted selection in the improvement ofquantitative traits. Genetics 124: 743–756.Langefors, J., Lohm, M. & Grahn, O. 2001. Association between major histocompatibility complexclass IIB alleles and resistance to Aeromonas salmonicida in Atlantic salmon. Proc. R. Soc. Lond. B268: 479–485.Lee, B.Y., Lee, W.J., Streelman, J.T., Carleton, K.L., Howe, A.E., Hulata, G., Slettan, A., Stern, J.E.,Terai, Y. & Kocher, T.D. 2005. A second-generation genetic linkage map of tilapia (Oreochromisspp). Genetics 170: 237–244.

Chapter 17 – Marker-assisted selection in fish and shellfish breeding schemes 357Li, Y., Byrne, K., Miggiano E., Whan, V., Moore, S., Keys, S., Crocos, P. & Lehnert, S. 2003.Genetic mapping of the kuruma prawn Penaeus japonicus using AFLP markers. Aquaculture 219:143–156.Liu, Y., Jansen, G.B. & Lin, C.Y. 2002. The covariance between relatives conditional on geneticmarkers. Genet. Select. Evol. 34: 657–667.Liu, Z.J. & Cordes, J.F. 2004. DNA marker technologies and their applications in aquaculturegenetics. Aquaculture 238: 1–37.Lohm, J., Grahn, M., Langefors, A., Andersen, O., Storset, A. & von Schantz, T. 2002.Experimental evidence for major histocompatibility complex allele-specific resistance to a bacterialinfection. Proc.R. Soc. Lond. B 269: 2029–2033.Lynch, M. & Ritland, K. 1999. Estimation of pairwise relatedness with molecular markers. Genetics152: 1753–1766.Marth, G.T., Korf, I., Yandell, M.D., Yeh, R.T., Gu, Z., Zakeri, H., Stitziel, N.O., Hillier, L., Kwok,P.Y. & Gish, W.R. 1999. A general approach to single-nucleotide polymorphism discovery. NatureGenet. 23: 452–456.Martinez, V. 2003. Quantitative genetic analysis of fish populations using molecular markers.University of Edinburgh, UK (Ph.D. thesis).Martinez, V. 2006a. Using marker data in conventional breeding programmes: a case of study inselective breeding programs of scallops. In Proc. Internat. Assocn. Genet. Aquaculture, p 66.Montpellier, France.Martinez, V. 2006b. Further insights of the variance component method to detect QTL in fishpopulations: relaxing the assumption of additive genetic effects. In Proc. Internat. Assocn. Genet.Aquaculture, p. 66. Montpellier, France.Martinez, V. 2006c. Importance and implementation of molecular technologies in selective breedingprograms for aquaculture species. In Proc. 8th Wrld. Congr. Genet. Appl. Livest. Prodn.,Communication 09-01. Minas Gerais, Brazil.Martinez, V. & di Giovanni. S. 2006. Breeding programmes of scallops: effect of self-fertilizationwhen estimating genetic parameters. In Proc. Internat. Assocn. Genet. Aquaculture, p.65.Montpellier, France.Martinez, V., Hill, W.G. & Knott, S.A. 2002. On the use of double haploids for detecting QTL inoutbred populations. Hered. 88: 423–431.Martinez, V., Neira, R. & Gall, G.A.E. 1999. Estimation of genetic parameters from pedigreedpopulations: lessons from analysis of alevin weight in coho salmon (O. kisutch). Aquaculture 330:22–30.Martinez, V., Sillanpaa, M., Thorgaard, G., Robison, B., Woolliams, J. & Knott, S. 2002. Evidenceof a pleiotropic QTL influencing components of early development in double haploid lines ofrainbow trout. Proc. 7th Wrld. Congr. Genet. Appl. Livestock Prodn., CD ROM Communication06-08. Montpellier, France.Martinez, V., Thorgaard, G., Robison, B. & Sillanpää, M. 2005. Simultaneous consideration ofenvironmental and genetic effects in QTL mapping: an application to early development in doublehaploid lines of rainbow trout. Genet. Res. 86: 209–221.Martinez, V., Kause, A., Mäntysaari, E. & Mäki-Tanila, A. 2006a. The use of alternative breedingschemes to enhance genetic improvement in rainbow trout: I. One-stage selection. Aquaculture254: 182–194.

358Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishMartinez, V., Kause, A., Mäntysaari, E. & Mäki-Tanila, A. 2006b. The use of alternative breedingschemes to enhance genetic improvement in rainbow trout: II. Two-stage selection. Aquaculture254: 195–202.McCune, A.R., Fuller, R.C., Aquilina, A.A., Dawley, R.M., Fadool, J.M., Houle, D., Travis, J.& Kondrashov, A.S. 2002. A low genomic number of recessive lethals in natural populations ofbluefin killifish and zebrafish. Science 296: 2398–2401.Meuwissen, T.H.E. 1997. Maximizing the response of selection with predefined rate of inbreeding. J.Anim. Sci. 75: 934–940.Meuwissen, T.H.E. 2003. Genomic selection the future of marker-assisted selection and animalbreeding. In Marker-assisted selection: a fast track to increase genetic gain in plant and animalbreeding? (available at www.fao.org/biotech/torino.htm).Meuwissen, T.H.E. & Goddard, M.E. 2001. Prediction of identity by descent probabilities frommarker-haplotypes. Genet. Select. Evol. 33: 605–634.Meuwissen, T.H.E. & Goddard, M.E. 2004. Mapping multiple QTL using linkage disequilibrium andlinkage analysis information and multitrait data. Genet. Select. Evol. 36: 261–279.Meuwissen, T.H.E. & Woolliams, J.A. 1994. Effective sizes of livestock populations to prevent adecline in fitness. Theor. Appl. Genet. 89: 1019–1026.Meuwissen, T.H.E., Hayes, B. & Goddard, M. 2001. Prediction of total genetic value using genomewidedense marker maps. Genet. 157: 1819–1829.Moav, R. & Wohlfarth, G. 1976. Two-way selection for growth rate in the common carp (Cyprinuscarpio L.). Genet. 82: 83–101.Moen, T., Hoyheim, B., Munck, H. & Gomez-Raya, L. 2004a. A linkage map of Atlantic salmon(Salmo salar) reveals an uncommonly large difference in recombination rate between the sexes.Anim. Genet. 35: 81–92.Moen, T., Kjersti, K.T., Fjalestad, H. & Gomez-Raya, L. 2004b. A multistage testing strategy fordetection of quantitative trait loci affecting disease resistance in Atlantic salmon. Genet. 167:851–858.Moen, T., Agresti, J.J., Cnaani, A., Moses, H., Famula, T.R., Hulata, G., Gall, G.A.E. & May, B.2004c. A genome scan of a four-way tilapia cross supports the existence of a quantitative trait locusfor cold tolerance on linkage group 23. Aquacult. Res. 35: 893–904.Morton, R. & Howarth, J.M. 2005. Estimation of breeding values using selected pedigree records.Biomet. 61: 422–432.Mousseau, T.A., Ritland, K. & Heath, D.D. 1998. A novel method for estimating heritability usingmolecular markers. Hered. 80: 218–224.Muir, W.M. 2005. Incorporation of competitive effects in forest tree or animal breeding programs.Genet. 170: 1247–1259.Muranty, H. 1996. Power of tests for quantitative trait loci detection using full-sib families in differentschemes. Hered. 76: 156–165.Ng, S.-H., Chang, S., Brown, A., Koop, B. & Davidson, W.S. 2005. Type I microsatellite markersfrom Atlantic salmon (Salmo salar) expressed sequence tags. Mol. Ecol. Notes 5: 762–766.Nichols, K.M., Young, W.P., Danzmann, R.G., Robison, B.D., Rexroad, C., Noakes, M., Phillips,R.B., Bentzen, P., Spies I., Knudsen, K., Allendorf, F.W., Cunningham, B.M., Brunelli, J., Zhang,H., Ristow, S., Drew, R., Brown, K.H., Wheeler, P.A. & Thorgaard, G.H. 2003. A consolidatedgenetic linkage map for rainbow trout (Oncorhynchus mykiss). Anim. Genet. 34: 102–115.

Chapter 17 – Marker-assisted selection in fish and shellfish breeding schemes 359Norris, A.T., Bradley, D.G. & Cunningham, E.P. 2000. Parentage and relatedness determination infarmed Atlantic salmon (Salmo salar) using microsatellite markers. Aquaculture 182: 73–83.O’Brien, S.J. 1991. Molecular genome mapping: lessons and prospects. Curr. Opin. Genet. Devel. 1:105–111.Ollivier, L. 1998. The accuracy of marker-assisted selection for quantitative traits within populationsin linkage equilibrium. Genet. 148: 1367–1372.Ozaki, A., Sakamoto, T., Khoo, S., Nakamura, K., Coimbra, M.R., Akutsu, T. & Okamoto, N.2001. Quantitative trait loci (QTLs) associated with resistance/susceptibility to infectious pancreaticnecrosis virus (IPNV) in rainbow trout (Oncorhynchus mykiss). Mol. Genet. Genom. 265:23–31.Panitz, F., Holm, L.-E., Høyheim, B., Hagen-Larsen, H., Taggart, J., Cairney, M., Davey, G.C.,Powell, R. & Bendicen, C. 2002. EST sequencing in Atlantic salmon. Proc. 7th Wrld. Congr. Genet.Appl. Livestock Prodn. CD-ROM Communication No 06-14. Montpellier, France.Pante, M.J.R, Gjerde, B. & McMillan, I. 2001. Effect of inbreeding on body weight at harvest inrainbow trout, Oncorhynchus mykiss. Aquaculture 192: 201–211.Pante, M.J.R., Gjerde, B., McMillan, I. & Mistzal, I. 2002. Estimation of additive and dominancegenetic variances for body weight at harvest in rainbow trout, Oncorhychus mykiss. Aquaculture204: 383–392.Paterson, S., Piertney, S.B. & Knox, D. 2004. Characterization and PCR multiplexing of novelhighly variable tetranucleotide Atlantic salmon (Salmo salar L.) microsatellites. Mol. Ecol. Notes4: 160–162.Pérez-Enciso, M. 2003. Fine mapping of complex trait genes combining pedigree and linkage disequilibriuminformation: a Bayesian unified framework. Genet. 163: 1497–1510.Pérez-Enciso, M. 2004. In silico study of transcriptome genetic variation in outbred populations.Genet. 166: 547–554.Pérez-Enciso, M., Toro, M.A., Tenenhaus, M. & Gianola. D. 2003. Combining gene expression andmolecular marker information for mapping complex trait genes: a simulation study. Genet. 164:1597–1606.Pfrender, M.E., Spitze, K., Hicks, J., Morgan, K.L. & Lynch, L.M. 2000. Lack of concordancebetween genetic diversity estimates at the molecular and quantitative-trait levels. Conserv. Genet.1: 263–269.Piepho, H-P. & Koch, G. 2000. Codominant analysis of banding data from a dominant markersystem by normal mixtures. Genet. 155: 1459–1468.Pong-Wong, R., Villanueva, B. & Woolliams, J.A. 2002. Comparison of direct and marker-assistedselection with optimised contributions. In Proc. 7th Wrld. Congr. Genet. Appl. Livestock Prodn.CD-ROM Communication No. 33-02. Montpellier, France.Pong-Wong, R., George, A.W., Woolliams, J.A. & Haley C.S. 2001 A simple and rapid methodfor calculating identity-by-descent matrices using multiple markers. Genet. Select. Evol. 33:453–471.Ponzoni, R.W., Hamzah, A., Tan, S. & Kamaruzzaman, N. 2005. Genetic parameters and responseto selection for live weight in the GIFT strain of Nile tilapia (Oreochromis niloticus). Aquaculture247: 203–210.Queller, D.C. & Goodnight, K.F. 1989. Estimating relatedness using genetic markers. Evol. 43:258–275.

360Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishQuinton, C., McMillan, I. & Glebe, B.D. 2005. Development of an Atlantic salmon (Salmo salar)genetic improvement program: Genetic parameters of harvest body weight and carcass qualitytraits estimated with animal models. Aquaculture 247: 211–217.Reid, D.P., Szanto, A., Glebe, B., Danzmann, R.G. & Ferguson, M.M. 2005. QTL for body weightand condition factor in Atlantic salmon ( Salmo salar): comparative analysis with rainbow trout(Oncorhynchus mykiss) and Arctic charr (Salvelinus alpinus). Hered. 94: 166–172.Rexroad, C.E., Rodriguez, M.F., Gharbi, K., Danzmann, R.G., DeKoning, J., Phillips, R. & Palti,Y. 2005. Comparative mapping of expressed sequence tags containing microsatellites in rainbowtrout (Oncorhynchus mykiss). BMC Genom. 6: 54.Rise, M.L., Von Schalburg, K.R., Brown, G.D., Mawer, M.A., Devlin, R.H., Kuipers, N., Busby,M., Beetz- Sargent, M., Alberto, R., Gibbs, A.R., Hunt, P., Shukin, R., Zeznik, J.A., Nelson,C., Jones, S.R., Smailus, D.E., Jones, S.J., Schein, J.E., Marra, M.A., Butterfield, Y.S., Stott,J.M., Ng, S.H., Davidson, W.S. & Koop, B.F. 2004a. Development and application of a salmonidEST database and cDNA microarray:data mining and interspecific hybridization characteristics.Genome Res. 14: 478–490.Rise, M.L., Jones, S.R., Brown G.D., von Schalburg, K.R., Davidson, W.S. & Koop, B.F. 2004b.Microarray analyses identify molecular biomarkers of Atlantic salmon macrophage and hematopoietickidney response to Piscirickettsia salmonis infection. Physiol. Genom. 20: 21–35.Robison, B., Wheeler, P. & Sundin, K. 2001. Composite interval mapping reveals a QTL of major effecton embryonic development rate in rainbow trout (Onchorhynchus mykiss). J. Hered. 92: 16–22.Robison, B., Wheeler, P. & Thorgaard, G. 1999. Variation in development rate among clonal lines ofrainbow trout (Oncorhynchus mykiss). Aquaculture 173: 131–141.Rodriguez, F.M., LaPatra, S., Williams, S., Famula, T. & May, B. 2004. Genetic markers associatedwith resistance to infectious hematopoietic necrosis in rainbow and steelhead trout (Oncorhynchusmykiss) backcrosses. Aquaculture 241: 93–115.Rye, M. & Mao, I.L. 1998. Nonadditive genetic effects and inbreeding depression for body weight inAtlantic salmon (Salmo salar L.). Livest. Prodn. Sci. 57: 23–32.Sakamoto, T., Danzmann, R.G., Gharbi, K., Howard, P., Ozaki, A., Khoo, S.K., Woram, R.A.,Okamoto, N., Ferguson, M.M., Holm, L-E., Guyomard, R. & Hoyheim, B. 2000. A microsatellitelinkage map of rainbow trout (Oncorhynchus mykiss) characterized by large sex-specificdifferences in recombination rates. Genet. 155: 1331–1345.Serapion, J., Waldbieser, G.C., Wolters, W. & Wolters, Z.J. 2004. Development of Type I markers inchannel catfish through intron sequencing. Anim. Genet. 35: 463–466.Singer, A., Perlman, H., Yan, Y., Walker, C., Corley-Smith, G., Brandhorst, B. & Postlethwait, J.2002. Sex-specific recombination rates in zebrafish (Danio rerio). Genet. 160: 649–657.Smith, B.R., Herbinger, C.M. & Merry, H.R. 2001. Accurate partition of individuals into full-sibfamilies from genetic data without parental information. Genet. 158: 1329–1338.Somorjai, I.M.L., Danzmann, R.G.& Ferguson, M.M. 2003. Distribution of temperature tolerancequantitative trait loci in Arctic charr (Salvelinus alpinus) and inferred homologies in rainbow trout(Oncorhynchus mykiss). Genetics 165: 1443–1456.Streisinger, G.,Walker, C.H., Dower, N., Knauber, D. & Singer, F. 1980. Production of clones ofhomozygous diploid zebra fish (Brachydanio rerio). Nature 291: 293–296.Sun, X. & Liang, L. 2004. A genetic linkage map of common carp (Cyprinus carpio L.) and mappingof a locus associated with cold tolerance. Aquaculture 238: 165–172.

Chapter 17 – Marker-assisted selection in fish and shellfish breeding schemes 361Sundin, K., Brown, K.H., Drew, R.E., Nichols, K.M., Wheeler, P.A. & Thorgaard, G.H. 2005.Genetic analysis of a development rate QTL in backcrosses of clonal rainbow trout, Oncorhynchusmykiss. Aquaculture 247: 75–83.Tanck, M.W.T., Palstra, A.P., van de Weerd, M., Leffering, C.P., van der Poel, J.J., Bovenhuis, H.& Komen, J. 2001. Segregation of microsatellite alleles and residual heterozygosity at single loci inhomozygous androgenetic common carp (Cyprinus carpio L.). Genome 44: 743–751.Tao, W.J. & Boulding, E.G. 2003. Associations between single nucleotide polymorphisms in candidategenes and growth rate in Arctic charr (Salvelinus alpinus L.). Hered. 91: 60–69.Taris, N., Baron, S. & Sharbel, T.F. 2005. A combined microsatellite multiplexing and boiling DNAextraction method for high-throughput parentage analyses in the Pacific oyster (Crassostrea gigas).Aquaculture Res. 36: 516–518.Thomas, S. C., Pemberton, J.M. & Hill, W.G. 2000. Estimating variance components in natural populationsusing inferred relationships. Hered. 84: 427–436.Thomas, S.C. & Hill, W.G. 2000. Estimating quantitative genetic parameters using sib-ships reconstructedfrom marker data. Genet. 155: 1961–1972.Thorgaard, G.H. & Allen S.K. 1987. Chromosome manipulation and markers in fishery management.In N. Ryman & F.M. Utter, eds. Population genetics and fishery management, pp. 319–331.Seattle, WA, USA, University of Washington Press.Toro, M. & Mäki-Tanila, A. 1999. Establishing a conservation scheme. In J.K. Oldenbroek, ed. Genebanks and the conservation of farm animal genetic resources, pp.75–90. Lelystad, Netherlands,DLO Institute of Animal Health.Toro, M., Barragán, C. & Óvilo, C. 2002. Estimation of coancestry in Iberian pigs using molecularmarkers. Conserv. Genet. 3: 309–320.Toro, J.E., Sanhueza, M.A., Winter, J.E., Aguila, P. & Vergara, A.M. 1995 Selection response andheritability estimates for growth in the Chilean oyster Ostrea chilensis (Philippi, 1845). J. ShellfishRes. 14: 87–92.Vasemägi A., Nilsson, J. & Primmer, C.R. 2005. Expressed sequence tag (EST)-linked microsatellitesas a source of gene associated polymorphisms for detecting signatures of divergent selection inAtlantic salmon (Salmo salar L.). Mol. Biol. & Evol. 22: 1067–1076.Villanueva, B.,Verspoor, E. & Visscher, P.M. 2002. Parental assignment in fish using microsatellitegenetic markers with finite numbers of parents and offspring. Anim. Genet. 33: 33–41.Villanueva, B., Pong-Wong, R. & Woolliams, J.A. 2002. Marker assisted selection with optimisedcontributions of the candidates to selection. Genet. Select. & Evol. 34: 679–703.Vos, P., Hogers, R., Bleeker, M., Reijans, M., Van De Lee, T., Hornes, M., Friters, A., Pot, J.,Paleman, J., Kuiper, M. & Zabeau, M. 1995. AFLP: A new technique for DNA fingerprinting.Nucl. Acids Res. 23: 4407–4414Waldbieser, G.C., Bosworth, B.G., Nonneman, D.J. & Wolters, W.R. 2001. A microsatellite-basedgenetic linkage map for channel catfish, Ictalurus punctatus. Genet. 158: 727–734.Walsh, B. & Henderson, D.J. 2004. Microarrays and beyond: What potential do current and futuregenomics tools have for breeders? Anim. Sci. 82: E292–E299.Wang, J. 2004. Sibship reconstruction from genetic data with typing errors. Genet. 166: 1963–1979.Wang, L., Song, L., Chang, Y. & Xu, W. 2005. A preliminary genetic map of Zhikong scallop(Chlamys farreri, Jones et Preston 1904). Aquaculture Res. 36: 643–653.

362Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishWard, R D., English, L. & McGoldrick, J. 2000. Genetic improvement of the Pacific oyster,Crassostrea gigas (Thunberg) in Australia. Aquaculture Res. 31: 35–44.Withler, R. & Beacham, T. 1984. Genetic consequences of the simultaneous or sequential addition ofsemen from multiple males during hatchery spawning of chinook salmon (Oncorhynchus tshawytscha).Aquaculture 126: 11–20.Wright, J.M. & Bentzen, P. 1994. Microsatellites – genetic markers for the future. Rev. Fish Biol. &Fisheries 4: 384–388.Wu, W.-R., Li, W.-M., Tang, D.-Z., Lu, H.-R. & Worland, A.J. 1999. Time-related mapping of quantitativetrait loci underlying tiller number in rice. Genet. 151: 297–303.Xu, S. & Atchley, W. 1995. A random model approach to interval mapping of quantitative trait loci.Genet. 141: 1189–1197.Xu, S. & Gessler, D. 1998. Multipoint genetic mapping of quantitative trait loci using a variablenumber of sibs per family. Genet. Res. 71: 73–83.Young, W.P., Wheeler, P.A., Fields, R.D. & Thorgaard, G.H. 1996. DNA fingerprinting confirmsisogenicity of androgenetically-derived rainbow trout lines. J. Hered. 87: 77–81.Young, W.P., Wheeler, P.A., Coryell, V.H., Keim, P. & Thorgaard, G.H. 1998 A detailed linkage mapof rainbow trout produced using doubled haploids. Genet. 148: 839–850.Yue, G.H., Ho., M.Y. & Orban, L. 2004. Microsatellites within genes and ESTs of common carp andtheir applicability in silver crucian carp. Aquaculture 234: 85–98.Zhang, W. & Smith, C. 1992. Computer simulation of marker assisted selection utilizing linkage disequilibrium.Theor. Appl. Genet. 83: 813–820.Zimmerman, A.M., Wheeler, P.A., Ristow, S.S. & Thorgaard, G.H. 2005. Composite interval mappingreveals three QTL associated with pyloric caeca number in rainbow trout. Aquaculture 247:85–95.

Section VISelected issues relevant toapplications of marker-assistedselection in developing countries

Chapter 18Marker-assisted selection in cropand livestock improvement: howto strengthen national researchcapacity and internationalpartnershipsMaurício Antônio Lopes

366Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishSummaryIt is generally recognized that marker-assisted selection (MAS) is a tool that breeders canuse to accelerate the speed and precision of crop and livestock breeding in developingcountries. However, its practical application has been more difficult than previouslyexpected. Although advances in molecular marker technology have uncovered manypossibilities for transferring genes into desired crops and livestock through MAS, moremethodological development and better planning and implementation strategies willbe needed for its successful and expeditious application to breeding programmes. Also,this technology should not be regarded as an end in itself, but as an interacting part ofcomplex strategies and decision-making processes. An appropriate mix of technologiesand capabilities together with effective approaches to networking must be viewed askey ingredients for its correct development and application to breeding programmes.This chapter describes some strategies to guide decisions about structures, methods andcapacities that may contribute to enhancing the access and successful use of MAS indeveloping countries.

Chapter 18 – Marker-assisted selection in crop and livestock improvement 367IntroductionThe tremendous advances made in molecularmarker techniques in the past twodecades have led to increased understandingof the genetic basis of many agriculturaltraits in a variety of plant and animal species.The use of these techniques has alsomade it possible to accelerate the transferof desirable traits among varieties and tointrogress novel genes from related wildspecies.DNA markers have many advantagesover conventional approaches available tobreeders. They are especially advantageousfor traits that are otherwise difficult to tag,such as resistance to pathogens, insects andnematodes, tolerance to abiotic stresses andquality parameters. They offer great scopefor improving the efficiency of conventionalbreeding by carrying out selectionnot directly on the trait of interest but onlinked genomic regions. Additionally thesemarkers are unaffected by environmentalconditions and are detectable during allstages of growth (Mohan et al., 1997).Molecular marker techniques have thereforemoved beyond their early projectedrole as tools for identifying chromosomalsegments and genes to uncovering manypossibilities for easing the transfer of genesinto desired cultivars and lines. MAS generatedgreat enthusiasm as it was seen as amajor breakthrough, promising to overcomemany limitations of conventionalbreeding processes (FAO, 2003). However,despite advances in the theory of MAS,direct utilization of the information it providesfor selecting superior individuals withcomplex traits is still very limited (Young,1999; Ferreira, 2003). Nevertheless, thereis still optimism about the contributionsof MAS, which is now balanced by therealization that genetic improvement ofquantitative traits using this tool may bemore difficult than previously considered(FAO, 2003). In 1999, Young reviewed thedevelopment of MAS, analysing in detailits main drawbacks, many of which remaintoday. He concluded that because MAStechnology was so challenging it should notbe a reason for discouragement but, instead,reason for more ingenuity and better planningand execution.Recent developments in high-throughputgenotyping, single nucleotide polymorphism(SNP) and the integration of genomictechnologies are advances that will play animportant role in the development of MAS asan effective tool for sustainable conservationand increased use of crop genetic resources(Ferreira, 2006). However, research teams,funding agencies, commodity groups andthe private sector will need to work togetherto develop MAS technology further andensure that breeders have the best availabletools. Also, the tools and strategies willneed to go beyond markers themselves toinclude genome-based knowledge derivedfrom model systems, high-throughput costeffective technology, as well as better technologiesand strategies for handling largevolumes of information.The purpose of this chapter is to discussthe access to and utilization of MAStechnology by breeding programmes,with special emphasis on strategies to helpstrengthen research capacity and partnershipsin developing countries. Wheneverpossible, recommendations are presentedto help guide decisions that may contributeto enhancing the access and successful useof MAS by national programmes.Perceptions about the use of MASin Crop and Livestock ImprovementAs MAS is still an evolving technology,there are not many detailed studies availabledescribing the state-of-the-art of its

368Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishapplication to breeding programmes. Also,there are very few prospective studies indicatingfuture trends in the application ofthis technology. The FAO BiotechnologyForum hosted an e-mail conference on“Molecular marker-assisted selection asa potential tool for genetic improvementof crops, forest trees, livestock and fishin developing countries”. This provideda comprehensive overview of the perceptionsof scientists from different parts ofthe world about key aspects of the applicationof MAS to genetic improvementin developing countries (www.fao.org/biotech/logs/c10logs.htm).As described in Chapter 21, this FAOconference was very inclusive, with a totalof 627 people subscribing. Eight percent ofthese (52 people) submitted 85 messages,which were received from all major regionsof the world, including Asia (33 percent),Europe (26 percent), Latin America and theCaribbean (14 percent), Africa (9 percent)Oceania (9 percent) and North America(8 percent). People from 26 different countriesparticipated, with a total of 50 messages(59 percent) from developing countries and35 messages (41 percent) from developedcountries. Institutional representation wasalso ample, including national research institutes,centres belonging to the ConsultativeGroup on International AgriculturalResearch (CGIAR), universities, consultants,farmer organizations, governmentagencies, non-governmental organizations(NGOs), etc. Although only 52 people outof 627 subscribers participated directly inthe conference, the number is significantconsidering the broad representation, thehigh level of the (moderated) discussionsand the number of relevant issues discussed(www.fao.org/biotech/logs/c10logs.htm).To prepare this chapter, a detailed reviewwas carried out of the conference results inan attempt to capture the main perceptionsand concerns related to access to and utilizationof MAS in developing countries.This analysis revealed a variety of ideas andcreative suggestions to overcome the problemsof MAS utilization. Although thereis a risk of narrowing views on importantissues discussed during the conference, fourmajor perceptions were clear from the richcontent of the discussions:Perception 1. There is a need for developmentof priority-setting mechanisms andcost benefit analysis to guide informeddecisions on how best to apply MAS andother technological innovations to crop andlivestock breeding in developing countries.Perception 2. MAS has to be understoodas part of a complex process.Complementarities, mix of technologies,integration of capabilities and networkingmust always be viewed as key ingredientsfor its correct application in breedingprogrammes.Perception 3. There is a need for an objectivedefinition of public-private functionsand responsibilities in relation to fundingand development of technological innovationin developing countries. Public-privateand north-south partnerships are essentialto accelerate progress and effectiveapplication of MAS and other innovationsto breeding programmes in developingcountries.Perception 4. Developing countries mustfocus on capacity building and humanresource development oriented to shapeeffective strategies of technologicalinnovation.In the following sections, possible strategiesand alternatives to deal with thechallenges and opportunities indicatedabove are outlined, including the need

Chapter 18 – Marker-assisted selection in crop and livestock improvement 369for objective priority-setting, developmentof partnerships, complementarities andcapacity building for compatible humanresource formation.MAS as Part of a ComplexProcess – Setting Priorities andTaking ActionBefore discussion of MAS as a technologicalalternative to increase the capacity of breedingprogrammes it is important to discuss andconsider the future of the breeding processitself. Until recently, selection was based onobservable phenotypes, without knowledgeof the genetic architecture of the selectedcharacteristics (Dekkers and Hospital,2002). However, advances in molecularmarker techniques and rapid advances inlarge-scale sequencing are creating newperspectives for exploiting the immensereservoir of polymorphism in genomes.Molecular genetic analysis of traits in plantand animal populations is leading to a betterunderstanding of quantitative trait genetics.More recently, the discovery and scoring ofsingle nucleotide polymorphisms (SNPs)using automated and high-throughputinstrumentation are already providing theincreased resolution needed to analyse setsof genes involved in complex quantitativetraits (Altshuler et al., 2000; De La Vega etal., 2002; Rafalsky, 2002, Lörz and Wenzel,2005; Ferreira, 2006).What impacts will all these developmentshave on breeding programmes? Asanticipated by Stuber, Polacco and Senior,in 1999, “when genomics is added to futurestrategies for plant and animal breeders,the projected outcomes are mind-boggling.There is every reason to believe that thesynergy of empirical breeding, MAS andgenomics will truly produce a greater effectthan the sum of the various individualactions.” Despite the positive view of manywho find technological development anopen venue for enhancement or completeredesign of traditional breeding, there aremany uncertainties about its future. The riseof genetic engineering and the bio-industry,and the widespread granting of intellectualproperty rights, followed by profoundchanges in the relationship between publicand private science make it very difficultto anticipate future developments in bothpublicly funded breeding research and thecommercial biotechnology industry.Unfortunately, very little effort has beendirected to thinking about the future ofbreeding, especially in developing countries(Castro et al., 2002, 2006). Many past andcurrent events are changing the performance,the relationships and the space that publicand private research organizations have inthe market, raising the need for a deeperunderstanding of their unfolding impacts onthe public activity of research (Price, 1999;Graff et al., 2003). The current scenario ofchanges and uncertainties has generatedthe necessity for strategic re-alignment ofpublic research in many parts of the world.Therefore, research organizations needinformation that is not currently availableabout such changes and influences andtheir impact on the future of key activities,such as crop and livestock breeding. Toobtain and to organize this information,prospective studies need to be developedon the present and future performanceof breeding programmes and their relatedproduction systems.The future configuration of breedingprogrammes depends on knowledgeto guide strategic decisions about structures,methods and capacities in order totake advantage of new opportunities andtechnological niches. Foresight methodologieshave been applied to this end,using systemic analysis of the past and

370Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishFigure 1Conceptual framework of a prospective study on genetic resources and breeding R&D in BrazilCurrent and Emerging EventsDevelopment of biotechnology, widespread granting of intellectual property, changesin the relationship between public and private science, the rise of geneticengineering, development of bio-industry, concentration in the seed market, etc.Public subsystem of geneticresources and breeding R&D(current state)Public subsystem of geneticresources and breeding R&D(future state)Private subsystem of geneticresources and breeding R&D(current state)Private subsystem of geneticresources and breeding R&D(future state)Dark arrows indicate the impact of current and emerging events on both public and private subsystems of R&Din genetic resources and breeding, at present and in the future, considering several alternative scenarios. Verticalarrows indicate the state of the relationship between the public and the private R&D subsystems as it is affectedby current and emerging events.Source: Castro et al., 2002, 2006.present performance of a research field,determining critical factors of performance(Linstone and Turoff, 1975; Castro, deCobbe and Goedert, 1995; Castro, de Limaand Freitas Filho, 1998; Castro et al. 2002,2006; Lima et al., 2000).An innovative model of a prospectivestudy was proposed and tested by Castroet al. (2002, 2006), based on the Braziliannational system of genetic resources andbreeding. The effort started with thedistinction between two component subsystems– public and private. The authorsconsidered that the two subsystems admittwo possible states or situations, currentand future, after the effect of current andemerging events (Figure 1). Prospectiveefforts based on this framework can be veryuseful to guide diagnosis of national programmes,identifying the main determinantsof current and past system performancethat can be used to guide decisions aboutthe configuration of genetic resources,breeding programmes and the associatedseed industry.This type of study can help identifychanges in the system and in the correspondingtechnology market, analysingtheir current and future impacts, determiningfuture opportunities and threatsto the strategic positioning of researchorganizations in the technology market.There is also the perspective of developingpossible alternative scenarios for therelationships between public and privateresearch, and of these with the market, toguide the strategic positioning of publicresearch. Results of this effort couldindicate new opportunities and niches forpublic breeding programmes, as well asareas of extreme value where the publicsector would have to acquire capacity inthe future. Key decisions on investmentsin new technologies and processes applied

Chapter 18 – Marker-assisted selection in crop and livestock improvement 371to genetic resources and breeding research,such as MAS, genomic tools, transgenictechnology and others, are better taken ifthese results are available.The results of this forward-lookingapproach developed in Brazil allowed theidentification of some important trendsthat must be considered by managers inthe process of adapting breeding effortsfor the future (Castro et al., 2002, 2005,2006). Current and emerging events identifiedin the process will certainly affectthe performance, methods, technologicalprocesses, portfolio of products and institutionalrelations in the public and privateR&D sectors dedicated to plant breeding inBrazil. This complexity indicates that it isquite dangerous for developing countries,pressured by market evolution and rapidexpansion of methods and technologies, toface the challenge of identifying priorityareas for investment without a minimumprospective effort.In summary, the ability to predictchanges that might affect the performanceof public and private R&D organizationsis essential for decision-makers and managersto guide adjustments in the focus ofthese sectors in a timely manner, avoidingthreats and promoting access to new toolsand opportunities. Although the same prospectivemethodology may be applied to awide range of countries, it is important topoint out that situations differ drasticallyfrom country to country, thereby requiringexamination of future configuration of asector on a case-by-case basis.MAS as Part of a ComplexProcess – Building Capacities,Complementarities andEnhancing NetworkingMAS cannot be considered an end in itselfor a tool detached from the complexitiesof breeding strategies. It has to beunderstood and analysed in the context ofan interacting mix of tools and strategiesthat have to be targeted towards crop andlivestock improvement in a coordinatedmanner. Independently of the outcome ofany priority-setting effort, the need for anexpanded networking approach to breedingand biotechnological research will alwaysbe an objective to be pursued. This needarises because networking and partnershipsare essential to enable organizationsto attain otherwise unattainable goals, addvalue to their products and processes andreduce costs. Also, the continuous demandfor efficiency and relevance presses R&Dprogrammes to move in the direction ofcooperation and alignment of efforts.One of the key problems limiting theuse of MAS and other advanced technologiesin developing countries is exactlythe difficulty of building effective teamsand networks. Unfortunately, very fewdeveloping countries have trained scientistsand advanced facilities concentrated in oneplace or institution. Usually, these scarceresources are scattered over different placesand institutions, and many times away ordisconnected from the relevant breedingprogrammes. This is a serious drawback asthe increasing interdependence of traditionaland upstream disciplines makes it necessaryto build and manage multidisciplinaryteams consisting of breeders, agronomists,molecular biologists, biochemists, pathologists,entomologists, physiologists, soilscientists, statisticians, etc. – a goal alwaysdifficult to achieve. In addition to the challengeof working within team alignmentsand cooperation, there is the pressing needto develop ways to share capacities, infrastructure,materials and information amongresearch teams located across a country, aregion, or even continents.

372Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishThe main problem in fostering collaborationand effective cooperation to achievecommon goals seems to be the difficulty ofrecognizing that different teams and organizationshave different general interestsand norms. For this reason, competitionusually prevails. While it has been wellaccepted that competition is one of the keyforces that keep industry competitive anddynamic, this view is being challenged bythe concept that many activities can benefitfrom a rational mix of competitionand cooperation that leads to complementaryproducts and expansion of possibilitiesthrough the formation of new relationshipsor even new modes of operation and management.Increasingly, the same is also truefor R&D organizations, which can benefitfrom working with partners (competitors)whose abilities make their own more attractivein the eyes of clients (Brandenburgerand Nalebuff, 1997). Also, faced withgrowing competition from industry andincreasing pressures and demands, publicR&D institutions must look at ways to domore with fewer resources. Collaborationthrough team nets and other networkingstrategies have the potential to reduce costs,add value and promote capacity to respondquickly to changes. Besides, with the newtools of information technology, collaborationwith any part of the world is possibleas this promotes information and otherresource sharing without the need for geographicalproximity (Lipnack and Stamps,1993).How should a R&D organization behavein a multifaceted relationship, when partnerscan be also competitors? Organizationsthat enter competitive collaboration mustbe aware that their partners may be out todisable them. This dilemma has been facedby a growing number of organizations,which rapidly understand that effectivenesswill be more and more a productof recognizing and using interdependence.With networks and interdependent teams,cooperation must be designed in the nameof mutual needs and with a clear sense ofsharing risks to reach objectives that arecommon to all partners (Lopes, 2000).In many parts of the world, includingin developing countries like Brazil, competitivefunding systems for agriculturalR&D are assuming growing importanceas new sources of funding and as driversfor cooperation among universities, R&Dinstitutes and the private sector, in manycases allowing collaboration even amonginstitutions that are traditional competitors(Lopes, 2000). Although the rules and proceduresgoverning the competitive grantingsystem indicate the need for partnership andthe general mode of interaction, experiencehas shown that industry/university/R&Dinstitutes cooperations succeed only if theyare founded on trust and understanding andpromise mutual benefits. Also, successfulexperiences have come from the clear recognitionof objectives and well structuredmanagement with intense communication.Two experiences are described belowthat rely heavily on cooperation and networkingdirected to effective applicationof advanced technologies, including MAS,to genetics and breeding. Both are excellentexamples of strategies that promoteeffective partnerships and collaboration byresearchers from different institutions, disciplinesor countries working on specifichigh-priority projects.The case of the CGIAR GenerationChallenge Programme: an internationalR&D network in genetic resources,genomics and breedingAs the number of stakeholders in the agriculturaldecision-making process increases and

Chapter 18 – Marker-assisted selection in crop and livestock improvement 373the agricultural research agenda expands,organizations must be able to respond to anincreasingly diverse and complex portfolioof priorities by strengthening interactionswithin the system and developing linksand partnerships with groups traditionallyoutside the system. Towards this end, theCGIAR has designed a strategy to nurturethe definition of objective R&D agendasin key themes and to guide scientists andteams worldwide towards integrated, synergisticinvolvement and operation. Thisstrategy became known as the “GlobalChallenge Programmes” (www.cgiar.org/impact/challenge/index.html).The strategy of the Global ChallengeProgrammes recognizes both that the costof conducting research is escalating andthat the complexity of the science neededfor agricultural research is increasing.Research in most fields requires not onlyspecialized equipment and facilities butalso highly trained technical support indiverse disciplines. Increasingly, multidisciplinaryteams of scientists will be requiredto address the complex issues facing agricultureand, in many cases, the professionalexpertise needed may have to be accessed indifferent parts of the world.One such Challenge Programme, entitled“Unlocking Genetic Diversity in Cropsfor the Resource-Poor”, also known as the“Generation Challenge Programme (GCP)”(CGIAR, 2003) is an international, multiinstitutional,cross-disciplinary publicplatform for accessing and developing newgenetic resources using advanced moleculartechnologies associated with conventionalmethods. Founded in July 2003 by theExecutive Council of the CGIAR withstart-up funding from the World Bank andthe European Commission, the GCP has amembership of twenty-two public researchinstitutions around the world, includingnine CGIAR centres, four advancedresearch institutes and nine national agriculturalresearch system institutions. Itsbudget in 2005 totalled at US$14 million(GCP, 2005).This platform was designed to ensurethat the advances of crop science and technologyare applied to the specific problemsand needs of resource-poor people whorely on agriculture for subsistence and theirlivelihoods. The GCP aims to “bridge thatgap by using advances in molecular biologyand harnessing the rich global stocks ofcrop genetic resources to create and providea new generation of plants that meetthese farmers’ needs”.The concrete objective of the GCP isto access and develop genomic and geneticresources as enabling technologies andintermediate products for crop improvementprogrammes. It will not produce andrelease finished crop varieties for farmers,but develop new genetic resources andmake the initial gene transfers to locallyadapted germplasm, and then transfer thederived materials to crop improvementprogrammes, particularly those conductedin national agricultural research systemsof developing countries, and to any otherentities that have crop improvement goals,especially those dedicated to resource-poorfarmers.The GCP is, to date, the most comprehensiveeffort to cover, in a well structuredand feasible manner, the complex interactionsbetween genetic resources, genomicsand breeding (Figure 2) in order to capturethe benefits of the revolutions in biologyand direct them to help solve some ofthe agricultural problems in the world’smost difficult and marginal environments.It addresses its three key component partsin a separate but interconnected manner:(1) genetic resource collections provide the

374Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishFigure 2Conceptual basis for the Generation Challenge Programme – Unlocking Genetic Diversityin Crops for the Resource-PoorReprinted by permission from the proposal for a CGIAR Challenge Programme (CGIAR, 2003).raw materials; (2) genomic science providesthe means to exploit genetic resources (i.e.identify new alleles); and (3) crop improvementapplies traditional and modernmethods of gene/allele transfer into functionalcrop varieties (CGIAR, 2003).The GCP is therefore an ambitious initiativeto put into action a complex mix oftools, capacities, concepts and strategies. Itis organized and managed to direct theseresources towards the pursuit of goals thatare not attainable through the disciplinaryand isolated modes of operation thatunfortunately prevail in the internationalagricultural R&D arena. As such, it is possiblythe best structured international effortfor development, adaptation and promotionof effective (and inclusive) access anduse of tools such as MAS.As part of its complex strategy, the GCPwill define protocols for more efficientgene transfer including molecular markersthat are closely linked to the genes for thedesired trait, rapid tests for phenotype recognition,and genetic transformation ofnew genes into locally adapted geneticmaterials, such as improved varieties andlandraces. All of these strategies dependon the adaptation and development ofmarker technology and marker-assistedprocedures, hopefully helping to consolidatea networking approach to breedingand biotechnological research with effectiveimpact, especially on resource-poorcountries.Research activities commenced in January2005 with the first round of competitiveresearch grants awarded for 17 three-yearprojects of approximately US$1 millioneach. In early 2005 a new round of commissionedgrants was started, which servedas the basis of the GCP platform of toolsand technologies for genetic studies andapplications. In total, the GCP initiated

Chapter 18 – Marker-assisted selection in crop and livestock improvement 375Figure 3Research priorities and partnerships among private companies, research institutesand universitiesRESEARCH PRIORITIESUsually strategic or or appliedTHERESEARCH PRIORITIESUsually basic, fundamentalor speculativeor speculativeRelatively short short time time horizons horizonsMUTUALBENEFITUltimately commerciallyUltimately beneficial commercially beneficialZONENo specific time horizonsor limitsor limitsGeneration and and disseminationof new knowledgeof new knowledgePRIVATE COMPANIESRESEARCH INSTITUTESUNIVERSITIESSource: adapted from Laider (1998).67 competitive and commissioned researchprojects and capacity-building activities in2005. The 2005 Annual Report and 2006Work Plan summarizes research progressand capacity building achievements in 2005and presents an overview of the competitiveand commissioned research portfolio andthe capacity-building and delivery activitiesfor 2006 (GCP, 2005, 2006).The Case of the GenolyptusProgramme in Brazil: a public/privatenetwork in genomics and molecularbreeding of Eucalyptus“The challenge and the opportunity forpublicly supported agricultural researchare not in duplicating the private sector’sresearch agenda, but in building uniquepublic/private partnerships or perhaps evenjointly supported consortia for agriculturalresearch” (CAST, 1994). Increasingly,agricultural research will be conductedthrough partnerships among private companies,public research institutes anduniversities (Figure 3). In forming such alliances,these organizations must recognizethat developing productive relationshipsinvolves non-competitive dialogue andunderstanding of each others’ abilities andlimitations. Partnerships will flourish onlyif founded on trust and understandingand if differences in drivers and objectivesare recognized and accommodated in initiativeswith a real perspective of mutualbenefits (Lopes, 2000).An example of a successful public/privatepartnership with clear understanding ofpartners’ abilities and limitations and cleardefinition of responsibilities and benefitsto be pursued is the Genolyptus Networkin Brazil (Grattapaglia, 2003). This R&Dnetwork was created to establish a foundationfor a genome wide understanding ofthe molecular basis of wood formation in

376Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishFigure 4The Genolyptus project – from phenotypes to genotypes in an integrated way√√√√Growth , floweringFIELD EXPERIMENTS24 connected full-sib familiesamong divergent treesLarge progeny sizes (> 1200individuals)Partial clonal replicationSeveral genetic backgrounds andlocationsPHYSICAL MAPPINGWOODPROPERTIESPHENOTYPESLINKAGEAND ASSOCIATIONMAPPINGDisease resistanceGENE DISCOVERY√ Several cDNA libraries√ EST sequencing√ Candidate genes√ SSRs and SNPs in cDNAs√ Full length cDNA database√ Bioinformatics: data mining√ Microarrays and SAGE√ SNP and association mapping√√√√√BAC libraryAnchoring with genetic mapComplete physical mapBAC end sequencingEST mappingFull gene and promoteridentificationQTL & candidate genesGENES & LINKEDMARKERS√√√QTL MAPPINGGenetic map construction inmultiple familiesGenome scan with 200+microsats in multiplexesOutbred QTL modelsSource: Grattapaglia, 2003.Eucalyptus, coupled with the translationof knowledge into improved tree breedingtechnologies.This programme relies heavily on thedevelopment of aligned R&D efforts ingenetic resources, genomics and molecularbreeding (Figure 4). It mobilizes capacitiesand infrastructure in constructing physicalmaps, developing expressed sequencetag (EST) databases, generating a databaseof expression profiling of genes that controlkey traits and developing methods forMAS for traits of high heritability in woodformation. Also, the network develops acapacity-building and training programmefor professionals in universities and forestrycompanies, targeting the integrationof genetics, genomics and breeding effortsof Eucalyptus.The rationale of the network is based onthe understanding that, even with the morepowerful tools allowing a much more globaland integrated view of genetic processes,genomics will only succeed in contributingto the development of improved eucalyptif it is deeply interconnected with intensivefieldwork and creative breeding. TheGenolyptus project therefore differs fromother plant genome initiatives in the intensity,refinements and scope of the effortdevoted to field experiments to generatethe diversity of phenotypes necessary tostudy gene function. Quantitative trait loci(QTL) detection, the development of SNPhaplotypes for association mapping andphysical mapping will link the phenotypesto genes that control processes of woodformation that define industrial level traits(Grattapaglia, 2003).A key feature of the Genolyptus networkis its pre-competitive nature. Theresearch programme was designed collaborativelywith no immediate intention ofmarketing its results, even although its

Chapter 18 – Marker-assisted selection in crop and livestock improvement 377planned outputs will eventually lead tothe creation of many new products andprocesses of commercial value. Thus, theactivities during its first phase are designedto resolve basic, common technologicalproblems – a sufficient reason to mobilizeseveral private companies (that arecompetitors in the market) and publicorganizations. After the first phase of sixyears, the network will have generated aconsolidated base of knowledge and toolsthat will promote the development of specificinterest projects, either in partnershipor individually, according to specific businessstrategies and market targets.Also, team organization and managementare based on modern tools andconcepts, involving a competent, highlyrespected scientist with talent to lead networkoperations, a steering committee anda technical committee for adequate planning,decision-making and follow-up, aswell as contract models and negotiationstrategies appropriate to the complexity ofthe network. Intellectual property rightsprovisions are based on access limited toparticipants, with all genetic materials andpatents produced being co-owned by the20 participating institutions. Scientific publicationsare highly encouraged.As in the Generation ChallengeProgramme, the Genolyptus network is anoriginal initiative to integrate and align acomplex mix of tools, capacities, conceptsand strategies. The ability to mobilize sucha wide range of organizations, including12 private companies operating in a highlycompetitive market space, indicates that thenetwork design was successful, while itspre-competitive nature, organization andmanagement strategy allowed the definitionof a “zone of mutual benefits” (Figure 3),facilitating the pursuit of goals that are notattainable through isolated research. TheGenolyptus network is therefore an excellentexample of the feasibility of developinga structured public/private effort for integratedand effective use of advanced toolssuch as MAS.Conclusions• Although advances in molecular markertechnology have uncovered many possibilitiesfor easing the transfer of genesinto desired crops and livestock throughMAS, there is still limited recordedimpact of these technologies in breedingprogrammes.• It is generally recognized that geneticimprovement of complex traits usingMAS is more difficult than previouslyconsidered. Therefore, more methodologydevelopment, better planning andimplementation strategies will be neededfor its successful and rapid application tobreeding programmes.• The future configuration of breedingprogrammes is dependent on knowledgeto guide strategic decisions aboutstructures, methods and capacities thattake advantage of new opportunitiesand technological niches. Unfortunately,there are very few efforts directed atthinking about the future of breedingprogrammes, especially in less developedcountries. Research organizations needinformation, which is not currently available,about changes and influences andtheir impact in the future on key activitiessuch as crop and livestock breeding.To acquire and organize this information,prospective studies on the presentand future performance of breeding programmesand their related activities willhave to be developed.• Priority-setting strategies, together withcost–benefit analysis are necessary toguide informed decisions on how best

378Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishto apply MAS and other advanced technologiesto crop and livestock breedingin developing countries.• MAS has to be understood and undertakenas part of a complex process. Complementaritiesand a mix of technologiesand capabilities, together with effectiveapproaches to networking, must beviewed as key ingredients for its appropriatedevelopment and application tobreeding programmes.• One of the key problems limiting the useof MAS and other advanced technologiesin developing countries is the difficultyof building effective teams and networks.Approaches to networking and partnershipsare key to enabling organizationsto attain new goals with less cost andto adding more value to their productsand processes. Also, the demand forefficiency and relevance presses R&Dprogrammes to move in the direction ofcooperation and alignment of efforts.• The present and future challenges andopportunities for agricultural researchorganizations are to build public/privatepartnerships or new types of consortiadedicated to innovation. In forming suchalliances, these organizations must recognizethat developing productive relationshipsinvolves non-competitive dialogueand understanding of each others’ abilitiesand limitations. In order to surviveand flourish, partnerships have to besustained on trust and understanding.• Developing countries must focus ontraining to build and shape capacities andeffective strategies to support researchin advanced biology applied to breeding.Also, new management strategies areneeded to deal with the complex natureof modern breeding programmes.ReferencesAltshuler, D., Pollara, V.J., Cowles, C.R., Van Etten, W.J., Baldwin, J., Linton, L. & Lander, E.S.2000. An SNP map of the human genome generated by reduced representation shotgun sequencing.Nature 407: 513–516.Brandenburger, A.J. & Nalebuff, B.J. 1997. Co-opetition, New York, USA, Doubleday.CAST (Council for Agricultural Science and Technology). 1994. Challenges confronting agriculturalresearch at land grant universities. Issue Paper 5. Ames, IA, USA.Castro, A.M.G., de Cobbe, R.V. & Goedert, W.J. 1995. Manual de prospecção de demandas para oSNPA. Brasília, Brazil, Embrapa.Castro, A.M.G., Lima, S.M.V. & Freitas Filho, A. 1998. Manual de Capacitación en Análisis deCadenas Productivas. Brasília, Brazil, Embrapa.Castro, A.M.G., Lima, S.M.V., Lopes, M.A. & Martins, M.A.A.G. 2002. Estratégia de P&D para oMelhoramento Genético em uma Época de Turbulência. Anais do XXII simpósio de gestão da inovaçãotecnológica, p.17. Salvador, BA, Brasil.Castro, A.M.G., Lopes, M.A., Lima, S.M.V. & Bresciani, J.C. 2005. Trends and technologicalstrategy in the Brazilian seed sector (Cenários do Setor de Sementes e Estratégia Tecnológica).Revista de Política Agrícola Brasília 3: 58–72.Castro, A.M.G., Lima, S.M.V., Lopes, M.A., Machado, M.S. & Martins, M.A.G. 2006. O Futuro doMelhoramento Genético Vegetal no Brasil – Impactos da Biotecnologia e das Leis de Proteção deConhecimento (The future of plant breeding in Brazil – impacts of biotechnology and intellectualproperty legislation). Brasília, DF, Embrapa Informação Tecnológica.

Chapter 18 – Marker-assisted selection in crop and livestock improvement 379CGIAR (Consultative Group on International Agricultural Research). 2003. Unlocking GeneticDiversity in Crops for the Resource-poor: A Proposal for a CGIAR Challenge Program (available atwww.generationcp.org/sccv10/sccv10_upload/CP_proposal.pdf ).Dekkers, J.C.M. & Hospital, F. 2002. The use of molecular genetics in the improvement of agriculturalpopulations. Nature Revs. Genet. 3: 22–32.De La Vega F.M., Dailey D., Ziegle J., Williams J., Madden D. & Gilbert D.A. 2002. New generationpharmacogenomic tools: a SNP linkage disequilibrium map, validated SNP assay resource,and high-throughput instrumentation system for large-scale genetic studies. Biotechniques Suppl32: S48–S54.FAO. 2003. Molecular marker-assisted selection as a potential tool for genetic improvement of crops,forest trees, livestock and fish in developing countries. Background document to Conference 9 ofthe FAO Biotechnology Forum (available at www.fao.org/biotech/C10doc.htm ).Ferreira, M.E. 2003. Melhoramento Genético do Arroz: Impactos da Genômica. In A. Borém, M.Giúdice & T. Sediyama, eds. Melhoramento genômico. Universidade Federal de Viçosa,Viçosa,Brazil.Ferreira, M.E. 2006. Molecular analysis of gene banks for sustainable conservation and increased useof crop genetic resources. In J. Ruane & A. Sonnino, eds. The role of biotechnology in exploringand protecting agricultural genetic resources. FAO, Rome (available at www.fao.org/docrep/009/a0399e/a0399e00.htm).GCP (Generation Challenge Programme). 2005. Generation Challenge Programme researchhighlights 2005. Mexico, DF. CIMMYT (available at http://www.generationcp.org/sccv10/sccv10_upload/Highlightslittle.pdf).GCP. 2006. Generation Challenge Program Annual Report and Year Three Workplan. Mexico, DF,(available at www.generationcp.org/comm/manual/doc_100.pdf).Graff, G., Cullen, S., Bradford, K., Zilberman, D. & Bennet, A. 2003. The public-private structureof intellectual property ownership in agricultural biotechnology. Nature Biotech. 21: 989–995.Grattapaglia, D. 2003. Genolyptus. In: A. Borém, M. Giúdice, & T. Sediyama, eds. Melhoramentogenômico, pp. 51–71. Universidade Federal de Viçosa,Viçosa, Brazil.Laider, D.A. 1998. Research partnerships between industry & universities. A guide to better practice.Runcorn, UK, ICI Technology.Lima, S.M.V., Castro, A.M.G., de Mengo, O. & Medina, M. 2000. Entorno y Enfoque Estratégico enI & D: Proyecto Nuevo Paradigma. Artigo submetido ao XXI Simpósio Internacional de GestãoTecnológica, FIA/ USP/PACTO.Linstone, H.A. & Turoff, M. 1975. The Delphi method: techniques and applications. Readings, MA,USA, Addison Wesley.Lipnack, J. & Stamps, J. 1993. The age of the network. New York, USA, Oliver WrightPublications.Lopes, M.A. 2000. Cooperation and competition in competitive grants: is “coopetition” beingachieved? In Proc. of the Competitive Grants in the New Millenium: A Global Workshop forDesigners and Practitioners, pp. 219–230. Brasília, Brazil, Embrapa/IDB/The World Bank.Lörz, H. & Wenzel, G. 2005. Molecular marker systems in plant breeding and crop improvement.Biotech. in Agric. & Forestry 55: XXII. Berlin, Germany, Springer.Mohan, M., Nair, S., Bhagwat, A., Krishna, T.G. & Yano, M. 1997. Genome mapping, molecularmarkers and marker-assisted selection in crop plants. Mol. Breeding 3: 87–103.

380Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishPrice, S.C. 1999. Public and private plant breeding. Nature Biotech. 17: 938.Rafalski, J.A. 2002. Novel genetic mapping tools in plants: SNPs and LD-based approaches. PlantSci. 162: 329–333,Stuber, C.W., Polacco, M. & Senior, M.L. 1999. Synergy of empirical breeding, marker-assisted selection,and genomics to Increase crop yield potential. Crop Sci. 39: 1571–1583.Young, N.D. 1999. A cautiously optimistic vision for marker-assisted breeding. Mol. Breeding 5:505–510.

Chapter 19Technical, economic and policyconsiderations on marker-assistedselection in crops: lessons fromthe experience at an internationalagricultural research centreH. Manilal William, Michael Morris,Marilyn Warburton and David A. Hoisington

382Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishSummaryMolecular markers and related technologies have been used extensively in geneticcharacterization and identification of loci controlling traits of economic importance inmany crop species. However, the application of such tools for crop improvement has notbeen extensive, at least in the public sector. Although there are clear advantages in usingmolecular markers as tools for indirect selection of traits of importance, available examplesindicate that their use is restricted to traits with monogenic inheritance or when theinheritance is conditioned by a few genes with large effects. Another important limitation oflarge-scale marker applications is the cost involved in marker assays, which may be beyondthe capacities of many public plant breeding enterprises. For an effective marker-assistedselection (MAS) activity to facilitate ongoing crop improvement programmes, especially inthe context of the developing countries, laboratories with adequate capacity and adequatelytrained scientific personnel as well as operational resources are required. Although recenttechnological advances such as single nucleotide polymorphisms (SNPs) and associatedassay protocols are likely to reduce assay costs significantly, for many of these operations,assay platforms with significant capital investments including computational capacity arerequired. Coupled with these limitations, private sector domination of biotechnologyresearch with proprietary rights to important products and processes with immediatebenefits to developing countries may further constrain the benefits these technologiesmay offer to resource-poor farmers. Policy-makers in different national programmesand international development and research agencies have a responsibility to sustain andaugment the capacity of national public agricultural research organizations to ensure thatbiotechnology tools and processes are infused appropriately into national research efforts.They must also ensure that any biotechnology efforts undertaken are well integrated withnational crop improvement activities.

Chapter 19 – Technical, economic and policy considerations on marker-assisted selection in crops 383IntroductionDue to their usefulness in characterizingand manipulating genetic factors responsiblefor qualitative as well as quantitativetraits, molecular markers are considered tobe valuable tools for crop improvement.These uses of molecular markers have beeninvaluable in helping researchers understandcomplex traits, dissect them intosingle Mendelian genetic factors, and establishtheir chromosomal locations via the useof linkage maps and/or cytogenetic stocks.Availability of well characterized geneticlinkage maps is a prerequisite for taggingimportant agronomic or other traits withmolecular markers, enabling their use inMAS related activities. To date, however,few practical applications have been publishedfrom these studies. This paucity ofpublished studies may indicate the longtermnature of this research, or it mightsimply reflect the fact that marker technologyhas been applied to plant breedingefforts mostly by scientists working in theprivate sector (Hoisington and Melchinger,2004).Maize was one of the first crop speciesfor which molecular linkage maps weredeveloped, and Gardiner et al. (1993)consolidated several individual maps into aconsensus map. Rice is another species forwhich high-density linkage maps have beendeveloped (reviewed in Gowda et al., 2003)while, due to its high ploidy level and largegenome (21 linkage groups, as opposed to 10in maize and 12 in rice), efforts to developwell characterized, saturated linkage mapswith wheat have lagged behind. Otherimportant cereals and legumes are at variousstages of linkage map development. Theavailability of well-defined linkage mapsand the extent of genetic studies conductedon them therefore vary among differentcrops, and this influences the feasibility ofany MAS-related activity. Thus, while it ispossible to carry out MAS to some degreein cereals such as rice, maize and wheat, andin legumes such as soybean, for species suchas cassava and sweet potato, the so-called“orphan crops”, genetic improvement withMAS may not yet be feasible. These cropspecies may benefit more readily fromgenetic modification arising from directintroduction of genes isolated from otherspecies or organisms, which is not the focusof this chapter.Citing practical lessons learned atthe International Maize and WheatImprovement Center (CIMMYT) as well asfindings of studies conducted elsewhere, thischapter describes some actual and potentialapplications as well as the advantages anddisadvantages of MAS, and outlines possibleapplications of MAS in developingcountry plant breeding programmes.Lessons Learned from CropsNumerous scientific reports describemolecular mapping and analysis of quantitativetrait loci (QTL) for nearly everyagronomic trait in a diverse array of cropspecies. The traits covered include manyparameters associated with tolerance todrought and other abiotic stresses, maturity,plant height, quality parameters, qualitativeand quantitative factors of disease and pestresistance, and numerous seed traits andyield. Although these efforts have resultedin a vast amount of knowledge and betterunderstanding of the underlying geneticfactors that control these traits, applicationof this knowledge to manipulate genes in aneffective or simple manner for improvingcrop species has had limited success. Thescientific community is faced with thechallenges of accurate and precise QTLidentification and application of the informationderived to successful MAS efforts.

384Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishScientific advances have been instrumentalin increasing the power and accuracyof computational parameters as well asdesigning ways of combining the informationgenerated from molecular geneticswith traditional crop improvement efforts.Numerous simulation studies have beenundertaken to evaluate the effectiveness ofMAS, taking into account the influence ofheritability, population size, linkage distance,etc. (Xie and Xu, 1998; Moreau etal., 1998; Ribaut, Jiang and Hoisington,2002), and MAS procedures have beenused to incorporate traits of interest fromexotic species including wild relatives intoelite cultivars through advanced backcrossQTL analysis (Tanksley and Nelson, 1996;Fulton et al., 2000).Manipulation of qualitative traitsMolecular markers that are tightly linkedto genes having a strong effect on theexpression of a trait can be used to introgressthe genes (and thus the trait) intodifferent backgrounds through backcrossbreeding schemes that rapidly and efficientlyimprove the recurrent parent forthe target trait. In conventional backcrossbreeding schemes and line conversionactivities, the donor parent containing thetrait of interest is crossed with the recurrentparent, normally a well-adapted varietylacking the trait of interest. The resultingprogeny are screened to identify the traitof interest, and individuals exhibiting thetrait are crossed to the recurrent parent.The entire process is repeated several times.For traits that are conditioned by recessivegene action, a cycle of selfing is alsorequired after each crossing cycle. Afterseveral cycles of backcrossing and a finalself-pollination, plant breeders are oftenable to recover lines that are nearly identicalto the recipient parent but also contain thetrait of interest. Compared with traditionalbackcrossing, the use of DNA markers enablesfaster recovery of the recurrent parentgenotype along with the introgressed targettrait in line conversion activities. Ribautand Hoisington (1998) reported that MASshould enable the recovery of the targetgenotype after three cycles of backcrossing,compared with a minimum of six cycleswith traditional approaches (Tanksley etal., 1989).CIMMYT has a long history of usingmolecular markers for certain traits inmaize improvement. Although maize iswidely used for both food and feed, maizekernels do not provide sufficient quantitiesof two essential amino acids, lysine andtryptophan. The opaque2 mutation, identifiedat Purdue University (United States ofAmerica) in the mid-1950s, confers elevatedlevels of these two amino acids. Althoughinitial efforts to introduce the opaque2mutation into breeding materials were notsuccessful (Villegas, 1994), researchers eventuallysucceeded in producing nutritionallyenhanced maize lines. These came to beknown as quality protein maize (QPM).CIMMYT breeders have used traditionalbackcrossing to transfer the opaque2 mutationand associated modifiers into elitelines. To perform phenotypic selection insegregating progenies for lines carrying theopaque2 mutation, it is necessary either towait until the plants produce mature ears,or to do random pollination on a largenumber of plants. Although reliable laboratoryscreening techniques are available,co-dominant microsatellite markers presentwithin the opaque2 mutation can be usedearlier in the growing season. Using thesemarkers in backcross progenies, plants heterozygousfor the opaque2 mutation can beselectively identified as a qualitative traitfor use in the next crossing cycle. Markers

Chapter 19 – Technical, economic and policy considerations on marker-assisted selection in crops 385are not used to select for the backgroundrecurrent parent genotypes, but only toselect lines carrying the opaque2 mutationallele. Although CIMMYT uses markersfor detecting the presence of the opaque2mutation, markers are not available to selectfor the modifiers, which are important indetermining seed texture and quality andfor which other traditional screening techniquesare being used.A well known example of markerassistedbackcrossing of a qualitative traitinvolves the introgression of the Bt transgeneinto different maize lines (Ragot etal., 1994). Whenever plant transformationtechniques are used to produce geneticallymodified organisms (GMOs), usually thereare some cultivars that are more receptiveto transformation procedures than others.When the cultivar with the best agronomictype is not the most receptive to transformation,it is often possible to transformanother cultivar that is receptive and thenuse the diagnostic marker that detects thetransgene to introgress it into more desirablebackgrounds. This type of MAS-aidedline conversion can be accomplished forany crop species. The presence of markersto detect the transgene enables the detectionof converted progeny with a highdegree of accuracy.Another MAS-related CIMMYT experienceinvolves the case of maize streakvirus (MSV) resistance, for which a majorQTL was identified on maize chromosome1 that explains 50–70 percent of total phenotypicvariation (Pernet et al., 1999a, b).As maize has a well-saturated molecularlinkage map, several microsatellite markersassociated with this QTL were identifiedin the specific chromosomal region. Thesemarkers were tested in three populationsgenerated using three different MSV tolerantlines crossed with one susceptibleline. After screening the F 2 progeny andF 3 families, lines identified by markerswere sent to Africa, where MSV is prevalent.By phenotypic screening of the linesselected by MAS, it was established thatMAS-selected lines were significantly moreresistant to MSV (J-M. Ribaut, personalcommunication).In legumes, resistance to soybean cystnematode (SCN) is one example of aneffective MAS approach. Routinely usedphenotypic assays for SCN screening takeapproximately five weeks and extensivegreenhouse space and labour. Successfulidentification of closely linked microsatellitemarkers has enabled transfer of theresistance gene rhg1 with about 99 percentaccuracy (Cregan et al., 1999; Young1999). Many public and commercial soybeancultivar improvement efforts usethese markers to screen for SCN resistance(Young, 1999). Another example ofsuccessful MAS in common beans wasreported by Yu, Park and Poysa (2000)who used markers associated with commonbacterial blight. These markers identified alocus that explained about 62 percent of thephenotypic variation and have been used inMAS experiments.As described earlier, linkage map constructionin wheat is more challengingthan in species such as rice or maize.The allohexaploid nature allows wheat towithstand chromosomal imbalances as theloss of one chromosome can be compensatedby the presence of a homologouschromosome. As a result, wheat can becrossed with a range of wild relatives (bothintergeneric and interspecific), enablingintrogression of genetic material possessingresistances to different biotic and abioticstresses. When translocations (especiallyintergeneric translocations) are present inwheat, markers can be readily developed

386Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishfor the translocated chromosome segments.If a translocated segment carries a trait ofimportance, markers can then be used totransfer it into different wheats. Diagnosticor perfect markers (i.e. markers with completelinkage to the genes of interest withno possibility of recombination) have beendeveloped for genes conferring resistanceto different biotic stresses in wheat.CIMMYT’s wheat improvement efforts usea set of markers routinely on a seasonalbasis for introgression of a set of genes intohigh-yielding backgrounds. Examples ofthe perfect markers that are currently inuse are:• Cereal cyst nematode (CCN) resistancegene Cre1 (2BL), identified in wheatlandrace AUS10894 and Cre3 (2DL),derived from Triticum tauschii (Lagudah,Moullet and Appels, 1997). These markersare used routinely in segregating populationsto enable selective advancementof lines containing the Cre genes targetedto all environments, but particularly tomarginal ones, where healthy root architectureis essential to allow plants to takeadvantage of minimal soil moisture. Phenotypicevaluation for CCN resistanceis labour intensive as well as expensive.Given that it is impossible to screenfor CCN resistance in Mexico (whereCIMMYT headquarters are located) dueto the lack of required screening facilities,the use of markers is essential forimproving this trait.• Barley yellow dwarf virus (BYDV) resistance,derived from a chromosome segmentintrogressed from Thinopyrumintermedium, on chromosome 7DL(Ayala et al., 2001). BYDV is an importantviral disease in certain wheat growingregions of the world. Environmentalinfluence makes field screening less reliable.The diagnostic marker for the translocatedchromosome segment allows thealien-derived resistance to be combinedwith the BYDV tolerance available inwheat.• Marker for Aegilops ventricosa-derivedresistance to stripe rust (Yr17), leaf rust(Lr37) and stem rust (Sr38 ) (O. Robert,personal communication). The translocationfrom Ae. ventricosa is present onchromosome 2AS. The diagnostic markerfor the translocation is used mainly inbread wheat x durum wheat crosses, toidentify the durum derivatives carryingthe translocation.In addition, CIMMYT uses a set oflinked markers for transferring a locus withmajor effects for boron tolerance (Bo-1),crown rot resistance, scab resistance andstem rust resistance in its MAS efforts.These efforts with linked genes are conductedwith the objective of increasing theallele frequency for desirable alleles in segregatingpopulations (William, Trethowanand Crosby-Galvan, 2007).Gene pyramiding/stackingMAS lends itself well to gene pyramidingefforts for disease resistance. When a cultivaris protected by one gene with majoreffects against a specific disease, it is oftennot possible to introgress additional genesconferring resistance to the same diseasebecause of the difficulty of phenotypicscreening for the presence of additionalgenes (as the plant already shows resistanceto the disease). However, if several genescan be tagged with closely linked molecularmarkers, MAS strategies can be used todevelop lines with stacked genes, giving thecultivar more durable protection than thatafforded by a single resistance gene.Resistance to bacterial blight providesan excellent example of using MAS for genepyramiding. Bacterial blight is caused by

Chapter 19 – Technical, economic and policy considerations on marker-assisted selection in crops 387Xanthomonas oryzae and is one of the mostimportant diseases of rice. Several genesthat confer resistance to bacterial blighthave been tagged with molecular markers.Huang et al. (1997) and Hittalmani et al.(2000) developed strategies for combiningfour resistance genes, namely Xa-4, Xa-5,Xa-13 and Xa-21, in a single cultivar usingpairwise combinations of the genes. Dueto the co-dominant nature of the markersused, the authors were able to select fromF 2 generations without having to performprogeny testing. The derived lines containingpyramided genes showed higherlevel of resistance and/or a wide spectrumof resistance compared with the parentalmaterial. Another gene pyramiding exampleusing MAS involves stacking of the resistancegenes rym4, rym5, rym9 and rym11for the barley yellow mosaic virus complexusing molecular markers and doubledhaploids (Werner, Friedt and Ordon, 2005).Other examples include pyramiding forbarley stripe rust resistance (Castro et al.,2003), and powdery mildew resistance inwheat (Liu et al., 2000) and, in MAS applicationsat CIMMYT, crosses have beenmade to combine two genes for cereal cystnematode resistance and three differentgenes for stem rust resistance (Sr24, Sr26and Sr25) in targeted wheat germplasm.Manipulation of quantitative traitsQuantitatively inherited traits are geneticallycomplex, are conditioned by a numberof genes each having relatively small effects,and their expression often depends on interactionsamong different genetic components(epistasis). The environment also has a highdegree of influence on the expression of thetrait, which confounds the interpretation ofQTL identification and often renders theresults obtained from QTL studies crossspecific.When it is necessary to manipulateseveral genomic regions simultaneously,each having different effects on the sametrait of interest, MAS-based approachesbecome more complicated and presentformidable challenges. Mapping studies conductedat CIMMYT identified five genomicregions associated with the anthesis-silkinginterval which is a parameter associatedwith drought tolerance in maize (Ribautet al., 1996, 1997). The drought tolerantparent was used in MAS experiments asthe donor parent to transfer the five QTLto CML 247, an elite inbred line with goodcombining ability that was drought-susceptiblebut high-yielding under favourableconditions. Markers were used to generate70 BC 2 F 3 lines containing the favourablealleles from the drought-resistant parentafter two backcrosses and two self pollinations.These lines were crossed withtwo testers for field evaluation. Field testsindicated that under severe drought stressconditions, the 70 MAS-derived lines weresignificantly better yielding than the controls.The differences were less prominentunder reduced drought stress (Ribaut andRagot, 2007).Other CIMMYT experiments aimed atcomparing MAS with phenotypic selectionhave been conducted for stem borers intropical maize (Willcox et al., 2002). In thecase of maize stem borer resistance, threeQTL identified through mapping experimentswere used in MAS. Three BC 2 S 2families that carried all three target QTLfrom the donor parent in homozygous statewere developed. Comparative studies withMAS and traditional phenotypic selectiondid not establish a clear advantage for MAS,but both approaches yielded significantgenetic gains in reducing leaf damage. MASis not being used currently on a routinebasis at CIMMYT for drought and stemborer resistance.

388Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishOther reports describing the manipulationof quantitatively inherited traitsinclude those of Bouchez et al. (2002)for introgressing favourable alleles atthree QTL for earliness and grain yieldin maize, and by Yousef and Juvik (2001)who reported on MAS for seedling emergenceand eating quality characters in sweetcorn. Also, Han et al. (1997) attempted toselect for barley malting traits using MAS.Additional scientific reports are availablethat describe MAS-related efforts for quantitativelyinherited traits.In general, manipulating several QTLassociated with multiple genomic regions insegregating progenies is considerably morechallenging. Often the success in geneticgains depends on the stability of these QTLas well as the cost efficiency of large-scaleMAS applications.Genetic diversity studiesIn addition to being used in MAS activities,molecular markers have been usedextensively for genetic diversity studies.Numerous scientific publications are availablethat describe the use of molecularmarkers in estimating the degree of relatednessof a set of cultivars in many cultivatedcrop species. In common with their use intrait manipulations, the practical outcomesof the numerous genetic diversity studiesusing molecular markers are not clear.Evaluation of genetic relatedness usingmolecular markers will have implicationson understanding the genetic structure ofexisting populations, enabling the design ofstrategies for proper acquisition of germplasmfor conservation purposes. Thegenetic uniqueness of accessions or populationsin germplasm collections can beaccurately estimated by the use of DNAprofiling (Brown and Kresovich, 1996;Smith and Helentjaris, 1996). Molecularmarkers have also been used for identifyingredundancies in existing germplasm collectionsin rice (Xu, Beachell and McCouch,2004) and sorghum (Dean et al., 1999). Incassava, Chavarriaga-Aquirre et al. (1999)used morphological traits, isozyme profilesand agronomic criteria to identify a core setof 630 accessions from a base collection ofapproximately 5 500 accessions.Modern farming in advanced countriesis based on high performing, geneticallyuniform new cultivars, which are generallyderived from well adapted, geneticallyrelated parental material. Tanksley andMcCouch (1997) have concluded that mostmodern soybean cultivars grown in theUnited States can be traced back to a verylimited number of strains from a small areaof northeastern China, while a majority ofhard red winter wheats is derived from a fewlines originated in Poland and the RussianFederation. The genetic basis of modernrice varieties grown in the United States isalso considered narrow (Dilday, 1990).Another application in the area ofgenetic diversity is the use of markersin identifying heterotic groups. Molecularmarkers have been used extensively in theconstruction of heterotic groups since the1990s in many different crop species ofeconomic importance. Heterotic groups areclusters of germplasm usually with similarcharacteristics and a high degree of relatednessthat, when crossed with materials fromanother heterotic group, tend to give riseto progeny with high levels of heterosis.Although markers randomly distributed inthe genome can be used to develop heteroticgroups, their usefulness in determininghybrid performance is not clear. While it isreasonable to assume that heterosis dependson the interactions among favourable allelesbelonging to the two parents, unless molecularmarkers that are known to be linked to

Chapter 19 – Technical, economic and policy considerations on marker-assisted selection in crops 389these favourable alleles are used in heteroticstudies, the predictive power of markers inestimating heterosis for practical applicationsmay not be very high.At CIMMYT, large-scale, rapid characterizationmethods for inbred lines andpopulations have been optimized usingup to 120 microsatellite markers spreadthroughout the maize genome. In the past,characterizing maize populations was costlyand time-consuming, given that as many as22 individuals had to be analysed individuallyto calculate allele frequencies for eachmarker. Currently, a bulking method inwhich 15 individuals from a populationare amplified in the same polymerase chainreaction (PCR) and run on an automaticDNA sequencer, provides a reliable estimateof the allele frequencies within thatparticular population. Between one andtwo bulks can now be used to fingerprintpopulations with considerable savings intime and resources. Other studies of maizegenetic diversity have been conducted forCIMMYT maize breeders as well as outsidecollaborators with objectives that include:determining how maize inbred lines fromdifferent national breeding programmesare related to each other (and to determinethe possibility of sharing among regionsor using lines from one region to expanddiversity in another); establishing heteroticgroups; determining levels of geneticdiversity present in synthetic varieties;determining how landraces and farmers’varieties from different regions are relatedto each other; monitoring homozygositylevels in inbred lines; and tracking changesin lines that have been intensively selectedfor a given trait.A core set of 100 microsatellite markershas been selected for wheat genetic diversitystudies. Recent fingerprinting studiesby CIMMYT and national programmescientists have been conducted to assist inregenerating gene bank accessions withoutlosing genetic diversity, measuring thecontribution of wild ancestors and exoticspecies in advanced backcross progeniesof synthetic bread wheat, and to track thechanges over time in diversity levels ofCIMMYT wheat cultivars from the originalGreen Revolution varieties to modernbreeding lines.Marker implementationTo facilitate the use of MAS activities inwheat and maize improvement efforts,CIMMYT has recently established amarker implementation laboratory. Thisprovides the facilities and technical expertiseto provide CIMMYT wheat and maizebreeders with access to biotechnologytools, including MAS. The laboratory carriesout two main MAS-related activities,marker adoption and research support. Thefirst includes constantly reviewing the literatureto identify markers developed bythird parties and verifying that these can beused to detect traits or genes of interest inCIMMYT germplasm improvement efforts,and developing efficient protocols for theirin-house use. The second consists of a rangeof routine tasks that include growth and/orsampling of plant tissue, DNA extraction,marker detection, data analysis and disseminationof results to breeders.Close cooperation between field andlaboratory staff is important to be able toapply molecular markers in crop improvementefforts. Ideally, laboratory staff shouldhave an understanding of field activities andfield workers should have basic knowledgeof different aspects of MAS-associatedlaboratory procedures. MAS is used whenthere is a high probability that markers willhelp plant breeders achieve genetic gainsfaster and more economically than field

390Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishor laboratory-based phenotypic selectionmethods. When perfect markers are availableto screen for a particular trait, suchmarkers are preferred. However, for traitsthat cannot be screened conveniently usingtraditional approaches and even when perfectmarkers are not available, if markers areavailable with close linkages to the trait(s)of interest, these can be used to increasethe desirable allele frequency for the targetgene. MAS-related activities in both wheatand maize at CIMMYT are conductedas collaborative projects involving bothbreeders and biotechnologists. The breedersuse information coming from wheat MASactivities to define better their parentalcrossing block materials and to make selectivecrosses using parents identified bymarkers. Moreover, segregating early generationprogenies in certain crosses areselected in the field based on whole plantphenotype, which are then further refinedby sampling leaf tissue from field-taggedplants and processing for MAS assays in thelaboratory. Only those entries that containthe target genes identified with MAS areadvanced to the next generation. This enablesbreeders to reduce population sizes forthe traits under evaluation and accumulatecertain gene combinations in elite backgrounds.The material thus generated isadvanced through several cycles of selfingand eventually used in field screening toidentify the best performing lines.Economics of MASEstablishing the capacity to conductMASFor MAS to be a viable option for a plantbreeding programme, adequately equippedlaboratory facilities must be in place andappropriately trained scientists must beavailable. Therefore, one of the first decisionsfacing research managers consideringMAS is whether to invest in biotechnologyresearch capacity.Economic theory suggests that the mostefficient level of research investment canbe determined with the help of a researchproduction function that relates researchinputs to research outputs. At the nationallevel, the research production function canbe thought of as a meta-function encompassingthe frontiers of many smallerfunctions, each representing a differentlevel of research capacity distinguished bycomplexity and scope (Figure 1) (Brennan1989; Byerlee and Traxler, 2001; Maredia,Byerlee and Maredia, 1999; Morris et al.,2001). Movement outwards along themeta-function, accomplished by addingsubprogrammes and thereby increasingthe number of researchers and the extentof available research infrastructure, is associatedwith changes in focus and increasesin the capacity of the national researchprogramme.For a plant breeding programme, addingnew biotechnology-based subprogrammesis equivalent to taking a series of discretesteps involving increased complexityand cost. These steps have the effect ofmoving the programme from one level ofresearch capacity to the next. These levelsof research capacity can be broadly characterizedas follows:• Biotechnology product user. Here, theresearch programme imports germplasmproducts developed using biotechnologyand incorporates them into its conventionalcrop improvement schemes, eitherby backcrossing them into local germplasmor by testing them for potentialimmediate release.• Biotechnology tools user where theresearch programme imports biotechnologytools and uses them, ifnecessary, after adapting them to local

Chapter 19 – Technical, economic and policy considerations on marker-assisted selection in crops 391Figure 1Biotechnology research production meta-functionResearch outputsP MCBiotechnologyinnovatorBiotechnologytools userBBiotechnologyproduct userAP 1P 2 P 30Source: Morris et al. (2001).Research inputscircumstances, to improve current cropimprovement practices.• Biotechnology methods innovator, inwhich the research programme establishesthe full capacity needed to developinnovative biotechnology tools andproducts.Moving from one level of biotechnologyresearch capacity to the next usually requiressignificant investments in laboratory facilitiesand staff training. The practical decisionfacing research managers is not to determinethe optimal level of research investment, butrather to select from among the differentlevels of biotechnology research capacitycharacterized by increasing complexity andcost (A or B or C in Figure 1). The choiceshould be based on whether a given level ofresearch capacity can be expected to generateenough additional benefits to justifythe additional expenditure. For most plantbreeding programmes, benefits consist ofvalue added to crop production enterprises.Therefore, the incentive to investin additional research capacity will tend toincrease with the size of the area plantedand/or the value of the crops expected tobenefit from the research.There are few published estimates ofthe cost of moving from one level of biotechnologyresearch capacity to the next,and new estimates are not provided here.Empirical estimates would quickly beoutdated, as the cost of biotechnologylaboratory equipment and materials continuesto change very rapidly. However,

392Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishfor the purposes of this chapter it is importantto point out that although establishingcapacity to develop new molecular markersrequires substantial investment, establishingthe capacity to use freely available existingmolecular markers requires only a modestinvestment.Variable cost of MASAt CIMMYT the capacity to carry outMAS on a reasonable scale has been developed,but the need now is to make thetechnology work on a high-throughputscale to reduce the cost per data point,while being able to handle large quantitiesof assays per growing season. In this regard,there are several challenges to consideras markers are not always cost-effectiveeven when they improve the precision ofselection. Depending on the nature of thetarget trait (quantitative or qualitative), thetype of gene (major or minor), the form ofgene action that controls expression of thetrait (dominant or recessive effect), and theease with which the trait can be measured(visually detected or more expensive fieldor laboratory analysis required), conventionalselection may be cheaper than MAS.The desirability of using genetic markerstherefore depends in part on the costsof genotypic versus phenotypic screening,which vary among applications.Information about the cost of usingMAS at CIMMYT for specific breedingprojects is available from case studies.For example, Dreher et al. (2002, 2003)examined the costs and benefits of usingMAS for a common application in maizebreeding. This study generated three noteworthyconclusions.First, for any given breeding project,detailed budget analysis is needed todetermine the cost-effectiveness of MASrelative to conventional selection methods.Although the costs of field operations andlaboratory procedures required for molecularmarker analysis may remain relativelyconstant across applications, every breedingproject is likely to involve unique phenotypicevaluation procedures whose costswill frequently differ.Second, direct comparisons of unitcosts for phenotypic and genotypic analysisprovide useful information to researchmanagers, but in many cases technologydecisions are not made solely on the basisof cost. Factors other than cost often influencethe choice of screening methods. Timeconsiderations are often critical, as genotypicand phenotypic screening methodsmay differ in their time requirements. Evenwhen labour requirements are similar, forapplications in which phenotypic screeningrequires samples of mature grain, genotypicscreening can often be completed muchearlier in the plant growth cycle.Third, conventional and MAS methodsare not always direct substitutes. Usingmolecular markers, breeders may be ableto obtain more information about what isgoing on at the genotypic level than they canobtain using phenotypic screening methods.For example, in conventional backcrossbreeding or line conversion projects (seesection Manipulation of qualitative traits),background molecular markers can be usedto identify those plants among a set ofprogeny that not only possess a desirableallele but also closely resemble the recurrentparent at the genetic level. Based onthis additional information, breeders areoften able to modify their entire breedingstrategy, with potentially significantimplications in terms of cost and/or timerequirements (this issue is discussed in thenext section).The CIMMYT case study thus confirmedwhat many practising plant breeders

Chapter 19 – Technical, economic and policy considerations on marker-assisted selection in crops 393intuitively know: namely, the costs andbenefits of MAS projects are likely to varydepending on the crop being improved, thebreeding objective being pursued, the skillof the breeder, the capacity of the researchorganization, the location of the workbeing carried out, the cost of key inputs,and many other factors.Economic trade-offsWhile caution is required when extrapolatingfrom the results of a case study,general conclusions regarding the costeffectivenessof molecular markers in cropgenetic improvement work can be drawnbased on the findings of the CIMMYTstudy and a number of other studies carriedout elsewhere. Broadly speaking, two typesof benefits associated with MAS can be distinguished:cost savings and time savings.Cost savingsFor certain applications, MAS methods cansubstitute directly for conventional selectionmethods, and for these applications the relativecost-effectiveness of the two methodscan easily be determined by comparing thescreening cost per sample. Generally, as thecost of phenotypic screening rises, markersare more likely to represent a cost-effectivealternative. For applications in whichphenotypic screening is easy and cheap(e.g. visual scoring of plant colour), MASwill not offer any obvious advantages interms of cost. However, for applicationsin which phenotypic screening is difficultor expensive (e.g. assessing root damagecaused by nematodes or for a disease that isnot present in the field site), MAS will oftenbe preferable.Time savingsCost is an important factor affecting thechoice of breeding technology, but it is notthe only one. Plant breeders worry aboutcontrolling costs, but they also worry aboutgetting products out quickly. Therefore, itis not sufficient to consider potential costsavings alone. The time requirements ofalternative breeding strategies must also betaken into account, because even when MAScosts more than conventional selection (asit does in some, although not all, cases),breeders who use it may be able to generatea desired output quicker. Acceleratedrelease of improved varieties can translateinto large benefits, especially for the privateseed industry, so time is an important considerationin addition to cost.For breeding applications in which MASoffers cost and time savings, the advantagesof MAS compared with conventionalbreeding are clear. More problematic, however,are the many applications in whichMAS methods cost more to implementthan conventional selection methods butalso reduce the time needed to accomplisha breeding objective. This commonlyhappens, for example, with inbred lineconversion schemes based on backcrossingprocedures. In such schemes, MAS methodscan often be used to derive convertedinbred lines containing one or more incorporatedgenes in much less time than wouldbe possible using conventional selectionmethods alone.In applications that involve a tradeoffbetween time and money, under whatcircumstances is the higher cost of MASrelative to conventional breeding justified?The choice of the plant breedingmethod can be viewed as an investmentdecision and evaluated using conventionalinvestment criteria (Sanders and Lynam,1982). Using data from the CIMMYT casestudy, Morris et al. (2003) explored therelationship between time and money asit relates to crop improvement research

394Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fish500 000Figure 2Stylized economic model of a plant breeding programmeVarietal adoption stage400 000Net annual benefits (US$)300 000200 000100 000ResearchinvestmentstageVarietalreleasestage0-100 0001 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20Time (years)by developing a simple model of a plantbreeding programme and using it to comparethe returns with alternative inbredline conversion schemes based on conventionalselection and MAS. Two measures ofproject worth were used: the net presentvalue (NPV) of the discounted streams ofcosts and benefits, and the internal rate ofreturn (IRR) to the investment.Figure 2 depicts the stylized “variety lifecycle” assumed by the model. The stream ofcosts and benefits associated with the development,release and adoption by farmers ofan improved variety can be divided intothree stages: a research stage during whichthe variety is developed; a release stageduring which the variety is evaluated andregistered for release, and commercial seedis produced; and an adoption stage duringwhich the variety is taken up and grownby farmers. During the first two stages,net benefits are negative, because costsare incurred without any benefits beingrealized. During the third stage, net benefitsturn positive as the variety is takenup and grown by farmers; they continueto increase until the peak adoption level isachieved and then decline when the varietyis replaced by newer varieties.The model was used to estimate theNPV and IRR of conventional and markerassistedinbred line conversion schemes.Research cost data were taken from theCIMMYT case study. Plausible values wereused for key parameters relating to the varietalrelease and adoption stages (for details,see Morris et al., 2003). Figure 3 shows thestreams of annual net benefits generated byeach of the two breeding schemes. Annualnet benefits are calculated as follows:NB t = (GB t - VR t - RC t )where:NB = net benefitsGB = gross benefits (calculated as areaplanted to the variety x incrementalbenefits associated with adoption)

Chapter 19 – Technical, economic and policy considerations on marker-assisted selection in crops 395Figure 3Annual net benefits, conventional vs. marker-assisted line conversion scheme400 000350 000300 000ConventionalselectionschemeAnnual net benefits (US$)250 000200 000150 000100 000MAS scheme50 0000-50 0001 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20Time (years)VR = varietal release expenses (cost ofevaluation trials, registration procedures,seed multiplication,advertising and promotion, etc.)RC = research investment costst = year (1…n)NPVs were calculated by adding the discountedstream of net benefits associatedwith each breeding scheme over the life ofthe variety (n years):where:nNPV = Σ ( GB t - VR t - RC t ) / (1 + r) tt = 1NPV = net present valuer = discount rateIRRs were calculated conventionally bysolving the discount rate that drives theNPV to 0.The profitability rankings of the twobreeding schemes, MAS and conventional,were found to differ dependingon the measure of project worth that wasused. The MAS scheme generated thehighest NPV, whereas the conventionalbreeding scheme generated the highestIRR on investment. These results, generatedusing a stylized model of a plantbreeding programme and plausible valuesfor varietal release and adoption parameters,provide an important insight into therelative cost-effectiveness of conventionalselection methods and MAS in applicationsinvolving trade-offs between timeand money. From an economic perspective,the relative attractiveness of conventionalversus MAS methods will depend on theavailability of research investment capital.If investment capital is abundant (meaningthat the breeding programme can afford toabsorb the higher up-front costs associatedwith MAS without curtailing other ongoingbreeding projects), MAS may become adesirable option, because it generates thelargest NPV. On the other hand, if investmentcapital is constrained (i.e. the breeding

396Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishprogramme cannot absorb the higher upfrontcosts associated with MAS, or thatit can absorb them only by forgoing otherpotentially profitable breeding projects), itmakes sense to choose conventional selection,because it generates the largest IRR.Implications for developingcountriesWhen discussing policy implications ofMAS efforts in developing countryscenarios, it is appropriate to consider theexperience gained over the past severaldecades, mainly in industrialized countries.In advanced countries, the private sectorhas made significant investments in MASefforts while there are a few publiclyfundedresearch groups using MAS inbreeding routinely and these are restrictedto a few target crops (Eagles et al., 2001;Dubcovsky, 2004; William, Trethowan andCrosby-Galvan, 2007). Information aboutthe traits and the breeding strategies usedin MAS applications in large agribusinessenterprises are not publicly available freely.To date, significant investments have beenmade in biotechnology applications onlyfor widely grown crop species such as rice,maize, wheat, soybean, cotton and canola.While GM crops and their implications arenot the focus of this chapter, it is reasonableto assume that technologies associated withGM crops offer significant potential foraddressing biotic and abiotic stress tolerancein widely grown cereals and legumes as wellas species that are important but thus farneglected such as tef, millets, yams and othertuber crops in the developing countries. Forexample, GM technologies that can makeone crop species perform better are likelyto be valuable with slight modificationsto enhance the performance of a neglectedcrop species. When useful GM varieties of aparticular crop are made available, they alsobecome prime candidates to apply MASbasedintrogression of the introduced geneconstruct/s to other well adapted cultivarsin different agro-ecological regions.Reports indicate that two rice varietieswith improved bacterial blight resistancehave been developed with MAS approachesand deployed in Indonesia (Toenniessen,O’Toole and DeVries, 2003). Moreover,rice varieties carrying multiple diseaseresistance genes are being developed byseveral national programmes with technicalbackstopping by the International RiceResearch Institute (IRRI) (Hittalmani etal., 2000). There are also reports describingthe use of MAS in China for improving certainquality traits in rice (Zhou, P.H. et al.,2003) and wheat (Zhou, W-C. et al., 2003)and fibre related traits in cotton (Zhang etal., 2003), but it is not clear whether theseare one-time research efforts or there iscontinued activity using MAS.Although it is not possible to obtainentirely reliable estimates of the costs,benefits and cost-effectiveness of MASapplications, the costs associated with MASare frequently considered as the main constraintto their effective use by many plantbreeders, especially in small- to mediumscalebreeding enterprises. However, newmarker technologies, especially thosebased on single nucleotide polymorphisms(SNPs) and associated ongoing large-scalegenome sequencing projects, should enablethe development and deployment of genebasedmarkers in the near future (Rafalski,2002). SNPs are defined as single base differenceswithin a defined segment of DNA atcorresponding positions. These SNP-basedpolymorphisms are known to be abundantlypresent in human as well as in plantgenomes. Consequently, the potential existsto develop SNP markers associated withmany important traits in a diverse array of

Chapter 19 – Technical, economic and policy considerations on marker-assisted selection in crops 397economically important crop species. Forspecies such as maize, rice and soybeans,robust SNP-based assay platforms alreadyexist in the private sector as well as in somepublic sector enterprises. The added advantageof SNP-based marker systems is thatthey avoid gel-based allele separations forvisualization and have the potential forautomation in high-throughput assay platforms.These ongoing research efforts willinevitably lead to the development of morerobust, high-throughput assays that areboth simple and cost effective (Jenkins andGibson, 2002).When is it advantageous to use MAS?In addition to the cost and time savingsdescribed above, for a number of breedingscenarios, MAS methods are likely to offersignificant advantages compared with conventionalselection methods. These scenariosassume the availability of markers for multipletraits and take into consideration theadvantages of MAS under optimum situations(Dreher et al., 2002; Dudley, 1993).• Gene stacking for a single trait. MASoffers potential savings compared withconventional selection when it allowsbreeders to identify the presence of multiplegenes/alleles related to a single trait,and the alleles do not exert individuallydetectable effects on the expression ofthe trait. For example, when one geneconfers resistance to a specific diseaseor pest, breeders would be unable to usetraditional phenotypic screening to addanother gene to the same cultivar in orderto increase the durability of resistance.In such cases, MAS would be the onlyfeasible option, provided markers areavailable for such genes.• Early detection. MAS offers potential savingscompared with conventional selectionwhen it allows alleles for desirabletraits to be detected early, well beforethe trait is expressed and can be detectedphenotypically. This benefit can be particularlyimportant in species that growslowly, for example, tree crops.• Recessive genes. MAS offers potential savingscompared with conventional selectionwhen it allows breeders to identifyheterozygous plants that carry a recessiveallele of interest whose presence cannotbe detected phenotypically. In traditionalbreeding approaches, an extra step ofselfing is required to detect phenotypesassociated with recessive genes.• Heritability of traits. Up to a point, gainsfrom MAS increase with decreasing heritability.However, due to the difficultiesencountered in QTL detection, the gainsare likely to decline beyond a certainthreshold heritability estimate.• Seasonal considerations. MAS offerspotential savings compared with conventionalselection when it is necessaryto screen for traits whose expressiondepends on seasonal parameters. Usingmolecular markers, at any time of theyear, breeders can screen for the presenceof an allele (or alleles) associatedwith traits that are expressed only duringcertain growing seasons. For example,CIMMYT’s wheat breeding stationin northern Mexico is usually used forscreening segregating germplasm for leafrust resistance. However, expression ofleaf rust is not uniform in all growingseasons. The same concept is true forfield screening for drought tolerance.When there are seasons with low expressionof leaf rust or less intense droughtdue to unexpected rainfall, markers, ifavailable, can be a valuable alternative asa tool for screening.• Geographical considerations. MAS offerspotential savings when it is necessary

398Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishto screen for traits whose expressiondepends on geographical considerations.Using molecular markers, breeders inone location can screen for the presenceof an allele (or alleles) associated withtraits expressed only in other locations.• Multiple genes, multiple traits. MASoffers potential savings when there is aneed to select for multiple traits simultaneously.With conventional methods, it isoften necessary to conduct separate trialsto screen for individual traits.• Biological security considerations. MASoffers potential advantages over selectionbased on the use of potentially harmfulbiological agents (e.g. artificial viralinfections or artificial infestations withinsect pests), which may require specificsecurity measures.In view of the above-mentioned factors,it is desirable to consider MAS approacheson a case-by-case basis, taking into accountfactors such as the importance of a trait inthe overall breeding scheme, the amount ofavailable resources in terms of both staff andoperational expenditures, and the nature ofthe breeding materials. There are no “onesize fits all” recommendations that can bemade for MAS approaches. Usually, nobreeding scheme focuses on improving justone trait. At current levels of capacity, MASis likely to be used to achieve genetic gainsfor single traits such as host plant resistanceto pests and/or diseases. Therefore,MAS activities should be integrated into anoverall breeding programme.Challenges for developing countriesThe rapid expansion of agricultural biotechnologyis generating a wide array ofmethodologies with potential applications,and therefore national programmesin developing countries face the difficultchallenge of identifying priority areas forinvestment. To complicate matters further,the private sector dominates many fields ofbiotechnology research and therefore hasproprietary rights to many technologies andproducts that have immediate applicationsin developing countries (e.g. transgenictechnology). This is quite different fromconventional plant breeding technologies,most of which were developed by publiclyfundedresearch programmes and thus haveremained more accessible.There is no single answer to meetingthese challenges, especially as developingcountries are not uniform in their publicagricultural research capacities. Broadlyspeaking, developing countries fall into thefollowing categories:• countries (a few) with strong publicsector research infrastructure enablingbiotechnology applications, as well asupstream research capability to developtools for their own specific needs;• countries with intermediate capacity inapplied plant breeding, as well as inusing biotechnology tools that are publiclyavailable or can be acquired throughbilateral partnerships with the privatesector;• countries (a considerable number) withmoderate plant breeding capacity andpractically no, or very little, capacity forbiotechnology applications.More advanced developing countrieswith major commercial farming sectors aremore likely to succeed in adopting agriculturalbiotechnology. In addition, thepresence of commercial opportunities willattract investment by private industry andthus allow the country to benefit fromfuture advances in biotechnology. This isnot always a positive outcome for the publicsector because, as competition increases,it may be more difficult to justify largepublic investments in biotechnology. This

Chapter 19 – Technical, economic and policy considerations on marker-assisted selection in crops 399has occurred to some degree in maize biotechnology,even in the United States.Developing countries, in which agricultureis still dominated by subsistencefarming and where there is limited or nocapacity for biotechnology research, areat an added disadvantage. Resource-poorfarmers in such countries rarely offer adequatemarket incentives for the privateindustry that dominates biotechnologyresearch. For example, the involvement ofthe private sector in research and developmentactivities for root crops or grainlegumes is doubtful as these crops aregrown mainly by small-scale farmers inpoorer regions of the world and therewould be potentially low returns oninvestment. Therefore, it is important thatinternational development agencies ensurethat neither the “orphan commodities”yielding broad socio-economic benefits,nor the less advantaged and least developedcountries, are left out from the prospectof harnessing potential benefits associatedwith biotechnology. In doing so, theymust evaluate what biotechnology toolscan be of immediate benefit to such cropsand countries and then develop strategiesleading to successful adoption by the targetgroups. This can only be accomplishedif the efforts made are serious, long-termand sustainable. Many examples can becited where international aid agencies haveinvested in purchasing equipment designedfor biotechnology research in developingcountries but, when the aid programmesterminate their short-term involvement,the capital investments either have not beenoptimally utilized or have remained idle.Policy-makers in different nationalprogrammes must also bear in mind thatsustained capacity in public agriculturalresearch is a pre-requisite for successfulapplication of biotechnology tools includingMAS for crop improvement. Biotechnologytools can be used to enhance genetic gainsfor a few traits in a few crops, but theirultimate impact depends on how well theyare adopted and integrated into existingplant breeding activities. This is a soberingthought, because in many developing countriespublic sector research capacity is beingeroded and public sector extension servicesare being severely curtailed.Other factors essential for the successfulapplication of biotechnology tools aretraining and capacity building. Many biotechnologyapplications require learningnew skills, some research infrastructureand effective operational capacity. It isespecially important to train and nurturenational scientists capable of usingemerging technologies. In general, it maynot be possible for older plant scientiststo acquire the capacity for biotechnologyapplications. Therefore, policy-makers indeveloping countries have to consider longterminvestments in training and nurturinga new generation of scientific talent. Theyalso need to consider how to utilize thistalent effectively by providing adequateresources and optimum work environments.Specialized technical training mustin turn be underpinned by complementarygovernment investments in basic education,e.g. by including biotechnology-relatedsubjects in national university curricula.Although it is widely assumed thatenormous investments are needed to establisha capacity to carry out MAS, this is notalways true. Certainly, a minimum level ofinvestment is needed for laboratory facilities,equipment and trained staff. However,considering that most MAS work in developingcountries is likely to be gearedtowards the use of existing markers ratherthan the development of new ones, investmentsin facilities and capital need not be

400Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishlarge. Developing countries are likely tohave difficulty obtaining the required laboratorymaterials including consumablesthat are manufactured mostly in the industrializedworld. Other factors such as localsupport for servicing and maintaining laboratoryequipment and reliable basic servicessuch as an uninterrupted power supply canalso be challenging. In the less advanceddeveloping countries, international researchorganizations and development assistanceagencies will have a more significant roleto play in ensuring the availability of thetechnology as well as the capacity to use iteffectively, though on a limited scale.Many developing countries are likely touse genetically modified cultivars with valueadded traits in the near future. Associatedwith transgenic technology are the complex,yet important, issues of biosafetyand management of intellectual property.Policy-makers should therefore also considerways of increasing the efficiency ofpublicly-funded research efforts, as wellas finding opportunities and providingincentives for formulating productivepublic–private sector partnerships. As mosttools of biotechnology that have potentialpractical applications are developed andpatented by private industry, policy-makershave the challenges of addressing the needto forge research partnerships that allowthe competitive private sector to maintainits interest in financial rewards while permittingtechnologies to be used by publicsector researchers in relevant areas to servefarmers in species of importance that haveso far been neglected. Coupled with thesepartnerships is the requirement to manageintellectual property issues.In many situations, international developmentagencies are able to play a rolein areas such as biotechnology prioritysetting,raising funds for establishing therequired biotechnology infrastructure andmaintenance capacity, supporting public–private sector partnerships, and assisting intechnology transfer and capacity building.International agricultural research institutes,which have had long-term involvement withnational programmes in a large numberof developing countries, should play arole in identifying key areas for contributingfurther in helping relevant nationalprogrammes identify, optimize and adoptMAS tools when it is feasible. Internationalresearch centres can also play an active rolein capacity building by identifying areaswhere it is needed and by providing necessarybackstopping.Novel marker systems based on SNPplatforms are likely to bring the costs associatedwith MAS applications to an affordablelevel by many breeding programmes andit will be challenging to establish thesetechnologies based on robotics and otherautomated, large-scale, screening platformsin many developing countries asthe technology development and associatedintellectual property rights remain in largeprivate sector enterprises. This is an areawhere developing country policy-makers,together with international aid bodies andresearch organizations, should ideally worktogether to find partnerships with the privatesector to devise ways of infusing thesetechnological breakthroughs and associatedbenefits to the developing countries, at leaston a limited scale.In conclusion, MAS technologies havematured to the extent that they can beused for making genetic gains in certaintraits and in some important crop species.National programmes in developingcountries should evaluate the feasibilityof applying MAS approaches for cropimprovement as, despite the considerablelimitations that exist in many developing

Chapter 19 – Technical, economic and policy considerations on marker-assisted selection in crops 401countries, the technology can be used ata relatively low operational cost. At leastfor major crops such as rice, maize, wheatand soybean, significant numbers of linkedmarkers have been identified for genes ofinterest, and ongoing selection programmeshave found them to be useful for makingrapid genetic gains. Incorporating thesetools into active breeding strategies willallow more rapid and efficient improvementof varieties for target traits.As national programmes in developingcountries vary in their capacities to absorbbiotechnology tools, priority-setting andidentification of MAS strategies should bedone on a case-by-case basis, ideally supportedby strong breeding programmes.Individual national programmes will haveto be selective in their choice of technologiesand markers to ensure that the level ofinvestment is appropriate to justify the costsand produce the most rapid returns. Thismeans that, while fully functioning biotechnologylaboratories may not be feasible inall countries, initiating MAS is an importantfirst step towards using modern biotechnologyapproaches in plant improvement.As the success of biotechnology applicationsdepends on the existence of strongcrop improvement programmes, policymakersand international developmentagencies must ensure that the limited fundsallocated to traditional agricultural researchare not curtailed to support biotechnologyactivities. International aid agencies andagricultural research institutes should playa role in building research capacity withinnational programmes, encouraging public–private sector partnerships, and promotingtechnology transfer.ReferencesAyala, L., Henry, M., Gonzalez-de-Leon, D., van Ginkel, M., Mujeeb-Kazi, A., Keller, B. &Khairallah, M. 2001. A diagnostic molecular marker allowing study of Th. Intermedium derivedresistance to BYDV in bread wheat segregating populations. Theor. Appl. Genet. 102: 942–949.Bouchez, A., Hospital, F., Causse, M., Gallais, A. & Charcosset, A. 2002. Marker assisted introgressionof a favorable alleles at quantitative trait loci between maize elite lines. Genetics 162:1945–1959.Brennan, J.P. 1989. An analysis of economic potential of some innovations in a wheat breeding programme.Austr. J. Agric. Econ. 33 (1): 48–55.Byerlee, D. & Traxler, G. 2001. The role of technology spillovers and economies of size in theefficient design of agricultural research systems. In P. Pardey & M. Taylor, eds. Agricultural sciencepolicy: changing global agendas, pp. 161–186. Baltimore, MD, USA, The Johns HopkinsUniversity Press.Brown, S.M. & Kresovich, S. 1996. Molecular characterization for plant genetic resource conservation.In A.H. Paterson, ed. Genome mapping in plants, pp. 85–93. Austin, TX, USA, R.G. Landers Co.Castro, A.J., Chen, X., Corey, A., Filichkina, T., Hayes, P., Mundt, C., Richardson, K., Sandoval-Islas, S. & Vivar, H. 2003. Pyramiding and validation of quantitative trait loci (QTL) allelesdetermining resistance to barley stripe rust: effects on adult plant resistance. Crop Sci. 43:2234–2239.Chavarriaga-Aquirre, P., Maya, M.M., Tohme, J., Duque, M.C., Iglesias, C., Bonierbale, M.W.,Kresovich, S. & Kochert, G. 1999. Using microsatellites, isozymes and AFLPs to evaluate geneticdiversity and redundency in the cassava core collection and to assess the usefulness of DNA basedmarkers to maintain germplasm collections. Mol. Breeding 5: 263–273.

402Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishCregan, P.B., Mudge, J., Fickus, E.W., Danesh, D., Denny, R. & Young, N.D. 1999. Two simplesequence repeat markers to select for soybean cyst nematode resistance conditioned by the rhg1locus. Theor. Appl. Genet. 99: 811–818.Dean, R.E., Dahlberg, J.A., Hopkins, M.S. & Kresovich, S. 1999. Genetic redundancy and diversityamong ‘Orange’ accessions in the US sorghum collection as assessed with simple sequence repeat(SSR) markers. Crop Sci. 39: 22–32.Dilday, R.H. 1990. Contribution of ancestral lines in the development of new cultivars of rice. CropSci. 30: 905–911.Dreher, K., Morris, M.L., Khairallah, M., Ribaut, J-M., Pandey, S. & Srinivasan, G. 2002. Ismarker-assisted selection cost-effective compared with conventional plant breeding methods? Thecase of quality protein maize. In R. Evenson, V. Santaniello & D. Zilberman, eds. Economic andsocial issues in agricultural biotechnology, pp. 203–236. Wallingford, UK, CABI.Dreher, K., Khairallah, M., Ribaut, J-M. & Morris, M.L. 2003. Money matters (I): Costs of fieldand laboratory procedures associated with conventional and marker-assisted maize breeding atCIMMYT. Mol. Breeding 11: 221–234.Dubcovsky, J. 2004. Marker assisted selection in public breeding programs: the wheat experience.Crop Sci. 44: 1895–1898.Dudley, J.W. 1993. Molecular markers in plant improvement: manipulation of genes affecting quantitativetraits. Crop Sci. 33: 660–668.Eagles, H.A., Bariana, H.S., Ogbonnaya, F.C., Rebetzke, G.J., Hollamby, G.H., Henry, R.J.,Henschke, P.H. & Carter, M. 2001. Implementation of markers in Australian wheat breeding.Austr. J. Agric. Res. 52: 1349–1356.Fulton, T.M., Grandillo, S., Beck-Bunn, T., Fridman, E., Frampton, A., Lopez, J., Petiard, V.,Uhlig, J., Zamir, D. & Tanksley, S.D. 2000. Advanced backcross QTL analysis of a Lycoperssiconesculentum x Lycopersicon parvifolium cross. Theor. Appl. Genet. 100: 1025–1042.Gardiner, J., Melia-Hancock, S., Hoisington, D.A., Chao, S. & Coe, E.H. 1993. Development of acore RFLP map in maize using an immortalized F 2 population. Genetics 134: 917–930.Gowda, M., Venu, R.C., Roopalakshmi, K., Sreerekha, M.V. & Kulkarni, R.S. 2003. Advances inrice breeding. Mol. Breeding 11: 337–352.Han, F., Romagosa, I., Ullrich, S.E., Jones, B.L., Hayes, P.M. & Wesenberg, D.M. 1997. Molecularmarker-assisted selection for malting quality traits in barley. Mol. Breeding 3: 427–437.Hittalmani, S., Parco, A., Mew, T.V., Zeigler, R.S. & Huang, N. 2000. Fine mapping and DNAmarker assisted pyramiding of the three major genes for blast resistance in rice. Theor. Appl. Genet.100: 1121–1128.Hoisington, D.A. & Melchinger, A.E. 2004. From theory to practice - marker assisted selection inmaize. In: H. Lorz & G. Wenzel, eds. Biotechnology in agriculture and forestry: molecular markersystems, pp. 335–352. Berlin, Heidelberg, Germany, Springer-Verlag.Huang, N., Angelels, E.R., Domingo, J., Mgapantay, G., Singh, S., Zhang, G., Kumaravadivel,N., Bennett, J. & Kush, G.S. 1997. Pyramiding of bacterial blight resistance genes in rice: markerassisted selection using RFLP and PCR. Theor. Appl. Genet. 95: 313–320.Jenkins, S. & Gibson, N. 2002. High-throughput SNP genotyping. Comp. Funct. Genom. 3: 57–66.Lagudah, E.S., Moullet, O. & Appels, R. 1997. Map-based cloning of a gene sequence encodinga nucleotide binding domain and a leucine-rich region at the Cre3 nematode resistance locus ofwheat. Genome 40: 659–665.

Chapter 19 – Technical, economic and policy considerations on marker-assisted selection in crops 403Liu, J., Liu, D., Tao, W., Li, W., Wang, C.P., Cheng, S. & Gao, D. 2000. Molecular marker-facilitatedpyramiding of different genes for powdery mildew resistance in wheat. Plant Breeding 119:21–24.Maredia, M., Byerlee, D. & Maredia, K. 1999. Investment strategies for biotechnology in emergingresearch systems. Paper presented at 2 nd Ann. Conf. Internat. Consort. Agricult. Biotech. Res.(ICABR), Rome, Italy.Moreau, L., Charcosset, A., Hospital, F. & Gallais, A. 1998. Marker-assisted selection efficiency inpopulations of finite size. Genetics 148: 1353–1365.Morris, M.L., Ribaut, J-M., Khairallah, M. & Dreher, K. 2001. Potential impacts of biotechnologyassistedselection methods on plant breeding programs in developing countries. In P.G. Pardey,ed. The future of food: biotechnology markets and policies in an international setting, pp.197–218.Washington, DC, Johns Hopkins and IFPRI.Morris, M.L., Ribaut, J-M., Dreher, K. & Khairallah, M. 2003. Money matters (II): Costs of maizeinbred line conversion schemes at CIMMYT using conventional and marker-assisted selection.Mol. Breeding 11: 235–247.Pernet, A., Hoisington, D., Franco, J., Isnard, M., Jewell, D., Jiang, C., Marchand, J-L., Reynaud,B., Glaszmann, J-C. & Gonzalez-de-Leon, D. 1999a. Genetic mapping of maize streak virusresistance from the Mascarene source. I. Resistance in line D211 and stability against different virusclones. Theor. Appl. Genet. 99: 524–539.Pernet, A., Hoisington, D., Dintinger, J., Jewell, D., Jiang, C., Khairallah, M., Letourmy, P.,Marchand, J-L., Galszmann, J-C. & Gonzalez-de-Leon, D. 1999b. Genetic mapping of maizestreak virus resistance from the Mascarene source. II. Resistance in line CIRAD390 and stabilityacross germplasm. Theor. Appl. Genet. 99: 540–553.Rafalski, A. 2002. Applications of single nucleotide polymorphisms in crop genetics. Curr. Opin.Plant Biol. 5: 94–100.Ragot, M., Biasiolli, M., Delbut, M.F., Dell’orco, A., Malgarina, L., Thevenin, P., Vernoy, J.,Vivant, J., Zimmermann, R. & Gay, G. 1994. Marker assisted backcrossing: a practical example.Techniques et utilizations des marqueurs moleculaires, Les Coloques 72, pp. 45–56. Paris, INRA.Ribaut, J-M. & Hoisington, D.A. 1998. Marker-assisted selection: new tools and strategies. Trendsin Plant Sci. 3 (6): 236–239.Ribaut, J-M. & Ragot, M. 2007. Marker-assisted selection to improve drought adaptation in maize:the backcross approach, perspectives, limitations and alternatives. J. Exp. Bot. 58: 351–360.Ribaut, J-M., Jiang, C. & Hoisington, D.A. 2002. Simulation experiments on efficiencies of geneintrogression by backcrossing. Crop Sci. 42: 557–565.Ribaut, J-M., Hoisington, D.A., Deutsch, J.A., Jiang, C. & Gonzalez-de-Leon, D. 1996.Identification of quantitative trait loci under drought conditions in tropical maize: 1. Floweringparameters and the anthesis-silking interval. Theor. Appl. Genet. 92: 905–914.Ribaut, J-M., Jiang, C., Gonzalez-de-Leon, D., Edmeades, G.O. & Hoisington, D.A. 1997.Identification of quantitative trait loci under drought conditions in tropical maize: II. Yield componentsand marker assisted selection strategies. Theor. Appl. Genet. 94: 887–896.Sanders, J.H. & Lynam, J.K. 1982. Definition of the relevant constraints for research resource allocationin crop breeding programs. Agric. Admin. 9 (4): 273–284.Smith, S. & Helentjaris, T. 1996. DNA fingerprinting and plant variety protection. In A.H. Peterson,ed. Genome mapping in plants, pp 95–110. Austin, TX, USA, R.G. Landers Co.

404Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishTanksley, S.D. & McCouch, S.R. 1997. Seed banks and molecular maps: unlocking genetic potentialfrom the wild. Science 277: 1063–1066.Tanksely, S.D. &. Nelson, J.C. 1996. Advanced backcross QTL analysis: a method for the simultaneousdiscovery and transfer of valuable QTLs from unadapted germplasm into elite breeding lines.Theor. Appl. Genet. 92: 191–203.Tanksley, S.D., Young, N.D., Paterson, A.H. & Bonierbale, M.W. 1989. RFLP mapping in plantbreeding: new tools for an old science. Biotech. 7: 257–264.Toenniessen, G.H., O’Toole, J.C. & DeVries, J. 2003. Advances in plant biotechnology and its adoptionin developing countries. Curr. Opin. Biotech. 6: 191–198.Villegas, E. 1994. Factors limiting quality protein maize (QPM) development and utilization. In B.A.Larkins & E.T. Mertz, eds.. Proc. Intl. Symp. on Quality Protein Maize, 1964-1994, pp 79–88. SeteLagoas, Minas Gerais, Brazil, Embrapa/CNPMS.Werner, K., Friedt, W. & Ordon, F. 2005. Strategies for pyramiding resistance genes against thebarley yellow mosaic virus complex (BaMMV, BaYMV, BaYMV-2). Mol. Breeding 16: 45–55.Willcox, M.C., Khairallah, M., Bergvinson, D., Crossa, J., Deutsch, J.A., Edmeades, G.O.,Gonzalez-de-Leon, D., Jiang, C., Jewell, D.C., Mihm, J.A., Williams, W.P. & Hoisington, D.A.2002. Selection for resistance to southwestern corn borer using marker assisted and conventionalbackcrossing. Crop Sci. 42: 1516–1528.William, H.M., Trethowan, R. & Crosby-Galvan, E.M. 2007. Wheat breeding assisted by markers:CIMMYT’s experience. Eyphytica (In press).Xie, C. & Xu, S. 1998. Strategies of marker aided recurrent selection. Crop Sci. 38: 1526–1535.Xu, Y., Beachell, H. & McCouch, S.R. 2004. A marker based approach to broadening the genetic baseof rice in the USA. Crop Sci. 44: 1947–1959.Young, N.D. 1999. A cautiously optimistic vision for marker-assisted breeding. Mol. Breeding 5:505–510.Yousef, G. & Juvik, J. 2001. Comparison of phenotypic and marker-assisted selection for quantitativetraits in sweet corn. Crop Sci. 41: 645–655.Yu, K., Park, S.J. & Poysa, V. 2000. Marker-assisted selection of common bean for resistance tocommon bacterial blight: efficacy and economics. Plant Breeding 119: 411–415.Zhang, T., Yuan, Y., Yu, J., Guo, W. & Kohel, R.J. 2003. Molecular tagging of a major QTL forthe fiber strength in upland cotton and its marker-assisted selection. Theor. Appl. Genet. 106:262–268.Zhou, P.H., Tan, Y.F., He, Y.Q., Xu, C.G. & Zhang, Q. 2003. Simultaneous improvement for fourquality traits of Zhenshan 97, an elite parent of hybrid rice, by molecular marker-assisted selection.Theor. Appl. Genet. 106: 326–331.Zhou, W-C., Kolb, F.L., Bai, G.H., Domier, L.L., Boze, L.K. & Smith, N.J. 2003. Validation of amajor QTL for scab resistance with SSR markers and use of marker-assisted selection in wheat.Plant Breeding 122: 40–46.

Chapter 20Impacts of intellectual propertyrights on marker-assisted selectionresearch and application foragriculture in developing countriesVictoria Henson-Apollonio

406Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishSummaryAlthough the impact of marker-assisted selection (MAS) in commercial and public sectorbreeding programmes in developing countries is to date limited to a few crops and traits,the potential benefits of using markers linked to genes of interest in breeding programmesfor improving the productivity of crops, livestock, forest trees and farmed fish issubstantial. While more recent methods associated with the use of MAS are technicallydemanding and often expensive, most applications of basic MAS were initially describedin the literature, and thus will likely have very few intellectual property (IP) restrictionsassociated with their use, irrespective of the agricultural sector involved. For example,isolating DNA, amplifying specific gene sequences from that DNA (with most availableprimers), separating fragments using gel/polyacrylamide electrophoresis and imagingof fragments with standard techniques are likely to be available without restriction toscientists and breeders in the developing world, even as part of a commercial service.Problems arise when there is a need to use or develop high-throughput modes, whichrequire more sophisticated technologies. For high-throughput use, a breeder will wantto use the most efficient techniques that are currently available. This means that the moreadvanced processes/methods, reagents, software applications/simulations and equipment,which provide the most effective means to exploit MAS fully, are most likely covered byintellectual property rights (IPRs) such as patent rights, confidential information (tradesecrets) and copyrights, both in industrialized countries and also in many developingcountries such as Brazil, China and India. In situations where breeders wish to use cuttingedge technologies and the most efficient markers, care must be taken to avoid activities thatmay infringe IPRs when using MAS methodologies.

Chapter 20 – Impacts of intellectual property rights on marker-assisted selection 407IntroductionOther chapters in this book describe theusefulness and applicability of MAS fordeveloping germplasm with superior qualities,in a timely manner. Markers have beendeveloped and used by plant and animalbreeders (Dekkers, 2004), for fish and shellfish(Consuegra and Johnston, 2006) andfor forest trees (Kellison, McCord andGartland, 2004; Lee, A’Hara and Cottrell,2005). Introduction of MAS to developingcountry scientists has been taken up by avariety of projects such as the GenerationChallenge Programme (cgiar.org/exco/exco8/exco8_generation_report), supportedby the Consultative Group on InternationalAgricultural Research (CGIAR) and MASjamborees sponsored by the SyngentaFoundation for Sustainable Development(syngentafoundation.org/pdf/Report%20Nairobi%20meeting%20.pdf and T. St.Peter, personal communication). MAS isalso being used by many of the centresbelonging to the CGIAR, notable examplesbeing the International Center for TropicalAgriculture (CIAT), the InternationalPotato Centre (CIP), the InternationalCrops Research Institute for the Semi-AridTropics (ICRISAT) and the InternationalInstitute for Tropical Agriculture (IITA),as well as the International Maize andWheat Improvement Centre (CIMMYT),in programmes such as the Asian MaizeBiotechnology Network (AMBIONET),and the International Livestock ResearchInstitute (ILRI) in the areas of livestock productionand health through the BiosciencesFacility for east and central Africa (BecA).In this chapter, a brief review of generalintellectual property law is used to introducea variety of aspects regarding intellectualproperty potentially associated with the useof the techniques, reagents and equipmentthat are necessary for implementing MAS.This intellectual property “primer” isfollowed by a description of specific casesand some recommendations regardingsteps that should be taken by scientistsand breeders in developing countries whoare contemplating using MAS in breedingprogrammes to avoid restrictions orincurring risks of infringing the intellectualproperty rights (IPRs) of others.Intellectual property rights andpublic access to innovationIPRs are awarded on the basis of nationallaws. There are, however, a few examples ofregional cooperation institutions grantingIPRs on a regional basis, such as the AfricanRegional Intellectual Property Organization(ARIPO), the Gulf CooperationCouncil (GCC) and the European PatentOffice. In addition, under the PatentCooperation Treaty (PCT), an internationalagreement administered by the WorldIntellectual Property Organization (WIPO)that facilitates patent filing, a single internationalapplication can be filed in a nationalPCT-receiving office, which can then subsequentlybe submitted to all PCT membernational patent offices. In addition, anexample situation is given in Box 1 thatillustrates the, perhaps unexpected, “farreach” of national patent law.IPRs comprise original and novel assetsthat involve the use of human intellect.The awarding of such rights is intendedto balance the needs of society to accessand use the products of human ingenuity,with rewards for the endeavours going tothe individuals from whom these intellectualassets originated. Obviously, thereis a certain amount of tension in thisSee www.wipo.int/pct/en/texts/pdf/pct_paris_wto.pdf for a list of countries that are members ofimportant IP international treaties, including thePCT.

408Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishbalance between private rights and the needsof society (Murchie et al., 2006). Society ispresumed to benefit from public disclosurein the form of patent disclosure requirementsand copyrights, which are awardedto creative works that have been fixed (madetangible). In addition, through the combinationof the requirement of full disclosurein the written description (manifested in apatent application), and the time limitationover patents rights, inventions are put intothe public domain when the rights expire.The pharmaceutical industry’s experiencewith the success of “generics” is a testamentto the value of “expired” inventions (CBO,1998). In some specific cases, however, suchas patents on certain drugs, rights may beextended for a certain period of time uponrequest to compensate for long delays inobtaining authorization for drug commercialization,e.g. the “Hatch-Waxman” actin the United States of America. The filing,prosecuting and maintenance of patents arebusiness decisions that are put into placeas a part of the strategy for bringing productsto the consumer. An additional partof such a strategy can also include a planfor maintaining profitability when patentrights expire (Smyth, 2006). For example,depending upon the creativity of inventors,it may be that improvements allowfor the filing of additional patents to coverthese improvements, thus having the effectof extending patent rights for additionalterms. This is a fact-based process, in thatthe improvement must meet the requirementsof invention. Not that efficient and sincere disclosure is notwithout problems – see Fromer (2007).Note that concerns regarding the abuse of thepatenting of incremental changes versus incrementalimprovements are often raised by a practice of patenttime extension called “evergreening”. For furtherdiscussion of evergreening from the view of genericpharmaceutical manufacturers, see Hore (2004).The balance between public and privaterights is considered by some to be tiltedin favour of private rights, leaving elementsof some societies wondering if IPRsystems work at all except to protect themonopolies that they award (Epp, 2004).A number of civil society organizations aremonitoring the potential effect of changingIndia’s patent law to include patents overpharmaceutical products and agriculturalchemicals (Sreedharan, 2007).Intellectual property law as itrelates to MASThe standard steps employed in MASgenerally include: selecting individuals tobe tested; harvesting material; extractionof DNA from the material; polymerasechain reaction (PCR) amplification of theDNA to enrich for gene sequences/fragmentsassociated with a particular trait orphenotype; separation of these fragments;visualization/identification of DNA fragments;and interpretation and utilizationof the information. Each of these stagesinvolves certain methods and the use ofparticular reagents and/or equipment associatedwith the particular methodologicalsteps. For the purposes of this chapter, aseries of tables (numbers 1–3) has been preparedto exemplify the types of intellectualproperty and associated IPRs that exist formaterials and/or processes within each ofthese seven steps.There is a general set of categories ofIPRs that are awarded in most countries/jurisdictions. These include industrial orutility patent rights, plant variety protection/plantbreeders’ rights, copyrights,rights of appellation/geographic indications,trademarks and secrecy rights (tradesecrets) associated with undisclosed or confidentialinformation. Other types of patentrights can be awarded in many jurisdictions.

Chapter 20 – Impacts of intellectual property rights on marker-assisted selection 409For example, in addition to utility patents,two others types of categories of patentsare available to inventors in the UnitedStates: a design patent for a new originalor ornamental design for an article of manufacture,granted to protect the externalappearance rather then the function of aproduct, and plant patents, awarded for theinvention or discovery of a cultivated plantvariety that can be asexually reproduced,(except via tubers, but including graftsand spores). Other countries have additionalcategories regarding subject matter(e.g. designs, plants) and also with respectto examination rigour and length of thepatent rights grant (e.g. “short-term” patentsin Belgium and the Netherlands (seee.g. www.ipr-helpdesk.org/docs/docs.EN/invencionesTecnicasBP.pdf), and innovationpatents in Australia (www.ipaustralia.gov.au/patents/what_innovation.shtml).Patent rights are awarded to inventionson the basis of criteria associated with usefulness(industrial applicability), originality(newness or novelty), and an “inventivestep” (non-obviousness to persons withtechnical skills in the particular field wherethe invention is applicable). There are alsorules governing the subject matter of theinvention for utility patent rights to beawarded. For example, all countries’ patentrights prohibit the awarding of patent rightsfor elucidating the “laws of nature”. Thus,the fact that scientists have described lawsof chemistry and physics, natural selection,or other such natural laws, does notrender them as products of a person’s intellectin intellectual property law. However,an innovation that applies one of these lawsmay well qualify for protection. Similarly,in many countries a new plant variety, avariety, type or breed of livestock used forfood production, or computer softwarecannot be the subject of patent rights. Japanand the United States are notable exceptionsin this regard. While the EuropeanUnion (EU) (Directive 98/44/EC of theEuropean Parliament and of the Council,1998 on the legal protection of biotechnologicalinventions) does not permit thepatenting of plant and animal varieties,it does allow patents for inventions concerninganimals or plants the feasibility ofwhich is “not confined to a particular plantor animal variety”. The fact that the term“variety” is not well defined in the contextof animal breeding means that the scope ofthis exemption is far from clear.Irrespective of whether one is dealing withpatent rights, plant breeders’ rights (PBRs),copyrights, trademarks, trade secrets, etc.,the type of IPR sought or awarded varieswith the type of intellectual asset over whichprotection is being sought. It is also possiblefor one asset to be protected by severaltypes of rights, depending upon the law inthe applicable territory. For example, it isnot unusual to have “double protection”,i.e. for an invention to be patented and theproduct resulting from that invention to betrademarked. The trademark for Aspirin®for the formerly patent-protected acetylsalicylicacid is such a case in many parts of theworld. It is not uncommon for a process ora piece of machinery to be treated in a similarfashion. This situation pertains to IPRsassociated with MAS, two notable examplesbeing “Selective restriction fragmentamplification: a general method for DNAfingerprinting”, a patented process pairedwith rights associated with the AFLP®trademark or the “Methods for genotypingby hybridization analysis” patent and theassociated DArT trademark.PatentsPatent rights are awarded on the basis ofclaims based on the inventor’s description

410Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishto explain the new, non-obvious patentablesubject matter in a way that clearly distinguishesits novel characteristics from allother available solutions. This explanation iscalled a patent “claim”, and using the wordsof the patent drafter a claim will describethe “metes and bounds” (Gallagher, 2002)of the invention. Patent drafters are usuallylicensed patent agents, patent attorneys,scientists working for legal firms in thiscapacity or, rarely, the inventors themselves.Drafting patent claims is an arcaneart that requires detailed knowledge of thescientific and technical basis of the inventionas well as a current understanding ofthe state-of-the-art, regarding the judicialinterpretation of claims, in the context ofnational patent law.One patent can have many claims. Infact, patent law requires that every patentmust contain at least one claim. Each claimis “directed to” an invention, ranging fromits broad use, to the most narrow use forwhich an inventor may wish to seek rights.For example, a broad claim could be forthe use of an enzyme class to perform atype of function (where this combinationis not found in nature). A narrow claimcould then specify the particular enzyme,the quantity of enzyme and/or the specificfunction. A distinction should be madebetween a patent application (often numberedin a different style such as the “WO”designation for PCT-filed patent applications),and an issued patent (generallynumbered with a country prefix, e.g. CA2172863, a patent issued by the CanadianPatent Office) to avoid confusion.Patent applications contain claims thatare untested and unexamined and theseclaims are therefore often very broad.During the patent prosecution process,the patent examiner seeks to limit claimsto the new invention held by the applicantat the time the patent was filed. The claimsare accompanied by written descriptionsthat would allow someone else familiarwith technology in the same general area(“person having ordinary skill in-the-art”or “PHOSITA”), to understand how tomake and carry out or “work” the claimedinnovation. This useful written descriptionaccompanying claims is directed by law toprovide “enablement”, and is a requiredpart of a patent disclosure, in order to makethe invention “available to the public”(this is part of the social contract to balanceprivate rights and public good). Thewritten descriptions can also be importantfor interpreting the exact limits of patentclaims. Patent rights are given to inventionsthat cover the reduction of ideas and conceptsto practical use, and these rights mayalso extend to other treatments/variationsthat are of a nature sufficiently similar tobe equivalent to the patented innovation.Such a “doctrine of equivalents”, as it iscalled in patent lingo, means that ideas/concepts that are the basis of the usefulinnovation are a part of the patent claimcoverage. Therefore, it is often stated thatpatents cover conceptual ideas as well asthe practical application of the idea (seewww.dwalkerlaw.com/patent.asp). Thismeans that it is often difficult to discernwhether a party is committing infringementwithout the interpretation of a court.Literal infringement, whereby the inventionis practised exactly as it is described ina claim, can usually be identified without aproblem. Equivalent infringement is oftenused as a strategic business tool by eitherthe patent rights holder and/or the infringer.This confusion over the exact limits ofpatent claims can often lead to companymergers or buy-outs, just to minimize therisk associated with the IPRs (Fulton andGiannakas, 2001; Kattan, 2002).

Chapter 20 – Impacts of intellectual property rights on marker-assisted selection 411Box 1Territoriality of patent rightsDeveloping country scientists and breeders should be aware that patent rights are onlyenforceable within the jurisdiction of the country or countries where the patent rights havebeen awarded. The caveat to this is that patent laws in most countries cover material that isimported into a country when patent rights exist on that material in the country where theimportation would take place. The language that is included in such patent laws contains theterms: “making”, “selling” or “using” within a country’s boundaries. For example, if patentrights over the formula for a particular herbicide had been awarded in Country AA, but nopatent rights over this same herbicide composition had been awarded in Country BB, then theherbicide could be made in Country AA only with the permission of the patent rights holder.However, the herbicide could be made in Country BB without permission of the rights holder inCountry AA; no infringement would be possible in Country BB. If someone wanted to importthe herbicide that was made in Country BB into Country AA, then the importer in Country AAwould need to obtain permission (a licence) from the rights holder in Country AA.The situation for Argentinian soybean containing a transgene covered by patent rights issuedto Monsanto in Europe is a good illustration of the territoriality of patent rights. Monsantoholds plant breeders’ rights over the variety, but does not have patent protection for the genein Argentina. Many farmers in Argentina are growing herbicide resistant soybeans developedby Monsanto, (often using seed multiplied by companies that do not have a licence fromMonsanto). The company has taken the strategy of preventing the importation of Argentiniangrownsoybeans or products made from Argentinian-grown soybean into any country whereMonsanto has patent rights by informing potential buyers of Argentinian-grown soybeans thatthey will be infringing Monsanto’s patent rights if they bring such material into a country suchas the United States or an EU country, where Monsanto has patent rights over the technologyembedded in the seed or over the seed itself (Balch, 2006), and therefore also present in thesoybean imported grain. Monsanto’s patent covers the final product, that is the gene, andextends its protection to the seed and the grain containing the gene sequence. The EuropeanCommission (EC), in fact, recognizes the right of Monsanto to prevent import of the soybeangrain, but not the soybean flour, where the gene sequence can no longer be expressed.What, however, is the relevance of such action to MAS, where there is no technologyembedded in the seed, remaining in the seed itself? Patent law is usually interpreted to coverany material where a patented technology was used to produce a product, even though sucha product does not literally contain the technology. This means that in most situations, ifpatent-protected techniques, methods, processes or products are used in a MAS scheme, theresulting products are covered by these patent rights. Of course, this type of infringement canbe very difficult to prove and therefore is rarely the subject of a legal suit, but the risk is presentand occasionally is enforced (AsiaLaw, 2004). However, for developing country farmers whoare not going to be exporting a product to an industrialized country, in actuality, the risk ofan infringement is minimal (Binnebaum et al., 2003). Nevertheless, the situation of using apatented invention without permission of the patent rights holder is not straightforward and, ifsuch a course involves public resources, it should only be embarked upon on the advice of anIP counsel or an IP lawyer.

412Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishTable 1Examples of patents relevant to MASTechnique Selected patent examples 1 Public domain equivalent Status of selected patent example ImplicationsHarvesting DNA Use of silica particles US 5 234 809 Many, e.g. Doyle andDoyle, 1987 and Saghai-Maroof et al. 1984Amplification ofspecific DNA seqsIdentification ofmarker genesEquipment“Matrix Mill”US 6 063 616 Many, e.g. Edwards,Johnstone andThompson, 1991.Combination withcentrifuge tubeIn effect in the United States;related patents in effect in: Austria,Australia, Canada, Denmark,Germany, EU, Greece, Japan,Republic of Korea, Netherlands,South Africa and SpainReagents US 4 683 195 None Expired in all countries (thereforein public domain)Primers/genes Primers for identifyingSoybean Sudden DeathResistanceUS 6 300 541Equipment Applied Biosystems ThermalCyclerUS 5 656 493Reagents Agarose, no applicablepatents foundEquipment Charge-coupled deviceimaging apparatusMany e.g., Röder et al.,1998. gwm493 in wheatOther equipment isavailable; contentiouslegal issues associatedwith manyLicence needed; often suppliedwith reagents, kits and/orequipment such as thermal cyclersIn effect in the United States If specialized equipment is used,licence may be needed. Likelihoodthat coverage would extend todeveloping country areasAdvance or improvement likelywill require licensing, many evenin developing countriesIn effect in the United States Sequence(s) to be used shouldbe checked by a patent searchersuch as Gene-IT.com if breedingproduct is valuable and would begrown for exportIn effect is the United States andmost other developed countries,and a few developing countriesincluding Brazil, China, Republicof Korea, South AfricaPolyacrylamide No patent rights on traditional gel/acrylamide mediaCameras Many systems that are no longerunder rights protectionMAS methods, ingeneralMethods andcompositionsUS 5 672 881Use of selective DNAfragment amplificationproducts for hybridizationbasedgenetic fingerprinting,MAS, and high-throughputscreening. US 6 100 030Numerous In effect in the United States Likely defensive patents. Could beproblematic with imports to theUnited StatesQTL mapping in plantbreeding populations.US 6 399 8551There will inevitably be innovative improvements or technological advancement associated with each of these methods and materials, many of which will have been awarded IPRs to theinventor and/or the inventor’s company.

Chapter 20 – Impacts of intellectual property rights on marker-assisted selection 413Examples of published patents whererights have been awarded in the area ofMAS include the basic PCR amplificationprocess patents in the United States, USPatent nos. 4 683 195, 4 683 202 and 4 965188, originally issued to the Cetus Companyand then assigned to Hoffman-Roche in1992, on the use of DNA polymerase basedon the Taq polymerase enzyme isolatedfrom the organism Thermus aquaticus. Asthese amplification patents expired worldwidein March 2006, when only the basictechniques and reagents covered by thesepatents are used, one does not now haveto be concerned with infringement of thesepatents anywhere. However, the equipmentused to control the reaction conditions mayalso carry IPRs on its own and most PCRtechniques currently used are patented asimprovements to the basic technology. Forexample, Applied Biosystems’ PCR andreal-time instrument patents and otherPCR-related patents such as US Patent no.5 656 493, are still in effect. A licence tothese instruments and other patents maybe needed in the United States in order touse their thermal cyclers to carry out PCR,although this is normally granted as partof the purchase price of the equipment andreagent kits. Table 1 contains additionalexamples of selected patents that are associatedwith MAS.Another strategy that should be pointedout is the concept of “defensive” patents.Patent rights may be awarded in most jurisdictionsover processes (actions/processes),and machines, manufactures and compositionsof matter (things). Enforcement ofpatent rights, e.g. bringing a lawsuit againsta person or forcing a licensing situationwhen a person is practising your invention(infringing your rights) without permissionis less equivocal when the infringementinvolves making, using, possessing orselling an object or composition. However,the detection of infringement of methodsclaims is often much less straightforward.A patent owner would need to have insightinto or gain access to how something wasmade or formed by the other party (potentialinfringer), in order to know whetherhis/her patented process or method wasbeing used. This means that it can be evenmore costly and time-consuming to pursuepotential infringers of methods claims thanlawsuits involving infringement of making,buying or selling a patent-protected materialor composition. Thus, sometimes acompany or institution will decide to filea patent application, seeking rights overa method where such a filing will simplyrepresent an attempt to preclude a competitorfrom preventing the company fromcarrying out a method, without concernsof infringement. Such a method or processpatent would likely never be enforcedexcept in blatant infringement and is onlysought to provide insurance for the filingorganization to lower the risk that theorganization will be sued by someone else.The distinction between a patent that isfiled defensively and one that is filed to preventsomeone from practising the claimedinvention can be very subtle. A discussionof patenting strategies including defensivepatents can be found at www.271patent.blogspot.com/2006/09/valuing-patentsand-patent-paradox-why.html.This is anarea of patent law that is always in flux andenforcement can be very complicated andexpensive.CopyrightsThese rights are awarded for creativeinnovations that are “fixed” in a printed,video, audiotape or other recorded form.Copyrights only cover the form of thefixation, and not the ideas or concepts

414Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishTable 2Examples of copyrighted software relevant to MASTechnique Use Selected software examples Licensing conditions CommentsAnalysis of QTLs For use primarily in analysinganimal pedigree associationsLokiwww.stat.washington.edu/thompson/Genepi/Loki.shtmlList of open source or freeware www.stat.wisc.edu/~yandell/qtl/software/Analysis of fragment patterns For use with ABI electrophoresisequipmentVery liberal, freeware-type oflicencewww.stat.washington.edu/thompson/Genepi/license.shtmlTo be downloaded only if licence isaccepted by userOpen source or as freeware Source code provided.Genotyper® Usually licensed with ABIequipment purchase. (AppleraCorporation). Additional individualpersonal copies cost ~ US$1 500.Source code is not provided;explicit prohibition in licenceStand-alone copy costs ~ US$5 000.Creation of binary table offragment patternsFor use with Genotyper® PeakMatcherAnalysis of fragment patterns For use with electrophoresis withfluorescently labelled markersGenotyping software forlinkage mapping applicationsFor use with ABI electrophoresisequipmenthttp://crop.scijournals.org/cgi/content/full/42/4/1361Genographerhttp://hordeum.msu.montana.edu/genographer/Software manual is also licensedwith softwareLicensed under GNU-GPL v 2 Source code is providedLicensed under GNU-GPL v 2 Source code is providedGeneMapper® Licensed by ABI (Applera corp.)with equipment.Source code is not provided;explicit prohibition in licenceSimulation of biophysicalprocesses in farming systemsManual is licensed with software.Manual has own independentcopyrightPredictive software ApSim See, www.apsru.gov.au/apsru/Products/APSIM/Access%20and%20Pricing%20Policy.pdfReduced licensing fee (on a caseby-casebasis for NARS)Simulation platform forquantitative analysis ofgenetic modelsPredictive software QuGene Original Reference:http://bioinformatics.oxfordjournals.org/cgi/reprint/14/7/632.pdfAlso an annual licence feeNow only available under licencefrom University of Queensland/CSIROReduced licensing fee (on a caseby-casebasis for NARS)

Chapter 20 – Impacts of intellectual property rights on marker-assisted selection 415associated with the innovation (as is thecase with patents). Although articleswritten about MAS, drawings of breedingschemes and the like would be productsfor which copyrights are awarded, it wouldbe quite rare for someone to be concernedabout infringing copyright in carrying outMAS. However, most MAS as currentlypractised, especially at high-throughputlevels, involves the use of computer softwareto analyse the often complex data thatresult from marker detection. While softwareapplications can be patented in a fewcountries, most jurisdictions only allowsoftware to be covered by copyrights. (Inmany jurisdictions, there is ongoing discussionregarding whether software code is anappropriate matter to be covered by copyrights.While in Europe, the EC Directiveon the Protection of Computer Programs(91/250EEC) has clearly established that inthe EU, computer programs are protectedon the same basis as literary works, othercountries have a more checkered history[Starkoff, 2001].) Such copyrights are usedas the basis for “Open Source” licensing ofsoftware. Most software used in conjunctionwith MAS must be licensed before itcan be utilized in MAS breeding schemesor analysis.The ethical aspects of copyright shouldalso be understood. For example, breedersneed to be respectful and careful when givingtalks or other presentations to ensure thatthe material they use is original, or that theowner of the copyright has given permissionfor its use. Just because there is no “©”sign on an article, drawing, slide, picture,etc. does not mean that the material has notbeen copyrighted. Copyright is attached toalmost any fixation with immediate effect.There is no need for an author or creator(or employer of the creator), to apply forcopyright in most countries because ofthe conditions set forward in the BerneConvention for the Protection of Literaryand Artistic Works (1886), which requiresits signatories to protect the copyright onworks of authors from other signatorycountries in the same way it protects thecopyright of its own nationals. A mainprinciple of the Berne Convention, andincorporated into the WTO’s Agreementon Trade Related Aspects of IntellectualProperty Rights (TRIPs), is the generalprinciple of national treatment, “whichrequires each member state to accordto nationals of other member states thesame level of copyright protection providedto its own citizens” (www.wipo.int/treaties/en/ip/berne/summary_berne.html). There are exceptions, e.g. publicationsthat originate from the United StatesFederal Government cannot be covered bycopyright, although sometimes copyrightowners will register a copyrighted articlewith the government to take advantage ofgovernmental assistance in infringementcases. Table 2 provides some examples ofcopyrighted materials that have relevanceto the practice of MAS.TrademarksThese are registered marks given toan applicant as a result of a trademarkapplication being made with a fee payment,and such an application withstanding asearch by a trademark examiner for similarmarks and use of marks (along with anopportunity for opposition to the awardingof the exclusive use by anyone in the public,based on use of the mark by someone elseprior to the application to the trademarkoffice). Trademark rights are different frompatents, plant variety rights and copyrights,in that they are renewable, and thus, ifnational procedural rules are followedcorrectly, can likely last indefinitely. As

416Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishTable 3Examples of trademarks relevant to MASMark Holder Product Use CommentsAFLP® KeyGene Associated with theAFLP process/methodand reagentsDArT® CAMBIA Diversity arraytechnologyABI® Applera Corp. Instruments such ascapillary electrophoresisCreation of polymorphicmarkers based ondifference in DNAsequenceSelection of markersbased on variation fromreference panelsVarious electrophoresisequipment, sequencers,etc.Sybr® Invitrogen Fluorescent dyes Visualization of DNAfragmentsGeneChip® AffymetrixMicroarray on glasssubstrateMicroarrays can be usedfor detection of nucleicacid sequences – DNA orRNAWidely used system; developingcountry institutions oftennegotiate a low/no cost-licenceon a case-by-case basisProprietary technology, oftenlicensed under a BiOS licenceWidely used, associated withmany patented technologiesWidely used, patent on originaldye in this series has expired inmany jurisdictionsWidely used methodology.Affymetrix one of the leaders inthis fieldmentioned earlier, “AFLP” is one exampleof a trademark. This means that in practice,when this method is referred to, the “®”symbol should accompany the term, i.e. thecorrect use of the term would be AFLP ® .Another relevant example would be theCertified Angus Beef ® protected by federaltrademark law in the United States. Inaddition, the names of new markers, newvarieties or types of crops, livestock, etc.would need to be checked by a professionaltrademark searcher if a breeder wished tobe sure that no trademark infringementmight occur by such naming. This is notprecluding the fact that PBRs legislationrequires the breeder to give its candidatevariety a denomination that cannot beregistered as a trademark, as it remainsthe generic denomination of the variety.Table 3 contains examples of trademarksthat are often used as “brand” names,associated with products/processes usedin MAS technologies. Commercial MASpractitioners need to be aware that use ofa trademarked name in conjunction witha product requires the permission of thetrademark holder.Trade secrets and confidentialinformationThese are not registered and, although consideredto be non-statutory IPRs, they areprotected by trade secret law in most countries.Crop breeders have used this approachfor many years to protect the parent linesand information used to produce hybridseeds for sale, and similar approaches areadopted in the poultry and pig industries.This type of IP is defined as commerciallyuseful information that can be saidto have the qualities of being any method,technique, process, formula, programme,design or other information that may beused in the course of production, sales oroperations. It must also meet requirementssuch as not being known to persons generallyinvolved in information of this type;having an actual or potential economicvalue due to its secretive and useful nature;and the owner has taken reasonable measuresto maintain its secrecy. Infringementor non-authorized disclosure/use or misappropriationof a trade secret can result incriminal penalties. These rights might be ofconcern to scientists and breeders who are

Chapter 20 – Impacts of intellectual property rights on marker-assisted selection 417working under conditions that require theuse of confidentiality agreements or nondisclosureagreements (NDAs). Examplesinclude MAS work being carried out byan employee of a company that requiresemployees to sign confidentiality agreements,or MAS carried out as part of jointwork where breeders have been required tosign confidentiality agreements.This is a very common type of protectionused by commercial breedingcompanies involved in the developmentand use of markers and software in all sectorsof agriculture. If a company becomesconcerned that a trade secret risks beingexposed, it may file a defensive patentapplication to ensure that a competitor willnot obtain rights that would preclude useof its own trade secret. Obviously, whena patent application is filed on an inventionthat includes confidential information,the information will no longer be a tradesecret. The applicant presumably wouldonly resort to such a move if the possibilityof “independent invention” were high, andthus the risk of disclosure in a patentapplication balances the risk of having thecompetition “know” of your trade secret.This will happen because of the way inwhich patent examiners normally decideif an invention is “new”. Often such decisionsare based upon the national IP law’sdefinition of “new”, as in the United Stateswhere there is a grace period of one year tofile a patent application after an invention ismade public and also where only use withinthe United States is considered to renderan invention “not” new. A patent examinercannot know that an invention has beenused or described prior to the filing of apatent application if the invention is kept asconfidential information. Therefore patentrights could be awarded to someone whoactually copies a trade secret and companiesmust then consider filing for a patent or runthe risk that a secret will be the subject of acompetitor’s patent.Why would a company not simply filea patent application for each marker thatit identifies? There are several strategicreasons. It is expensive to file for patentprotection and, also, the applicant must disclosethe invention and all of the specifics ofthe invention to satisfy the written descriptionrequirement of enablement. For amarker, this means that the applicant wouldneed to disclose its nucleic acid sequenceif it is known and, by wanting the rightsover the use of the marker in MAS, alsothe trait(s) that is(are) associated with thepresence (or absence) of detection of themarker, etc.Obviously, it is impossible to list specifictrade secrets that exist in MAS technology,although one indication of the existence ofthese can be a reference to a “personal communication”as, for example, in the case ofthe “15PICmarq” marker listed in Table 1of the paper by Dekkers (2004). However,there are examples of information thatis of the opposite nature, i.e. informationthat is publicly available and that canbe used without permission because it isin the public domain such as informationpublished by the United States FederalGovernment, or because no attempts aremade to enforce rights. The companywww.resgen.com, for example, sells kitscomprising simple sequence repeat (SSR)primers mainly for use as MAS markersfor many different species and based onsequences that have been published. Thesemay therefore be covered by copyright, butthese rights are not enforced.Contractual arrangementsAn additional, “non-statutory” systemof rights (Ricketson, 1984 as referenced

418Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishin Drahos, 2005), such as rights/requirementscovered by conditions associatedwith a contract is often described as an IPR,although technically these types of rightsor conditions are not the subject of IP lawin most countries, but rather are a part oflegal codes that deal with private rights.These requirements might be of concernto breeders working under conditions thatrequire the use of contracts such as materialtransfer agreements (MTAs). Conditionsthat result from entering into agreementsor contracts could carry a minimum levelof awareness of the duties or responsibilitiesincurred by one agreeing to theterms. Other “non-statutory” rights couldinclude contractual/legal terms, such asthose included in a licence or a “TechnologyUse Agreement” (TUA). Enforcement andpractice associated with contract law varyin all jurisdictions and can even vary at thelocal level. O’Connor (2006) has recentlypointed out the degree to which MTAsare used to confer a licence to both patentrights and biological materials themselves.He refers to this arrangement as a “leaselicence”model wherein the IPRs and thephysical property rights are “woven”together. Again, if the documents are readcarefully, these conditions will not takeanyone by surprise, in that they are a partof a contract or licence or other permissiongranted by an owner or provider of material.However, sometimes this permissionmay be agreed to in a manner that does notmake a strong impression on a recipient.For example, the so-called “shrink-wrap”licence that accompanies software, or the“click-wrap” licence that covers softwareor other material downloaded from theInternet, may be too subtle for most peopleto be really aware that they have agreed toa licence. In agriculture, “bag-tag” or “seedwrap” licences exist that have the samesort of connotation (Kershen, 2004). Manycourts have looked at the enforcementof these licensing/contract issues, withslightly varying results. The web site www.lex2k.org/shrinkwrap/shrinkwraprev.htmldescribes individual cases and discussesthese cases with regard to enforceability of“shrink-wrap” contracts in different jurisdictionsand conditions.Examples of IPR practicesassociated with the use ofMAS and recommendations forscientists and breedersThe type of formal IPRs most likely to causea problem with the utilization of MAS arepatent rights. Some examples of patents inthis area are given in Table 1. Patents/patentapplications are also listed in the paperby Concibido, Diers and Arelli (2004).Also, as mentioned in the preceding section,contractual arrangements/obligations mayinterfere with unfettered use of productsand processes associated with MAS.Patent rights have been awarded formost of the materials and methods thatare involved in practising MAS within allfields of agricultural production. A carefulresearcher will choose methods and markersequences that have been published andthen carry out at least a cursory searchof patent databases such as the EuropeanPatent Database (http://ep.espacenet.com) to make a first pass for determiningthe likelihood that the method and/orsequence(s) of choice are not covered bypatent rights in the jurisdiction where theywork. Depending upon the level of riskthat one is willing to assume, for work thatcould result in a commercial product, moreinvestigation is likely needed and perhapsthe services of a patent information specialist(see www.piug.org/) or an IP lawyerwill be required.

Chapter 20 – Impacts of intellectual property rights on marker-assisted selection 419Most patents will be of concern primarilyto those in developed countries,particularly the United States where manyprivate companies have their base. Forexample, taking the company Pioneer, 209US patents assigned to Pioneer are identifiedwhen the US Patent Database issearched for the terms “breeding” in thepatent and “marker” in the claims. Thisis reduced to eight when the additionalterm “assisted” is searched in the claims ofthese 209. At the time of writing, Pioneerhad 46 published US patent applicationscovering the “breeding”+“marker” category;reduced to one with the addition of“assisted”. Monsanto, Bayer and Syngentahave utilized MAS practices for a numberof years and accumulated patent portfoliosand very likely many trade secrets inperfecting MAS techniques for their particularuses (Cahill and Schmidt, 2004).Monsanto announced in February 2007that it would begin sharing its markersfor soybean cyst nematode (SCN) resistancewith academic and public institutionresearchers worldwide. According to theannouncement, “Academic researchers andpublic institutions who request access willbe given a royalty-free licence for usingthe rhg1 marker under a patent that wasgranted to Monsanto in December 2006(US Patent no. 7 154 021)”. It is of interestto note that the company, Genome andAgricultural Biotechnology, LLC, with fiveissued US patents and five US patent applicationscovering SCN inventions, has beensued for patent infringement in the use ofSCN markers in conjunction with MAS(Genome and Agricultural Biotechnologyhad sought patent protection in order toestablish “freedom-to-operate” testingservices for material supplied by breederswho lack the facilities to perform MAStechniques for assessing the presence of particulardisease-resistance alleles [www.siuc.edu/~psas/faculty/pubs/lightfoot_achv.htm]). As this situation indicates, personswishing to establish their rights to usemarkers, by filing patent applications andeven obtaining patent rights, need to understandthat one cannot presume that anissued patent means that one then can practisethe inventions, described in the claims,without concern that one may also beengaging in infringement of another patentor set of claims that have been allowed inother patents.As of February 2007, a cursory searchof the US Patent Database as an indicatorof overall patenting activity related to MASand plants revealed 372 issued patents and112 published US patent applications. Ofthese 112 US patent applications, 79 wereassociated with plant breeding and 33 withanimal MAS.These numbers do not include mostof the patents covering equipment, PCRand PCR-related technologies like AFLP®,such as US Patent no. 6 045994 that maybe especially useful for generating markers.Also, analysis of the data indicates anincrease in the number of applicationssubmitted over the four years up to 2005,but most of these applications (58 percent)are for IPRs over specific plant varietiesand sets of markers that allow identificationof the germplasm variety. In recent yearsmany patents have been granted thatcover genes and markers associated witheconomically important traits in livestockspecies (Rothschild, Kim and Anderson,2006; Barendse and Reverter-Gomez,2007).Potential commercialization of suchinventions was predicted by Rothschild,Plastow and Newman (2004), as well as theassociated development of inventions formethods covering breeding management

420Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishBox 2Representative claims that illustrate the breadth of patent claims oversequence informationUS 6 235 972 , “Maize Rad23 genes and uses thereof” issued 22 May 2001What is claimed is:1. An isolated RAD23 polynucleotide comprising a member selected from the groupconsisting of:(a) a polynucleotide having at least 85 percent sequence identity to the polynucleotide ofSEQ ID NO: 1; wherein the percent sequence identity is based on the entire regioncoding for SEQ ID NO: 2 and is calculated by the GAP algorithm under defaultparameters;(b) a polynucleotide encoding the polypeptide of SEQ ID NO: 2;(c) a polynucleotide encoding the polypeptide of SEO ID NO: 4;(d) a polynucleotide amplified from a Zea mays nucleic acid library using primers whichselectively hybridize, under stringent hybridization conditions, to loci within thepolynucleotide of SEQ ID NO: 1;(e) a polynucleotide which selectively hybridizes, under stringent hybridization conditionsand a wash in 0.1.times.SSC at 60 degree C., to the polynucleotide of SEQ ID NO: 1;(f) the polynucleotide of SEQ ID NO: 1;(g) the polynucleotide of SEO ID NO: 3;(h) a polynucleotide which is complementary to a polynucleotide of (a), (b), (d), (e), or (f);(i) a polynucleotide which is complementary to a polynucleotide of (c) or (g); and(j) a polynucleotide comprising at least 75 contiguous nucleotides from a polynucleotideof (a), (b), (d), (e), (f), or (h); wherein the polynucleotides of parts (a), (d)-(e), (h)-(j)each encode monocot Rad23 polypeptides.US 6 815 578 “Polynucleotide encoding MRE11 binding polypeptide and uses thereof”issued 9 November 2004Claim 9. An isolated polynucleotide comprising of polynucleotide selected from the groupconsisting of:(a) a polynucleotide encoding a polypeptide having at least 95 percent sequence identity overits entire length to SEQ ID NO: 2; as determined by the GAP program under defaultparameters, wherein the encoded polypeptide binds to a MRE11 polypeptide; and(b) a polynucleotide which is fully complementary to the polynucleotide of (a).and breeding–related computer applications(Schaeffer, 2002).Previous search of patent literatureAs mentioned earlier, and irrespective of theagricultural sector in which they are operating,breeders and scientists should adopta habit of checking online patent databasessuch as the database of the EuropeanPatent Office and the US Patent Database(www.uspto.gov) for patents and patentapplications that may cover informationand/or innovations relevant to their area ofbreeding and research.It can be quite difficult to search forsequences and combinations of SSRs that

Chapter 20 – Impacts of intellectual property rights on marker-assisted selection 421might be covered in patents and patentapplications. It is beyond the scope ofthe non-professional patent searcherto state definitively whether or not aparticular sequence is covered by patentrights. Searching patents for specific DNAsequence coverage is not quite as easy asit may seem because of the peculiarity ofthe language used in drafting patent claims.A few example claims taken from twoUS Patents, numbers 6 235 972 and 6 815 578are reproduced in Box 2 to illustrate thecomplexity of this type of claim language.However, there are companies, such asGene-IT, that have developed software tosearch for all possible matches that mightoccur in any patent (available in electronicform), and where unlicensed use wouldbe considered an infringement. A goodpatent drafter will attempt to cover as muchground as possible when writing a patentclaim as the broader the claim, the larger itstechnical spread over the landscape of thatparticular area of science/technology. Thisresults in claims to a sequence and its usesbeing written so that the inventor claimsthe sequence and any sequences that areclosely similar. Just how broadly a claimis written is a matter of how much thepatent drafter/prosecutor can get a patentexaminer to accept. Without the assistanceof sophisticated computer software, it canbe difficult to determine whether the use ofa particular genetic sequence would infringeexisting patents. Fortunately, however,biotechnology patents are now examined bybiologists and molecular geneticists, insteadof, as in the “early days”, by chemists.Copyright aspectsOthers have thought that copyrights wouldbe of little concern to the breeder or scientistinterested in using MAS, in that copyrightinfringement might only occur if a materialsuch as text, a design, photograph or videowas copied and re-used without permission,such as in a publication or video that was tobe distributed widely or sold. However,most results of marker testing need to beanalysed by a computer program for thebreeder to obtain maximum value fromsuch testing. Most software is covered by(at least) copyrights and therefore mustbe licensed from the rights holder. Evensoftware that is distributed under an “OpenSource” type of licence is indeed licensed,and the conditions of the licence must beadhered to when the product is used and/or improved.In addition, care should be taken bypersons creating training materials that willbe distributed widely or sold as part ofa workshop to either refrain from usingmaterials written and created by others orto obtain permission before use, especiallyif such use might be part of a coursewhere participants pay for instructionor must buy the training materials, orwhere materials might be distributed in anelectronic format.Trademarks aspectsIn general, the same is true for trademarksas for copyright. A minor point wouldbe to remind authors that terms such asAFLP ® and “Breeding by Design TM ”,both trademarks of Keygene, Inc., shouldcarry the “®” or “ TM ” designation. Inthis regard, breeders would be primarilyconcerned with the correct use of their owntrademarks, both by themselves and others.When naming varieties, etc., care shouldbe taken to ensure that the trademarkof another entity is not being infringed.Those responsible for creating namesshould therefore check public trademarkdatabases such as the UK TrademarksDatabase (www.patent.gov.uk/tm/dbase/),

422Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishand the services of a professional trademarksearcher or attorney should be soughtbefore proceeding with the registration ofa “new” trademark.Plant breeders’ rights aspectsBreeders using basic MAS protocolswith non-proprietary breeding materials(e.g. germplasm that does not qualify as an“Essentially Derived Variety” [Wendt andIzquierdo, 2001]) generally do not need tobe concerned with using materials coveredby PBRs for breeding purposes.Contractual aspectsIt is very important that licences, contractsand agreements are monitored for restrictionsas these often contain provisionsdealing with IPRs that last until a contractexpires or is renegotiated. Permission to useequipment and associated reagents is normallygranted as a licence granted as a partof the purchase price. However, this typeof licence may often contain limitations onthe use of equipment, reagents and kits fornon-research applications. As an example,and as stated in its legal information Webpage, Applied Biosystems has an exclusivelicence with Roche/Hoffman-La Roche forits PCR patents: “Applied Biosystems isthe exclusive licencee of Roche MolecularSystems, Inc., and F. Hoffmann-La Roche,owner of the basic PCR process and reagentpatents, for the field of research anddevelopment, and for applied fields suchas quality assurance and control, environmentaltesting, food testing, agriculturaltesting (including plant disease diagnostics),forensics and identity testing in humans(other than parentage testing), and animalidentity and breeding applications.” Thismeans that when a researcher buys (orhas legal access to) and uses an AppliedBiosystems machine, the rights to use thismachine for certain specified purposes(rarely commercial) are included in thepurchase agreement. Note, however, thatthe use of kits or other products of AppliedBiosystems that involve any processes orreagents licensed from Roche/Hoffman-LaRoche to carry out MAS is not specificallymentioned as a “field of use” in the terms ofthis licence. While it could be assumed thatuse for MAS is possible under the AppliedBiosystems licence, if it was considerednecessary to have the lowest probable levelof risk associated with the use of equipment/reagentsfor MAS, then legal advicein the jurisdiction of the user should besought.An equipment or reagent licence couldalso contain provisions for what are called“reach-through” rights. These arise whenimprovements are made to an existing technology.When such innovations come aboutthrough use of the existing technology therights to them may have to go back to theowner of the original existing technology.Such a transfer of sharing of the rights iscalled “reach-through rights”. Some argue,for example, that the requirement in someOpen-Source licences for improvementsgoing back to the original creator of thesoftware for distribution are a form of“reach-through”.Agreements to purchase and “packageinsert” licences should therefore be routinelychecked to ensure that these sorts oflicence are avoided.MTAs can also cause problems,depending upon the conditions that areset down in such agreements. Laboratorypersonnel need to make sure that MTAsare only signed by persons authorized todo so and that efforts are made to checkMTA language for provisions that restrictor interfere with the intended use of thegermplasm that is produced using MTA-

Chapter 20 – Impacts of intellectual property rights on marker-assisted selection 423associated materials. A practical explanationof MTAs is available in COGR (2003).Breeders and scientists need to keep afile and/or database of all licences, packageinserts, purchase agreements and MTAs aspart of their routine record keeping. Theyalso need to learn to reject documentsthat contain provisions that indicate anassertion of rights or include a restriction,to negotiate for terms that they require,or source replacement brands/materials.Contracts can be enforced long after patentrights expire.Of all the types of IPRs/proprietaryrestrictions that could affect scientists andbreeders in developing countries, licencesand agreements have the most potentialto impede the use of MAS technologies,unless a sophisticated, high-throughputlaboratory is sought. MAS has considerablepotential and relevance to developingcountry breeding systems for capturingdesirable characteristics from widelydisparate germplasm. IPRs should not holdthis back.ReferencesAsiaLaw. 2004. Process patent litigation and the collection of evidence in China. (available at www.asialaw.com/default.asp?Page=20&PUB=68&ISS0=10970&SID=433837).Balch, O. 2006. Seeds of dispute: it’s Argentina v Monsanto in the battle for control over GM soytechnology. (available at www.guardian.co.uk/gmdebate/Story/0,,1715331,00.html).Barendse, W. & Reverter-Gomez, A. 2007. A method for assessing traits selected from longissimusdorsi peak force, intramuscular fat, retail beef yield. WO2007012119 (available at www.wipo.int/pctdb/en/wo.jsp?LANGUAGE=ENG&KEY=07/012119&ELEMENT_SET=F).Binnenbaum, E., Nottenburg, C., Pardey, P.G., Wright, B.D. & Zambrano, P. 2003. South-Northtrade, intellectual property jurisdictions, and freedom to operate in agricultural research on staplecrops. Economic Development & Cultural Change, 51(2): 309–335.Cahill, D.J. & Schmidt, D.H. 2004. Use of marker-assisted selection in a product developmentbreeding program. Proc. 4 th Internat. Crop Sci. Congr. (available at www.cropscience.org.au.).CBO (Congressional Budget Office). 1998. How increased competition from generic drugs hasaffected prices and returns in the pharmaceutical industry. Washington, DC, Congress of theUnited States (available at www.cbo.gov/showdoc.cfm?index=655&sequence=0).COGR (Council on Governmental Relations). 2003. Materials transfer in academia. (www.cogr.edu/docs/MTA_final.pdf).Concibido, V.C., Diers, B.W. & Arelli, P.R. 2004. Review & interpretations: a decade of QTL mappingfor cyst nematode resistance in soybean. Crop Sci. 44: 1121–1131.Consuegra, S. & Johnston, I.A. 2006. Polymorphism of the lysyl oxidase gene in relation to musclecollagen cross-link concentration in Atlantic salmon. Aquaculture Res. 37(16): 1699–1702.Dekkers, J.C.M. 2004. Commercial application of marker- and gene-assisted selectionin livestock: strategies and lessons. J. Anim. Sci. 82: E313–328 (available at http://jas.fass.org/cgi/reprint/82/13_suppl/E313).Doyle, J.J. & Doyle, J.L. 1987. A rapid DNA isolation procedure for small quantities of fresh leaftissue. Phytochem. Bull. 19: 11–15.Drahos, P.A. 2005. Intellectual property rights in the knowledge economy. In D. Rooney, G.E.Hearns & A. Ninan, eds. Handbook on the knowledge economy, pp. 139–151. Northhampton,MA, USA, Edward Elgar.

424Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishEdwards, K., Johnstone, C. & Thompson, C. 1991. A simple and rapid method for the preparationof plant genomic DNA for PCR analysis. Nucleic Acids Res. 19(6): 1349.Epp, T.D. 2004. Four-wheeling through the soybean fields of intellectual property law: A practitioner’sperspective. Washburn Law J. 43: 669–679.Fromer, J.C. 2007. Invigorating the disclosure function of the patent system. NYU Law School,Public Law Research Paper Series, 2 March 2007 (available at http://ssrn.com/abstract=967560) .Fulton, M. & Giannakas, K. 2001. Agricultural biotechnology and industry structure. AgBioForum,4(2): 137–151.Gallagher, G.J. 2002. Recent development: the Federal circuit and claim construction: resolving theconflict between the claims and the written description. North Carolina J. Law & Technol. 4(1):121–142.Hore, E. 2004. Patently absurd: evergreening of pharmaceutical patent protection under the PatentedMedicines (Notice of Compliance) Regulations of Canada’s Patent Act. Canadian GenericPharmaceut. Assocn. (available at www.canadiangenerics.ca/en/issues/patently_absurd_04.pdf).Kattan, J. 2002. Evaluating patent infringement and validity in antitrust analysis (available atwww.ftc.gov/opp/intellect/020514kattan.pdf#search=%22%22merger%22%2B%22infringement%22%22).Kellison, R., McCord, S. & Gartland, K.M.A. 2004. Forest biotechnology in Latin America. Proc.Forestry Biotech. Workshop, 3–5 March 2004, Concepcíon, Chile (available at www.forestbiotech.org/pdffiles/ChlePDFfinal31Jan2005.pdf).Kershen, D.L. 2004. Of straying crops and patent rights. Washburn Law J. 43: 575–610.Lee, S., A’Hara, S. & Cottrell, J. 2005. The use of DNA technology to advance the Sitka sprucebreeding programme. Forestry Res. Report (available at www.forestresearch.gov.uk/pdf/FR_report_2005-6_dnatech.pdf).Murchie, B.J., Renaud, A.B., Horne, L.E.T. & Hussey, D.T. 2006. Canada’s balancing act. (availableat www.lexpert.ca/500/lb.php?id=120).O’Connor, S.M. 2006. The use of MTAs to control commercialization of stem cell diagnostics andtherapeutics. Berkeley Technology Law J. (available at : http://ssrn.com/abstract=921170).Röder, M.S., Korzun, V., Wendehake, K., Plaschke, J., Tixier, M.H., Leroy, P. & Ganal, M.W. 1998.A microsatellite map of wheat. Genetics. 149(4): 2007–2023.Rothschild, M.F., Plastow, G.S. & Newman, S. 2004. Patenting in animal breeding and genetics. InA. Rosati, A. Tewolde & C. Mosconi, eds. WAAP Book of the Year 2003: a review on developmentsand research in livestock systems, pp. 269–278. Wageningen, Netherlands, Academic Publishers(also available at www.psas-web.net/documents/Info/3.5%20Rothschild.pdf).Rothschild, M.F., Kim, K.S. & Anderson, L.L. 2006. Ghrelin alleles and use of the same for geneticallytyping animals. US Patent Number 7,074,562.Saghai-Maroof, M.A., Soliman, K.M., Jorgensen, R.A. & Allard, R.W. 1984. Ribosomal DNAspacer-length polymorphisms in barley: Mendelian inheritance, chromosomal location, and populationdynamics. Proc. Nat. Acad. Sci. 81: 8014–8018.Schaeffer, L.R. 2002. Dairy cattle test day models: a case study. In M. Rothschild & S. Newman, eds.,Intellectual property rights in animal breeding and genetics. Wallingford, UK, CABI.Smyth, S. 2006. Implication and potential impacts from the expiry of patents on herbicide tolerantcanola varieties. Saskatchewan Canola Development Commission (available at www.saskcanola.com/pdfs/scdc-patent-report.pdf).

Chapter 20 – Impacts of intellectual property rights on marker-assisted selection 425Sreedharan, S.K. 2007. Coming of age – the Indian product patent regime. Intellectual AssetManagement Magazine. Special report on “From Innovation to Commercialisation 2007” (availableat www.iam-magazine.com/issues/article.ashx?g=e6c6910d-4788-4006-aed5-b9ad1aa69937.).Starkoff, D. 2001. Copyright: Law and Practice in a Digital Age. Honour’s Thesis, The Universityof Queensland (available at www.itee.uq.edu.au/~cristina/students/dbs/davidStarkoffHonoursThesis01.pdf).Wendt, J. & Izquierdo, J. 2001. Biotechnology and development: a balance between IPRprotection and benefit-sharing. Electronic J. Biotech. (available at http://ejb.ucv.cl/content/vol4/issue3/issues/01/index.html).

Chapter 21Marker-assisted selection asa potential tool for geneticimprovement in developingcountries: debating the issuesJonathan Robinson and John Ruane

428Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishSummaryMarker-assisted selection (MAS) is a complementary technology, for use in conjunctionwith more established conventional methods of genetic selection, for plant and animalimprovement. It has generated a good deal of expectations, many of which have yetto be realized. Although documentation is limited, the current impact of MAS onproducts delivered to farmers seems small. While the future possibilities and potentialimpacts of MAS are considerable, there are also obstacles to its use, particularly indeveloping countries. Principal among these are issues relating to current high costs ofthe technology and its appropriateness, given that publicly funded agricultural research inmany developing countries is suboptimal and development priorities do not necessarilyinclude genetic improvement programmes. Other potential obstacles to the uptake ofMAS in developing countries include limited infrastructure, the absence of conventionalselection and breeding programmes, poor private sector involvement and lack of researchon specific crops of importance in developing countries. Intellectual property rights mayalso be an important constraint to development and uptake of MAS in the developingworld. It is hoped that through partnerships between developing and developed countryinstitutions and individuals, including public–private sector collaboration, MAS costs canbe reduced, resources pooled and shared and capacity developed. With the assistance of theConsultative Group on International Agricultural Research (CGIAR) and internationalorganizations such as FAO, developing countries can benefit more from MAS. These weresome of the outcomes of a moderated e-mail conference, entitled “Molecular Marker-Assisted Selection as a Potential Tool for Genetic Improvement of Crops, Forest Trees,Livestock and Fish in Developing Countries”, that FAO hosted at the end of 2003. Duringthe four-week conference, 627 people subscribed and 85 messages were posted, about 60percent coming from people living in developing countries. Most messages (88 percent)came from people working in research centres (national or international) or universities.The remainder came from people working as independent consultants or from farmerorganizations, government agencies, non-governmental organizations (NGOs) or UnitedNations (UN) organizations.

Chapter 21 – Marker-assisted selection as a potential tool for genetic improvement: debating the issues 429IntroductionFAO, an intergovernmental organizationwith 189 Member Nations and oneMember Organization, was foundedin 1945 with a mandate to raise levelsof nutrition and standards of living, toimprove agricultural productivity, and tobetter the condition of rural populations.One of FAO’s primary roles is to serve asa knowledge network sharing informationon agriculture. It uses the expertise ofits staff, agronomists, foresters, fisheriesand livestock specialists, nutritionists,social scientists, economists, statisticiansand other professionals, to collect, analyseand disseminate data that aid development.The information is made available using anumber of different strategies, e.g. providingdocuments on the FAO Web site that canbe freely downloaded, publishing hundredsof newsletters, reports and books, andhosting dozens of electronic fora. In thiscontext, FAO has also been playing anactive part in disseminating informationand promoting information exchangeregarding biotechnology. For example, in2000 it established the FAO BiotechnologyForum (www.fao.org/biotech/forum.asp),with the aim of providing quality balancedinformation on agricultural biotechnologyin developing countries and making aneutral platform available for people toexchange views and experiences on thissubject.At the end of 2003, the FAOBiotechnology Forum hosted a four-weeklong e-mail conference entitled “MolecularMarker-Assisted Selection as a PotentialTool for Genetic Improvement of Crops,Forest Trees, Livestock and Fish inDeveloping Countries”. The conferencewas open to everyone and 627 people subscribed.Each of them received a ten-pagedocument providing easily understandablebackground information on the conferencetheme, so that those with little knowledgeof the area could understand what the themewas about. The conference was moderated(by John Ruane) and participants wererequested to introduce themselves briefly intheir first posting to the conference and tolimit their messages to 600 words. Duringthe conference, 85 messages were posted,each numbered in chronological order. Ofthe 627 subscribers, 52 (8 percent) submittedat least one message. Messages werereceived from each of the different worldregions, 28 of the 85 messages (33 percent)were posted from Asia, 26 percent fromEurope, 14 percent from Latin America andthe Caribbean, 9 percent each from Africaand Oceania and 8 percent from NorthAmerica. Messages were posted from peopleliving in 26 different countries, the largestnumbers from India (25 percent), followedby Australia (9 percent), United States ofAmerica (8 percent), United Kingdom(7 percent) and Peru (6 percent), with theremainder from Argentina, Austria, Benin,Brazil, Chile, Cyprus, Egypt, Finland,France, Germany, Ireland, Israel, Kenya,Madagascar, Mexico, Netherlands, Nigeria,Philippines, Spain, Syrian Arab Republicand Turkey. Fifty messages (59 percent)were contributed from people in developingcountries and 35 (41 percent) indeveloped countries. The majority of messagescame from people working in researchcentres (52 percent), including ConsultativeGroup on International AgriculturalResearch (CGIAR) centres, and in universities(37 percent). The remainder workedas independent consultants or for farmerorganizations, government agencies, NGOsor UN organizations.This chapter summarizes the main issuesthat were discussed during the conference,based on the messages posted by the

430Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishparticipants. These included some generaltopics regarding MAS (such as its costs, itsactual impact to date on products deliveredto farmers, whether it should be a priorityin developing countries and whetherit was necessary to have an establishedbreeding programme in place before introducingMAS), as well as some MAS-relatedissues that were more technical, such aswhich traits are suitable for MAS andthe importance of tight marker-gene linkages.Other kinds of issues raised includedintellectual property rights, public–privatesector linkages, the differences in capacitybetween developing countries with respectto MAS and the role of the CGIAR andinternational organizations. Throughoutthe chapter, specific references to messagesposted are provided, giving the participant’ssurname and message number. All the individualmessages are available at www.fao.org/biotech/logs/c10logs.htm.During the conference, contributionswere not evenly spread across the fouragricultural sectors of the conference.MAS for crop and livestock geneticimprovement dominated the discussions,with issues relating to forest trees andaquaculture mentioned much less, possiblyindicating differences in uptake of thisrelatively new technology among thefour sectors. Nonetheless, many of theissues and concerns raised were generalin nature and applicable across sectors.These issues included considerations ofcosts and gains, intellectual property rightsand the benefits of partnerships to allowdeveloping countries greater opportunitiesfor developing and using MAS.Murphy (1) began the conference with arequest that MAS be viewed dispassionatelyas a potential tool for crop improvement to bedeployed alongside conventional methods.Sokefun (64) referred to conventionalselection methods as “soft” technologiesand the newer technologies, such as MAS,as “hard” technologies, and suggestedthat the hard would not replace the softtechnologies and that a fusion of both wouldachieve the best results. In contrast to moreupstream technologies (including geneticmodification, mutagenesis and protoplastfusion), which generate additional variationin plant populations, Murphy (1) describedMAS as a “downstream technology” that,like conventional phenotypic selection, canbe used to select the optimal variants in apopulation.The conference discussion was balancedand the topic of the potential of MAS didnot evoke a strong reaction among theparticipants, although many had reservationsabout it. There was consequently littleindication of a substantial dichotomy ofopinion whereby participants could be putinto pro- and anti-MAS camps. This is insharp contrast to many debates that havebeen held about genetic modification. Asstated by Muralidharan (7), MAS differsfrom genetic modification in being morewidely acceptable.There was considerable agreementamong the participants on the perceivedopportunities and constraints associatedwith MAS and the usefulness and applicabilityof the technology in developingcountries. Olori (21) thought that successfulapplication of MAS in a well structuredbreeding programme in any developingcountry would yield the same benefitsas in developed countries. However, assuggested by Montaldo (18) for geneticimprovement in animals, it would benecessary to make case-by-case studies,taking into account not only biologicalissues, but also social, political and economicones, before making recommendations onapplication of MAS.

Chapter 21 – Marker-assisted selection as a potential tool for genetic improvement: debating the issues 431Main themes discussedWhether MAS should be a priority indeveloping countriesThe general opinion was that MAS couldbe usefully applied for genetic improvementof plants and animals in developingcountries, but that it would not necessarilyrepresent a priority. Gianola (6) pointedout that in order for MAS to be taken upin developing countries, because of thescarcity of resources the returns to investmentshould be far superior compared withthose for a developed country, given thesignificant opportunity costs. Africa wasmentioned as facing major constraints toagricultural production, including droughtstress, low soil fertility and pests, whichwere not easily and economically amenableto MAS. Koudandé (68) and Seth (26)stressed the importance of priority-settingin the context of national agricultural economies.Crop diversification and research onunderutilized species were also mentionedas other possible priorities for addressingproblems of the expanding human population(Priyadarshan, 11 and 71). Murphy (1)suggested that tremendous gains could bemade in agricultural development withoutresorting to applications of biotechnology,by addressing issues of management andinfrastructure. For example, in the case ofBrazil, a priority might be improvementsin the road system to facilitate export cropsreaching the ports (Murphy, 1).Costs of MASThe cost associated with MAS was acommon theme during the conferenceand several participants, including Collard(9), considered it to be the most importantissue for developing countries. Itwas pointed out (e.g. De Koning, 13) thatalthough costs associated with MAS canbe high, conventional genetic improvementprogrammes can also be expensive. Gianola(2) called for an economic analysis of MASin comparison with conventional methods,specifically requesting estimates of internalrates of return. He (6) also warned that therewas a risk that some investments in MASmight be wasted given the advances beingmade in post-genomics. For Weller (4),“with respect to the economic questions,MAS is no different from any other technologythat increases rates of genetic gain,but also increases costs”, concluding thatthe investments required for MAS couldbe massive, but so also could the long-termeconomic gains. However, as pointed outby Montaldo (18), the economics of MASis based on the value of the selected traitsand most importantly, each case shouldbe looked at individually. De Koning (13)highlighted the major economic benefitsthat could be gained by breeding livestockfor resistance to trypanosomiasis.Various stages in the MAS developmentand application process were regarded asbeing costly. Labour and DNA extractionwere viewed by Williams (37) as representingthe major costs, but Collard (45) consideredequipment, consumables and infrastructureto be among the most costly items ina MAS programme. Genotyping (Toro,67), marker development (El Ouafi, 77;Wallwork, 59) and patenting (Ganunga,69) were other areas that represented largecosts that might constrain MAS use indeveloping countries. It was suggested thatfarmers in the developing world could notbe expected to pay for MAS (Chávez, 33),while Muralidharan (74) suggested that costsin a country like India would eventually bea lot cheaper than in developed countries.Participants, including Buijs (58),pointed out that technologies becomecheaper as knowledge accumulates andcapacity is built up, citing the example of

432Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishtissue culture. Buijs (22) also felt that thecosts of MAS should be put in perspectivewith those from other related researchareas, pointing out that plant varieties oranimals bred by MAS do not require costlysafety regulations, in contrast to those bredusing genetic modification. Toro (50) andMuralidharan (74) suggested that MASwould become cheaper due to automation/robotics,and Varshney (82) reportedthat microsatellite marker development hasbecome cheaper as a result of bioinformatics.Many participants suggested thatdeveloping countries could make the bestuse of MAS through collaborative ventures(Olori, 21, 65; Acikgoz, 66; Saravanan,73), formation of multidisciplinary teams(Sridhar, 76; William, 70; Muchugi, 49) andwithin national and regional frameworks(Montaldo, 18). Collaboration wouldspread resources and reduce costs.Figures for the costs of genotyping mentionedin the conference ranged from US$4per marker for MAS in pigs (Toro, 79) tounder US$0.2 for durum wheat (El Ouafi,77). Discussion of such exact figures forcosts is at best indicative in the face of continuouschanges in the world economy,particularly in exchange rates and purchasingpower. Suffice to say that as costsare reduced, the value of MAS rises and itpossibly becomes more widely applicable.Putting MAS in contextAlthough MAS has generated a good deal ofexpectations, leading in some cases to overoptimismand in others to disappointmentbecause many of the expectations have notyet been realized, participants in the conferenceaimed to consider MAS rationallyand to put it in the context of the wholeagricultural picture. As Murphy (1) wrote,MAS “should be viewed dispassionately asa potential tool for crop improvement to beusefully deployed alongside conventionalphenotype selection for certain crops andfor certain characters”.Good genetic improvement strategieswere considered by many to be among themost important prerequisites for successfulimplementation of MAS. Montaldo (18)said that, with respect to livestock improvement,MAS would not substitute forchoosing the right breeding objectives andthe starting point of a programme incorporatingMAS should be a sound breedingstrategy founded on traditional selectionmethodology. Wallwork (59) thought thatmany of the criticisms of MAS (e.g. see DeLange, 57) stemmed from poor researchand development strategies and not necessarilyfrom shortcomings in the technology.El Ouafi (77) stated plainly that if a successfulconventional breeding programmecould not be established, MAS would nothelp, and Olori (21) suggested that theabsence of “any real sense of the needfor a genetic improvement programme” indeveloping countries would hinder applicationof MAS. Such practical strategicconsiderations balance the hyperbole andover-optimism that has sometimes beenassociated with MAS. De Lange (57) arguedthat because of its high costs and relativelymoderate results to date, MAS seemed to be“yet another over-hyped gene technology”and questioned, like Ackigoz (66), whetherMAS should be a primary considerationfor developing countries. Bhatia (8) wasamong several participants to comment onthis issue and believed that the hyperboleto some extent reflected fashion and vendorbias, as for all new technologies.MAS in relation to conventional breedingprogrammesThe need for an established breeding programmeto be in place for MAS to be

Chapter 21 – Marker-assisted selection as a potential tool for genetic improvement: debating the issues 433usefully introduced represented one ofthe main points debated in the conference.Many participants (e.g. Montaldo,18) explicitly stated the need for a conventionalprogramme to be operational prior toimplementation of MAS and others inferredit. Notter (25), on the other hand, suggestedthat animal recording need not precedeimplementation of MAS, and he proposedthey could be implemented together.Referring to animal trypanosomosis inAfrica, De Koning (13) commented thatlack of routine recording of production andhealth traits, with limited national molecularresearch facilities, presented a structuralproblem to implementing a breeding programmeusing MAS. De Koning (20) alsosaid that when livestock were mainly keptby smallholders, each with a handful of animals,there would be no routine recording.Makkar (52) too suggested that in the lowinput systems that characterize many developingcountries, phenotype and pedigreeinformation were often not available, andthis would make it difficult to realize thevalue of MAS. Notter (25) proposed, however,that MAS (or related technologies)could act as a lever to promote implementationof animal recording. He also noted that“MAS without recording is unlikely to bevery beneficial for most traits”.For crops, Singh (61) suggested thatMAS should be an integral part of thebreeding strategy, but Acikgoz (66) wascritical of situations where scientistswithout any experience of traditional plantbreeding programmes entered directly intoMAS. Sridhar (76) and El Ouafi (77), whileacknowledging the importance of MAS, bothsuggested that meaningful breeding programmeswere necessary to make progresswith MAS and Dulieu (23) doubted thattraditional selection methods could easilybe replaced by MAS. Priyadarshan (11) alsobelieved that more basic biological knowledgeabout the intricacies of nature wasneeded to improve selection proceduresfor plants and Montaldo (18) pointed outthat knowledge of genetic control of someimportant traits remained incomplete.MAS in aquaculture in developing countrieswas only briefly discussed in theconference, although Priyadarshan (71)argued that aquaculture merited moreemphasis. Martinez (63) suggested that, foraquaculture, application of DNA technologiesand MAS was scarce even in developedcountries because of the lack of integrationbetween quantitative and moleculargenetics, and that the only successful applicationin aquaculture was that described byToro (50), who said that molecular markerscould be used to assist classical geneticimprovement programmes by constructingpedigrees needed for genetic evaluationin trees and fish where pedigree informationwas otherwise lacking. Martinez (63)noted, however, that economic analysis ofthis strategy compared with individuallyidentifying fish using electronic devices wasscarce. Krause (75) gave an example wheremolecular marker information could beused to reduce the costs of a fish breedingprogramme. Normally, electronicallytagged back-up copies of nucleus breedingpopulations of fish are made as an insuranceagainst loss of a deployed population. Thisis an expensive process that can be avoidedby taking tissue samples from sires anddams that are analysed for the presence ofestablished molecular markers if a nucleusstock is destroyed. This allows a nucleusstock to be regenerated relatively easily andcheaply, if and when necessary.While the merits of applying MAS togenetic improvement of trees in developingcountries were appreciated (e.g.Muralidharan, 7), participants suggested

434Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishthere are many problems that detract fromits usefulness. Principal among these isthe poor state of current tree breeding ingeneral, and in developing countries inparticular. Simons (28) listed a numberof problems concerning genetic improvementof tropical trees, including dioecy,undocumented origins and uncertainty ofgenetic control of traits. However, Galvez(10) mentioned that MAS had been usedto assist in selection of coconut parentsfor breeding. Priyadarshan (11) consideredMAS to be helpful for rubber improvement,at least theoretically, and Badr (47)seemed to be looking forward to MASreducing the time needed for evaluationof fruit trees in Egypt, obviating the needfor grafting to see the products of breedingefforts. Forest trees, perhaps more thanother genetic resources used by humans,are at, or still very near, their wild state(Muralidharan, 7), which indicates thattremendous improvement can probablybe made quite rapidly based on selectionamong existing genotypes. Muchugi (49)recognized the potential of MAS for treespecies improvement, seeing it as a techniquebest placed to help select and upgradetropical tree species where the first fruitingmay take as long as twenty years.Technical details of MAS useThere were several contributions to theconference regarding technical aspects ofMAS, and how to use MAS effectively ingenetic improvement programmes. Mota(14) raised the issues of molecular markerslocated far from the target gene, increasingthe probability of recombination takingplace between them, resulting in reducedefficiency of MAS and, secondly, of falsepositive marker-gene associations. Dulieu(23) also emphasized the importance oftight marker-gene linkage to minimizelosses through recombination. Weller (15)acknowledged the importance of both issuesraised by Mota (14) and proposed that thebest solution to the problem of false positivesis to employ the false discovery rate,to get an idea about the expected numberof false positives. De Koning (16) supportedthe use of the false discovery rateand also referred to recent research resultssuggesting there were benefits in MASfrom using a relaxed threshold for QTL(quantitative trait loci) detection. Mota(36) concluded that developing countriesshould only use MAS in their breeding programmeswhen there is complete linkagebetween the marker and the gene of interest,to avoid wasting precious resources. Dulieu(42) commented on this, pointing out theadvantages of using flanking markers (i.e.where markers are located on either side ofthe gene of interest) in MAS.Singh (44) described the usefulness ofMAS in backcrossing programmes, bygrowing large BC1 populations (BC1 isthe first backcross generation), rejecting50–60 percent based on phenotype (conventionalscreening) and analysing the remainderwith MAS. This could be repeated in thesecond backcross population, saving considerabletime and resources. The usefulnessof this approach was confirmed by Dulieu(53), and Sridhar (54) explained how threegenes for rice bacterial blight resistancewere pyramided into adapted germplasmusing MAS in a backcrossing programme.Which traits for MAS?Referring to crop improvement, Murphy(1) noted that not all crops and traits wereamenable to MAS. A Dutch perspectiveon the type of traits amenable to MASto date was provided by De Lange (57),who indicated that single gene controlledtraits had received most attention, but little

Chapter 21 – Marker-assisted selection as a potential tool for genetic improvement: debating the issues 435progress had been made with multiple genetraits. Makkar (52) stated that many MASstudies had adopted a single trait approach,pointing out that with a multitrait breedingobjective, response for one trait often goesat the expense of another. He also suggestedthe utility of MAS when heritabilityfor the trait was low. Singh (41) indicatedthat “breeders are not much thrilled aboutMAS for simply inherited traits, and notmany QTL (especially the productivityrelated ones) with tightly linked markersare available”.Several other participants mentionedtraits that would be amenable to MAS,including Priyadarshan (11) working withrubber trees, Williams (37) who providedthe case of root nematodes and William(70) who mentioned work being done onbarley yellow dwarf virus resistance incereals, rust diseases, nematode resistanceand root health. Rakotonjanahary (78) alsosuggested that MAS be used when conventionalapproaches to selection were difficultor impossible. For example, Reddy (62)proposed MAS be used for traits where it isdifficult to get phenotypic data, like qualitytraits, and William (70) indicated that proteinassays to develop quality protein maizewere expensive compared with markerassays. Slaughter traits in livestock werealso considered to be amenable to MAS asthe desired traits are otherwise difficult tomeasure without killing the animal (Makkar,52). Muchugi (49) suggested the potentialusefulness of MAS in selecting for medicinaltraits and growth rate in tropical trees.Introgression of genes from wild intocultivated germplasm was proposed to bea good use of MAS (Bhagwat, 46). Notter(25) also commented on the opportunitiesmolecular markers provide for screeningpopulations of animals with favourableor unfavourable genotypes, giving as anexample scrapie in sheep. Krause (75) mentionedother genetic examples, such as asperm defect in pigs and the halothane geneimplicated in low pork quality, that couldbe screened out using MAS. Sex-linkedtraits were also mentioned as being suitablefor MAS (Makkar, 52).Galvez (10) suggested that molecularmarkers could be also useful for work withtransgenic crops, for characterizing GMplants and tracking movement of the transgenein the gene pool. William (70) alsomentioned the use of MAS for transferringa desirable transgene, such as a gene fromBacillus thuringiensis, from one cultivar toanother.In addition to discussing traits consideredamenable to MAS, brief mention wasmade of traits not considered amenable toMAS. It was realized that more progresshad been made with single genes that wererelatively easily transferred, but that therewas potential for facilitating QTL transfer,although this was still relatively undeveloped.Traits that are highly influenced bythe environment or production system,including crop yield (Priyadarshan, 11),were not considered easily amenable toMAS. Williams (37) pointed out that amajor problem associated with MAS waslack of polymorphism at the DNA level,which would render a trait not amenable toMAS. Inadequate coverage of the geneticmap with molecular markers was viewedby Dulieu (23) as an obstacle to applyingMAS. He also detailed other conditionsrelating to the nature of the trait that shouldbe considered for MAS to be efficient:single versus multigene, additive versusdominant, expressivity and penetrance.Practical applications of MASSome participants considered the actualimpact of MAS on genetic products deliv-

436Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishered to farmers. Although documentationwas limited, the current impact seemedsmall while the future impact was likely tobe far more substantial.Priyadarshan (11) indicated that biotechnologyresearch had been supportedactively for over 17 years in India, butwas doubtful about the impact on varietiesreleased to farmers. He believed thatresearch on MAS and other biotechnologieshad remained largely in journal articlesand had not significantly boosted conventionalplant breeding efforts on the ground.Kirti (12) lamented that there was no comprehensivedocumentation regarding thesuccessful use of MAS for breeding newcrop varieties or developing breeding material,as this information would be importantfor evaluating the technology. Collard (45),while noting that MAS had been successfulin cereal crops in his country, Australia,said he was not aware of many examples ofMAS-derived cultivars grown in Australiadespite the wealth of publications fromAustralian institutions on the technology.Sridhar (48) suggested that, in India, mostproducts of MAS are still in the hands ofresearch institutions undergoing evaluation.He suggested that MAS products require a“fast track” evaluation system to expeditethe release of promising germplasm.According to Makkar (52), success indemonstrating genetic gain in the laboratorydid not always equate with successunder field conditions. However, some realsuccesses were reported, including transferof important resistance genes into adaptedrice germplasm for Indian farmers (Sridhar,35 and 54), indicating that more successesmight be in the pipeline. Williams (51) saidthat molecular markers had been used for atleast five years in Australia in some wheatand barley improvement programmes andthat “it is likely that in Australia all breedingprogrammes with industry funding andprobably also the private breeding companiesare currently using MAS to someextent”. However, the potential of the newtechnology has to be weighed against thesuccess achieved using traditional methods.Acikgoz (66) pointed out that the Turkishrice cultivar Tokak was still being solddespite having been released in 1937, andquestioned how much impact populationgenetics studies, popular 20–30 years ago,had on cultivar development, let alone theimpact of biotechnology applications.Buijs (58) mentioned tissue culture, onceregarded as a modern, relatively expensivetechnology, which is now relatively inexpensiveand widely used in developingcountries. It will only be known retrospectivelywhether MAS evolves similarlyto become a standard tool of the plant andanimal breeder in developing countries.Intellectual property rights issuesSome participants felt that intellectualproperty rights (IPRs) were an importantconstraint to development and uptakeof MAS in the developing world. Corva(29) raised the issue of the use of licensedgenomic technology by public institutionsin developing countries, mentioning thatmany useful cattle markers were becomingavailable, but which were patented, andthat there was therefore a demand for practicalinformation about IPRs and violationof IPRs. Weller (30) pointed out that patentsare only valid in the country wherethey are granted, that research tends to beexempted from patent restrictions and thatthere can be long delays between filing ofpatent claims and their eventual granting.Saravanan (31) argued strongly for thefreedom of researchers to use patented biotechnologytools. Storlie (32) argued thatfarmers in the developing world should

Chapter 21 – Marker-assisted selection as a potential tool for genetic improvement: debating the issues 437be concerned about being constrained by“corporate patents on particular genes,which may require a company’s authorizationfor possession and use”. William (70)noted that development of useful markersfor MAS was already a significant challengein developing countries and felt that if theiruse was restricted due to IPRs “their usewould be really limited”. Both Williams(51) and Sarla (80) stressed that new geneticinformation has to be kept as much in thepublic domain as possible to ensure thatthere is equal access to it.Fairbanks (60) described a case demonstratinghow some of the limitationsimposed by IP issues, including transferof germplasm across international boundaries,could be overcome, while alsoavoiding some of the economic obstaclesfaced by scientists in developing countries.Microsatellite markers for quinoa werebeing developed at an American universityin a joint programme with a Bolivian foundation,to be then sent to Bolivia for use byBolivian scientists in their quinoa breedingand conservation programmes.Differences in capacity between developingcountriesFrom the conference it was clear that thereis enormous diversity in terms of capacity,opportunities and constraints among developingcountries that would have a bearingon development and application of MAS.There are substantial differences in factorsincluding the state of public sectorresearch, the involvement of the privatesector in research, development and marketingcapabilities, perceived priorities fordevelopment, the social and agriculturalsystems of the country, the state of educationalsystems and the degree to whichinformation and technology remain in thepublic domain.Many participants, including Buijs (22)and Corva (29), commented on developingcountries lagging behind developedcountries in uptake of new technologies,and Sokefun (3) expressed concern that alack of resources should not result in thedeveloping world being bypassed. Davila(81) suggested that developing countrieslike Brazil, where MAS can be used relativelyeasily, could help other developingcountries with MAS development, throughsouth-south cooperation. Roughly a quarterof messages posted in the conference camefrom India, and it was apparent that thisis another developing country that hasinvested substantially in MAS, among otherbiotechnologies. Such are the trends incapacity and infrastructure there that it wasindicated that Indian institutions might beable to provide MAS services more cheaplythan in developed countries (Muralidharan,74). This is an important consideration, asBhatia (8) suggested that breeders shouldask whether MAS-related analytical workcould be outsourced. Reddy (62) believedthat MAS would only be economical indeveloping countries like India.Role of the CGIAR and internationalorganizationsCollaboration between the developing anddeveloped world was inferred to be theonly way for the developing world to participaterealistically in the development ofMAS and avail itself of the opportunitiesit represented (Sokefun, 3; Galvez, 38).Fasoula (84) expressed the need for developingcountries to play an active role indeveloping MAS, particularly in makingthe associations between markers and traits,although Koudandé (68) considered thatfor economic reasons developing countriescould simply import required technology.Many other participants voiced the need for

438Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishinternational cooperation. One demonstrationof the extent to which scientists fromdeveloping countries are contributing toresearch on, and application of, MAS is thatmany participants were from developingcountries but studying and/or workingabroad. Contributions came from nationalinstitutions hosting foreign researchers andalso from centres of the CGIAR that promotecollaborative research and training.Olori (65) described the many ways thatdeveloping country individuals and institutionsare contributing to the developmentof MAS by participating in internationalagricultural research. Gianola (24), however,questioned the apparent altruism ofdeveloped countries in sponsoring collaborativeMAS efforts, fearing that it mighthide motives for developing biomedicalapplications from the results.Partnerships between the CGIAR andnational researchers led to some successes inMAS mentioned in the conference. Sridhar(35) reported on the collaboration betweenan Indian rice research institute and theInternational Rice Research Institute, andWallwork (59) on cooperation betweenan Australian institution, the InternationalCenter for Agricultural Research in theDry Areas and the International Maize andWheat Improvement Center.There was a strong call from many participantsfor the CGIAR and internationalorganizations such as FAO to play an activerole in the area of MAS development andapplication. For example, Murphy (1) suggestedthat the CGIAR and FAO shouldfacilitate international collaboration in thisarea, while Priyadarshan (11) suggestedthat the CGIAR might manage a centralizedfacility for routinely doing MAS.Acikoz (66) envisaged a role for FAO inaddressing issues of classical plant breedingat regional and national levels, which hesaw as being more of a priority than MAS,while Muralidharan (74) thought FAO to besuited to playing the role of coordinator forMAS research among laboratories workingon the same crop. Rakotonjanahary (78)proposed a similar role for FAO and theCGIAR as facilitators in the exchange ofinformation and genetic material obtainedfrom MAS. Sarla (80) suggested that FAOcould play a catalytic role in marker-aidedallele mining and facilitate capacity buildingfor applying MAS, especially for crops ofregional importance.Public–private sector linkagesVarious additional constraints to usingMAS in plant and animal improvementprogrammes in developing countries werediscussed in the conference. Notter (25)stated that the history of public fundingin developing countries was not good andFairbanks (60) commented that agriculturalresearch in developing countries was notwell coordinated. Australia has investedheavily in MAS in its breeding programmesbut, as pointed out by Collard (45) regardingplant breeding, the major target cropshave been cereals produced for export.Moreover, there has been considerablesupport from private industry for researchand development of MAS. For example,the Grains Research and DevelopmentCorporation (GRDC) of Australia wasset up to serve farmers and is maintainedthrough a levy collected from them. Incontrast, in the developing world, mostimportant crops are usually produced forsubsistence and there is often little private–public cooperation (Murphy, 1). Developingcountry farmers are unlikely to be ableto support the activities of a dedicatedresearch and development organizationequivalent to the GRDC (Collard, 45).Similarly, Notter (25) pointed out that there

Chapter 21 – Marker-assisted selection as a potential tool for genetic improvement: debating the issues 439was a scarcity of private animal breedinginitiatives in developing countries andlittle or no commercial sector. MAS, in hisopinion, would not change this situation.Nicol (19) highlighted the importance ofextension agencies in assisting uptake ofcommercially available DNA marker tests.Koudandé (68) noted that in developedcountries, most of the applied MASin breeding is undertaken by companies,and wondered which companies in Africawould be wealthy enough to support MASdevelopment and application. An additionalfactor is that MAS requires that molecularmarkers are available for particular cropsand important traits, but most of the publiclyavailable markers are for the majorcrops (Collard, 9), which are not necessarilyof primary importance in developingcountries. Some crops are also very regionspecific, such as black gram mentioned byGopalakrishna (72), and are unlikely to bethe target of research leading to developmentof MAS technologies. There seemedto be general support for a collaborativeapproach to MAS research and application,including public–private sector linkages,which would represent the best opportunityto facilitate development of, andaccess to, MAS in developing countries.Unfortunately, private sector contributionsto this e-mail conference were limited andthe discussion would have benefited frominputs by more of them.ReferencesThe author, number and title of messages referenced in the chapter – all messages areavailable at www.fao.org/biotech/logs/c10logs.htmAcikgoz, N. (66). Is MAS a little luxurious fordeveloping countries?Badr, A. (47). Tropical fruit breedingBhagwat, A. (46). Polyploid cropsBhatia, C.R. (8). Indicators of utility of MAS inplant breedingBuijs, J. (22). MAS and other ongoing researchBuijs, J. (58). Locally adapted technologyChávez, J. (33). MAS and animal breedingCollard, B. (9). Cost of MAS - the biggestbarrierCollard, B. (45). Investment makes MAS feasiblein developed countriesCorva, P. (29). Use of licensed genomictechnologyDavila, A. (81). MAS and bioinformatics fordeveloping countriesDe Koning, D-J. (13). MAS for livestock indeveloping countriesDe Koning, D-J. (16). Re: Crossing-over //false-positive markersDe Koning, D-J. (20). Re: MAS for livestock indeveloping countriesDe Lange, W. (57). Experiences with MAS so far- Netherlands, plantsDulieu, H.L. (23). Genetic conditions for MASefficiencyDulieu, H.L. (42). Re: When the marker is thegeneDulieu, H.L. (53). Re: marker assistedbackcrossingEl Ouafi, I. (77). Some of the main topics fordiscussionFairbanks, D. (60). Collaborative internationalresearch on MASFasoula, D. (84). Re: Integration of molecularmarkers with plant breedingGalvez, H. (10). MAS for tree crop improvementand transgenicsGalvez, H. (38). Re: When the marker is thegeneGanunga, R. (69). Flanking markers/ Patenting

440Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishGianola, D. (2). Economic impact of MASGianola, D. (6). Re: economic impact of MASGianola, D. (24). International collaborativeefforts - MASGopalakrishna, T. (72). MAS technologyKirti, P.B. (12). Re: indicators of utility of MASin plant breedingKoudandé, D. (68). Role of developing countriesin MASKrause, A. (75). Costs of alternative MAS - fish,animalsMakkar, H. (52). Gene-based technologies forimproving animal production and health indeveloping countriesMartinez, V. (63). Marker technologies aquaculture,some thoughts....Montaldo, H. (18). Re: MAS for livestock indeveloping countriesMota, A. (14). Crossing-over/false-positivemarkersMota, A. (36). When the marker is the geneMuchugi, A. (49). Applying MAS in indigenoustropical African tree speciesMuralidharan, E.M. (7). MAS and forestrycropsMuralidharan, E.M. (74). Costs of MASMurphy, D. (1). MAS for crop improvement indeveloping countriesNicol, D. (19). MAS in livestock- extension/traitsNotter, D. (25). Use of MAS and related technologiesin livestock improvementOlori, V. (21). Suitability of MAS for livestockimprovement in developing countriesOlori, V. (65). Participation of developing countriesin MAS developmentPriyadarshan, P.M. (11). Areas that need to bedebated - plant breedingPriyadarshan, P.M. (71). MAS, new crops anddeveloping countriesRakotonjanahary, X. (78). When MAS shouldbe used in a developing countryReddy, V.L.N. (62). Developing countries canuse information from developed countriesSaravanan, S. (31). Re: use of licensed genomictechnologySaravanan, S. (73). Active role in the developmentof MAS technologySarla, N. (80). MAI from wild species, otherissuesSeth, A. (26). Re: Use of MAS and related technologiesin livestock improvementSimons, T. (28). Utility of MAS in tropical treesfor small-holdersSingh, K. (41). Crops - IndiaSingh, K. (44). Marker-assisted backcrossingSingh, K. (61). Pedigree selection, plantsSokefun, O. (3). Re: MAS for crop improvementin developing countriesSokefun, O. (64). Fusion of soft and hardtechnologiesSridhar, R. (35). MAS approach yields diseaseresistant rice linesSridhar, R. (48). Re: successful use of MAS -cereal cropsSridhar, R. (54). Gene pyramiding - cropsSridhar, R. (76). MAS for developing nationsStorlie, E. (32). Corporate patents on particulargenesToro, M. (50). Pedigree - linkage disequilibriumToro, M. (67). Costs of genotypingToro, M. (79). Re: Costs of genotypingVarshney, R. (82). Re: MAS and bioinformaticsfor developing countriesWallwork, H. (59). Re: Experiences with MASso far - Netherlands, plantsWeller, J. (4). Re: Economic impact of MASWeller, J. (15). Re: Crossing-over/false-positivemarkersWeller, J. (30). Re: Use of licensed genomictechnologyWilliam, M. (70). Marker applications - wheat,maizeWilliams, K. (37). Successful use of MAS - cerealcropsWilliams, K. (51). Re: successful use of MAS -cereal crops

Chapter 22Marker-assisted selection: policyconsiderations and optionsfor developing countriesJames D. Dargie

442Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishSummaryPolicy options for research, development and diffusion of the products of markerassistedselection (MAS) depend on the development objectives and priorities of theagricultural sector, its various subsectors and cross-cutting activities dealing with scienceand technology (S&T), including biotechnology and the management of genetic resources.The policy agenda in each of these areas has shifted from the traditional focus of “raisingproductivity” to a broader agenda of improving rural livelihoods in both economic andnon-economic terms in support of the Millennium Development Goals (MDGs). Securingfinancial commitments from national governments and donors to invest in MAS and relatedmolecular approaches requires more active engagement by national agricultural researchand extension systems (NARES) in the processes of revising poverty reduction strategypapers (PRSPs), and in developing policies and strategies for agricultural development,S&T and genetic resources. Agriculture and agricultural S&T are undergoing rapid changebut few developing countries have either agricultural S&T or biotechnology policies. Theyneed to develop these to build coherence across the agricultural sector including delineatingthe roles of public and private sector entities, and as a means to strengthen accountabilitywith respect to priority setting, monitoring and evaluating the outcomes and impacts ofboth research and practical applications of MAS. Options are provided for developingand implementing MAS programmes and projects, for setting priorities and evaluatingoutcomes and impacts. Given the uncertain nature of technical change and the long timeframes that often characterize translation of the research and extension services providedby NARES into sustainable improvements in productivity and livelihoods through geneticenhancement, it is concluded that greater emphasis needs to be placed on research toanalyse systematically the critical paths involved in successfully transforming researchoutputs into development outcomes and impacts and that this can best be achieved usingan innovation systems approach.

Chapter 22 – Marker-assisted selection: policy considerations and options for developing countries 443IntroductionThis book provides a comprehensivedescription and assessment of MAS forincreasing the rate of genetic gain in a widerange of agriculturally important speciesusing DNA-based markers for both simpleand quantitative traits. Its various chaptersattest to the remarkable progress that hasbeen made in researching this approach.This progress has only been possiblethrough the determined pursuit of multidisciplinarity,i.e. by bringing together intoteams the skills and knowledge of individualswho could: innovate around thesuite of techniques provided by advances inmolecular biology to isolate, multiply, identifyand insert DNA sequences; produceinnovations in electronics and engineeringto miniaturize, automate and providehigh sample preparation and analyticalthroughput; use statistical and computerscience to analyse and manage the information(bioinformatics) obtained; extendknowledge of the mechanisms that regulatephysiological processes in plants andanimals; and use quantitative genetics inassociation with conventional and novelbreeding and selection approaches. Thisresearch has contributed enormously tothe processes of adapting the basic techniquesand tools of molecular biology tostudy the genetic make-up of agriculturallyimportant species at the molecular level,and to accumulating knowledge of the linkagesbetween DNA sequences and in somecases genes and traits that are importantfor the livelihoods of farmers, foresters andfisherfolk.Yet, while recognizing this admirableprogress, for most species and most traitsthat are important for both large commercialenterprises in the industrializedworld and more particularly for small-scaleand resource-poor production systems thatconstitute the livelihoods of the majorityof the rural poor in developing countries,MAS has still to deliver on its undoubtedpotential, and on the claims in academicand other circles that it would “revolutionize”the way advantageous varietiesand breeding stock are developed. As aresult, there is still a substantial mismatchbetween “the field” and the expectations ofpolicy-makers, social scientists, communitygroups and non-government organizations(NGOs), etc. In fact, the reality is that,while the approach has certainly transformedlaboratory operations, apart fromits use by the private sector in backcrossingof transgenes into elite inbred lines ofmaize and other crops and some commercialapplications in livestock, the impactsof MAS on rural livelihoods have to datefallen well short of expectations.This chapter does not dwell on thescientific and technical issues underpinningMAS, nor does it challenge eitherthe need for continued research or theunquestionably greater opportunities forscientific and technical breakthroughs andsocio-economic benefits that will surelyarise from sequencing and post-genomicsresearch – provided levels of investment ingathering and analysing phenotypic datakeep pace with molecular developments.Its focus is on the evolving political, policyand institutional settings (both nationallyand internationally) within which agricultureand agricultural S&T institutions andextension services are operating in developingcountries, and on some of the optionsopen to governments and public sectorinstitutions in these countries to engagemore forcefully in MAS-related R&D andthe diffusion of genetically improved productsgenerated through this approach toproducers. It argues that the challengesand opportunities for doing so cannot be

444Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishdivorced from the policies and objectivesunderpinning country-based and donorassistedstrategies for achieving the targetsset by the World Food Summit (WFS), theMDGs, and as described in national PRSPsand in national and regional programmesfor food security. Policies and strategiesfor successful implementation of MAS arealso inextricably linked to those for thesector as a whole and its various subsectors,and encompass cross-cutting issueslike the management of S&T includingmodern biotechnology, genetic resourcesand other developments in the internationalpolicy and regulatory arenas thatcross lines of national sovereignty. Policyconsiderations and options for MAS aretherefore described within these broaderframeworks.The chapter begins by outlining thesocial and economic contexts within whichthe agricultural sector currently operates,the challenges it faces, the main politicalforces driving change, and both theprocesses and considerations involved indeveloping comprehensive agriculturaldevelopment policies. It then goes on todiscuss and provide options available tocountries for formulating policies for agriculturalresearch, S&T, biotechnology andgenetic resources for food and agriculture(GRFA), arguing that in most countriesthere is substantial scope for greater “joinedup” thinking and coherence of action informulating, implementing and monitoringthe outcomes and impacts of programmesand projects involving MAS. Based onthe author’s interpretation of the informationprovided by other contributors to thisbook, there then follows a section coveringsome general points that policyanddecision-makers should consider beforeembarking on MAS, and this is followed bysections dealing respectively with considerationsfor priority-setting and options forimplementing MAS. The chapter concludesby looking at the future of MAS, stressingthe need for greater effort in building politicalsupport, in setting priorities and betterdelineating the roles and responsibilities ofdifferent stakeholders, in fostering partnershipsand in creating more effective deliverymechanisms.ContextHunger, poverty and agricultureThe number of people who go hungryeach day in developing countries standsat around 820 million and around 24 percentof the people in developing countriesare absolutely poor, living on less thanUS$1 a day (FAO, 2006). Hunger and povertyin the midst of plenty are the centralchallenges in today’s global economy andsociety, but if the trends of the past decadeare extrapolated forward, there will still be582 million undernourished people by 2015(FAO, 2006). This is well short of the targetof 412 million that was set at the time of theWFS in 1996, although possibly on track tomeet the somewhat less ambitious MDGsthat were set in 2000. More than half ofthe 582 million will be in South Asia andEast Asia, with 203 million and 123 millionrespectively, while sub-Saharan Africa willbe home to 179 million hungry. The challengeis not only to provide food security in2015 for the present 820 million malnourished,but for the additional 600 millionpeople born over the coming nine years andthe nine billion people projected to makeup the world’s population by the middle ofthis century.The nature and causes of hunger andpoverty are many, varying widely betweenand within countries; they are also evolvingand often interlinked. Even so, the factthat they are most concentrated in rural

Chapter 22 – Marker-assisted selection: policy considerations and options for developing countries 445areas where people’s livelihoods dependon agriculture (including fisheries and forestry)and the non-farm small and mediumagro-industrial processing and servicingindustries that are connected to it, meansthat investing in agriculture and morebroadly in rural development must beat the heart of any strategy for hungerand poverty reduction. While the measuresneeded certainly go well beyond theissue of producing more food and agriculturalproducts, achieving greater yields andhigher value products from the same plotof land or enterprise must be a key ingredientfor the great majority of developingcountries.How to do this at a lower cost toimprove household access to food andthe competitiveness of small-scale farmerswhile maintaining or improving producerincentives, the sustainability of farmingsystems, and the many services providedto societies by both managed and naturalland and aquatic ecosystems, poses hugechallenges. Particularly challenging is tacklingsituations where agricultural potentialis low, resources poor and markets distant.Any investment in MAS needs to bejustified on the basis of its potential to contributein an effective and efficient mannerto these challenges.The evolving context of agriculturalgrowth and policyThe situation facing rural producers,households and public institutions now isquite different from that of 20 years ago.Political support and consequently publicsector investments in agriculture and ruraldevelopment have fallen both nationallyand from international donors and financialinstitutions. Privatization has been theoverarching policy response, but often theprivate sector has failed to fill the gaps,leaving many producers with no or significantlyreduced flows of the inputs andservices critical for both production andaccess to markets such as technologies,extension and credit. Additionally, marketaccess for poor producers has deteriorateddue to greater integration of the globaleconomy and other market distortions,and the need to conform to internationalsanitary, phytosanitary and food safetystandards as well as to product accreditationschemes established by supermarketchains and others. Complicating the situationfurther are pests and diseases, naturalresource degradation and climate change.On the positive side, there is irrefutableevidence of a deepening politicalcommitment within governments and theinternational community to tackle poverty,hunger and environmental degradationurgently and in a concerted manner. Atthe global level, this includes the Plan ofAction that emerged from the WFS in1996, the set of eight MDGs that followedthe United Nations Millennium Summitin 2000, and the Plan of Implementationfrom the World Summit on SustainableDevelopment in 2002. Regionally, itincludes the vision and strategic frameworkdocument for the New Partnershipfor Africa’s Development (NEPAD) andits underpinning Comprehensive AfricaAgriculture Development Programme(CAADP) (NEPAD, 2002).Nationally, the most notable examplesare: the development of revised countrydrivenPRSPs, which aim to link nationalpublic actions, donor support and theresults needed to support the MDGs, andwhich provide the basis for World Bank andInternational Monetary Fund’s concessionallending and for debt relief under the heavilyindebted poor countries (HIPC) initiative;national development strategies, plans

446Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishand programmes; sector-wide approaches(SWaps) that are aligned with the MDGs;and national and regional programmes forfood security that are supported by FAOand its donors. In one form or another,these documents describe national macroeconomic,structural and social policiesand programmes to promote growth andreduce poverty, and map out plans for theirattainment and priorities for both domesticand external assistance. With respect tothe latter, the 2005 Paris Declaration onAid Effectiveness provides a multidonorcommitment to improve aid effectivenessthrough harmonization, alignment andmanaging for results (see, for example,www.oecd.org/document/18/0,2340,en_2649_3236398_35401554_1_1_1_1,00.html).Significant within essentially all of theseprocesses is the shift from the traditionalagenda of “raising productivity” for agricultureto the broader agenda of improvingrural livelihoods in both economic andnon-economic terms. Also significant is thefact that despite their positive track records,agriculture and, even more so, agriculturalresearch are not high on the list of prioritiesin country PRSPs. This points clearly to theneed for agricultural ministries and NARESto engage more actively in the process ofrevising future PRSPs. However, successin elevating the priority given to agriculturewithin PRSPs will only be achievedby formulating and delivering agriculturalR&D policies, programmes and activitiesthat are coherent within the agriculturalsector, with national PRSPs and with theMDGs. Noteworthy here is that in 2004 the Economic and Social Council of theUnited Nations (ECOSOC) underscoredUN Economic and Social Council (ECOSOC)Resolution 2004/68 “Science and Technology forDevelopment” (E/2004/INF/2/Add.3).that most developing countries are unlikelyto meet the MDGs without a clear politicalcommitment to making S&T among the toppriorities in their development agendas, andrecommended that governments increaseR&D expenditure to at least one percent ofgross domestic product (GDP).Comprehensive agriculturaldevelopment policiesDefining these policies basically entailsdetermining the broad-based objectivesof the sector, which, in the new contextof development, also needs to includegoals for enhancing social equity and naturalresource sustainability. Essential tothe process is “mapping the terrain”. Thisrequires evidence-based stocktaking of pasttrends within each subsector (e.g. withrespect to production, marketing and legislation),and identifying barriers to realizingopportunities for expansion, strategic alternativesfor moving forward based on anassessment of what the future is likely tohold, and the instruments and means fortheir implementation (e.g. through newlegislation, administrative decrees, publicand/or donor investment, and participationby the private sector and civil society).Preparation then moves on to developan integrated sector-wide package of policiesto guide implementation, including aninvestment programme. Policies directedtowards rural poverty reduction throughagriculture must be based on (a) determiningwhere poor people intersect mostprominently with agriculture and the majorrisks they face (e.g. drought, salinity, diseaseoutbreaks); (b) the types of productionsystems and commodities they produce;In many countries, forestry and fisheries areseparate sectors with ministries responsible fordeveloping National Forestry or Fisheries Actionor Development Plans

Chapter 22 – Marker-assisted selection: policy considerations and options for developing countries 447and (c) the linkages they have to markets,research and extension systems, etc. Someapproaches for obtaining this informationare described later.From the standpoint of planning investmentsand strategies, it should set the scenefor how the government intends to pursuereductions in poverty and food insecuritythrough agriculture, answering questionssuch as: will greater emphasis be placedon self-sufficiency and, if so, for whatcommodities; to what extent does the governmentintend to promote production bysourcing seeds and planting materials, fertilizers,breeding and broodstock, and feedsfor livestock, fish and shellfish from abroador relying on its own genetic resources andresearch and dissemination systems; does itforesee greater private sector involvementand, if so, in what areas and how will thisbe achieved; what is its attitude towardsmodern biotechnology – does it intend topursue this and, if so for what purposesand how? For example, for internal political,external trade, cost and technical skillconsiderations, do investments in MASappear more attractive than pursuing thedevelopment and/or importation of geneticallymodified organisms (GMOs), andshould investments in MAS and/or GMOsbe given priority over conventional geneticselection approaches?The value of having an agriculturaldevelopment policy in place lies not onlyin the end result, i.e. a description of thecourse of action that a country intends totake to move the sector and its various subsectorsforward over a given time frame. Italso comes from the process itself, which ifdone with commitment to detail and rigourin analysing both past trends and future sce-Referred to as LMOs (living modified organisms) inthe Cartagena Protocol on Biosafety (2000) to theUN Convention on Biological Diversity (1992).narios for the sector, and inclusiveness andtransparency to ensure the broadest possiblestakeholder participation and buy-in, leadsto policies and strategies that are coherentwithin and between subsectors and betweenagriculture and rural development. It alsohas better prospects of securing consensuswithin the sector and endorsement for itsimplementation from other ministries witha stake in rural development.There is clearly no ready-made modelfor conducting this process or for how thepolicy itself is implemented, monitoredand the lessons learned are fed back forupdating, but sound leadership and commitmentare critical for preparing relevantand objective inputs through analysis andsynthesis of information available withinthe ministry itself and from other relevantministries. Ideally, information is alsoprovided by research, extension, highereducation and other service institutionsand bodies in the public and private sectorsincluding from civil society throughdocumentation and/or organizing meetingsat central, local and even communitylevels, and by outside advisers. In otherwords, both “top-down” and “bottom-up”approaches are essential to achieve balanceand consensus with respect to goals andobjectives.Agreeing and implementing policiesand strategies that are mutually supportiveand where the “sum” is greater than the“parts”, to meet the needs of the hugediversity of farming systems that exist indeveloping countries while targeting poorproducers and consumers is clearly a hugechallenge requiring negotiation, compromiseand realism. It is also something thatcan neither be rushed nor “set in stone”,and a fully integrated “rolling” policy thatis updated at regular intervals representsthe ideal. Developing an agricultural policy

448Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishreform package in Honduras, for example,involved around 80 meetings over a yearin which both Campesino and large-scaleproducer organizations participated, whilein Guyana more than 100 meetings of civilsociety task forces were held over severalyears (www.fao.org/docs/up/easypol//354/agrc_pl_str_cnpt_prct_031EN.pdf).Policies for agriculturalresearch and extensionThe absence of systematic and comparativedata on the benefits arising from theuse of different technologies in agricultureprecludes attaching priority to anyone approach. However, the high rates ofreturn and the reductions in both economic(Pardey and Beintema, 2001; Evansonand Gollin, 2003; Raizer, 2003) and noneconomicpoverty (Meinzen-Dick et al.,2004) are impressive by any standards,and justify many of the past investmentsmade in research and technology transfer.Consequently, it is hardly surprising thatmobilizing and directing national institutionsand skills towards capitalizing moreforcefully on the opportunities availablethrough S&T to meet the MDGs and otherglobal, regional and national targets areconsistently stated commitments withindifferent sectors and themes. Most noteworthyin this regard are the plan to achievethe MDGs prepared by the UN MillenniumProject and the underlying report of itsTask Forces on Science, Technology andInnovation (UN Millennium Project,2005a and 2005b), Africa’s S&T consolidatedplan of action (NEPAD, 2005), andthe CAADP and InterAcademy Councilreports on realizing the promise and potentialof agriculture in Africa (NEPAD, 2002;IAC, 2004).Yet, here again, judged by the content ofcurrent PRSPs and agricultural developmentplans, the vision of how S&T can contributeto enhancing the value of GRFA and therebyto achieving national economic and socialobjectives is invariably either missing, orthe belief is projected that “on the shelf”technologies and existing knowledge onlyhave to be adapted to local circumstances tomeet the challenges ahead.Unquestionably, policies and strategiespromoting adaptive research anddissemination of existing technologies witha successful track record must have highestpriority in the short term. However, prioritiesthat will respond to the needs ofsmall producers and rural households fornew technologies in 10 or 20 years time(and requiring more upstream strategic andapplied research) also need to be identifiednow because (and despite the claimsof some scientists), even with the availabilityof more advanced R&D methods andtools, there is no reason to believe that theuptake of new agricultural technologies willcontinue to be other than slow and incremental(described by Pardey and Beintema,2001, as “slow magic”). In this regard, theimportance of countries having pluralistic,participatory, client-focused, decentralizedand gender-sensitive advice on the processesof technology diffusion and adoptionshould be emphasized.Maximizing the relevance and futurecontributions of new technologies tooverall agricultural development alsorequires greater attention to planning anddecision-making about the direction andmanagement of the scientific techniquesand tools as well as the genetic resources towhich they will be applied in food and agricultureagainst the backdrop of current andlikely future driving forces of change.These forces include:• political policies, where, as outlinedpreviously, the new framework for

Chapter 22 – Marker-assisted selection: policy considerations and options for developing countries 449agricultural policy focuses on alleviationof rural poverty. By shifting futureinvestments in research towards anemerging paradigm of “research fordevelopment”, the research agenda isbroadened to ensure functional linkagesto national development policies and toinclude the wider dimensions of livelihoodstrategies in both planning and assessingthe impact of projects and programmes.It also shifts past emphasis on inputbasedtechnology supply by scientiststo demand and need-driven innovationsystems involving many other actors.Political commitment is also crucial toensure sustainability of funding;• advances in science, and most notablyin the computing and biological sciences,as these have provided new techniquesand tools for researchers to locate betterand therefore target production systemsand communities most associated withpoverty and food insecurity, and newtechnologies in the form of seeds, breedingstock, vaccines, etc. with the potentialto increase productivity within agriculturalsystems and wider food chains andimprove economic and social well-being.They are also helping to overcome barriersto wider social engagement in decision-making;• growing acceptance of the importance ofoptimizing system productivity by bettermanagingthe interactions among diversifiedfarm enterprises and sustainableresource management and ensuring accessto markets, rather than maximizing individualcrop or animal performance;• expanded intellectual property rights(IPRs) for biological innovations (seeChapter 20) and changed norms foraccessing and sharing the benefits ofgenetic resources in general and plantgenetic resources in particular, supportedby international agreements, conventionsand treaties (see later);• increased private investment in S&T ingeneral, and within agriculture, throughboth R&D directed primarily towardscrop, livestock and fish genetic improvement,and the delivery of productsthrough multinational and national seedand breeding companies and their franchises;• expanded public–private sector collaborationin research, development and extension,in some countries supported bylegislation;• increased public awareness of the relevanceof the uptake of new technologiesand their significance for improving thelivelihoods of rural people.These changes have already been feltmost forcefully in industrialized and largedeveloping countries such as Brazil, China,India and South Africa, where the demandpullcreated for products of R&D isgreatest. However, their impact is increasinglyspilling over into others, includingthe low-income food-deficit countrieswith much less capacity to benefit fromor otherwise adjust to the new realities ofconducting S&T in a globalized world. Animportant issue for all countries is thereforehow to adapt their NARES to respondbetter to both the current and likely futureneeds of their agricultural sectors, and in sodoing to consider their S&T “futures”, oneof which is clearly modern biotechnology.Nonetheless, judging by the content of bothPRSPs and national agricultural developmentpolicies and strategies, few developingcountries appear to have started along thisroad by producing an integrated agriculturalresearch and extension policy. Thissituation is hardly conducive to obtainingpolitical and financial support for R&Don approaches such as MAS, which, as

450Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishis clear from other chapters in this book,still remains largely in the laboratoriesand experimental stations of research institutesin both industrialized and developingcountries.National S&T and biotechnologypoliciesLike the comprehensive agriculture developmentpolicy, the rationale for having anational policy on S&T and, within thator separately, on biotechnology, is to providea framework for government andkey stakeholders to work together in acoherent and mutually supportive way toensure that developments are captured fornational benefit. The principles involvedand mechanics of how it can be developedand managed are essentially the sameas those described earlier for agriculturaldevelopment policy planning, the main differencebeing in the breadth of governmentinvolvement – being cross-cutting issues,developing and implementing nationalS&T and modern biotechnology policiesare clearly cross-sectoral responsibilitieswith coordination normally assigned to theMinistry of Science and Technology. Theexamples given in Box 1 illustrate optionsfor pursuing the development of a biotechnologypolicy. More information onnational biotechnology policies in individualcountries is available at www.fao.org/biotech/country.asp.While many countries have an overallS&T policy in place, and these and someother developing countries now havebiotechnology policies (the most recentexamples being Bangladesh, Kenya,Malaysia and Nigeria), the vast majoritydo not. Most national agricultural R&Dinstitutions therefore lack the compassprovided by the process of developingan overall national policy to guide thedevelopment and management of anagricultural biotechnology policy andfrom there, to formulate programmesand projects specific to the agriculturalsector. This paralysis in policy-makingonly serves to promote supply-driven, atthe expense of demand-driven, prioritysettingand hence targeting of investmentstowards questionable needs. It also leadsto fragmented and uncoordinated activities,and in some cases to delays in the adoptionof technologies that could help improvethe efficiency of agricultural research andprovide products and services that directly orindirectly improve livelihoods. Indeed, thesurvey conducted by FAO on applicationsof MAS in the crop subsector (Chapter 2)illustrates well both the dearth of skills inpriority-setting and coordination withinmany countries that adopt this approachand the complete lack of such activities inmany others. While this can be explainedto some extent by the relative novelty ofbiotechnology applications, as far as MASis concerned, the paucity of informationon the actual or potential economic andsocial benefits of the products arising fromits application in the different agriculturalsubsectors is surely a major stumblingblock to priority-setting and investment.National agricultural research and biotechnologypoliciesEven in the absence of an overall nationalbiotechnology policy, countries have anumber of options for improving the strategicplanning, monitoring and evaluationof modern biotechnology applications,including MAS, within their agriculturaland wider rural development sectors.The preferred approach is for theMinistry of Agriculture in associationwith other relevant ministries (particularlyHigher Education) to champion the process

Chapter 22 – Marker-assisted selection: policy considerations and options for developing countries 451Box 1National biotechnology policies: Thailand and South AfricaIn Thailand, the National Biotechnology Policy Framework (2004-2009) was prepared by aNational Biotechnology Policy Committee chaired by the Prime Minister. This then led tothe setting up of a national centre specifically devoted to biotechnology (the National Centrefor Genetic Engineering and Biotechnology) under the National Science and TechnologyDevelopment Agency. It is both a granting and research agency with its own researchlaboratories and is funded from a combination of government finances, revenue from servicesand commercial projects and competitive grants from national and international sources. Ithas major activities in agricultural biotechnology including: on genome mapping and markerassistedbreeding of rice; on cassava improvement where a database of cassava expressedsequenced tags (ESTs) is currently being developed and employed in the study of starchbiosynthesis; and shrimp, with major projects on ESTs and genome studies for applicationin breeding, disease diagnostics and shrimp domestication. Noteworthy also is that throughjoint government-private sector funding, Thailand will host “Biotechnology Asia 007” withthe focus firmly on agriculture.In the case of South Africa, the National Biotechnology Strategy (2001) arose from agovernment request and the work of an interdepartmental committee led by the Department ofArts, Culture, Science and Technology with participation of the Departments of Agriculture,Health, Trade and Industry, and Environmental Affairs and Tourism. This committee set upan Expert Panel to provide specific inputs based again on “mapping the terrain” in terms ofcurrent applications, legislation and finance, and participation by all key stakeholder groups,etc. Arising from the policies proposed within the strategy document, a National BiotechnologyAdvisory Committee was established in 2006 and the Department of Science and Technologycreated a Biotechnology Unit. Since then, three Biotechnology Research and InnovationCentres and a National Bio-informatics Network have been established, interdepartmentalcooperation has been promoted, and bilateral agreements have been signed. Again, agriculturalapplications of biotechnology receive high priority in this national strategy.of establishing an open learning process fora national agricultural S&T policy dialogueincluding biotechnology, leading eventuallyto a planning document and a processof monitoring and evaluating outcomes andimpacts. This could be achieved by establishinga national committee that wouldthen define terms of reference and setup various task forces/working groups ina participatory and pluralistic manner toreport on specific subsector and thematicissues.In common with other planning procedures,the first step should involve adiagnostic study and analysis of existingS&T policies as well as of the national,regional and international S&T landscape.Bijker (2007) provides an excellent descriptionof the criteria for building an S&Tpolicy via a policy dialogue and a methodologyfor carrying out a diagnostic study.Essential for promoting a well thoughtoutpolicy and its effective managementis the closest possible involvement of all

452Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishplayers with a stake in research, developmentand diffusion of genetic material(ministry personnel, representatives ofNARES, private companies, NGOs, farmer’sgroups, etc.). However, its focus mustbe on developing new national agriculturalS&T (including biotechnology) policiesand strategies to: (a) support institutionalreforms, including intensifying cooperationat national, regional and internationallevels; (b) strengthen national capacities;and (c) identify new funding mechanisms.Within this process, countries need toidentify priorities and appropriate levelsof resources to assign to biotechnology inlight of their socio-economic conditionsand cultural contexts and, in situations ofno-growth budgets, they need to decide onwhat is to be diverted from other importantproblems. A critical issue is also reachingagreement on the roles and responsibilitiesof public and private sector entities.Suggestions are given later for consideringMAS within this overall context.An alternative – or preferably as part ofthe process of preparing a national agriculturalS&T policy – is to draw up specificsubsector strategies. This has the advantageof focusing minds and resources withinthat subsector in a holistic manner, forexample, in the case of the crop subsector,by bringing together stakeholders dealingwith breeding, conservation and seed production/dissemination.Another possibility,very attractive from both the S&T angle andfor avoiding the creation of new structures,is to cover modern biotechnology policydevelopment and programming throughexisting structures for managing geneticresources (see below for more details onrationale). Least attractive and cost-effective,but unfortunately all too often thecase, is for individual research institutionsand universities to draw up and implementpolicies and programmes that lack coordinationwith others dealing with the same orclosely related subject matter, in particulargenetic resources management.Whatever the path chosen and notwithstandingthe need to ensure nationalownership of the process, advice (if neededand requested) on the actual or potentialrole of MAS within the agriculturalS&T landscape should be sought fromindependent sources. These include theConsultative Group on InternationalAgricultural Research (CGIAR), FAO, theWorld Bank, the International Council forScience (ICSU), the InterAcademy Council(IAC), the InterAcademy Panel (IAP),regional academies such as the Federationof Asian Scientific Academies and Societies,and national academies.National policies on genetic resourcesin food and agricultureMAS needs access to DNA-based techniques,constructs, tools, databases,statistical packages, etc., and Chapter 20describes the IPRs surrounding these whichinclude patents, copyrights, trademarksetc., as well as providing suggestions toNARES about acquiring these technologicalresources. However, successful applicationof MAS also depends on accessing newbreeding techniques and, as many nationalcollections may lack sufficient diversity(e.g. to reduce vulnerability to pests anddiseases), they may need to acquire geneticresources that are available in other countrieswithin landraces, wild ancestors andrelatives, parental and breeding lines, protectedvarieties, breeds and broodstock.Additionally, as knowledge grows of thelinkages between phenotype and genotype,awareness increases of the potentialvalue of genetic resources and, as participatoryprocesses involving local communities

Chapter 22 – Marker-assisted selection: policy considerations and options for developing countries 453become more prevalent, so the demandsfor both germplasm exchange and sharingthe benefits of the final products that aregenerated from R&D will increase. In fact,over the last 30 years, and due to a combinationof the new possibilities openedup by molecular biology and some wellpublicized cases of “biopiracy”, governmentshave increasingly come to appreciatethe actual and potential value of geneticresources. This has resulted in an expansionof legally-binding global and regionalinstruments, and national laws, regulationsand policy concerning issues of access,ownership and control of genetic resourcesand the sharing of benefits arising fromtheir use or enhancement.For the further pursuit and future successof MAS, policy- and decision-makers as wellas individual scientists need to be aware ofthe requirements for international exchangesof genetic resources such as those describedin the CBD (1992), the International Treatyon Plant Genetic Resources for Food andAgriculture (ITPGRFA) (see Stannard et al.,2004 and Bragdon, 2004) and its standardMaterial Transfer Agreement (MTA), andthe World Trade Organization (WTO)Agreement on Trade-Related Aspects ofIntellectual Property Rights (TRIPs), particularlyArticle 27.3 (b), which states thatwhile members may exclude plants andanimals from patentability, if they chooseto do so in the case of plants, they mustprovide an effective “sui generis” systemof protection such as the 1978 and 1991 versionsof the International Convention forthe Protection of New Varieties of Plantsadministered by the International Unionfor Protection of Plant Varieties (UPOV),or a combination of the two (IPGRI, 1999;Le Buanec, 2004; Donnenworth, Grace, andSmith, 2004; FAO, 2005a; Tripp, Eaton andLouwaars, 2006).They should also be aware that internationalexchange of germplasm carrieswith it the risk of introducing diseasesand pathogens through plants and animalsand their parts such as seeds andpropagules, semen and embryos, andthat sanitary and phytosanitary certificatesare required to facilitate the safeexchange of genetic resources between, andunder some circumstances, within countries.Familiarity is therefore needed withthe WTO Agreement on the Applicationof Sanitary and Phytosanitary Measures(SPS), the WTO Agreement on TechnicalBarriers to Trade (TBT) and the instrumentsrelevant to standard setting withinthese, including: the International PlantProtection Convention with its objectiveof preventing the introduction and spreadof plant and plant product pests, and theAnimal Health Code implemented by theWorld Organisation for Animal Health(OIE) that covers both livestock and fish.This international policy and regulatoryframework is both complex andcontinuously evolving. Hence, apart fromthe scientific and technical challengesinvolved in MAS, developing countriesface formidable difficulties in crafting andimplementing legal and regulatory frameworksthat facilitate exchange of GRFAas well as the range of tools used in MASfor both research and commercial uses. Italso challenges national policy-makers tokeep abreast of the international policymakingprocesses and all that these implyin terms of both coordination betweenministries of agriculture, trade, environment,and S&T, and in human and financialresources. However, the consequences ofnot being knowledgeable about these matters,and in particular about the appropriatenational laws of countries from whichgenetic resources and scientific techniques

454Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishand tools are sought, could be serious forindividual researchers and their institutions.All of these aspects must thereforebe managed in a well coordinated, efficientand fair manner if countries are torealize fully the potential offered by MASto contribute to national food security andagricultural development.Most developing countries have activitiesdealing with specific aspects of plantgenetic resources for food and agriculture(PGRFA) and substantial numbers haveestablished cross-institutional programmesand coordination mechanisms includingcrop-specific bodies and networks to setpriorities, evaluate progress and in generalto promote the more effective use of geneticresources. In the case of livestock, and asnoted in the draft report on the State of theWorld’s Animal Genetic Resources for Foodand Agriculture, presented at the FourthSession of the Intergovernmental TechnicalWorking Group on Animal GeneticResources for Food and Agriculture, heldat FAO headquarters in December 2006(www.fao.org/ag/againfo/programmes/en/genetics/documents/AH473e00.pdf), apartfrom a few (mainly northern hemisphere)countries with well-developed commerciallivestock enterprises, interest in the topicis often limited to isolated departmentswithin institutes that are rarely involvedin animal genetic resources-related activities.In the case of forestry, nationalprogrammes have been established withconsultative fora and lead institutions oftenoutside the Ministry of Agriculture, andessentially similar arrangements operatewith fisheries. India, for example, has aNational Bureau within the Indian Councilfor Agricultural Research (ICAR) devotedto fish genetic resources, which undertakeswork with microsatellite markers for populationgenetic analysis and determininggenetic variation among and within inlandspecies.With respect to national biodiversityprogrammes, by being Parties to the CBD,countries have committed themselves toestablishing policies, legal and regulatoryframeworks and programmes for conservingand using both wild and non-wildbiodiversity in a sustainable manner and, indoing so, to establish national BiodiversityStrategies and Action Plans (BSAPs)through focal points invariably coordinatedby Ministries of Environment.At present, however, few countriesemphasize the conservation and sustainableuse of GRFA in their PRSPs and agriculturaldevelopment plans, or have strategicand holistic roadmaps as to how theseresources should be managed within a particularsubsector, let alone how this willbe accomplished across subsectors. Suchdeficiencies are not conducive to prioritysetting and reaching agreement on specificprogrammes and projects, and thereforejeopardize funding of the most criticalneeds and the attainment of national objectives.Since genetic resources are the rawmaterials to which molecular methods andapproaches such as MAS are applied, thesedeficiencies inevitably also lead to ineffectiveand inefficient integration of modernbiotechnology into national programmes.To put management of GRFA on abetter footing, including through theappropriate integration of MAS and relatedapproaches, countries need to establish orstrengthen existing organizational structuresand programmes that respond tonational development objectives. Thismeans ensuring that: they are well linked tothe wider policies and programmes drawnup for agriculture, biodiversity and biotechnology;that they take into accountthe perspectives both of public institutions

Chapter 22 – Marker-assisted selection: policy considerations and options for developing countries 455dealing with research, genebank operationsand the supply of seeds or breeding stock,and those of wider stakeholder groups suchas farmer and community groups, privatesector entities, breed societies, etc.; and theyrecognize the interdependencies betweennational, regional and global policies andlaws concerning access to these resourcesand sharing the benefits from their use.It is beyond the scope of this chapterto deal with the setting up and coordinationof systems or programmes forthe management of genetic resources atthe species, subsector, sector and widerlevels. These aspects are covered comprehensivelyfor PGRFA by Spillane et al.(1999), and the principles involved areequally relevant to the livestock, forestryand fisheries subsectors. However, whilerecognizing the essentiality of having anational biodiversity system or programmethat is overseen, for example, by a highlevelinterministerial coordinating body forpursuing national development objectivesand reporting through the CBD process,there is simply no substitute for specializedgenetic resources knowledge withineach of the agricultural subsectors to promoteeffective and efficient planning andimplementation of MAS, including throughawareness building and advocacy withinnational and international policy forumsand interactions with donors.MAS: general considerations forpolicy- and decision-makersOne of the main take-home messages fromthe experts contributing to this book is thatMAS can be demanding in its requirementsfor specialized equipment, consumables,electricity supplies, laboratory design andmanagement, data handling and statistics,and Internet connectivity. Another is thatMAS is a complement to and not a substitutefor skills in conventional breedingand selection. Embarking on MAS shouldtherefore never be considered as a paradigmreplacing classical crossbreeding and phenotypicscreening programmes, which inmany developing countries are in any caselimited in terms of species coverage and theavailability of, for example, temperatureand humidity-controlled greenhouses andgrowth chambers and field sites, and fragilein terms of staffing and funding levels (see,for example, Chapter 8).Yet another message is that efficientand effective application of MAS requireswell-qualified staff. First and foremost, itneeds staff who have the knowledge to leaddecision-making on when and when notto embark on MAS. This has to be donestrictly on a case-by-case basis, bearingin mind that MAS may accelerate geneticprogress in some traits better than others,and that the costs and benefits of usingMAS in a production system need to beweighed up in the same way as any otherinput. It also needs leaders who give the“end product” rather than the “laboratory/research process” the main consideration,and staff with substantial design, technical,analytical and problem-solving skills andwho are up to date with developments inthe field. Furthermore, it demands a sustainablefunding base. What should neverbe forgotten is the bottom line – namely,the investment made will ultimately bejudged on the number of people benefitingfrom planting improved plant germplasmor keeping improved farm animals or fish.Another key message is the absolutenecessity of ensuring effective coordinationbetween breeders and the people workingin molecular biology laboratories. Whileit is not essential for all of these to belocated physically within the same institution,policy- and decision-makers need to

456Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishknow that investments in staff and infrastructurefor the “molecular component”of MAS are wasted if they are not linked tothe “breeding and selection” components.Apart from countries with technologicallyadvanced NARES (Type 1 and Type2 described by Byerlee and Fischer, 2001),getting all of the above elements together isa big task for the vast majority of developingcountries, particularly against the backgroundof current and often deterioratinglevels of public funding for agriculturalR&D (Pardey et al., 2006). So big indeedthat, while recognizing the need/opportunitiesfor molecular MAS, they mayconsider, in the first instance, other veryvaluable applications of molecular-basedtechniques such as the polymerase chainreaction (PCR) for plant, livestock and fishdisease diagnosis (see, for example, Viljoen,Nel and Crowther, 2005), estimating geneticdistances between varieties, strains, linesand breeds, conducting variety and parentagetesting (De Vicente, 2004; Chapters14 and 17) and for GMO characterizationand detection. These applications are notconsidered further here since they fall outsidethe core subject matter of this book.Also, while recognizing the increasing roleof the private sector, the options describedbelow for pursuing MAS are based on theassumption that the public sector will continueto be the major investor in R&Dfor small-scale producers and increasingthe access of poorer sections of society toaffordable food and agricultural products.Additional options are available throughpublic–private partnerships and these arediscussed later.A final consideration is that, unlike thedevelopment and release of GMOs, MASdoes not require the establishment andthe enforcement of a specific legislativeframework. Apart from avoiding the needfor specific capacities in public administration,this certainly reduces the finalcosts of adopting MAS-derived varietiesand breeds.Priority setting for MASTargeting the farming systems, speciesand traits linked most to poverty andhungerInvesting in MAS has to be based on strikingan appropriate balance between needs andopportunities for combating hunger andpoverty through genetic enhancement.Essential to that process is determiningwhere the greatest concentration of povertyand hunger exists and the causal factors.There are essentially four approaches forpursuing this.• Poverty and hunger mappingAlthough still relatively new, this approachis gaining increasing acceptance in nationaland international development circles. Oneof the major challenges faced by all countriesin targeting their development, andhence research efforts, towards the foodinsecure and poor lies in the diversity oftheir farming systems and socio-economicconditions. However, using a combinationof survey and census information(e.g. household surveys), administrativedata (e.g. markets, roads), geographicalinformation systems (GIS) and small areaestimation maps, it is becoming increasinglypossible to develop correlations andmaps that link population densities, welfaredata and crop and livestock production andlivelihood systems; in effect, to pinpointwhere poor people live and the productionand livelihood systems associated with highlevels of poverty.Increasingly, through programmes suchas the Inter-Agency Programme on FoodInsecurity and Vulnerability Information

Chapter 22 – Marker-assisted selection: policy considerations and options for developing countries 457and Mapping Systems (FIVIMS) whichworks both locally and internationally,Ministries of Planning are developingdisaggregated poverty maps to help targettheir interventions for greatest benefitsto the poor. Recent examples include thehigh resolution Kenyan poverty mapsdeveloped by the Bureau of Statisticswithin the Ministry of Planning with theassistance of the International LivestockResearch Institute (ILRI), the RockefellerFoundation, the World Bank and the WorldResources Institute (WRI) (Ndeng’e et al.,2003), and the International Rice ResearchInstitute’s work linking poverty and ricesystems in Bangladesh (www.irri.org/science/progsum/pdfs/DGReport2000/FP1.pdf).• Rapid rural appraisalsThese are systematic but semi-structuredactivities conducted by teams with bothtechnical and social science backgrounds,usually as part of farming systems research(see below and Crawford, 1997). Theirchief characteristics are that they take onlya short time to complete, tend to be relativelycheap to carry out and make use ofmore “informal” data collection procedures.The techniques rely primarily onexpert observation coupled with semistructuredinterviewing of farmers, localleaders and officials. In substance, the techniquesof rapid rural appraisals (RRA) areexecuted over a period of weeks, or at mostmonths, rather than extending over severalyears. To date, RRA has mainly been usedin the field of rural development as a shortcut method to be employed at the feasibilitystage of project planning.• The farming systems approachThis groups farm households with similarcharacteristics and constraints andtherefore from a R&D perspective has thepotential of promoting technology andknowledge spillovers. Unquestionably,the most authoritative study of the linkbetween farming systems and poverty isprovided by Dixon, Gulliver and Gibbon(2001). These authors describe 72 majorfarming systems throughout the developingworld based on available natural resources,patterns of farm activities and householdlivelihoods, intensity of production andtheir relationship to markets. They alsodescribe the needs of those living withinthem (with an average agricultural populationof about 40 million inhabitants), thelikely challenges they face and opportunitiesopen to them in the next 30 years, andthe relative importance of different strategiesfor escaping from poverty and hunger.In sub-Saharan Africa for example, of the15 major farming systems identified, boththey and the IAC (2004) gave priorityto four systems based on the economicvalue of production and the extent of malnutrition,namely: the maize-mixed; thetree-crop based; the cereal/root crop based;and irrigated systems. However, NARESneed to undertake similar priority assessmentsto complement such analyses.• The “rural worlds” conceptThis categorizes rural people as capitalintensive farmers, mixed commercial/subsistencefarmers, the near or totally landlessand those without any economic activity(OECD, 2006).While each of these approaches hasmerits and limitations for targeting interventionsbased on geography and population,they all embrace the principle of engagingfarmers and rural consumers/households indiagnosing problems and identifying possiblesolutions adapted to their particularcircumstances.

458Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishAnalysis of the needs andopportunities for MASAs noted earlier and in an ideal world,the needs and opportunities for embarkingon MAS should emerge first and foremostthrough the policy dialogue processesthat lead to country priorities and objectivesfor agriculture and for agriculturalS&T. In any case, careful analysis is neededto determine whether, given the currentS&T, socio-economic and cultural landscapeand government/community plansfor the foreseeable future, the use of MASwill realistically contribute to hunger andpoverty reduction. This requires a team ofcompetent analysts to conduct an ex anteimpact assessment that makes the best use ofexisting knowledge to determine whether:• the principal barrier to sustainable intensificationor diversification of the productionsystem(s) as a whole could beovercome by introducing a new plantor animal genotype or by changing theenvironment, e.g. introducing better soil,water and nutrient management practices,draught power, vaccination, tsetsefly or other disease/integrated pest managementpractices. Also to be consideredare the management changes that wouldinevitably be needed following the introductionof such genotypes. For example,increasing the prolificacy of local sheepor goats through MAS brings with it therequirement, inter alia, for an improvedfeed resource base. Does the system havethe potential to provide this, and is therea market demand for the animals andtheir products? Foresight and total systemsthinking are clearly required here;• the species x trait(s) combination(s)required is not available in locally availablegermplasm (or breeds/broodstock)or in varieties/pre-breeding materialsdeveloped by and available from theInternational Agricultural Research Centres(IARCs), or other countries growingthe same crops within similar productionsystems and located in similar agro-ecologicalzones;• the species x trait combination cannot bedeveloped more easily, and/or at less cost,through phenotypic selection. A numberof chapters in this book provide excellentguidance on the factors that are importanthere, which include, inter alia: thespecies involved; the genetic complexityand heritability of the trait(s) required(the current focus for most crop andmany animal species is heavily on diseaseand pest resistance); the availability ofmarkers for the trait(s) in question andability to scale up their usage, whetherthe trait is sex-limited (livestock); and theavailability of reliable phenotypic data,etc.;• there is already an existing nationalbreeding programme(s) for the species inquestion;• the national breeding programme(s) forthe species in question has the infrastructureand levels of human and financialresources needed to sustain selection andbreeding activities;• national infrastructures and capacities inmolecular biology match the scientific,technical and information requirementsfor effectively supporting MAS;• professional legal advice is available concerninglaws, agreements, licences, etc.for the acquisition and diffusion of boththe tools or enabling techniques, andthe starting- and end-products of MAS(see Chapter 20 and earlier in relationto access and benefit sharing of geneticresources);• efficient mass propagation systems (e.g.seed multiplication or semen productionprogrammes) are in place;

Chapter 22 – Marker-assisted selection: policy considerations and options for developing countries 459Box 2Centralized national centres of excellence and sectoral/subsectoralinstitutionsMore technologically advanced developing countries such as Brazil, China, India and South Africaand others have established one or a number of cutting-edge centres for biotechnology workingon both basic and strategic techniques and tools, and providing analytical and other support tonational sectoral or subsectoral centres working on more applied and adaptive research projects.For example, the African Centre for Gene Technologies (ACGT) is an initiative by the SouthAfrican Council for Scientific and Industrial Research (CSIR) and the University of Pretoriato create a national centre of expertise and a world-class platform in gene and genome analysis.Its focus is on using genetic markers to understand disease resistance in plants and nitrogenmetabolism in cattle under harsh and arid environmental conditions. It supports the moredownstream work of the Forestry and Agriculture Biotechnology Institute as well as various cropcentres. ACGT is a member of the Southern African Network on Biosciences (SANBIO) andpart of the NEPAD-sponsored African Biosciences initiative.• adequate technical advisory services areable to support technically the disseminationof the improved variety or breed;and• effective delivery, monitoring and evaluationstrategies are in place to bring theproducts of MAS-related R&D to usersand beneficiaries.Country options forimplementing MASCountries with high-quality personneland facilities for phenotypicevaluation and selection and inmolecular biologyIndividually or collectively and for anumber of crop species, public institutionsin these countries have the skills both tochoose the appropriate parental and segregatingmaterials and to apply routinely andwith high throughput the full range of techniquesavailable (including those requiringsequence information) to develop molecularmarkers. Through the establishmentof centralized centres of excellence andsectoral/subsectoral institutions (Box 2),they have the potential to validate putativemarkers by combining their use withthe detailed and comprehensive phenotypicinformation available on large numbers oflines for multiple traits to produce geneticlinkage maps for identifying genomicregions controlling variations in simple andquantitative traits (QTL), and to use theright combination of trait-linked markersto improve the efficiency of parental selectionand breeding programmes.These countries tend to have concentratedtheir MAS activities on theintrogression of a few traits (for instancethose encoded by transgenes) and in afew crops, although markers themselvesare being used for many of the non-MASapplications mentioned earlier. However,they also contribute effectively to globaland regional crop genomic projects directedtowards developing and validating geneticand linked markers and testing their usefulnessfor MAS in breeding programmes.They may also have some skills in applying

460Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fish(some of) these methods and approaches tolivestock, fish and especially forest speciesresearch. Generally, however, these effortsare of a small-scale experimental nature and,particularly in the case of livestock wheremost traits, even for disease resistance, arecomplex, they are unlikely to move beyondthe research stage in the near term becauseof the large numbers of animals required,the limited amount of structured phenotypicdata available and the long generationintervals of many animal species.These countries have both considerablepotential and many options to focusresources for MAS on poverty and hungeralleviation, including:• mobilizing the techniques, tools, geneticresources and phenotypic data alreadyavailable nationally and internationally(e.g. parental lines and segregating populationsfrom international and othernational programmes), tapping the vastand rapidly increasing molecular andgenetic knowledge available internationallythrough collaboration with internationaland advanced agricultural researchcentres, and contributing to genomicsconsortia (e.g. the International RiceGenomics Consortium). Applicationsinclude extending non-transgenic andtransgenic approaches by developing andvalidating markers based on fine genomicmapping of QTL (i.e. by identifying singlenucleotide polymorphisms, [SNPs])for more complex traits like drought,salinity and heat tolerance and nutritionalquality in major food crops;• pursuing MAS for both simple and complextraits in crops that although relativelyminor and scientifically neglectedare of tremendous importance to manypoor households;• recognizing the increasing importance oftrees, livestock and aquaculture to theirrural economies, strengthening effortsto characterize genetic diversity throughboth classical phenotypic and molecularmarker approaches and then developing,validating and eventually using markersfor improving economically importanttraits such as host resistance to diseases.Countries with reasonable capacitiesfor phenotype evaluation andselection and some capacities to applymolecular marker methodsThese countries have less comprehensivebreeding programmes and therefore cancover fewer species. They will likely havebeen relatively “late starters” in MAS andmay not have the latest high-throughputequipment, which is invariably located inone or a number of institutes supportinga particular subsector. Neither of thesefeatures is a major limitation providedthe country prioritizes its work appropriately.This means pursuing justifiable anddoable genetic enhancements to the limitednumber of species for which it has an effectiveselection and breeding programme to:(a) provide the foundation for developingsegregating populations from parentallines and for characterizing and validatingmarkers for the trait(s) in question; and(b) evaluate populations in the environmentsfor the traits that are prioritized.Also, while recognizing the need toadapt specific molecular techniques tolocal circumstances, and markers for particulartraits to their own genotypes, thesecountries should take full advantage of“lessons learned” with respect to both themolecular methods themselves and howbest to integrate these into selection andbreeding programmes. With the caveat thatthese conditions are satisfied, countries inthis general category have the followingoptions:

Chapter 22 – Marker-assisted selection: policy considerations and options for developing countries 461Strengthening existing national scientificand technical capacities and infrastructuresin molecular laboratory(ies) andcoordination with selection and breedingprogrammesThis can be achieved by:• relying on their own germplasm andsegregating populations and/or partneringwith IARCs and advanced researchinstitutes to obtain these. Using lesssophisticated and largely “manual” samplepreparation and analytical equipmenteven through to the point of sequencingsince large-scale and high-throughputgenetic analysers, accessories and otherequipment are necessary only after theinitial development and implementationof markers;• taking advantage of the many kits,biological and other materials andthe “how to do” and “what to avoid”protocols and manuals, statisticalpackages, bioinformatics freeware,software and analysis programmes.These, as well as specific markers that areavailable in the form of DNA clones foruse as probes in restriction and amplifiedfragment length polymorphism (RFLPand AFLP) analysis, PCR primers foruse as SSR (microsatellite) markers, andsequence information available in publicdatabases that can be used to synthesizeand clone specific markers are availablecommercially or for free from severalof the IARCs belonging to the CGIAR,and from advanced research institutionsand universities in developed anddeveloping countries (e.g. Brazil, Chinaand India). All of these resources helpto avoid “reinventing the wheel” and to“short-cut” the process by assisting ingetting round bottlenecks, e.g. the needto establish facilities and expertise incloning; and• taking advantage of the many opportunitiesavailable to upgrade scientific andtechnical expertise in molecular biologyitself and in linking molecular and phenotypicapproaches through species andtheme-specific networks, workshops,training courses, scientific visits, etc. (seeBox 3).Using regional centres of excellenceWhile there is no real substitute for buildingnational institutions and capabilities inMAS research, product development andtransfer, countries with very limited infrastructuresand skilled human resources canstill engage in meaningful research by usingthe state-of-the-art analytical, bioinformaticsand computing facilities located inregional centres of excellence (Box 4) andin advanced national institutes.Countries with limited capacities inphenotypic evaluation and selectionand no capacities to apply moleculartechniquesThese countries fall into the category ofType 3 NARES described by Byerlee andFischer (2001). Unless the governmentcommits itself to increasing substantially itslevel of investment in essentially all aspectsof genetic resources management in one or anumber of the different agricultural subsectorsor at least one species, but particularlyin selection and breeding programmes andin capacity-building for employing moleculartechniques, their options are:• to partner with institutes of the CGIARsystem and other advanced institutionsin developed and developing countriesand import varieties and advanced breedinglines developed by these institutesthrough MAS that contain the neededtrait(s). After testing these or their crosseswith local varieties or landraces for

462Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishBox 3Regional networksThe East African Regional Programme and Research Network for Biotechnology, Biosafety andBiotechnology Policy Development (BIO-EARN) is supported by the Swedish InternationalDevelopment Cooperation Agency and involves Ethiopia, Kenya, Uganda and the UnitedRepublic of Tanzania. It has a Governing Board and a Programme Advisory Committee thatprovides technical advice to the Board on project proposals, progress and continuing funding.MAS is a priority theme within BIO-EARN, being used to look for resistance markers for plantviruses and fungi (including sweet potato, maize, banana and sorghum) and genotype variationin coffee and banana. The programme has contributed substantially to the improvementof scientific and technical capacities, research infrastructure, equipment and stocking ofconsumables. Connectivity at all BIO-EARN Network institutions has been achieved andthis has not only greatly improved communication among network members, but also accessto information from the Internet to keep abreast with new biotechnology developments inthe world. BIO-EARN Ph.D. students have developed close links with each other throughcommon workshops and annual BIO-EARN student meetings and created a strong basis forfuture regional collaboration.Box 4Biosciences eastern and central Africa hub and networkLocated on the campus of ILRI in Nairobi, Kenya, the Biosciences eastern and central Africa(BecA) hub is the first of four regional networks of centres of excellence in biosciencessponsored by the New Partnership for Africa’s Development (NEPAD). The establishment ofBecA is funded mainly by the Canada Fund for Africa (CFA) of the Canadian InternationalDevelopment Agency (CIDA). It consists of a hub that provides a common biosciencesresearch platform, research-related services and capacity building and training opportunities,and a network of regional nodes and other participating laboratories distributed throughouteastern and central Africa for conducting research on priority problems affecting Africa’sdevelopment. It has a Steering Committee and a Scientific Advisory Committee responsible forthe quality and relevance of the programme. The genomics platform includes state-of-the artequipment for genotyping, DNA sequencing, transcriptomics and bioinformatics, and currentactivities include microsatellite and EST marker development, genetic linkage mapping, MAS,and fingerprinting for distinctness, uniformity and stability and plant variety protection. Itcurrently supports work being conducted on MAS by the International Maize and WheatImprovement Center (CIMMYT), the International Crops Research Institute for the Semi-Arid Tropics (ICRISAT), the International Institute of Tropical Agriculture (IITA), and ILRIand their national partners. Further information is available at www.biosciencesafrica.org.

Chapter 22 – Marker-assisted selection: policy considerations and options for developing countries 463their suitability for local environments,they can be released to producers;• to increase awareness among policyanddecision-makers of the importance ofimproving GRFA through a multidisciplinaryapproach including molecularmethods to their national economies andpoverty reduction strategies. Opportunitiesfor doing so include throughadvocacy within national policy dialogueprocesses and within FAO’s Commissionon Genetic Resources for Food and Agriculture,and by its country representativesand staff in regional and subregionaloffices during the processes of revisingPRSPs and agricultural developmentpolicies; and• subject to increased investment for highpriority activities, training and capacitybuildingin selection and breeding proceduresshould precede the introductionof molecular approaches. Both should beinitiated through close collaboration withinternational, regional and/or nationalcentres.MAS: Other policy considerationsand optionsFew people would question the stark realitiesof doing any kind of R&D in the vastmajority of developing countries and ofgetting the products generated from it tothe rural poor and hungry. ConductingR&D directed towards MAS raises the barconsiderably in terms of its requirementsfor organizational, scientific, technical andlegal skills, as well as for physical infrastructureand financial resources. Funds,however, for public sector agriculturalR&D in all but a handful of developingcountries are becoming ever more scarce.While data on spending and humanresources for modern biotechnologyapplications in agriculture are not available,inflation-adjusted spending on agriculturalR&D as a whole is now growing at muchlower rates than in the 1970s and currentlyruns at around US cents 53 for every US$100of agricultural output (Pardey et al., 2006).In developed countries, public researchfunding actually fell by 6 percent per yearduring the 1990s, but is still running atthe rate of US$2.36 per US$100 worth ofagricultural output. This reflects a strongshift in funding priorities away from publicR&D by both governments and donors.However, the big differences betweenthese groups of countries lie in two broadand interconnected areas. First, in theirlevels of private investment. In developingcountries, this runs at between 8 percent (inAsia and Pacific, but in only a few countries)and 2 percent (in sub-Saharan Africa, with66 percent of that being in South Africa),and by and large is devoted to export cropsand conducted by locally-owned companiesor affiliates of multinationals. In countriesof the Organisation for Economic Cooperationand Development (OECD), suchinvestments now form around 55 percentof their total agricultural R&D spending(Pardey et al., 2006), with 93 percent of thatR&D being performed in these countries.The second difference lies in theorganization/orientation of their research.In developed countries, there is a muchclearer division of labour between thepublic and private sectors. This generallyconforms to the notions of “public goods”and profit/market-oriented R&D, althoughfor MAS this demarcation differs acrosscommodities and is often characterized bypublic–private sector research collaboration.For example, MAS-related R&D activitiesconducted by public sector institutionsare very much oriented towards basic orstrategic research to develop and validatenew knowledge, methods and procedures

464Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishfor variety, strain or breed selectionthrough markers and quantitative genetics,attending to minor species and removingbottlenecks. However, for some crops suchas maize (see, for example, Chapter 8),wheat, soybeans and cotton, and for somelivestock (Narrod and Fuglie, 2000) andaquaculture species, the private sector is asignificant player in both the upstream andapplied molecular biology and quantitativegenetics components.This situation reflects the varying incentivesprovided for private agriculturalresearch by a combination of incomedrivendemand-led market growth for thecommodity(ies) or value chains in question,new technologies, changes in IPR regimes,market structure and the globalization ofagricultural input markets, and public scienceand investment policies that have bothsupported private and undercut publiclyfundedresearch. However, the impulsesprovided by effective demand-led marketgrowth of commodity chains and by policiesto promote private sector investmentin R&D are much weaker in lower incomecountries, and governments that simplylack the cash are left to pick up the total bill.This, in turn, blurs the focus of the R&Dconducted by their NARES which, ratherthan directing resources more towards science-orientedpre-product research (suchas molecular marker development and validationfor selection), attempt, often withinone or two institutes, to cover the wholespectrum from strategic, applied and adaptiveresearch, through to development andon to diffusion of products and services.At the same time, the wider andinterlinked contexts within which the agriculturalsector now operates are increasinglyrequiring ministries, and the research institutesresponsible for agriculture that areunder them, to forward proposals for policies,legislation, programmes and projectsthat are not only sound, convincing andprioritized within and between subsectors,but also aligned with the needs perceivedby other ministries, e.g. health, education,trade and the environment. Critically, inpreparing plans for both domestic anddonor finance, they must provide convincingevidence of engagement with thoserepresenting the interests of agriculturalproducers and other sectors of rural societyincluding women’s groups and the poor,private commercial and non-governmentorganizations (in addition to involving theirown officials and technical experts).In other words, the pressure is real andgrowing both nationally and internationallyfor more “joined up” governance andgreater participatory diagnostic and decision-makingin helping to define, implementand assess the outcomes and impacts ofpublic sector interventions, the expectationbeing that this will focus both minds andfunds on tackling the problems of greatestrelevance to the largest number of poorpeople in rural areas. This raises the issueof how NARES can better ensure that theiragendas, including plans for using modernbiotechnological approaches like MAS (thatclearly requires long-term budgetary support),better meet the needs of the poor.Besides macro and sectoral policiesthat provide appropriate price and marketincentives to agricultural producers andservice providers, developing countrieshave a number of options for creating themore conducive and enabling environmentnecessary for MAS research and the developmentand adoption of the products thatemanate from it.Building political supportThe biggest policy gap in many developingcountries is perhaps the lack of official

Chapter 22 – Marker-assisted selection: policy considerations and options for developing countries 465appreciation of the importance of S&T formeeting their socio-economic objectivesthrough agriculture. Hence, the necessityfor agricultural research and extension institutionsto engage in dialogue during theprocesses of revising PRSPs, comprehensiveagricultural development and relatedpolicies and strategies cannot be underestimated.This promotes learning and capacitydevelopment among and between policyandlaw-makers, technical experts and civilsociety, as well as greater appreciation ofpoverty and its different dimensions andthe trade-offs between different approachesto its amelioration. It results in strongerlinkages between S&T, national povertyreduction and agricultural developmentobjectives, and greater awareness amonghigh-level policy-makers of the contributionsthat S&T can make to achieve theseobjectives. The merits of such engagementalso include greater need-driven prioritysetting,elimination of duplication, moreinformed decision-making both on theroles of the public and private sectors, andpartnership identification. All of this helpsto create more efficient and coordinatedactivities within and between the differentagricultural subsectors and their supportingNARES. Well-conceived studieson the socio-economic impacts of cropsand breeds developed through MAS wouldalso assist decision-making on S&T investmentallocation as this is hardly available inthe literature (FAO, 2005b).Creating S&T policies for drivingstronger priority-setting and betterdelineating roles and responsibilitiesMany developing countries continue towork with outdated, isolated and highlyfragmented NARES, each with their ownset of rules, fiscal arrangements and governmentoversight, and they have weakor non-existent linkages between publicinstitutions, the IARCs and private firms.Some possess many of the essential componentsincluding cutting-edge equipment,but they are not maximizing their potentialto develop MAS capacity. While substantialprogress has been made by a number ofcountries to establish a single frameworkfor managing agricultural R&D, includingmechanisms for setting and evaluating priorities,in most developing countries these arerare. Making a case for MAS, unsupportedby a well thought out (evidence/diagnostic-based)S&T policy framework thatpromotes mutually supportive actions bythe different actors, is a recipe for continuingwith a science and supply-drivenresearch agenda, minimal interaction amongthe different institutions involved, underfundingand tinkering around the edges bysome dedicated individuals.There are, however, a number of optionsopen to governments to reinvigorate theiragricultural S&T systems and make way fornew technologies:• use wider or agriculture-specific S&T legislationto promote enhanced collaborationamong public sector institutions andbetween public and private sector entities,and to establish a national funding agencyand/or agricultural research council thatis independent from any specific ministryand provides competitive grants forresearch and fellowship training withinthe public sector. Thailand, through itsNational Science and Technology DevelopmentAgency established under a S&TDevelopment Act, and Brazil with itsNational Council and National Fund forS&T Development and its Sectoral Funds(Box 5), are two good examples for othercountries to consider following, e.g. someAfrican countries that under NEPADhave committed themselves to increasing

466Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishBox 5Brazil’s sectoral fundsTo promote high-quality research and development in Brazil’s industrial sector, the nationalgovernment established a programme of “sectoral funds” in which a percentage of corporatetaxes are targeted to funding specific research and development objectives. The sectoral fundsprogramme serves four major national objectives:• stability of financial resources for medium- and long-term research and development;• transparency in funding decisions, merit review and evaluation;• reduction of regional inequalities; and• interaction between universities, research institutes and companies.The selection of strategic sectors, their respective shares of the funds’ resources, the blendof basic and applied research, the required overall budget, and sources of support are alljointly decided upon by the indigenous academic community, private sector, and government.No new taxes are involved, just the redirection of already-established government levies. Acomprehensive set of 14 funds has been established. It includes agriculture, biotechnology,informatics and university-industry research.their S&T spending from 0.5 to 1 percentof GDP.• use wider or specific agriculture legislation,tax breaks, public S&T funds andfunds from donors to provide incentivesfor public and private sector involvementin MAS and public–private, civil societycollaboration.Governments and development agencieshave shown increasing interest inpartnerships as a mechanism to promotemarket-driven development. Byerlee andFischer (2001) provide an excellent reviewon the subject of enhancing public–privatepartnerships for transfer of genes andconstructs for GM crops, and some of theprinciples and mechanisms described arealso relevant to MAS. However, outside ofthe Latin American region where Hartwich,Gonzalez and Veira (2005) conducted astudy of 124 cases of such partnerships inagricultural innovation, including a numberdealing with basic and applied researchon plant breeding, the evidence that thesemake research more efficient or “deliver”more or better products to small-scaleproducers is not strong. Indeed, ensuringthat these partnerships comply with publicneeds was found to be one of the majorchallenges in these partnerships.Nevertheless, there are clear signs ofsuch partnerships flourishing for both basicand applied MAS-related R&D activitiesin industrialized countries. With strongand enlightened leadership on both sidescoupled with matching interests, opportunitiesshould also be available for NARESto benefit both from the molecular technologyplatforms and from the selectionand breeding experiences of industrializedcountry-based private sector entities(Chapter 8), as well as from cooperationwith their developing country affiliates,local private companies, NGOs and producerorganizations in taking the productsfrom the laboratory to the field.At the same time, the nature and scopeof IPRs for genetic resources, the research

Chapter 22 – Marker-assisted selection: policy considerations and options for developing countries 467tools used in MAS and the breeding linesand varieties that are developed from itmay be significant barriers to its furtheranceby NARES, private sector entities andpublic–private partnerships. An excellenttreatise of the influence of IPRs for plantbreeding and the seed sector in developingcountries, including the possible implicationsfor MAS, is given by Tripp, Eaton andLouwaars (2006). By providing an empiricalanalysis of IPR developments in China,Colombia, India, Kenya and Uganda, thisreport provides government and institutionalpolicy- and decision-makers withdetails of the different challenges faced bythese countries in developing their IPRregimes, the options they have chosenthrough policies and legislation to developthese, and the lessons learned in implementingthem. It concludes that while IPRregimes in developing countries requireurgent attention, the supporting legislationand regulations should be the product ofopen debate among different stakeholders,and that, even if legislation is already inplace, many countries will find that theyhave sufficient options for interpretationand application to warrant a thoroughreview of procedures and priorities. Interms of hunger and poverty reduction,the importance of segmenting markets intoexport and non-export crops, and intomajor and orphan crops, should not beoverlooked for gaining preferential accessto molecular tools, breeding lines and varieties(Spillane, 2000).• make greater use of regional and bilateralagreements and organizations to fosterinternational collaboration and obtaincomplementary assets for the furtheranceof MAS.Diplomatic level S&T agreements,knowledge exchange networks andresearch consortia (including those of anational and regional nature) can all buildknowledge within and between molecularlaboratories, genetic resources managementprogrammes and organizations involvedin product delivery. The benefits to developingcountries of both the formal andinformal networked world of collaborativeresearch in molecular biology, geneticimprovement and agricultural S&T in generalare potentially enormous, providingsources of funding and making knowledgeeasier to access and researchers and policymakersmore interconnected (safeguardsare needed, however, to minimize disadvantages/risks).Developing countries arenot sufficiently linked to these resources,the CGIAR’s Generation Challenge andthe EU Research Framework Programmesand competitive grants with partnersfrom individual or groups of developedand developing country research institutionsfunded by bilateral donors beingjust some examples. Their governmentsneed to help them do so by providingfunds for building broadband connections,establishing databases and informationsystems, and attending conferences. Also,some functions of a R&D system such asaccessing IP-encumbered technology maybe accessed virtually or even shared withneighbouring countries (Box 6).Creating effective delivery strategiesto bring the products of MAS-relatedR&D to users and beneficiariesThe channels through which the productsof agricultural research reach producershave undergone major structural changesworldwide, and there is now a wide range ofpublic, private and non-government organizationsinvolved in providing extensionservices. At the same time, those responsiblefor funding and supporting R&D havecome to realize that getting technology and

468Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishBox 6Partnerships for technology transfer: the African Agricultural TechnologyFoundationThe mission of the African Agricultural Technology Foundation (AATF) is to promote foodsecurity and reduce poverty. It is a not-for-profit foundation set up in 2002 with the help of theRockefeller Foundation, the United States Agency for International Development (USAID)and the United Kingdom’s Department for International Development (DFID) to identifyopportunities for royalty-free transfers of technologies useful to resource-poor smallholderfarmers in Africa. In pursuing its mission, it negotiates access to technologies, enters intocontractual arrangements to facilitate their deployment and provides stewardship over theirdeployment. It is the responsible party for addressing the concerns about technology ownerswhile protecting the interests of smallholders, handling intellectual property management,regulatory compliance, liability, licensing and freedom-to-operate assessments. In effect itis a “one-stop-shop” for structuring and accessing agricultural technologies and know-how.Among its priorities is genetic improvement of cowpea. This is being tackled through aNetwork for the Genetic Improvement of Cowpea for Africa, involving African and UnitedStates’ universities, IITA, the Kirkhouse Trust, which is a United Kingdom charity, andMonsanto. Part of the project involves developing ready-to-use molecular marker kits forcowpea breeding teams in Africa. The markers are being selected from the cowpea SNPprogramme in IITA and the cowpea genome-sequencing programme, with polymorphismsdetected using agarose gel systems.knowledge to the field does not followa linear and top-down transfer path thatbegins with research, moves on to developmentand production, and ends with thesuccessful introduction of new products orprocesses. Instead, it involves continuousfeedback loops between researchers, extensionagents and farmers within which thedevelopment, fine-tuning and adoption ofthe products of research take place withina specific context. All too often, technologieslie “on the shelf”, lost between researchand its transformation into useful products,because of the lack of understanding of thefunctional linkages between research institutions,extension services and farmers.Policy- and decision-making aboutsupporting agricultural research and technologydevelopment therefore need to shiftaway from traditional and often laboratorybasedresearch and the supply of technologyper se, towards fostering an innovation systemsapproach to understand better theways in which the producers and users oftechnology interact, and thereby identifyand get round the obstacles faced in transformingresearch outputs into developmentoutcomes and impacts. For example, whyhas artificial insemination, a technologycentral for driving improved genetics intomany species of farm animals, been successfulin some countries and localities andnot in others and what are the implicationsof this for applying MAS?Unrealistic expectations of whatagricultural research and agricultural biotechnologyin particular can do to alleviatehunger and poverty have had a negative

Chapter 22 – Marker-assisted selection: policy considerations and options for developing countries 469impact on the fiscal policies of nationalgovernments, financial institutions anddonors for more than 20 years. It will beimportant, therefore, for policy-makers toprovide incentives for getting the facts rightbefore deciding on priorities and investments.They can do so by mandating thatgreater emphasis is placed on research tounderstand better the critical pathwaysinvolved in technical change, including thereasons for the long time frames betweenthe research and extension efforts oftheir NARES and sustainable improvementsin farm productivity through geneticenhancement. This can be achieved interalia by requiring greater accountability,for example, through up-front specificationof the R&D delivery strategy, andthe introduction of monitoring and evaluationprocesses for research outputs andoutcomes that use an innovation systemsapproach to promote information flow, andthrough this, to understand and improvecurrent needs and priority assessments andlevels of customer satisfaction.ReferencesBijker, W.E. 2007. Science and technology policies through policy dialogue. In L. Box & R.Engelhard, eds. Science and technology policy for development: dialogues at the interface. London,Anthem Press (in press).Bragdon, S. 2004. International law of relevance to plant genetic resources: a practical reviewfor scientists and other professionals working with plant genetic resources. Issues in GeneticResources 10. Rome, IPGRI (available at www.bioversityinternational.org/Publications/pubfile.asp?ID_PUB=937)Byerlee, D. & Fischer, K. 2001. Assessing modern science: policy and institutional options for agriculturalbiotechnology in developing countries. IP Strategy Today 1: 1–27 (available at www.biodevelopments.org/ip/ipst1n.pdf).Crawford, I.M. 1997. Rapid rural appraisals, Chapter 8. In Marketing research and information systems.Rome, FAO (available at www.fao.org/docrep/W3241E/w3241e09.htm)De Vicente, M.C. 2004. The evolving role of genebanks in the fast-developing field of moleculargenetics. Issues in Genetic Resources 11. Rome, IPGRI (available at www.bioversityinternational.org/Publications/pubfile.asp?ID_PUB=986).Dixon, J., Gulliver, A. & Gibbon, D. 2001. Farming systems and poverty: improving farmers’ livelihoodsin a changing world. Rome, FAO and Washington, DC, World Bank.Donnenworth, J.M., Grace, J. & Smith, S. 2004. Intellectual property rights, patents, protection andcontracts. IP Strategy Today 9: 19–33 (available at www.biodevelopments.org/ip/ipst9.pdf).Evanson, R.E. & Gollin, D. 2003. Assessing the impact of the green revolution. Science 300: 758–762.FAO. 2005a. The legal framework for the management of animal genetic resources, by A. Ingrassia,D. Manzella and E. Martyniuk. FAO Legislative Study 89. Rome (also available at www.fao.org/docrep/009/a0276e/a0276e00.htm).FAO. 2005b. Annotated bibliography on the economic and socio-economic impact of agricultural biotechnologyin developing countries (available at ftp://ftp.fao.org/sd/SDR/SDRR/bibliography1.pdf).FAO. 2006. The state of food insecurity in the world 2006: eradicating world hunger-taking stock10 years after the World Food Summit (available at www.fao.org/docrep/009/a0750e/a0750e00.htm).

470Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fishHartwich, F., Gonzalez, C. & Viera, L-F. 2005. Public-private partnerships for innovation-ledgrowth in agrichains: a useful tool for development in Latin America? ISNAR Discussion Paper 1.Washington DC, IFPRI (available at www.ifpri.org/divs/isnar/dp/papers/isnardp01.pdf).IAC. 2004. Realizing the promise and potential of African agriculture. Amsterdam, InterAcademyCouncil (available at www.interacademycouncil.net/CMS/Reports/AfricanAgriculture.aspx).IPGRI. 1999. Key questions for decision-makers: protection of plant varieties under the WTOAgreement on Trade Related Aspects of Intellectual Property Rights. Decision tools. Rome (availableat www.bioversityinternational.org/publications/pubfile.asp?ID_PUB=41).Le Buanec, B. 2004. Protection of plant-related innovations: evolution and current discussion. IPStrategy Today 9: 1–17 (available at www.biodevelopments.org/ip/ipst9.pdf).Meinzen-Dick, R., Adato, M., Haddad, L. & Hazell, P. 2004. Science and poverty: an interdisciplinaryassessment of the impact of agricultural research. Washington, DC, IFPRI (available at www.ifpri.org/pubs/fpr/pr16.pdf).Narrod, C.A. & Fuglie, K.O. 2000. Private investment in livestock breeding with implications forpublic research policy. Agribusiness: an International Journal, 16: 457–470.Ndeng’e, G., Opiyo, C., Mistiaen, J. & Kristjanson, P. 2003. Geographic dimensions of well-being inKenya, Volume 1. Nairobi, The Regal Press Kenya Ltd.NEPAD. 2002. Comprehensive Africa Agriculture Development Programme (CAADP) (available atwww.nepad.org/2005/files/progressrepport.php).NEPAD. 2005. Africa’s science and technology consolidated plan of action (available at www.nepadst.org/doclibrary/pdfs/ast_plan_of_action.pdf).OECD. 2006. Promoting pro-poor growth: agriculture. Paris, OECD Publishing (available at www.oecd.org/dataoecd/9/60/37922155.pdf).Pardey, P.G. & Beintema, N.M. 2001. Slow magic: agricultural R&D a century after Mendel.Washington, DC, IFPRI (available at www.ifpri.org/pubs/fpr/fpr31.pdf).Pardey, P.G., Beintema, N., Dehmer, S. & Wood, S. 2006. Agricultural research: A growing globaldivide? Washington, DC, IFPRI (available at www.ifpri.org/pubs/fpr/pr17.pdf).Raitzer, D.A. 2003. Benefit-cost meta-analysis of investment in the international agricultural researchcentres of the CGIAR. FAO, Rome, CGIAR Science Council Secretariat (available at http://impact.cgiar.org).Spillane, C. 2000. Could agricultural biotechnology contribute to poverty alleviation? AgBiotechNet2: 1–72.Spillane, C., Engels, J., Fassil, H., Withers, L. & Cooper, D. 1999. Strengthening national programmesfor plant genetic resources for food and agriculture: planning and coordination. Issuesin Genetic Resources 8. Rome, IPGRI (available at www.bioversityinternational.org/Publications/pubfile.asp?ID_PUB=29).Stannard, C., van der Graaff, N., Randell, A., Lallas, P. & Kenmore, P. 2004. Agricultural biologicaldiversity for food security: shaping international initiatives to help agriculture and the environment.Howard Law J., 48: 397–430.Tripp, R., Eaton, D. & Louwaars, N. 2006. Intellectual property rights; designing regimes to supportplant breeding in developing countries. Washington DC, The World Bank (available at http://siteresources.worldbank.org/INTARD/Resources/IPR_ESW.pdf).

Chapter 22 – Marker-assisted selection: policy considerations and options for developing countries 471UN Millennium Project. 2005a. Investing in development: a practical plan to achieve the MillenniumDevelopment Goals. London and Sterling, VA, USA, Earthscan (available at www.unmillenniumproject.org/reports/fullreport.htm).UN Millennium Project. 2005b. Report of the task force on science, technology and innovation.London and Sterling, VA, USA, Earthscan (available at www.unmillenniumproject.org/documents/Science-complete.pdf).Viljoen, G.J., Nel, L.H. & Crowther, J.R. 2005. Molecular diagnostic PCR handbook. Dordrecht,Netherlands, Springer.

This book provides a comprehensive description andassessment of the use of marker-assisted selection forincreasing the rate of genetic gain in crops, livestock,forestry and farmed fish, including the related policy,organizational and resource considerations. It continuesFAO's tradition of dealing with issues of importance toagricultural and economic development in amultidisciplinary and cross-sectoral manner. As such it ishoped that the information and options presented andthe suggestions made will provide valuable guidance toscientists and breeders in both the public and privatesectors, as well as to government and institutional policyanddecision-makers.ISBN 978-92-5-105717-99 7 8 9 2 5 1 0 5 7 1 7 9TC/M/A1120E/1/5.07/3000

marker-assisted selection in wheat - ictsd

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?