marker-assisted selection in livestock – case studiesused to enhance selection, have raised high...

Section iii

marker-assisted selection in livestock – case studies

Chapter 10

Strategies, limitations and opportunities for marker-assisted

selection in livestock

Jack C.M. Dekkers and Julius H.J. van der Werf

Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fish168

SummaryThis 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 forinclusionofmarkerandQTLinformationingeneticevaluation.Practicalissuesregardingimplementationofmarker-assistedselection(MAS)forselectioninbreedcrossesandforselectionwithinbreedsaredescribed,alongwithlikelyroutestowardsachievingthatgoal.Opportunitiesandchallengesarealsodiscussedfortheuseofmolecularinformationforgeneticimprovementoflivestockindevelopingcountries.

Chapter 10 – Strategies, limitations and opportunities for marker-assisted selection in livestock 169

introduCtionSincethe1970s,thediscoveryoftechnologythat enables identification and genotypingof large numbers of genetic markers, andresearch that demonstrated how this tech-nology could be used to identify genomicregionsthatcontrolvariationinquantitativetraitsandhowtheresultingQTLcouldbeusedtoenhanceselection,haveraisedhighexpectations for the application of gene-(GAS) or marker-assisted selection (MAS)inlivestock.Yet,todate,theapplicationofGASorMASinlivestockhasbeenlimited(see e.g. review by Dekkers, 2004 and thecasestudychaptersthatfollow).However,recent further advances in technology,combined with a substantial reduction inthe cost of genotyping, have stimulatedrenewedinterest inthe large-scaleapplica-tionofMASinlivestock.

Successful application of MAS inbreedingprogrammesrequiresadvancesinthefollowingfiveareas:• Gene mapping: identification and map-

ping of genes and genetic polymor-phisms.

• Marker genotyping: genotyping of largenumbersofindividualsforlargenumbersofmarkersatareasonablecost forbothQTL detection and routine applicationforMAS.

• QTL detection:detectionandestimationof associations of identified genes andgeneticmarkerswitheconomictraits.

• Genetic evaluation: integration of phe-notypic and genotypic data in statisticalmethods to estimate breeding values ofindividualsinabreedingpopulation.

• MAS:developmentofbreedingstrategiesandprogrammesfortheuseofmoleculargeneticinformationinselectionandmat-ingprogrammes.Thischapteroutlinesthemainstrategies

fortheapplicationofMASinlivestockand

identifiesanddiscussesthelimitationsandopportunities for successfulMAS in com-mercialbreedingprogrammes.ItconcludesbydiscussinglimitationsandopportunitiesforapplyingMASindevelopingcountries.

markerS and linkage diSequiliBriumOverthepastdecades,asubstantialnumberofalternatetypesofgeneticmarkershavebecomeavailabletostudythegeneticarchitectureoftraits and for their use in MAS, includingrestriction fragment length polymorphisms(RFLPs),microsatellites, amplified fragmentlength polymorphisms (AFLPs) and singlenucleotidepolymorphisms (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.Linkagedisequilibriumrelatestodependenceof alleles at different loci and is centralto both QTL detection and MAS. Thus, athorough understanding of LD and of thefactorsthataffectthepresenceandextentofLDinpopulationsisessentialforadiscussionofbothQTLdetectionandMAS.

linkage disequilibriumConsideramarkerlocuswithallelesMandmandaQTLwithallelesQandq that ison the same chromosome as the marker,i.e.themarkerandtheQTLarelinked.Anindividual that is heterozygous for bothloci would have genotype MmQq. Allelesat the two loci are arranged in haplotypes onthetwochromosomesofahomologous


pair that each individual carries. An indi-vidual with genotype MmQq could havethe following two haplotypes: MQ/mq,wherethe/separatesthetwohomologouschromosomes.Alternatively,itcouldcarrythe haplotypes Mq/mQ. This alternativearrangement of linked alleles on homolo-gous chromosomes is referred to as themarker-QTL linkage phase. The arrange-ment of alleles in haplotypes is importantbecause progeny inherit one of the twohaplotypes that a parent carries, barringrecombination.

The presence of linkage equilibrium(LE)ordisequilibriumrelatestotherelativefrequencies of alternative haplotypes inthe population. In a population that is inlinkage equilibrium, alleles at two loci arerandomlyassortedintohaplotypes.Inotherwords, chromosomes or haplotypes thatcarry marker allele M are no more likelytocarryQTLalleleQ thanchromosomesthat carry marker allele m. In technicalterms,thefrequencyoftheMQhaplotypesis 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 genotypebecauseitprovidesnoinformationonQTLgenotype. If the marker and QTL arein linkage disequilibrium, however, therewill be a difference in the probability ofcarrying Q between chromosomes thatcarryMandmmarkerallelesand,therefore,a difference in mean phenotype betweenmarkergenotypeswouldalsobeexpected.

The main factors that create LD in apopulation are mutation, selection, drift(inbreeding),andmigrationorcrossing.SeeGoddard and Meuwissen (2005) for fur-therbackgroundonthesetopics.ThemainfactorthatbreaksdownLDisrecombina-tion, which can rearrange haplotypes that

exist within a parent in every generation.Figure1showstheeffectofrecombination(r) on the decay of LD over generations.The rate of decay depends on the rate ofrecombinationbetweentheloci.Fortightlylinked loci, any LD that has been createdwillpersistovermanygenerationsbut,forlooselylinkedloci(r>0.1),LDwilldeclinerapidlyovergenerations.

population-wide versus within-family ldAlthoughamarkerandalinkedQTLmaybe in LE across the population, LD willalwaysexistwithinafamily,evenbetweenlooselylinkedloci.Consideradoublehet-erozygous sire with haplotypes MQ/mq(Figure 2). The genotype of this sire isidentical to that of an F1 cross betweeninbred lines. This sire will produce fourtypes of gametes: non-recombinants MQand mq and recombinants Mq and mQ.Asnon-recombinantswillhavehigherfre-quency, 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 (Figure1), becauseit has undergone only one generation ofrecombination. This specific type of LD,

FiGURe 1Break-up of ld over generations

Q

q

M

m

r

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0

r=.001

r=.01

r=.5

r=.2

r=.1

r=.05

LD is continuously eroded by recombination

Dis

equ

ilib

riu

m

Generation5 10 15 20 25


however, only exists within this family;progenyfromanothersire,e.g.anMq/mQsire, will also show LD, but the LD is inthe opposite direction because of the dif-ferent marker-QTL linkage phase in thesire (Figure2). On the other hand, MQ/mQ and Mq/mq sire families will not bein LD because the QTL does not segre-gate inthesefamilies.WhenpooledacrossfamiliesthesefourtypesofLDwillcanceleachotherout,resultinginlinkageequilib-rium across the population. Nevertheless,thewithin-familyLDcanbeusedtodetectQTLandforMASprovidedthedifferencesinlinkagephasearetakenintoaccount,aswillbedemonstratedlater.

qtl deteCtion and typeS of markerS for maSApplication of molecular genetics forgenetic improvement relies on the abilitytogenotypeindividualsforspecificgeneticloci. For these purposes, three types ofobservablegeneticlocicanbedistinguished,asdescribedbyDekkers,2004:• direct markers: loci for which the func-

tionalpolymorphismcanbegenotyped;• LD-markers:lociinpopulation-wideLD

withthefunctionalmutation;• LE-markers: loci in population-wide

linkage equilibrium with the functionalmutationbutwhichcanbeusedforQTLdetection and MAS based on within-familyLD.Forthesealternatetypesofmarkers,dif-

ferent strategies are appropriate to detectQTL in livestock populations. These aresummarizedinTable1andwillbedescribedinmoredetail.StrategiesforQTLdetectioninlivestockdifferfromthoseusedinplantsbecauseofthelackofinbredlines.

qtl detection using ld markers within crossesCrossingtwobreedsthatdifferinalleleand,therefore, haplotype frequencies, createsextensive LD in the crossbred popula-tion.ThisLDextendsover largedistances

FiGURe 2within-family marker-qtl ld

Within-family LD among progeny of a sire

Sire 1

Q

q

M

m

Sire 2

q

Q

M

m

Sire 3 Sire 4

Q

Q

M

m

q

q

M

m

Linkage phase can differ between families

table 1Summary of strategies for qtl detection in livestock

type of population within crosses outbred population

f2/Backcross advanced intercross

half- or full-sib families

extended pedigree

non-pedigreed population sample

type of markers lD markers le markers lD markersGenome coverage Genome-wide Genome-wide candidate gene

regionsGenome-wide

Marker density Sparse Denser Sparse More dense Few loci Densetype of lD used Population-wide lD Within-family lD Population-wide lDnumber of generations of recombination used for mapping 1 >1 1 >1 >>1

extent of lD around Qtl long Smaller long Smaller SmallMap resolution Poor better Poor better High


becauseithasundergoneonlyonegenera-tionofrecombinationintheF2(Figure1).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 to20cM).Thisapproachhasformedthebasisfor the extensive use of F2 or backcrossesbetweenbreedsorlinesforQTLdetection,inparticularinpigs,poultryandbeefcattle(see Andersson, 2001 for a review). Theextensive LD enables detection of QTLthat are some distance from the markersbut also limits the accuracy (map resolu-tion)withwhich thepositionof theQTLcanbedetermined.

More extensive population-wide LD isalso expected to exist in synthetic lines,i.e. lines thatwerecreated fromacross inrecent history. These can be set up on anexperimentalbasisthroughadvancedinter-cross lines (Darvasi and Soller, 1995) orbe available as commercial breeding lines.Depending on the number of generationssincethecross,theextentofLDwillhaveerodedovergenerationsandwill,therefore,span shorter distances than in F2 popula-tions (Figure 1). This will require a moredensemarkermaptoscanthegenomewithequivalentpowerasinanF2butwillenablemoreprecisepositioningoftheQTL.

qtl detection using le markers in outbred populationsAs linkage phases between the markerandQTLcandifferfromfamilytofamily,use of within-family LD for QTL detec-tionrequiresQTLeffectstobefittedona

within-familybasis, rather thanacross thepopulation. Similar to F2 or backcrosses,theextentofwithin-familyLDisextensiveand, thus, genome-wide coverage is pro-videdbya limitednumberofmarkersbutsignificant markers may be some distancefromtheQTL,resultinginpoormapreso-lution. Thus, LE markers can be readilydetected on a genome-wide basis usinglargehalf-sibfamilies,requiringonlysparsemarkermaps(~15to20cMspacing).Manyexamplesof successful applicationsof thismethodologyfordetectionofQTLregionsare available in the literature, inparticularfordairycattle,utilizingthelargepaternalhalf-sibstructuresthatareavailablethroughextensiveuseofartificialinsemination(seeWeller,Chapter12).

QTL detection using LE markers canalso be applied to extended pedigrees bymodelling the co-segregation of markersandQTL(FernandoandGrossman,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 genotypesthatareincludedintheanalysis,mapreso-lution will be better than with analysis ofhalf-sibfamiliesbecausemultipleroundsofrecombinationareincludedinthedataset.

qtl detection using ld markers in outbred populationsThe amount and extent of LD that existsinthepopulationsthatareusedforgeneticimprovementarethenetresultofallforcesthat create and break down LD and are,therefore, the result of the breeding andselectionhistoryofeachpopulation,alongwithrandomsampling.Onthisbasis,pop-ulations that have been closed for manygenerations are expected to be in linkageequilibrium,except forclosely linked loci.


Thus, in those populations, only markersthataretightlylinkedtoQTLmayshowanassociationwithphenotype(Figure1),andeventhenthereisnoguaranteebecauseofthechanceeffectsofrandomsampling.

Therearetwostrategiestofindmarkersthatareinpopulation-wideLDwithQTL(seeTable1):• evaluating markers that are in, or close

to,genes thatare thought tobeassoci-atedwiththetraitofinterest(candidategenes);

• a genome scan using a high-densitymarkermap,withamarkerevery0.5to2cM.The success of both approaches obvi-

ouslydependson the extentofLD in thepopulation. Studies in human populationshavegenerallyfoundthatLDextendsoverless than 1 cM. Thus, many markers areneededtoobtainsufficientmarkercoveragein human populations to enable detectionof QTL based on population-wide LD.Opportunities to utilize population-wideLDtodetectQTLinlivestockpopulationsmaybeconsiderablygreaterbecauseoftheeffectsofselectionandinbreeding.Indeed,Farnir et al. (2000) identified substantialLD in the Dutch Holstein population,whichextendedover5cM.Similar resultshavebeenobservedinotherlivestockspe-cies (e.g. in poultry, Heifetz et al., 2005).The presence of extensive LD in live-stockpopulationsisadvantageousforQTLdetection, but disadvantageous for iden-tifying the causative mutations of theseQTL;withextensiveLD,markersthataresomedistancefromthecausativemutationcanshowanassociationwithphenotype.

The candidate gene approach utilizesknowledge from species that are rich ingenome information (e.g. human, mouse),effects of mutations in other species, pre-viously identified QTL regions, and/or

knowledge of the physiological basis oftraits, to identify genes that are thoughtto play a role in the physiology of thetrait. Following mapping and identifica-tion of polymorphisms within the gene,associations of genotype at the candidategene with phenotype can be estimated(RothschildandPlastow,1999).

Whereas the candidate gene approachfocusesonLDwithinchosenregionsofthegenome, recent advances in genome tech-nology have enabled sequencing of entiregenomes,includingofseverallivestockspe-cies;thegenomesofthechickenandcattlehavebeensequencedandpublicsequencingof 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 thesequenceoftheRedJungleFowlwiththatofthreedomesticatedbreeds(InternationalChickenPolymorphismMapConsortium,2004).This,combinedwithreducingcostsof genotyping, now enables detection ofQTLusingLD-mappingwithhigh-densitymarkermaps.

qtl detection using combined ld and linkage analysis in outbred populationsAs markers may not be in complete LDwith the QTL, both population-wideassociationsofmarkerswithQTLandco-segregation of markers and QTL withinfamiliescanbeusedtodetectQTL.Usingthese combined properties of being bothLD and LE markers, methods have beendeveloped to combine LD and linkageinformation. These methods are furtherexplored under genetic evaluation modelsinwhatfollows.


inCorporating marker information in genetiC evaluation programmeSThe value of genotypic information forpredicting the genetic merit of animals isdependent on the predictive ability of themarkergenotypes.Thethreetypesofmolec-ularlocidescribedpreviouslydiffernotonlyinmethodsofdetectionbutalsoinmethodsof their incorporation in genetic evalua-tion procedures. Whereas direct and, to alesser degree, LD markers, allow selectionon genotype across the population, use ofLEmarkersmustallowfordifferentlinkagephases between markers and QTL fromfamilytofamily,i.e.LEmarkersarefamilyspecificandfamilyspecificinformationmustbederived.Asdiscussedlaterinthischapter,this makes LE markers a lot less attractivefor use in breeding programmes. In thissection, the different types of models thathave been proposed for genetic evaluationbasedonmarker informationaredescribedandthisisfollowedbyabriefdescriptionofsomepracticalissuesregardingimplementa-tionof suchmethods and the likely routestowardsachievingthatgoal.

modelling qtl effects in genetic evaluationByusingQTLinformationingeneticeval-uation, in principle, part of the assumedpolygenicvariationissubstitutedbyasep-arateeffectduetoageneticpolymorphismataknownlocus. Thishastheimmediateeffect of having a much better handle onthe Mendelian sampling process, as phe-notypicco-variancecanbeevaluatedbasedon specific genetic similarity rather thanon an average relationship. For example,on average two full sibs share 50percentof their alleles, but at a specific locus it isnow possible to know whether these fullsibscarryexactlythesamecompletegeno-type(bothpaternalandmaternalallelesare

incommon),oractuallyhaveacompletelydifferent genotype. The actual degree ofsimilarity of full sibs at a QTL can thusvarybetween0and1.Thisadditionalinfor-mationhelpstobetterevaluatethegeneticmerit due to specific QTL, and to betterpredictoffspringthatdonotyethavephe-notypicmeasurements.

Anumberofdifferentapproacheshavebeen described to accommodate markerinformationingeneticevaluation.Roughly,these methods can be distinguishedthroughtheirmodellingoftheQTLeffectand through the type of genetic markerinformation used. The QTL effect can bemodelled as random or fixed, while themolecularinformationcomesfromLE,LDordirectmarkers.

WithafixedQTLmodel,regressionongenotype probabilities would be used ingeneticevaluationtoaccountfortheeffectof QTL polymorphisms. In the simplestadditive QTL model, suitable for esti-matingbreedingvalues,simpleregressionscouldbeincludedontheprobabilityofcar-rying the favourable mutation. Regressioncan be on known genotypes (class vari-ables), or probabilities can be derived forungenotypedanimalsinageneralcomplexpedigree (Kinghorn, 1999). A fixed QTLmodel is sensible if few alleles are knownto be segregating, and where dominanceand/or epistasis are important. The modelalsoassumeseffectsbeing the sameacrossfamilies. The effects of various genotypescouldbefittedseparately,givingpowertoaccountfordominanceandepistasisincaseof multiple QTL. For selection purposes,a fixed QTL effect, if additive, would beaddedtothepolygenicestimatedbreedingvalues (EBVs), similar to breed effects inacross-breedevaluations.TheadvantageofafixedQTLmodelisthelimitednumberofeffectsthatneedtobefitted.


Alternatively, 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 poly-genic effects. With full knowledge aboutsegregation, this would effectively fit allfounder alleles as different effects. Therandom QTL model was first describedbyFernandoandGrossman(1989),wherefor each animal both the paternal and thematernalallelewerefitted.Withoutlossofinformation, theseeffectscanbecollapsedinto one genotypic effect for each animal(Pong-Wong et al., 2001). The randomQTL model makes no assumptions aboutnumberofallelesataQTLanditautomati-cally accommodates possible interactioneffects of QTL with genetic background(families or lines). Therefore, the randomQTL model is less reliant on assumptionsabout homogeneity of QTL effects. TherandomQTLmodelisanaturalextensiontotheusualmixedmodelandseemsthere-forealogicalwaytoincorporategenotypeinformationintoanoverallgeneticevalua-tion system. These models result in EBVsfor QTL effects along with a polygenicEBV. The total EBV is the simple sum oftheseestimates.Oneofthemaincomputa-tionallimitationsofthismethod,however,isthelargenumberofequationsthatmustbe solved, which increases by two peranimal for each QTL that is fitted. Thus,the number of QTL regions that can beincorporatedislimited.

Genetic evaluation using direct markersWhenthegenotypeofanactualfunctionalmutation is available, no pedigree infor-mation is needed to predict the genotypiceffect, as QTL genotypes are measured

directly.Whenthereisonlyasmallnumberofalleles,thenumberofspecificgenotypesis 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.Suchassumptionsmightbereasonableforabi-allelicQTLmodelinarelativelyhomo-geneouspopulation.Alternatively,randomQTLmodels couldbeusedwithdifferenteffectsfordifferentfounderalleles,orevenQTLbyenvironmentinteractions.Inbothfixed and random QTL models, genotypeprobabilitiescanbederivedforindividualswithmissinggenotypes.

Genetic evaluation using LE markersWhenthegenotypetestisnotforthegeneitself,butfora linkedmarker,QTLprob-abilities derived from marker genotypeswillbeaffectedby the recombination ratebetween marker and QTL and by theextentofLDbetweentheQTLandmarkeracross the population. If LD betweenthe QTL and a linked marker only existswithinfamilies,markereffectsor,atamin-imum,marker-QTLlinkagephasemustbedeterminedseparatelyforeachfamily.Thisrequiresmarkergenotypesandphenotypeson family members. If linkage betweenthe marker and QTL is loose, phenotypicrecords must be from close relatives oftheselectioncandidatebecauseassociationswillerodequicklythroughrecombination.With progeny data, marker-QTL effectsorlinkagephasescanbedeterminedbasedon simple statistical tests that contrast themeanphenotypeofprogenythatinheritedalternate marker alleles from the commonparent.AmorecomprehensiveapproachisbasedonFernandoandGrossman’s(1989)randomQTLmodel,wheremarker infor-mationfromcomplexpedigreescanbeused


toderiveco-variancesbetweenQTLeffects,yielding best linear unbiased prediction(BLUP) of breeding value for both poly-genic andQTLeffects.RandomeffectsofpaternalandmaternalQTLallelesareaddedtothestandardanimalmodelwithrandompolygenic breeding values. The variance-co-variance structure of the random QTLeffects,alsoknownasthegameticrelation-shipmatrix (GRM), isbasedonprobabilitiesof identity by descent (IBD), and is nowderivedfromco-segregationofmarkersandQTLwithinafamily.ProbabilitiesofIBDderivedfrompedigreeandmarkerdatalinkQTL allele effects that are expected to beequalorsimilar, thereforeusingdata fromrelatives to estimate an individual’s QTLeffects. For example, if two paternal half-sibsiandjhaveinheritedthesamepaternalalleleformarkersthatflanktheQTL(withrecombination rate r), theyare likely IBDfor thepaternalQTLalleleand thecorre-lationbetweentheeffectsoftheirpaternalQTL alleles will be (1-r)2. The method isappealing,butcomputationallydemandingforlarge-scaleevaluations,especiallywhennotallanimalsaregenotypedandcomplexproceduresmustbeappliedtoderiveIBDprobabilities.

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 geno-type across families, and is in a sense notvery different from a direct marker. Themost convenient way to include geno-typicinformationfrommarkerhaplotypesin genetic evaluation systems is through

the random QTL model. In their orig-inalpaper,FernandoandGrossman(1989)derivedIBDfromgenotypedataonsinglemarkers and recombination rates betweenmarker and QTL. However, the randomQTLmodelismoreversatile,andco-vari-ances based on IBD probabilities can alsouseinformationbeyondpedigree,basedonLD.Thelattercanbederivedfrommarkeror haplotype similarity, e.g. based on anumber of marker genotypes surroundingaputativeQTL.MeuwissenandGoddard(2001)proposedusingbothlinkageandLDinformation to derive IBD-based co-vari-ances (termedLDLanalysis).Leeandvander Werf (2005) showed that with densermarkers, thevalueof linkage information,and therefore pedigree, reduces. Hence,when QTL positions become more accu-rately defined, genetic information fromclose markers (within a few cM) can beusedincreasinglytoderiveLD-basedIBDprobabilities,therebydefiningco-variancesbetween randomQTLeffectswithout theneedfora familystructureor informationthroughpedigree.

LeeandvanderWerf(2006)haveshownthatLDinformationresultsinaverydenseGRM.Geneticevaluation,whichisusuallybased on mixed model equations that arerelatively sparse, is currently not feasiblecomputationally for the LDL method fora largenumberof individuals andalterna-tive models are needed. One approach isto model population-wide LD by simplyincluding the marker genotype or haplo-type as a fixed effect in the animal modelevaluation, as suggested by Fernando(2004). An advantage of modelling popu-lation-wideLDeffectsasfixedratherthanrandom is that fewer assumptions aboutpopulationhistoryareneeded.Adisadvan-tage is that estimates are not “BLUPed”,i.e.regressedtowardsameandependingon


theamountofinformationthatisavailabletoestimatetheireffects.Thiswillbeimpor-tant if someof thegenotypeorhaplotypeeffectscannotbeestimatedwithsubstantialaccuracybecausethenumberofindividualswiththatgenotypeorhaplotypeislimited.Haplotype effects could also be fitted asrandom, but more development is neededinthisarea.

Whole genome approach for genetic evaluation using high-density LD markersWithmoreandmoreQTLbeingdiscovered,the polygenic component will slowly bereplacedbymultipleQTLeffects,theinher-itance of each being followed by markerbracketsormoregenerallybyinformationonhaplotypes.Nejati-Javaremi,SmithandGibson(1997)presentedtheconceptofthetotalallelicrelationship,wheretheco-vari-ancebetween two individualswasderivedfromallelicidentitybydescent,orbystate(based on molecular marker information),witheachlocationweightedbythevarianceexplained by that region. This approachcontrasts with the average relationshipsderived from pedigree that are used inthenumerator relationshipmatrix.Nejati-Javaremi,SmithandGibson(1997)showedthatusingtotalallelicrelationshipresultedin a higher selection response than pedi-gree based relationships, because it moreaccuratelyaccountsforthevariationintheadditivegeneticrelationshipsbetweenindi-viduals. Therefore, the gain of followinginheritance at specific genome locationscontributestomoreaccurategeneticevalu-ation,and isable todealmorespecificallywith within and between loci interactionsand with specific modes of inheritance atdifferentQTL.

When large-scale marker genotypingbecomes cheap and available to breedersat low cost, this approach could even be

used for non-detected QTL and geneticevaluation could be based on a “wholegenome approach” (Meuwissen, Hayesand Goddard, 2001). In this approach,markerhaplotypesarefittedasindependentrandomeffectsforeach,e.g.1cMregionofthe genome. In the work by Meuwissen,HayesandGoddard(2001),variancesasso-ciated with each haplotype were eitherassumed to be equal for each chromo-somal region or estimated from the datausing Bayesian procedures with alternateprior distributions. In essence, this pro-cedure estimates breeding values for eachhaplotype, and EBVs of individuals arecomputed by simply summing EBVs forthehaplotypesthattheycontain.

Using this procedure, Meuwissen,Hayes and Goddard (2001) demonstratedthrough simulation, that for populationswithaneffectivepopulationsizeof100andaspacingof1or2cMbetweeninformativemarkers across the genome, sufficient LDwas present to predict genetic values withsubstantialaccuracyforseveralgenerationsbasedonassociationsofmarkerhaplotypeswith phenotype on as few as 500 individ-uals.Itshouldbenotedthat,intheapproachproposed by these authors, no polygeniceffect is included since all regions of thegenomeareincludedinthemodel.Itmay,however, be useful to include a polygeniceffect because LD between markers andQTLwillnotbe complete for all regions.In addition, this model assumes that hap-lotype effects are independent within andacross regions. Incorporating IBD prob-abilities to model co-variances betweenhaplotypeswithinaregionasinMeuwissenandGoddard(2000),andbyincorporatingco-variances between adjacent regionscausedbyLDbetweenregions,couldleadto further improvements but would alsoleadtoincreasingcomputationaldemands.


Ingeneral,forthepurposeofincreasedgenetic change of economically impor-tant quantitative 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 predic-tive value about genetic merit. Moreover,the benefit from the extra investment andtime spent on finding functional muta-tions might be superseded by the geneticchange that can be made in the breedingprogrammeinthemeantime.

Implementation of marker-assisted genetic evaluationIt is important to note that, for most ofthe gene marker tests currently on themarket, integration with existing systemsforgeneticevaluation isnotobvious.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 gen-otypes for their breeding animals. Thisinformation on individual genotype willbecome less relevant if more gene testsbecome available and if testing becomescheaper and more widespread. This mightstilltakesomeyears.Thus,asgenemarkertesting is gradually introduced, it is morelikely to create additional selection cri-teria to consider and it will take sometimebeforeQTLinformationisseamlesslyandoptimallyintegratedinexistinggeneticevaluation programmes. In particular, ifgeneticevaluation isbasedon informationfrom many different breeding units, suchas in cattle or sheep, genotyping informa-tion will initially be available for only a

small proportion of the breeding animals,possibly not justifying a total overhaul ofthe system for genetic evaluation. Simplead hoc procedures where QTL effects areestimatedandpresentedseparatelyasaddi-tional effects are initially a more likelyroutetoimplementation.

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 nopolygeniccomponent.

inCorporating maS in SeleCtion programmeSMolecular information can be used toenhance both the processes of integratingsuperior qualities of different breeds andwithin-breedselection.Thesestrategiesarefurtherdescribedbelow.

Between-breed selectionCrossing breeds results in extensive LD,whichcanbecapitalizeduponusingMASin a number of ways. If a large propor-tion of breed differences in the trait(s) ofinterestareduetoasmallnumberofgenes,gene introgression strategies can be used.If a larger number of genes is involved,MASwithinasyntheticlineisthepreferredmethodofimprovement.

Marker-assisted introgression Introgression of the desirable allele at atargetgenefromadonortoarecipientbreedis accomplished by multiple backcrossesto the recipient, followed by one or moregenerations of intercrossing. The aim ofthe backcross generations is to produce


individualsthatcarryonecopyofthedonorQTLallelebutthataresimilartotherecipientbreedfortherestofthegenome.Theaimofthe intercrossing phase is to fix the donorallele at theQTL.Marker information canenhancetheeffectivenessofthebackcrossingphaseofgeneintrogressionstrategiesby:(i)identifying carriers of the target gene(s)(foreground selection); and (ii) enhancingrecoveryoftherecipientgeneticbackground(backgroundselection).Theeffectivenessoftheintercrossingphasecanalsobeenhancedthrough foreground selection on thetarget gene(s). If the target gene cannot begenotyped directly, carrier individuals canbe identified based on markers that flanktheQTLat<10cM,becauseoftheextensiveLD in crosses. The markers must havebreed-specific alleles in order to identifylineorigin.Fortheintrogressionofmultipletargetgenes,genepyramidingstrategiescanbe used during the backcrossing phase toreduce the number of individuals required(HospitalandCharcosset,1997;Koudandéet al., 2000). For background selection,markers are used that are spread over thegenomeat<20cMintervals,suchthatmostgenes thataffect the traitwillbewithin10cMfromamarker.Combining foregroundandbackgroundselection,selectionwillbefor the donor breed segment around thetargetlocusbutforrecipientbreedsegmentsin the rest of the genome. Foregroundselectionwillresultinselectionforboththetarget locus and for donor breed loci thatarelinkedtothislocus,someofwhichcouldhaveanunfavourableeffectonperformance.Toreducethisso-calledlinkagedragaroundthetargetlocus,inthemolecularscoreusedfor background selection greater emphasiscan be given to markers that are in theneighbourhood of the target locus (apartfrom the flankingmarkers,which areusedinforegroundselection).

Most studies have considered marker-assistedintrogression(MAI)ofsingleQTL(e.g. Hospital and Charcosset, 1997) butoften several QTL must be introgressedsimultaneously. Koudandé et al. (2000)showed that large populations are neededtoobtainsufficientindividualsthatarehet-erozygousforallQTLinthebackcrossingphase.ThiswouldmakeMAInot feasibleinlivestockbreedingprogrammes.Inmanycases,however,immediatefixationofintro-gressed QTL alleles may not be required.Instead, the objective of the backcrossingphase canbe to enrich the recipientbreedwith the favourable donor QTL alleles atsufficiently high frequency for selectionfollowing backcrossing. The effectivenessof such strategies was demonstrated byChaiwonget al.(2002).

Marker-assisted improvement of synthetic linesIn MAI studies it is usually assumed thatthe aim is to recover the recipient breedgenotype, except for the donor QTL. Analternativeobjectivecouldbetoaimsimplyforindividualswithhighestmerit.SelectionwouldthenbeforQTLgenotypeaswellasEBV,estimatedacrossbreedsorlines.ThisEBV selection would replace backgroundselection,asrecoveryoftherecipientgen-otype is achieved through selection ongeneticmeritratherthanthroughselectingfor breed of origin. This strategy wouldbemorecompetitive if theoriginalbreedsoverlapinmerit,andindeed,aswasshownbyDominiket al.(2007),backgroundselec-tion based on anonymous markers wouldbelessprofitable.

Strategies for using markers to selectwithinahybridpopulationwerefirstpro-posed by Lande and Thompson (1990).These strategies capitalize on population-wide LD that initially exists in crosses


betweenlinesorbreeds.Thus,marker-QTLassociationsidentifiedintheF2generationcanbeselectedforseveralgenerations,untilthe QTL or markers are fixed or the dis-equilibrium disappears. Zhang and Smith(1992)evaluatedtheuseofmarkersinsucha situation with selection on BLUP EBV.Althoughbothstudiesconsideredtheidealsituationofacrosswithinbredlines,therewill be opportunities to utilize a limitednumber of markers to select for favour-able QTL regions that are detected incrossesbetweenbreeds,therebyenhancingthe 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 substantiallygreaterresponsestoselectionthanselectiononidentifiedQTLregionsonly.DuetothemuchgreaterLD,wholegenomeselectioninacrosscanbeaccomplishedwithamuchsmallernumberofmarkerscomparedwiththe number required for whole genomeselectioninanoutbredpopulation.

within-breed selectionThe procedures described previously forincorporatingmarkersingeneticevaluationresultinestimatesofbreedingvaluesasso-ciatedforQTL,togetherwithestimatesofpolygenicbreedingvalues.Alternatively,ifmolecular data are not incorporated intogenetic evaluations, as will be the casefor more ad hoc approaches and for genetests for Mendelian characteristics, sepa-rate selectioncriteriawillbeavailable thatcapturethemolecularinformation.Thefol-lowingthreeselectionstrategiescanthenbedistinguished(Dekkers,2004):• selectontheQTLinformationalone;

• tandemselection,withselectiononQTLfollowedbyselectiononpolygenicEBV;

• selection on the sum of the QTL andpolygenicEBV.Selection on QTL or marker infor-

mation alone ignores information that isavailable on all other genes (polygenes)thataffectthetraitandisexpectedtoresultin 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 esti-mate marker effects, and can be attractivewhen phenotype is difficult or expensiveto record (e.g.disease traits, meat quality,etc.).SelectiononthesumoftheQTLandpolygenicEBVisexpectedtoresultinmax-imumresponseintheshortterm,butmaybe suboptimal in the longer term becauseof losses in polygenic response (Gibson,1994).IndexesofQTLandpolygenicEBVcan be derived that maximize longer-termresponse (Dekkers and van Arendonk,1998)oracombinationofshort-andlonger-termresponses(DekkersandChakraborty,2001). However, if selection is on mul-tipleQTLandemphasis isonmaximizingshorter-termresponse,selectiononthesumofQTLandpolygenicEBVisexpectedtobe close to optimal. Optimizing selectionon a number of EBVs, indexes and geno-types, while also considering inbreedingrate and other practical considerations isnotatrivialtask.Kinghorn,MeszarosandVagg (2002) have proposed a mate selec-tionapproachthatcouldbeusedtohandlesuchproblems,anditcanbeexpectedthatwith more widespread use of genotypicinformationforalargernumberofregions,specific knowledge about individual QTLbecomes less interesting and will simplycontribute topredictionofwholeEBVorwholegenotype.


Meuwissen and Goddard (1996) pub-lishedasimulationstudythatlookedatthemaincharacteristicsdeterminingefficiencyof MAS using LE markers. They foundthatMAScouldimprovetherateofgeneticimprovementupto64percentbyselectingon the sum of QTL and polygenic EBV.TheirworkalsodemonstratedthatMASismainly useful for traits where phenotypicmeasurement is less valuable because of:(i)low heritability; (ii) sex-limited expres-sion; (iii) availability only after sexualmaturity;and(iv)necessitytosacrificetheanimal (e.g. slaughter traits). Selection ofanimals based on (most probable) QTLgenotypewillallowearlierandmoreaccu-rate selection, increasing the short- andmedium-termselectionresponse.

Most simulation studies have assumedcompletemarkergenotypeinformationbutinpracticeonlya limitednumberof indi-viduals will 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,HenshallandvanderWerf(2002)lookedatstrategiestominimizegenotypingcostinasheepbreedingprogramme.Closeto maximal gain could be achieved whengenotyping was undertaken only for highranking males and animals whose markergenotypeprobabilitycouldnotbederivedwith enough certainty based on informa-tion on relatives. Marshall, van der WerfandHenshall(2004)alsolookedatprogenytesting of sires to determine family-spe-cificmarker-QTLphasewithinabreedingnucleus.Again,testingofalimitednumberofmalesprovidedalotofinformationaboutphase for several generations of breeding

animals, as progeny tested sires have rela-tionships with descendants. However, inbreeding programmes for more extensiveproductionsystems(beef,sheep),pedigreerecording is often incomplete and only asmallproportionofanimalsaregenotyped.Moreover,thesegenotypedanimalsarenotnecessarily the key breeding animals. Theutilityoflinkedmarkerswillbeevenmorelimited if pedigree relationships cannotbe used to resolve genotype probabilitiesand marker-QTL phase of un-genotypedindividuals.

Asecondpointofcautionisthatmanystudies on MAS have taken a single-traitapproach and shown that genetic markerscouldhavealargeimpactonresponsesfortraits thataredifficult to improvebyphe-notypic selection. However, within thecontext of a multitrait breeding objective,theoverall impactof suchmarkerson thebreedinggoalmaybelessbecauseagreaterresponse forone traitoftenappearsat theexpense of another. For example, geneticmarkersforcarcasstraitsimprovetheabilityto select (i.e. earlier, with higher accu-racy)forsuchtraits,butselectionemphasisfor other traits is reduced. Therefore, theoveralleffectofMASonthebreedingpro-gramme will generally be much smallerthanpredictedforsingletraitMAS-favour-ablecases.ThemaineffectsofMASwouldbetoshifttheselectionresponseinfavourof themarkedtraits, rather thanachievingmuch additional overall response. Hence,while it will be easier to select for carcassand disease resistance, further improve-mentforthesetraitswillbeattheexpenseof genetic change for production traits(growth,milk).

The impact of MAS on the rate ofgenetic gain may be limited in conven-tional breeding programmes (ranging upto perhaps 10percent extra gain) unless


the variation in profitability is domi-nated by traits that are hard to measure.However, new technologies often lead toother breeding programme designs beingcloser to optimal. Genotypic informationhasextravalueinthecaseofearlyselectionand where within-family variance can beexploited,which isparticularly thecase inprogrammeswherereproductivetechnolo-gies are used. Reproductive technologiesusually lead to early selection and moreemphasis on between-family selection.DNAmarkertechnologyandreproductivetechnologies are therefore highly syner-gistic and complementary (van der WerfandMarshall,2005)andgenemarkershavemuch more value in such programmes.Gene marker information is also clearlyvaluable in introgression programmes, asdemonstratedbysimulation(Chaiwonget al.,2002;Dominiket al.,2006)aswellasinpractice (Nimbkar, Pardeshi and Ghalsasi,2005). Yet, although these examples arefavourable to the value of gene markerinformation, the added value of MAS stillreliesheavilyonahighdegreeoftraitandpedigreerecording.

opportunitieS for maS in developing CountrieSComplete phenotypic and pedigree infor-mation is often only available in intensivebreedingunits.Therefore,inthecontextoflowinputproductionsystems,someques-tionscanberaisedconcerningthevalidityand practicality of the simulation studiesdescribed above, and it would be moredifficult to realize the value of markerinformation.Itwouldbeharderandmoreexpensivetodeterminethelinkagephaseinthecaseofusinglinkedmarkers.Moreover,

evenifthegeneticmarkerwereadirectorLDmarker,itseffectonphenotypewouldhave to be estimated for the populationand the environment in which it is used.Thiswouldrequirephenotypesandgeno-typesonasampleofaratherhomogeneouspopulation to avoid spurious associationsthat could result from unknown popula-tionstratification.Therefore,agenemarkerfor a QTL is likely to be most successfulinanenvironmentwith intensivepedigreeand performance recording. Nevertheless,in low input environments, direct andLD markers will be more useful than LEmarkers because the latter require routinerecordingofphenotypesandgenotypestoestimateQTLeffectswithinfamilies.

InadditiontoMASwithinlocalbreeds,severalotherstrategiesforbreedimprove-ment could be pursued in developingcountries,includinggeneintrogressionandMAS within synthetic breeds. This wouldbemostadvantageousforintroducingspe-cific disease resistance alleles into breedswith improved production characteristicstomakethemmoretoleranttotheenviron-mentsencounteredindevelopingcountries.Geneintrogressionis,however,alongandexpensive process and only worthwhilefor genes with large effects. MAS withinsyntheticbreeds,e.g.acrossbetweenlocaland improved temperate climate breeds,can allow development of a breed that isbasedonthebestofbothbreeds(e.g.ZhangandSmith,1992).BecauseoftheextensiveLD within the cross, a limited number ofmarkers would be needed. Care should,however, be taken to avoid the impactof genotype x environment interactions ifMASis implemented inamorecontrolledenvironment.


referenCeSAndersson, L.2001.Geneticdissectionofphenotypicdiversityinfarmanimals.Nature Revs. Genet.

2:130–138.Chaiwong, N., Dekkers, J.C.M., Fernando, R.L. & Rothschild, M.F. 2002.Introgressingmultiple

QTLinbackcrossbreedingprogramsof limitedsize.Proc. 7th Wld. Congr. Genet. Appl. Livest. Prodn.ElectronicCommunicationNo.22:08.Montpellier,France.

Darvasi, A. & Soller, M.1995.Advancedintercrosslines:anexperimentalpopulationforfinegeneticmapping.Genetics141:1199–1207.

Dekkers, J.C.M. & van Arendonk, J.A.M. 1998.Optimizing selection forquantitative traitswithinformationonanidentifiedlocusinoutbredpopulations,Genet. Res.71:257–275.

Dekkers, J.C.M.2004.Commercialapplicationofmarker-andgene-assistedselection in livestock:strategiesandlessons.J. Anim. Sci.82:E313–E328.

Dekkers J.C.M. & Chakraborty, R.2001.Potentialgainfromoptimizingmulti-generationselectiononanidentifiedquantitativetraitlocus.J. Anim. Sci.79:2975–2990.

Dominik, S., Henshall, J., O’Grady, J. & Marshall, K.J.2007.Factorsinfluencingtheefficiencyofamarkerassistedintrogressionprograminmerinosheep.Genet. Sel. Evol.Inpress.

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.Extensivegenome-widelinkagedisequilibriumincattle.Genome Res.10:220–227.

Fernando, R.L. 2004.Incorporatingmolecularmarkersintogeneticevaluation.SessionG6.1.Proc. 55th Meeting of the European Association of Animal Production. 5–9 September 2004, Bled,Slovenia.

Fernando, R.L. & Grossman, M.1989.Marker-assistedselectionusingbestlinearunbiasedpredic-tion.Genet. Sel. Evol.21:467–477.

Gibson, J.P.1994.Short-termgainattheexpenseoflong-termresponsewithselectionofidentifiedloci,Proc. 5th Wld. Cong. Genet. Appl. Livest. Prodn. CD-ROMCommunicationNo.21:201–204.UniversityofGuelph,Canada.

Goddard, M.E. & Meuwissen, T.H.E.2005.Theuseoflinkagedisequilibriumtomapquantitativetraitloci.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.Extentandconsistencyacrossgenerationsoflinkagedisequilibriumincommerciallayerchickenbreedingpopulations.Genetics 171:1173–1181.

Hospital, F. & Charcosset, A.1997.Marker-assistedintrogressionofquantitativetraitloci.Genetics 147:1469–1485.

International Chicken Polymorphism Map Consortium.2004.Ageneticvariationmapforchickenwith2.8millionsinglenucleotidepolymorphisms.Nature432:717–722.

Kinghorn, B.P.1999.Useofsegregationanalysistoreducegenotypingcosts.J. Anim. Breed. Genet.116:175–180.

Kinghorn, B.P., Meszaros, S.A. & Vagg, R.D. 2002. Dynamic tactical decision systems foranimalbreeding,Proc. 7th Wld. Congr. Genet. Appl. Livest. Prodn.CommunicationNo.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.


Lande, R. & Thompson, R. 1990. Efficiency of marker-assisted selection in the improvement ofquantitativetraits.Genetics124:743–56.

Lee, S.H. & van der Werf, J.H.J. 2005.Theroleofpedigreeinformationincombinedlinkagedis-equilibriumandlinkagemappingofquantitativetraitlociinageneralcomplexpedigree.Genetics 169:455–466.

Lee, S.H. & van der Werf, J.H.J.2006.AnefficientvariancecomponentapproachimplementinganaverageinformationREMLsuitableforcombinedLDandlinkagemappingwithageneralcomplexpedigree.Genet. Sel. Evol.38:25–43.

Marshall, K., Henshall, J. & van der Werf, J.H.J. 2002.Response frommarker-assisted selectionwhenvariousproportionsofanimalsaremarkertyped:amultipletraitsimulationstudyrelevanttothesheepmeatindustry.Anim. Sci.74:223–232.

Marshall, K.J., van der Werf J.H.J. & Henshall, J.2004.Exploringmajorgene-markerphasetypingstrategiesinmarkerassistedselectionschemes.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.Finemappingofquantitativetraitlociusinglinkagedis-equilibriawithcloselylinkedmarkerloci.Genetics155:421–430.

Meuwissen, T.H.E. & Goddard, M.E. 2001. Prediction of identity by descent probabilities frommarkerhaplotypes.Genet. Sel. Evol.33:605–634.

Meuwissen, T.H.E., Hayes, B. & Goddard, M.E. 2001. Prediction of total genetic value usinggenome-widedensemarkermaps.Genetics157:1819–1829.

Nejati-Javaremi, A., Smith, C. & Gibson, J.P.1997.Effectoftotalallelicrelationshiponaccuracyandresponsetoselection.J. Anim. Sci.75:1738–1745.

Nimbkar, C., Pardeshi, V. & Ghalsasi, P. 2005.EvaluationoftheutilityoftheFecBgenetoimprovetheproductivityofDeccanisheep inMaharashtra,India.pp.145–154.InH.P.S.Makkar&G.J.Viljoen,eds.Applications of gene-based technologies for improving animal production and health in developing countries.Netherlands,Springer.

Pong-Wong, R., George, A.W., Woolliams, J.A. & Haley, C.S.2001.Asimpleandrapidmethodforcalculatingidentity-by-descentmatricesusingmultiplemarkers.Genet. Sel. Evol.33:453–471.

Pyasatian, N., Fernando, R.L. & Dekkers, J.C.M.2006.Genomicselectionforcompositelinedevel-opment using low density marker maps. In Proc. 8th Wrld. Congr. Genet. Appl. Livest. Prodn.,Paper22–65.BeloHorizonte,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.Combininggene-basedmethodsandreproductivetech-nologies to enhance genetic improvement of livestock in developing countries, pp. 131–144. InH.P.SMakkar&G.J.Viljoen,eds.Applications of gene-based technologies for improving animal production and health in developing countries.Netherlands,Springer.

Zhang, W. & Smith, C.1992.Computersimulationofmarker-assistedselectionutilizinglinkagedis-equilibrium.Theor. Appl. Genet.83:813–820.

Chapter 11

marker-assisted selection in poultry

Dirk-Jan de Koning and Paul M. Hocking


SummaryAmonglivestockspecies,chickenhasthemostextensivegenomicstoolboxavailablefordetectionofquantitativetraitloci(QTL)andmarker-assistedselection(MAS).TheuptakeofMAS is thereforenot limitedbytechnical resourcesbutmostlybytheprioritiesandfinancialconstraintsofthefewremainingpoultrybreedingcompanies.Withthecostofgenotypingdecreasingrapidly,anincreaseintheuseofdirecttrait-singlenucleotidepoly-morphism(SNP)-associationsinMAScanbepredicted.

Chapter 11 – Marker-assisted selection in poultry 187

Current StatuS of ChiCken Breeding programmeS Poultry production has been the fastestgrowing livestock industry over the lastdecades especially in middle- and low-income countries (Taha, 2003). In 2001,poultry production accounted for 70mil-liontonnesofpoultrymeatand47milliontonnes of eggs (Arthur and Albers, 2003).Amongpoultry,chickenaccountfor85per-centofmeatproductionand96percentofegg production (Bilgili, 2001; Arthur andAlbers, 2003; Taha, 2003). While chickenshave been domesticated and selectedfor thousands of years, modern poultrybreeding startedduring the1950s.Oneofthe most notable features is the diversi-fication between chickens bred for meatproduction (broilers) and those bred fortable egg production (layers). This is aresultofthenegativegeneticcorrelationinchickenbetweengrowthand reproductivetraits. Within breeds, there is a separationintomaleandfemalelinesthatarecrossedtoproducecommercialhybrids.Inbroilers,malelinesareselectedforgrowthandcar-cass quality whereas in female lines lessemphasis is placed on growth and moreonreproductivetraitssuchaseggproduc-tion and hatchability. In table egg-layingchickens,malelinesareselectedforhigheggproductionandhigheggweightwhereasinfemale lines selection may emphasize rateof lay with less attention to egg size. Inboth broiler and layer lines the primaryselection goal is the improvement of feedefficiencyandeconomicgain.

Significantheterosis for fitness traits inpoultryiswellestablishedandallcommercialpoultry (chickens, turkeys and ducks) arehybridsthatareproducedinaselectionandmultiplicationpyramidthatisillustratedinFigure 1. Crossing male and female linesmaximizes heterosis at the grandparent

and parent levels of the hierarchy, andallows traits that have been geneticallyimprovedindifferentlinestobecombinedinthecommercialbirds.Thepowerofthisstructuretodeliverlargeeconomicgainsinchickensisaresultoftheirhighreproductiverate and short generation interval and isclearly illustrated by this example of anegg-layingimprovementprogramme.Evengreater numerical efficiency is possiblein broilers: a single pen containing tenfemales and one male at the nucleus levelmight produce 150great-grandparentsafter selection (lineD of Figure 1); thesewill produce 50 female offspring each or7500grandparents in a year and thesegrandparents will generate 375000femaleparent stock during the succeeding year.These hybrid parent females will eachproduceover130maleandfemaleoffspringandgeneratenearly50millioncommercialbroilers or 70000tonnes 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 largenumbersofcommerciallayers.

The existence of this breeding struc-tureresultsinrapidtransmissionofgeneticchange to commercial flocks (about fouryears), including traits that might beimproved by MAS. Conversely, undesir-ablegeneticchangecanalsobedisseminatedvery quickly to a very large number ofbirds. In practice, far more birds are keptatthenucleuslevelthanshowninFigure1wherethenumberspresentedarepurelyforillustrativepurposes.

StatuS of funCtional genomiCS in ChiCkenAmongthevariouslivestockspecies,chickenhasthemostcomprehensivegenomictool-


box. The chicken genome consists of 39pairs of chromosomes: eight cytologi-cally distinct macrochromosomes, the sexchromosomes Z and W and 30pairs ofcytologically indistinguishable microchro-mosomes. Linkage maps were developedinitially using three separate mappingpopulations (Bumstead and Palyga, 1992;Crittenden et al., 1993;Groenenet al., 1998)that were later merged to provide a con-sensus map with 1 889 markers (Groenenet al., 2000).Agoodoverviewofthecon-sensus linkage map and the cytogeneticmap can be found in the First Report of Chicken Genes and Chromosomes2000andits 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).ThechickengenomesequencecanbebrowsedviaanumberofWebsites,whichare summarized at www.chicken-genome.org/resources/databases.html.Thegenomesequenceeffortwasaccompaniedbypartialsequencingofthreedistinctpoultrybreeds(abroiler, a layer andaChineseSilky), toidentify SNPs between and among theseand the reference sequence of the RedJungleFowl.ThisresultedinanSNPmapconsistingofabout2.8millionSNPs(Wonget al., 2004). The chicken polymorphismdatabase (ChickVD) can be browsed at:http://chicken.genomics.org.cn/index.jsp

FiGURe 1multiplication pyramid for a four-strain commercial hybrid layer

Male lines Female lines Generation Time

B 10 x 100

A 1 x 10

C 10 x 100

D 100 x 1 000

A 200

B 2 000

C 2 000

D 20 000

Pure line selection

Grandparents Parents

X X

AB 40 000

CD 400 000

ABCD 32.2 million

Commercial hybrids

X

Year 0 Year 1 Year 2 Years 3 to 4

♂ ♂

♂ ♂

♂

♂ ♂

♀

♀♀

♀

♀♀

♀

adapted from bowman, 1974.

the 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 numbers in lines b, c and D, that are required to generate hybrid laying hens at the commercial level and the numbers that result from this process. the time scale is indicated on the right hand side of the figure assuming a generation interval of one year.


(Wang et al., 2005). The SNP map willfacilitatethedevelopmentofgenome-wideSNPassays,containingbetween5000and20000SNPsperassay.

Forthestudyofgeneexpression, therearevariouscomplementaryDNA(cDNA)microarrays available, varying from tar-geted arrays (immune, neuroendocrine,embryo) to whole genome generic arrays.Recently, a whole-genome Affymetrixchip was developed in collaboration withthe chicken genomics community (www.affymetrix.comandwww.chicken-genome.org/resources/affymetrix-faq1.htm).Altogether, this provides a very compre-hensive toolbox to study the functionalgenomics of chicken, whether this be anindividualgeneortheentiregenome.

Current uptake of maS in ChiCken Implementation of MAS requires knowl-edge of 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,meatquality,eggproduction,dis-ease resistance (both infectious diseasesand production diseases) and behaviour.Thesestudieshaverecentlybeenreviewed

(Hocking, 2005). A total of 27 papersreported114genome-widesignificantQTLfromexperimentalcrosseslargelyinvolvingWhite Leghorn and broiler lines. A sum-maryoftheQTLthathavebeendetectedispresentedinTable1.Whiletheabundanceof QTL would indicate ample opportu-nityforMASinchicken,itmustbenotedthat nearly all studies were carried out inexperimental crossesandhence the resultsdo not reflect QTL within selected popu-lations.However, these resultsdoprovidea good starting point to search for QTLwithincommercialpopulations,asdemon-stratedforgrowthandcarcasstraitswheremanypublishedQTLalsoexplainedvaria-tion within a broiler dam line (de Koninget al., 2003;deKoninget al., 2004).Totheauthors’ knowledge, there are no otherQTL studies within commercial lines ofpoultryinthepublicdomain.OftheQTLfrom experimental crosses, only a smallnumberhasbeenfollowedupbyfinemap-ping analyses and the responsible genemutationhasonlybeendescribedforsomediseaseresistanceQTL(Liuet al., 2001a,b;Liuet al., 2003).

A good example of how QTL map-pingcombinedwithfunctionalstudiescanidentify functional variants is for Marek’sdisease. Marek’s disease (MD) is an infec-

table 1quantitative traits and chromosomal locations in experimental chicken crosses

trait Chromosome (number of qtl) total qtl number of papers

behaviour/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 2

Source: Hocking, 2005.


tious viral disease caused by a memberof the herpes virus family and costs thepoultry industry about US$1 000 millionper annum. An F2 cross between resistantandsusceptiblelineswaschallengedexperi-mentallyandgenotyped,providingthedataforaQTLanalysis that resulted ina totalof seven QTL for susceptibility to MD(Vallejo et al., 1998; Yonash et al., 1999).Subsequently, the founder lines of the F2cross were used for a micro-array studyto identify genes that were differentiallyexpressedbetweenthetwolinesfollowingartificial infection. Fifteen of these geneswere mapped onto the chicken genomeandtwoofthemmappedtoaQTLregionforresistancetoMD(Liuet al., 2001a).Atthe same time, protein interaction studiesbetween a viral protein (SORF2) and achicken splenic cDNA library revealed aninteraction with the chicken growth hor-mone (GH) (Liu et al., 2001b). This ledtothedetectionofapolymorphismintheGH gene that was associated with differ-ences in the number of tumours betweenthesusceptibleandtheresistantline(Liuet al., 2001b).GHcoincidedwithaQTLforresistance and was differentially expressedbetweenfounderlines(Liuet al., 2001a).

AlongsidethevariousgenomescansforQTL, a large number of candidate genestudieshavebeencarriedout.Themajorityof studies summarized in Table 2 havebeenconductedonWhiteLeghornstrainsand have utilized restriction fragmentlength polymorphisms (RFLPs), SNPs orsinglestrandconformationpolymorphisms(SSCPs). These techniques require boththatthegeneisknownandthattheexperi-menterisabletosequencepartofthegeneto detect polymorphisms that distinguishtheexperimentallines.

Candidate gene studies have been usedin two ways. First, candidate genes may

be used merely as a marker for a trait(typically disease) based on prior knowl-edge and, second, and much less often,to search for the mutation within a genethatisassociatedwithphenotypicvariationin 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.locatinggenesthatareinthelocationoftheQTLbasedoncommonareasofthegene-rich genomes of different species, usuallyhumanandmouse).Thereare likely tobemanymoreofthesecondtypeofcandidategenestudiesasinformationfromlarge-scalegeneexpressionandproteomicexperimentsbegintosuggestnovelgenecandidatesfortraitsofcommercialandbiological impor-tance. It should also be noted that thereis good evidence that genetic variation isnotlimitedtogenomicDNA:associationsbetween polymorphisms in mitochondrialgenesandMDresistance,bodyweightandeggshellqualitywerereportedbyLiet al. (1998a,b).

Despite great enthusiasm for breedingcompanies to be involved in functionalgenomics research in poultry, there arevery few applications of MAS in com-mercial poultry breeding. One existingexampleistheuseofbloodgroupmarkersto improve resistance toMDwhere selec-tion of haplotypes B21 and B12 based onconventional serological tests has beenwidely used (McKay, 1998). In discus-sionswiththeindustryitisclearthatmostinterest is in QTL or candidate genes forresistancetodiseases likeMDorascites,agenetic condition associated with pulmo-nary hypertension, leading to mortalityin fast growing birds. There is also con-siderable interest among breeders of layerlines for egg quality, especially egg shellqualitybecauseof its importance for food


safety.Forproductiontraitssuchasgrowthand egg numbers, breeders make suffi-cient progress using traditional selectionmethods, and they expect little improve-ment from MAS for such traits unlessmarkerscanbeusedtoincreasetheaccuracyof selection. Nonetheless, among breedersofbroilerstockthereisinterestinmarkersfortraitsthataredifficulttomeasuresuchasfeedefficiencyandmeatqualityinaddi-tiontodiseaseresistance.

potential for maS in ChiCkenThe technical aspects and potentialimplications of implementing MAS inlivestock are discussed in Chapter 10and Dekkers (2004), and van der Beekand vanArendonk (1996) evaluated thetechnical aspects of MAS in poultrybreeding.AreviewofthepotentialofMAS

inpoultry isprovidedbyMuir (2003)butthis includes many of the technical issuesthat are common across livestock species.Thischapterthereforefocusesonpoultry-specific issues, and readers are referredto Chapter 10 or Muir (2003) for a morecomprehensive overview of applicationsandlimitationsofMAS.

Muir (2003) identified twocaseswhereMAScouldincreasetheselectionintensityinpoultrybreeding:(i)traitsthataremeas-ured later in life or are costly to measure(suchaseggproductionandfeedefficiencyfor broiler breeders); and (ii)selectionwithin full-sib families for sex-limitedtraits(e.g.malechicksforeggproduction).Accuracyofselectioncanalsobeimprovedvia MAS when selecting between full-sibfamiliesforsex-limitedtraitsandtraitsthatcannotbemeasureddirectlyononeorboth

table 2association of candidate genes with quantitative traits in poultry

trait Chromosomes1 gene symbols references

age at first egg 1,2,3 GH, nPY, oDc Feng et al., 1997; Dunn et al., 2004; Parsanejad et al., 2004

Disease resistance (E. coli) 16 MHc1, MHc4, taP2 Yonash et al., 1999

Disease resistance (MD2) 1,nK GH, lY6e Kuhnlein et al., 1997; liu et al., 2001a,b and 2003

Disease resistance (Sal3) 4,6,7,16,19, 1,17,nK

tnc, PSaP, nRaMP14, MHc1,caSP1, iaP1, tlR4, tlR5

Hu et al., 1997; lamont et al., 2002; leveque et al. 2003; liu and lamont, 2003; iqbal et al., 2005

Double yolked eggs 10 GnRHR Dunn et al., 2004

egg production Z,1,20 GHR, GH, PePcK Feng et al., 1997; Kuhnlein et al., 1997; Parsanejad et al., 2003

egg weight 1 iGF1 nagaraja et al., 2000

eggshell quality 1,3,20 iGF1, oDc, PePcK nagaraja et al., 2000; Parsanejad et al., 2003, 2004

body fat 1,1,5,Z GH, iGF1, tGFβ3, GHR Feng et al., 1998; Fotouhi et al., 1993; li et al., 2003; Zhou et al., 2005

Feed efficiency 3,20 oDc, PePcK Parsanejad et al., 2003 and 2004

body weight/carcass quality 1,3,5,Z,1,1,1 iGF1, oDc, tGFβ3, GHR, aPoa2, Pit1

Feng et al., 1998; li et al., 2003; Jiang et al., 2004; Parsanejad et al., 2004; li et al., 2005; Zhou et al., 2005

organ weight (spleen) 3,5,32 tGFβ2, tGFβ3, tGFβ45 li et al., 2003

Skeletal traits 1,3,5,32 iGF1, tGFβ2, tGFβ3, tGFβ45 li et al., 2003; Zhou et al., 2005

1 nK = gene has not yet been assigned to a chromosome.2 Marek’s Disease.3 Salmonellosis.4 now known as Slc11a1.5 tGFβ4 in the paper is now known to be tGFβ1.


sexes and/or have a low heritability (e.g.egg production, disease resistance, carcassqualityandwelfaretraits).

Limiting factors for application ofMAS (Muir, 2003) include biological fac-tors (reproductive capacity) and manytheoretical considerations related to theeffectiveness of MAS (e.g. diverting selec-tion pressure from polygenes to a singlemarked gene), which are generally appli-cabletoMASin livestock(Dekkers,2004;Chapter10).OneoftheconcernsofMuir(2003) is the expected lack of major QTLfor traits that have been under selectionfor many generations (following simula-tionresults).However,recentQTLstudieswithin commercial lines of pigs (Evanset al., 2003; Nagamine et al., 2003) andpoultry(deKoninget al., 2003,2004)havedemonstratedthatmanysizeableQTLarestillsegregatingincommercialpopulationsdespitedecadesofselection.

There is strong academic interest inchickengenomicsoutsideagriculturefrom,amongothers,developmentalbiologistsandevolutionary geneticists, and this has con-tributedgreatlytothedevelopmentof thecurrentfunctionalgenomicstoolboxavail-ableforchicken.Amonglivestockspecies,chickens are best placed to pioneer newapproaches where QTL studies are com-plementedbygeneexpressionstudies(Liuet al., 2001a) or where they become fullyintegratedwithin“geneticalgenomics”(deKoning, Carlborg and Haley, 2005; deKoningandHaley,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 arethreelevelsofMAS:gene-assistedselection

(GAS) where the functional mutation anditseffectsareknown;linkagedisequilibriumMAS (LD-MAS) where a marker (ormarker haplotypes) is in population-widedisequilibrium with a QTL; and linkageequilibriumMAS(LE-MAS)wheremarkersare in Hardy-Weinberg equilibrium withtheQTLatthepopulationlevel,butlinkagedisequilibrium 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 withouttargetinganyindividualQTLormeasuring(expensive)phenotypesoneverygeneration(Meuwissen, Hayes, and Goddard, 2001).Integrating current evaluations with MASismost straightforward forGASandLD-MASbecausetheQTLeffectcanbeincludedin routine evaluations as a fixed effect(Chapter10).LE-MAS,ontheotherhand,requires extensive genotyping and fairlycomplicated statistical procedures (Wang,FernandoandGrossman,1998),whileGW-MASreducesthegenometoa“black-box”but does not require selection of QTLusing arbitrary thresholds. Furthermore,the dense marker information requiredfor GW-MAS may dispense with oftenfaultypedigreerecordsbecauseallpedigreeinformationisencodedinthegenome-widegenotypes.

In termsofquantitativegenetic theory,there are ongoing developments in thetoolsrequiredtodetectandevaluateQTLin arbitrary pedigrees, moving away fromstrictly additive-dominance models toepistasis and parent-of-origin effects (Liu,Jansen and Lin, 2002; Shete and Amos,2002).Atthesametime,thetechnologytoanalysemorethan10000SNPsinasingleassay is available, and a cost of as little asUS$0.02pergenotypeislikelyforchicken


SNPsinthenearfuture.However,thefinemappingandcharacterizationof identifiedQTL remain costly and time-consumingprocesses and are often restricted to themostpromisingQTL,resultinginhundredsof QTL that will never make it past thestage of mapping to a 30 cM confidenceinterval.

WhilecurrentresearchanddevelopmentsinpoultryfunctionalgenomicsarerelevanttoallfourpossibleapplicationsofMAStolivestock, poultry breeders need to decideatwhatleveltheywanttoexploitmolecularinformationandforwhichtraits.

The emerging picture is that breedersare more comfortable with known genemutationsasthisprovidesaneasyroutetoimplementationaswellasknowledgeabouttheunderlyingbiology.Furthermore,thereis concern that the marker-trait linkagewillbreakdownoverarelativelyfewgen-erations of selection in large commercialflocks.Whilecandidategenestudieswouldprovidethequickestroutetoimplementa-tion, finemappingandcharacterizationofQTL (e.g. using expression studies) mayreveal gene variants that are not obviouscandidategenesforquantitativetraits.

potential for maS in poultry in developing 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 poultrybreedingindustryandallpoultrybreedingis privately owned. Fifty years ago therewere many primary breeders in each andevery industrialized country, but not solongagotherewereonly20breedingcom-panies worldwide. Today, three groups ofprimary breeders dominate the interna-tionallayermarket.Equally,inthechicken

meatindustry,therearefourmajorplayersin broiler breeding worldwide (Flockand Preisinger, 2002). The concentrationprocessisprobablynowcomplete,andthepresent players are sufficient to meet theglobal supply for 700000million eggs asfinalproducts.Asimilar trend isexpectedin the pig industry, where internationalbreedingcompaniesofhybridproductsareincreasing 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 pres-sure. With increasing poultry productionin developing countries, breeding compa-nies may give priority to using breedingandmoleculartools toaddresstheseaddi-tional challenges. While chickens are veryefficient in converting grain into valuablemeat and egg protein, and smallholderchicken production can be valuable forsustainingthelivelihoodsoffarmersinthedevelopingworld,thistypeofpoultrypro-ductionwouldrequirerobustdual-purpose(meat and egg) birds, rather than special-ized broiler and layer lines. It is unlikelythat the commercial breeders will developsuch lines but there may be scope fornational or international research organi-zations to do so. Any MAS would havetobedoneat the institutional levelwherethelineisdevelopedandwouldnecessitateprior knowledge of trait-marker associa-tionsatthefarmlevel.TheimplementationofwholegenomeSNPapproachestofarmlevel recordingmight facilitateprogress inthisareabut thechallenges,bothpracticalandtheoretical,areformidable.

ConCluding remarkSAmong livestock species, chicken haveby far the most comprehensive genomictoolbox. However, uptake of MAS will


depend strongly on whether the industrywishes to supplement its current selectionprogramme with a known gene variantor whether it is prepared to restructurebreeding programmes around MAS.Comparedwith,forinstance,thedairycattleindustry, thepoultrybreedingcommunitymaybeslower toembraceemergingcom-plexapproachestoMAS.Thisissomewhatsurprising because the closed structure ofthepoultrybreedingpyramidoffersmuchbetter protection of intellectual propertythanthedairycattleindustrywheresemenfrom highest ranking bulls is available forall.Ontheotherhand,thefactthatbloodgroupshavebeenusedtoselect forresist-ancetoMDsuggeststhatpoultrybreedershave some experience and skills in thistype of selection. Poultry breeding com-panies contribute 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 currentknowledgeofthegenesormarkersassoci-atedwiththesetraits,andtheirassociationwithothereconomicselectioncriteria.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 reservationsmustbetheveryevidentsuccessofcurrentbreeding programmes in achieving manydesirablecommercialgoals.

aCknowledgementSThe Roslin Institute is supported by a corestrategic grant from the Biotechnologyand Biological Sciences Research Council(BBSRC).Theauthorsgratefullyacknowledgeinformationonbreedingobjectivesfromrep-resentativesofthepoultrybreedingindustry.

referenCeSArthur, J.A. & Albers, G.A.A.2003.Industrialperspectiveonproblemsandissuesassociatedwith

poultrybreeding.InW.M.Muir&S.E.Aggrey,eds.Poultry genetics, breeding and biotechnology, pp.1–12.Wallingford,UK,CABIPublishing.

Bilgili, S.F. 2001.Poultryproductsandprocessingintheinternationalmarketplace.Proc. Internat. Anim. Agric. & Food Sci. Conf.Indianapolis,IN,USA(availableatwww.fass.org/fass01/pdfs/Bilgili.pdf).

Bowman, J.C.1974.An introduction to animal breeding.TheInstituteofBiology’sStudiesinBiologyNo46.London,EdwardArnold.

Bumstead, N. & Palyga, J.1992.Apreliminarylinkagemapofthechickengenome.Genomics13:690–697.

Crittenden, L.B., Provencher, L., Sanatangello, L., Levin, I., Abplanalp, H., Briles, R.W., Briles, W.E. & Dodgson, J.B.1993.CharacterizationofaRedJungleFowlbyWhiteLeghornbackcrossreferencepopulationformolecularmappingofthechickengenome.Poultry Sci.72:334–348.

de Koning, D.J., Carlborg, O. & Haley. C.S. 2005.Thegeneticdissectionofimmuneresponseusinggene-expressionstudiesandgenomemapping.Vet. Immun. & Immunopath.105:343–352.

de Koning, D.J. & Haley, C.S. 2005.Geneticalgenomics inhumansandmodelorganisms.Trends Genetics21:377–381.

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.


de Koning, D.J., Windsor, D., Hocking, P.M., Burt, D.W., Law, A., Haley, C.S., Morris, A., Vincent, J. & Griffin, H. 2003.Quantitativetraitlocusdetectionincommercialbroilerlinesusingcandidateregions.J. Anim. Sci.81:1158–1165.

Dekkers, J.C.M.2004.Commercialapplicationofmarker-andgene-assistedselection in livestock:strategiesandlessons.J. Anim. Sc.82E-Suppl:E313–E328.

Dunn, I.C., Miao, Y.W., Morris, A., Romanov, M.N., Wilson, P.W. & Waddington, D.2004.Astudyofassociationbetweengeneticmarkersincandidategenesandreproductivetraitsinonegenerationofacommercialbroilerbreederhenpopulation.Heredity92:128–134.

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.Identificationofquantitativetraitlociforproductiontraitsincommercialpigpopulations.Genetics164:621–627.

Feng, X.P., Kuhnlein, U., Aggrey, S.E., Gavora, J.S. & Zadworny, D. 1997. Trait association ofgeneticmarkersinthegrowthhormoneandthegrowthhormonereceptorgeneinaWhiteLeghornstrain.Poultry Science76:1770–1775.

Feng, X.P., Kuhnlein, U., Fairfull, R.W., Aggrey, S.E., Yao, J. & Zadworny, D. 1998. A geneticmarkerinthegrowthhormonereceptorgeneassociatedwithbodyweightinchickens.J. Heredity89:355–359.

Flock, D.K. & Preisinger, R. 2002.Breedingplansforpoultrywithemphasisonsustainability.7th World Congress on Genetics Applied to Livestock Production, 19–23 August, 2002, Montpellier,France,Session24:Sustainablebreedingplansindevelopedcountries.Communication24-02.

Fotouhi, N., Karatzas, C.N., Kuhnlein, U. & Zadworny, D. 1993. Identificationof growth-hor-mone DNA 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.Aconsensuslinkagemapofthechickengenome.Genome Research10:137–147.

Groenen, M.A., Crooijmans, R.P., Veenendaal, A., Cheng, H.H., Siwek, M. & van der Poel, J.J.1998.Acomprehensivemicrosatellitelinkagemapofthechickengenome.Genomics49:265–274.

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.,


Elnitski, 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.Sequenceandcomparativeanalysisofthechickengenomeprovideuniqueperspectivesonvertebrateevolution.Nature432:695–716.

Hocking, P.M. 2005.ReviewofQTLresultsinchicken.Wld. Poultry Sci. J.61:215–226.Hu, J.X., Bumstead, N., Barrow, P., Sebastein, G., Olien, L., Morgan, K. & Malo, D.1997.Resistance

tosalmonellosisinthechickenislinkedtoNRAMP1andTNC.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.IdentificationandfunctionalcharacterizationofchickenToll-likereceptor5revealsafundamentalroleinthebiologyofinfectionwithSalmonella enterica serovartyphimurium.Infect. & Immunity73:2344–2350.

Jiang, R., Li, J.,Qu, Li, H. & Yang, N. 2004.Anewsinglenucleotidepolymorphisminthechickenpituitary-specifictranscriptionfactor(POULF1)geneassociatedwithgrowthrate.Anim. Genet.35:344–346.

Kuhnlein, U., Ni, L., Weigend, S., Gavora, J.S., Fairfull, W. & Zadworny, D. 1997.DNApoly-morphismsinthechickengrowthhormonegene:responsetoselectionfordiseaseresistanceandassociationwitheggproduction.Anim. Genet.28:116–123.

Lamont, S.J., Kaiser, M.G. & Liu, W.2002.CandidategenesforresistancetoSalmonella 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.AllelicvariationinTLR4islinkedtosusceptibilitytoSalmonella entericaserovartyphimuriuminfectioninchickens.Infect. & Immunity71:1116–1124.

Li, S., Aggrey, S.E., Zadworny, D., Fairfull, W. & Kuhnlein, U. 1998a. Evidence for a geneticvariation in themitochondrialgenomeaffecting traits inWhiteLeghornchickens.J. Hered. 89:222–226.

Li, S., Zadworny, D., Aggrey, S.E. & Kuhnlein, U. 1998b.Mitochondrialpepck:Ahighlypolymor-phicgenewithallelesco-selectedwithMarek’sdiseaseresistance inchickens.Anim. Genet.29:395–397.

Li, H., Deeb, N., Zhou, H., Mitchell, A.D., Ashwell, C.M. & Lamont, S.J. 2003.Chickenquantita-tivetraitlociforgrowthandbodycompositionassociatedwithtransforminggrowthfactor-betagenes.Poult. Sci.82:347–356.

Li, H., Deeb, N., Zhou, H., Ashwell, C.M. & Lamont, S.J. 2005.Chickenquantitativetraitlociforgrowthandbodycompositionassociatedwiththeverylowdensityapolipoprotein-IIgene.Poult. Sci.84:697–703.


Liu, W. & Lamont, S.J.2003.Candidategeneapproach:Potentialassociationofcaspase-1,inhibitorofapoptosisprotein-1,andProsaposingenepolymorphismswithresponsetoSalmonella enteri-tidischallengeorvaccinationinyoungchicks.Anim. Biotech.14:61–76.

Liu, H.C., Cheng, H.H., Tirunagaru, V., Sofer, L. & Burnside, J.2001a.Astrategytoidentifyposi-tionalcandidategenesconferringMarek’sdiseaseresistancebyintegratingDNAmicroarraysandgeneticmapping.Anim. Genet.32:351–359.

Liu, H.C., Kung, H.J., Fulton, J.E., Morgan, R.W. & Cheng. H.H.2001b.Growthhormoneinter-acts with the Marek’s disease virus SORF2 protein and is associated with disease resistance inchicken.Proc. Natl. Acad. Sci. USA98:9203–9208.

Liu, H.C., Niikura, M., Fulton, J.E. & Cheng. H.H. 2003. Identificationof chicken lymphocyteantigen6complex,locusE(LY6E,aliasSCA2)asaputativeMarek’sdiseaseresistancegeneviaavirus-hostproteininteractionscreen.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.Apoultrybreeder’sapproachtoavianneoplasia.Avian Pathology27:S74–S77.Meuwissen, T.H.E., Hayes, B.J. & Goddard, M.E. 2001. Prediction of total genetic value using

genome-widedensemarkermaps.Genetics157:1819–1829.Muir, W.M. 2003. Incorporating molecular information in breeding programmes: applications and

limitations.InW.M.Muir&S.E.Aggrey,eds.Poultry Genetics, Breeding and Biotechnology,pp.549–562.Wallingford,UK,CABIPublishing.

Nagamine, Y., Haley, C.S., Sewalem, A. & Visscher, P.M.2003.Quantitativetraitlocivariationforgrowthandobesitybetweenandwithinlinesofpigs(Sus scrofa).Genetics164:629–635.

Nagaraja, S.C., Aggrey, S.E., Yao, J., Zadworny, D., Fairfull R.W. & Kuhnlein, U. 2000.Traitasso-ciationofageneticmarkerneartheIGF-Igeneinegg-layingchickens.J.Heredity91:150–156.

Parsanejad, R., Torkamanzehi, A., Zadworny, D. & Kuhnlein, U.2003.Allelesofcytosolicphos-phoenolpyruvate carboxykinase (PEPCK): trait association and interaction with mitochondrialPEPCKinastrainofwhiteleghornchickens.Poultry Sci.82:1708–1715.

Parsanejad, R., Praslickova, D., Zadworny, D. & Kuhnlein, U.2004.Ornithinedecarboxylase:hap-lotypestructureandtraitassociationsinWhiteLeghornchickens.Poultry Sci.83:1518–1523.

Preisinger, R. 2004.InternationaleTendenzenderTierzüchtungunddieRollederZuchtunternehmen.Züchtungskunde76(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.Firstreportonchickengenesandchromosomes2000.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.Secondreportonchickengenesandchromosomes2005.Cytogen.& Genome Res.109:415–479.


Shete, S. & Amos, C.I. 2002. Testing for genetic linkage in families by a variance componentsapproachinthepresenceofgenomicimprinting.Am. J. Human Genet.70:751–757.

Taha, F.A.2003.The poultry sector in middle income countries and its feed requirements:the case of Egypt.OutlookReportNo.WRS03-02,pp.3–14.Washington,DC,EconomicResearchService,USDA.

Vallejo, R.L., Bacon, L.D., Liu, H.C., Witter, R.L., Groenen, M.A., Hillel, J. & Cheng, H.H.1998.GeneticmappingofquantitativetraitlociaffectingsusceptibilitytoMarek’sdiseasevirusinducedtumorsinF2intercrosschickens.Genetics148:349–360.

van der Beek, S. &. van Arendonk, J.A.M. 1996.Marker-assistedselection inanoutbredpoultrynucleus.Anim. Sc.62:171–180.

Wang, T., Fernando, R.L. & Grossman, M. 1998.Geneticevaluationbybestlinearunbiasedpredic-tionusingmarkerandtraitinformationinamultibreedpopulation.Genetics148:507–515.

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:asequencevariationdatabaseforthechickengenome.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.Ageneticvariationmapforchickenwith2.8millionsingle-nucleotidepolymorphisms.Nature432:717–722.

Yonash, N., Bacon, L.D. Witter, R.L. & Cheng. H.H.1999.Highresolutionmappingandidentifica-tionofnewquantitativetraitloci(QTL)affectingsusceptibilitytoMarek’sdisease.Anim. Genet.30:126–135.

Zhou, H., Mitchell, A.D., McMurtry, J.P., Ashwell, C.M. & Lamont, S.J. 2005.Insulin-likegrowthfactor-1genepolymorphismassociationswithgrowth,bodycomposition,skeletonintegrity,andmetabolictraitsinchickens.Poult. Sci.84:212–219.

Chapter 12

marker-assisted selection in dairy cattle

Joel Ira Weller


SummaryConsideringthelonggenerationinterval,thehighvalueofeachindividual,theverylimitedfemalefertilityandthefactthatnearlyalleconomictraitsareexpressedonlyinfemales,itwouldseemthatcattleshouldbeanearlyidealspeciesforapplicationofmarker-assistedselection (MAS). As genetic gains are cumulative and eternal, application of new tech-nologies that increase ratesofgeneticgaincanbeprofitable even if thenominal annualcostsareseveraltimesthevalueofthenominaladditionalannualgeneticgain.Completegenomescans forquantitative trait loci (QTL)basedon thegranddaughterdesignhavebeen completed for most commercial dairy cattle populations, and significant across-studyeffectsforeconomictraitshavebeenfoundonchromosomes1,3,6,9,10,14and20.QuantitativetraitlociassociatedwithtrypanotolerancehavebeendetectedinacrossbetweentheAfricanN’DamaandtheBoranbreedsasthefirststepintheintrogressionofthesegenesintobreedssusceptibletotrypanosomosis.Indairycattle,theactualDNApolymorphismhasbeendeterminedtwice,forQTLonBTA6andBTA14.Inbothcasesthepolymorphismcausedanon-conservativeaminoacidchange,andbothQTLchieflyaffectfatandproteinconcentration.MosttheoreticalstudieshaveestimatedtheexpectedgainsthatcanbeobtainedbyMAStobeintherangeofa5to20percentincreaseintherates of genetic gain obtained by traditional selection programmes. Applied MAS pro-grammeshavecommencedforFrenchandGermanHolsteins.InbothprogrammesgeneticevaluationsincludingQTLeffectsarecomputedbyvariantsofmarker-assistedbestlinearunbiasedprediction(MA-BLUP).

Chapter 12 – Marker-assisted selection in dairy cattle 201

introduCtionCompared with other agricultural species,dairy cattle are unique in terms of thevalue of each animal, their long genera-tion interval and the very limited fertilityof females. Thus unlike plant and poultrybreeding, most dairy cattle breeding pro-grammesarebasedonselectionwithinthecommercialpopulation.Similarly,detectionofquantitativetraitloci(QTL)andmarker-assisted selection (MAS) programmes aregenerallybasedonanalysisofexistingpop-ulations.Thespecificrequirementsofdairycattle breeding have led to the generationofverylargedatabanksinmostdevelopedcountries,whichareavailable foranalysis.In this chapter, dairy cattle breeding pro-grammes in thedevelopedanddevelopingcountriesarereviewedandcompared.TheimportantissuesintheapplicationofMASarethenoutlined.These includeeconomicconsiderations based on phenotypic selec-tion, the current status of cattle marker

maps,methodstodetectQTLandtoesti-mate QTL effects and location suitablefor dairy cattle, the current state of QTLdetectionindairycattle,methodstoincor-porate information from genetic markersin genetic evaluation systems, methods toidentifytheactualpolymorphismsrespon-sibleforobservedQTLanddescriptionofthe reported results, methods and theoryforMASindairycattle, thecurrentstatusof MAS and, finally, the future prospectsforMASindairycattle.

dairy Cattle Breeding programmeS in developed CountrieSIn most developed countries, dairy cattlebreeding programmes are based on the“progeny test” (PT) design. The PT isthedesignofchoice formoderate 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 programme

100 elite cows

200 elite cows

3 foreign bulls

4 local bulls

50 candidate

bulls

30 000

cows

5 000

daughters

records on daughters

5 bulls are selected

45 bulls are culled

120 000 cows

20 bulls


Thispopulationconsistsof approximately120000cows of which 90percent aremilk recorded. Approximately 20 bullsare used for general service. Each yearabout 300 elite cows are selected as bulldams.These aremated to the two to fourbest local bulls and an equal number offoreignbulls toproduceapproximately50bullcalves forprogenytesting.At theageof one year, the bull calves reach sexualmaturity, and approximately 1 000 semensamplesarecollectedfromeachyoungbull.These bulls are mated to approximately30000 first parity cows to produce about5000daughters, or 100daughters 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 asimilarnumberoftheoldprovensiresareculled. By this time the selected bulls areapproximatelyfiveyearsold.

Various studies have shown that ratesof genetic gain by a PT scheme are about0.1 to 0.2genetic 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,comparedwithbreedingschemesforother species, thePThas severalmajorweaknesses. 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 signifi-cantly reduce rates of genetic gain (IsraelandWeller,2000).Second,generationinter-vals, especially along the sire-to-dam and

sire-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 priceformalecalvesofelitecows,thesecowsareoften given preferential treatment in ordertoincreasetheirgeneticevaluations(Powelland Norman, 1988). The small number ofbulls actually used for general service, andthe even smaller number of bulls used asbullsires,tendstoreducetheeffectivepop-ulationsize,whichincreasesinbreedinganddecreasesgeneticvarianceinthepopulation.TheeffectivepopulationsizeoftheUnitedStates Holstein population with ten mil-lion cows has been estimated at about 100(Farniret al.,2000).Finally,thereisvirtuallyno selection along the dam-to-dam path.Generally, 70-80percent of healthy femalecalvesproducedareusedasreplacements.

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 populationconsistsofseveralhundredindividuals,andbullsarenotprogenytested.Instead,bullsareselectedbasedonthegeneticevaluationsoftheirdamsandsisters,whichshortensthegeneration interval on the sire-to-dam andsire-to-sirepaths,butreducesthereliabilitiesofthegeneticevaluations.Damsofbullsandcowsareselectedbasedchieflyontheirownproduction records, and MOET is appliedtoincreasethenumberofprogenyperdam.Astheselectionpopulationconsistsofonlyseveral hundred individuals, MOET costs


are manageable if costs are spread overthe entire national dairy industry. Ratesofgeneticgainwithin thenucleusare thushigher than can be obtained by a nationalPT design. This gain is transferred to thegeneralpopulationthroughtheuseofbullsfromthenucleuspopulation.Inadditiontothegreateroverall rateofgeneticgain, thenucleusschemehastheadvantagethat it isnecessarytocollectdataonamuchsmallerpopulation, which should reduce costsand increase accuracy. The disadvantagesof MOET are that overall costs and ratesof increase of inbreeding will be greaterunlessstepsaretakentoreduceinbreeding.However, these steps will also slightlydecreaseratesofgeneticgain.Inpractice,nocountryhasreplaceditsstandardPTschemewithanucleusbreedingprogramme.

dairy Cattle Breeding in developing CountrieSThe genus Bos includes five to seven spe-cies, of which Bos taurus and Bos indicusare the most widespread and economi-callyimportant.B. taurusisthemaindairycattle 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 toler-ance to environmental stress due to poornutrition, heat and disease challenge tosustain relatively high production levels(Cunningham, 1989). Tropical breeds areadaptedtothesestressesbuthavelowmilkyield, whereas more productive temperatebreedscannotwithstandtheharshtropicalconditions,tothepointofnotbeingabletosustain their numbers (de Vaccaro, 1990).Furthermore, most tropical countries aredevelopingcountries,whichlacksystematiclarge-scalemilkandpedigreerecording.

A number of studies have been con-ducted on crosses between imported andlocalbreedsinthetropics.Generally,theF1B. taurusxB. indicuscrossesareeconomi-cally superior to either of the purebredstrains (FAO, 1987). The heterosis effectof the F1 cross is due to genes for diseaseresistancefromthelocalparent,andgenesfor milk production from the importedstrain (Smith, 1988; Cunningham, 1989).However, this heterosis is lost in futuregenerationsiftheF1isbackcrossedtoeitherparental strain.Madalena (1993)presentedan F1 continuous replacement schemeto capitalize on its superiority. Recently,Kosgey, Kahi and van Arendonk (2005)proposed a closed adult nucleus MOETscheme to increase milk production intropicalcrossbredcattle.

eConomiC ConSiderationS in applying maS to dairy CattleForanynewtechniquetobeeconomicallyviable, overall gains must be greater thanoverall costs. This also applies to usingMAS within a dairy cattle breeding pro-gramme. However, unlike investment innewequipment,geneticgainsnever“wearout”,i.e.breedingisuniqueinthatgeneticgains are cumulative and eternal. Thus, asshownbyWeller (1994,2001) investmentsin MAS or other techniques that enhancebreeding programmes are economicallyviable even if “nominal” costs are greaterthan“nominal”gains.

Forexample,consideranongoingbreed-ingprogrammewithaconstantrateofgeneticgainperyear.Assumethattheannualrateofgeneticgainhasanominaleconomicalvalueof V. The cumulative discounted returns toyearT,Rv,willbeafunctionofthenominalannual returns, the discount rate, d, theprofit horizon, T, and the number of yearsfromthebeginningof theprogrammeuntil


firstreturnsarerealized,t.Rviscomputedasfollows(Hill,1971):

where rd=1/(1+d).For example,withd=0.08, T = 20 years, and t = 5 years, Rv =32.58V.That is, thecumulativereturnsareequaltonearly33timesthenominalannualreturns. For an infinite profit horizon,Equation{1}reducesto:

andRv=124.04V.The value of nominal annual genetic

gainwillnowbecomparedwiththeannualcostsofabreedingprogramme,assumingafixed nominal cost per year. Costs, unlikegeneticgain,onlyhaveaneffectintheyeartheyoccur.Assumingthatannualcostsareequal during the length of the breedingprogramme,andthatfirstcostsoccurintheyearafterthebaseyear,CT,thenetpresentvalueofthetotalcostsofthebreedingpro-grammeiscomputedasfollows:

where Cc = annual costs of the breedingprogramme. Using the same values for Tand d, CT = 9.82Cc. Thus, with a profithorizon of 20 years, cumulative profit ispositiveifV>0.31Cc.Foraninfiniteprofithorizon, CT = 12.5Cc, and profit will bepositiveifV>0.1Cc.

Therefore,abreedingprogrammecanbeprofitableevenifthenominalannualcostsare several times the value of the nominalannualgeneticgain.Forexample,consider

the United States of America dairy cattlepopulation, which consists of about tenmillioncows.Annualgeneticgainisabout100kgmilk peryear.Thevalueof a 1kggaininmilkproductionhasbeenestimatedatUS$0.1(Weller,1994).Thus,thenominalannualvalueofa10percentincreaseintherateofgeneticgain(10kgperyear)is:

V=(10kgpercowperyear)(US$0.1perkg)(10000000cows)=US$10000000peryear

The cumulative value with a profithorizon of 20 years and an 8percent dis-count rate would be US$326 million, andbreak-even annual costs for a technologythatincreasesannualgeneticgainby10per-cent are US$32 million per year. Thus, itwouldbeprofitabletospendquitealotforarelativelysmallgeneticgain.

The value of genetic gain to a specificbreeding enterprise will generally be lessthanthegaintothegeneraleconomy.Thisis because most of the gains obtained bybreedingwillbepassedontotheconsumers.Brascamp,vanArendonkandGroen(1993)considered the economic value of MASbased on changes in returns from semensales for a breeding organization oper-atinginacompetitivemarket.Inthiscase,a breeding firm that adopts a MAS pro-gramme can increase its returns either byincreasingitsmarketshareorincreasingthemeanpriceofasemendose.Althoughthevalueofgeneticgainwillbeless,relativelysmallchangesingeneticmeritcanresultinlargechangesinmarketshare.

Current StatuS of marker mapS in CattleCattlehave29pairsofautosomesandonepair of sex chromosomes. All the auto-somes are acrocentric, and map units are

{1}d

Td

d

Td

td

v rrtT

rrrVR

-+-

---

=++

1)1(

)1(

1

2

1

{2}222 )1()1( –+== t

d

t

v ddV

rR Vr

–

{3}d

Tddc

T rrrCC

−−

=1

)1(

{4}


scoredfromthecentromere.Chromosomesare denoted with the prefix “BTA” (B. taurus). Similar to other mammals, thebovine DNA includes 3×109 base pairs(bp), and themap length isapproximately3000cM.Thehumangenomeisestimatedto encode 20 000–25 000 protein-codinggenes (International Human GenomeSequencing Consortium, 2004), and it canbe assumed that the number of genes inothermammals,includingcattle,shouldbequite similar. Thus, a single map unit, onaverage,includesapproximatelyeightgenesandonemillionbp.

Asinotheranimalspecies,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 50000 SNPs have been identified inhumans, but only several thousand havebeenvalidatedincattle(www.afns.ualberta.ca/Hosted/Bovine%20Genomics/), andrates of polymorphism are generallyunknown.Withthecompletionofthesix-fold coverage of the bovine genome bytheBovineGenomeSequencingProjectatBaylor College of Medicine (www.hgsc.bcm.tmc.edu/projects/bovine/)manymoreSNPswillbeidentified.

Severalgeneticmapsareavailableontheinternet. The United States Meat AnimalResearch Center (MARC) (www.marc.usda.gov/) includes thousands of markers,chiefly microsatellites. The ArkDB data-base system, hosted at Roslin Institute,includes data from several published maps(www.thearkdb.org/). The CommonwealthScientificandIndustrialResearchOrganiza-tion (CSIRO) livestock industries cattlegenome marker map is built upon dataprovided by the University of Sydney’s

comparative location database (www.livestockgenomics.csiro.au/perl/gbrowse.cgi/cattlemap/).Thismapcombinedallpub-licly-availablemaps intoa single integratedmapthatcurrentlyincludes9400markers.

methodS of qtl deteCtion SuitaBle for CommerCial dairy Cattle populationSDetection of QTL requires generation oflinkage disequilibrium (LD) between thegeneticmarkersandQTL.Inplants,thisisgenerallyaccomplishedbycrossesbetweeninbred lines but, for the reasons noted intheintroduction,thisisnotaviableoptionfor dairy cattle in developed countries, inwhichall analysesmustbebasedonanal-ysisoftheexistingpopulation.DetectionofQTLindevelopingcountriesisconsideredbelow. For advanced commercial popula-tions,the“daughter”and“granddaughter”designs, which make use of the existenceof large half-sib families, are most appro-priateforQTLanalysis(Weller,KashiandSoller, 1990). Thesedesigns arepresentedinFigures2and3.

Bothdesignsaresimilartothebackcrossdesign for crosses between inbred lines inthatonly theallelesofoneparentare fol-lowedintheprogeny.Thus,similartothebackcross design, dominance cannot beestimated.Thesedesignsdifferfromcrossesbetween inbred lines in that several fami-liesareanalysedinwhichthelinkagephasebetween QTL and genetic markers maydiffer.Inaddition,anyspecificQTLwillbeheterozygousinonlyafractionofthefami-lies included in the analysis. Thus, QTLeffects must be estimated within families,and these designs are therefore less pow-erfulperindividualgenotypedthandesignsbasedoncrossesbetweeninbredlines.

The granddaughter design has theadvantage of greater statistical power per


individual genotyped. As each genotypeis associated with multiple phenotypicrecords, the power per individual geno-typed in the granddaughter design can befour-foldthepowerofthedaughterdesign(Weller,KashiandSoller,1990).Thedisad-vantageofthisdesignisthattheappropriatedatastructure(hundredsofprogenytestedbulls,sonsofa limitednumberofsires) isfoundonlyinthelargestdairycattlepopu-lations. Both daughter and granddaughterdesigns are less powerful per individualgenotyped than designs based on analysisof inbred lines. Furthermore, the half-sibdesignshavethedisadvantagethatprogenywiththesamegenotypeasthesireareunin-formative,becausetheprogenycouldhavereceivedeitherpaternalallele.

Additional experimental designs havealso been proposed. Coppieters et al.(1999) proposed the “great-granddaughterdesign”. One of the disadvantages of thegranddaughterdesignisthatthenumberofprogeny-testedsonsofmostsiresistoolowtoobtainreasonablepowertodetectQTLofmoderateeffects.Coppieterset al.(1999)proposed that power can be increased byalso genotyping progeny-tested grandsonsof the grandsire. Inclusion of the grand-sonsiscomplicatedbythefactthatthereisanothergenerationofmeiosisbetweenthegrandsireandhisgrandson.

Asignificantdrawbackofallthedesignsconsidered above is that they give noindication of the number of QTL allelessegregatinginthepopulationortheirrela-

FiGURe 2the daughter design

M

A

?

?

M

A

m

a

m

a

?

?

only a single family is shown, although in practice several families will be analysed jointly. the sire is assumed to be heterozygous for a Qtl and a linked genetic marker. the two alleles of the marker locus are denoted “M” and “m”, and the two alleles of the Qtl are denoted “a” and “a”. alleles of maternal origin are denoted by question marks.


tive frequencies. To answer this question,Weller et al. (2002) proposed the “modi-fied granddaughter design” presented inFigure4. Assume that a segregating QTLfor a trait of interest has been detectedand mapped to a short chromosomalsegmentusingeitheradaughteroragrand-daughter design. Consider the maternalgranddaughters of a grandsire with a sig-nificant contrastbetweenhis twopaternal

alleles. This grandsire will be denoted the“heterozygous grandsire”. Each maternalgranddaughterwill receiveoneallele fromhersire,whoisassumedtobeunrelatedtotheheterozygousgrandsire, andoneallelefrom her dam, who is a daughter of theheterozygous grandsire. Of these grand-daughters, one-quarter should receive thegrandpaternal QTL allele with the posi-tive effect, one-quarter should receive the

FiGURe 3 the granddaughter design

M

A

m

a

M

A

?

?

m

a

?

?

the grandsire is assumed to be heterozygous for a Qtl and a linked genetic marker. as in Figure 2, only a single family is 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 granddaughters because they were not genotyped.


negativegrandpaternalQTLallele,andhalfshouldreceiveneithergrandpaternalallele.Inthethirdcase,thegranddaughterreceivedone 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 populationwithrespecttotheallelicdistributionoftheQTL.Allgeneticandenvironmentaleffectsnotlinkedtothechromosomalsegmentinquestionare assumed tobe randomlydis-tributedamongthegranddaughters,orareincludedintheanalysismodel.Thus,unlikethedaughterorgranddaughterdesigns,itispossible tocompare theeffectsof the twograndpaternal alleles with the mean QTLpopulationeffect.

AssumingthattheQTLis“functionallybiallelic” (i.e. there are only two alleleswith differential expression relative to thequantitative trait), and that allele origin

can be determined in the granddaughters,the relative frequencies of the two QTLallelesinthepopulationcanbedeterminedby comparing the mean values of thethree groups of granddaughters for thequantitative trait. Using the modifiedgranddaughterdesign it is alsopossible 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 frequencyof theQTLallelethatincreasesfatandproteinconcentrationonBTA6intheIsraeliHolsteinpopulationas0.69and0.63,relativetofatandproteinpercent, by the modified granddaughterdesign. This corresponded closely to thefrequency of 0.69 estimated for the Y581allele of the ABCG2 gene for cows bornduringthesametimeperiod(Cohen-Zinderet al.,2005).

FiGURe 4the modified granddaughter design

Q1 Q2 M1

Q1 M1 Q2 M2H1 H2

H1 Q1 H2 M1 H3 M2

H3 H4

Q2 H4

M2

only alleles for the Qtl are shown. alleles originating in the heterozygous grandsire are termed “Q1” and Q2”. alleles originating in the grand-dams are termed “M1” and “M2”. alleles originating in the sires are termed “H1”, “H2”, “H3” and “H4”.


methodS to eStimate qtl effeCtS and loCation in dairy CattleIfasignificanteffectonaquantitativetraitis associated with a genetic marker, thedifference between the means of markergenotype classes will be a biased estimateof the QTL effect due to recombinationbetween theQTLand thegeneticmarker.Weller(1986)firstdemonstratedthatmax-imumlikelihood(ML)methodologycouldbe used to obtain estimates of QTL loca-tionandeffectunbiasedbyrecombination,whileLanderandBotstein(1989)proposedintervalmapping,basedonMLforaQTLbracketedbetweentwomarkers.HaleyandKnott (1992) and Martinez and Curnow(1992) proposed an interval mappingmethod based on non-linear regression,whichwaseasier toapplythanML.Theirmethodsarenotdirectlyapplicabletohalf-sib designs because, as noted previously,linkagerelationshipsbetweentheQTLandthegeneticmarkerswillbedifferentacrossfamilies,andinsomefamiliesthecommonancestorwillbehomozygousfortheQTL.Furthermore, if multiple QTL alleles aresegregating in the population, or if theobserved effect is due to several tightlylinked QTL, the magnitude of the effectwillalsodifferacrossfamilies.

Amethodsuitableforintervalmappingthat accounts for theseproblemshasbeendevelopedbyKnott,ElsenandHaley(1996)andhasbeenappliedtonearlyalldaughterand granddaughter design analyses. Theirmethodisamodificationofthenon-linearregression method, and assumes a singleQTLlocationforallfamilies,butestimatesaseparateQTLeffectforeachfamily.Thismethodhastheadvantagethat,unlikeML,it can readily deal with missing and unin-formative genotypes for some markers.MackinnonandWeller(1995)proposedanMLmethodtoestimatebothQTLlocation

and effect for half-sib designs under theassumption thatonly twoQTLalleles aresegregating in the population. Using thismethoditisalsopossibletoestimateQTLgenotype of the common parent of eachfamily. However, these determinations areaccurateonlyforrelativelylargeQTL.ThemethodofMackinnonandWeller(1995)ismoredifficulttoapplythanthemethodofKnott,ElsenandHaley(1996),andhasnotcomeintogeneralusage.

Lander and Botstein (1989) proposedthe LOD-score (logarithm of the odds tothe base 10) drop-off method to estimateconfidenceintervalsforQTLlocation,butseveral studies have shown that this seri-ously underestimate 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 confi-dence intervals. Bennewitz, Reinsch andKalm(2003)proposedimprovementstothebootstrapmethodthatresultinshorterbutstillunbiasedconfidenceintervals.

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 appro-priate threshold to declare significance. Ifnormalpoint-wisesignificancelevelsof5or1percentareused,manymarker-traitcombi-nationswill show“significance”bychance.WhilethisisaproblemforallQTLgenomescans,it isevenmoreseverefordairycattlein which multiple half-sib families are ana-lysed, in addition to multiple markers andtraits.Severalsolutionstothisproblemhavebeenproposed,noneofwhichiscompletelysatisfactory. The only solution to deal ade-quatelywithbothmultipletraitsandfamiliesin addition to multiple markers is the falsediscoveryrate(Welleret al.,1998).


The QTL effects derived from eitherdaughterorgranddaughterbyMLornon-linear regression will still be biased forseveral reasons. First, the usual assump-tions of interval mapping, a single QTLsegregatingwithinthemarkerintervalandno 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 andWeller (1998)dem-onstrated that QTL effects derived fromanalysisofeithergeneticevaluations,yielddeviationsorDYDwillbeunderestimated.In addition to this downward bias, thereare two sources of upward bias for QTLeffects. First, the direction of the effectsis generally arbitrary, and therefore abso-lute values are retained and all effects are>0. Third, only the effects deemed “sig-nificant”areretained,andthisisaselectedsample (Georges et al., 1995). Bayesiananalysis methods that account for bias ofQTL effect due to selection have recentlybeendevelopedbyWeller,SchlezingerandRon(2005).

Current StatuS of qtl deteCtion in dairy CattleGenome scans by the granddaughterdesign have been completed for Holsteinsfrom Canada (Nadesalingam, Plante andGibson, 2001), the Netherlands (Spelmanet al.,1996;Schrootenet al.,2000),France(Bennewitz, et al., 2003a; Boichard et al.,2003), Germany (Bennewitz, et al., 2003a;Kuhnet al.,2003a),NewZealand(Spelmanet al.,1999),andtheUnitedStates(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 et al., 1999);FinnishAyrshires (Vilkkiet al.,

1997; Viitala et al., 2003; Schulman et al.,2004);FrenchNormandeandMontbeliardecattle (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(Mosiget al.,2001;Ronet al.,2004).Moststudies have considered the five economicmilkproductiontraits:milk,fatandproteinproduction,andfatandproteinconcentra-tion, although a number of studies havealso considered somatic cell score (SCS),female fertility, herd life, calving traits,health traits, temperament and conforma-tiontraits.TheSCSisalogbase2functionof the concentration of somatic cells, andhasbeenshowntobeausefulindicatorofudder health. Results are summarized inTable1.

Results for milk, fat and protein pro-duction,fatandproteinconcentration,andSCSfrommostof thestudies listedaboveare summarized at www.vetsci.usyd.edu.au/reprogen/QTL_Map/. Results fromthese traits, and many others includingmeatproduction,aresummarizedathttp://bovineqtl.tamu.edu.Significanteffectswerefoundonall29autosomes,butmosteffectswerefoundonlyinsinglestudiesandhavenot been repeated. Khatkar et al. (2004)performedameta-analysis,combiningdatafrom most of these studies, and foundsignificantacross-studyeffectsonchromo-somes1,3,6,9,10,14and20.

methodS of inCorporating information from genetiC markerS in genetiC evaluation SyStemSHeritabilities of most economic traits indairy cattle are low to moderate. Geneticevaluation of dairy cattle is complicatedby confounding between genetic and


environmental factors. Cows are scatteredover many different herds with differentmanagementlevels,anddistributionofsiresacrossherds isnotrandomororthogonal.Furthermore, cows generally producemultiple lactations that are correlated. Inordertoaccountforthelimitedheritability,and co-variances among relatives, geneticeffects are generally assumed to be

random, while most environmental effectsare assumed to be fixed. Thus, geneticevaluation is performed by the mixedmodelusingbestlinearunbiasedprediction(BLUP)methodology(Henderson,1984).

Beginning in the late 1980s, the modelof choice for genetic evaluation for milkproductiontraitswastheindividualanimalmodel,inwhichageneticeffectiscomputed

table 1Summary of dairy cattle genome scans

experimental design

Breed Country traits analysed references

Granddaughter ayrshire Finland Milk production1 Vilkki et al., 1997; de Koning et al., 2001; Viitala et al., 2003

ScS2, mastitis, other treatments Schulman et al., 2004

Jersey new Zealand conformation Spelman, Garrick and van arendonk, 1999

Holstein canadian Milk production Plante et al., 2001

France Milk production boichard et al., 2003

Germany Milk production thomsen et al., 2001

Functional Kuhn et al., 2003

conformation, temperament, milking speed

Hiendleder et al., 2003

netherlands conformation, ScS, fertility, calving, milking speed, gestation, birth weight, temperament

Schrooten et al., 2000

new Zealand conformation Spelman, Garrick and van arendonk, 1999

USa Milk production ashwell et al., 1998b; ashwell and tassell 1999; ashwell et al., 1997, 2004; ashwell, Van tassell and Sonstegard, 2001;Georges et al., 1995; Heyen et al., 1999; Zhang et al., 1998

ScS ashwell et al., 1996, 1997, 1998b; ashwell and Van tassell, 1999; Heyen et al., 1999

Herdlife Heyen et al., 1999

conformation ashwell et al., 1998a, 1998b; ashwell and Van tassell, 1999

Fertility ashwell et al., 2004

Montbeliarde France Milk production boichard et al., 2003

normande France Milk production boichard et al., 2003

norwegian norway Milk production olsen et al., 2002

ScS, mastitis Klungland et al., 2001

Swedish Sweden ScS, mastitis, other diseases Holmberg and andersson-eklund, 2004

Daughter Holstein israel Milk production, ScS, fertility, herdlife

Ron et al., 2004

% protein Mosig et al., 2001

1 Milk, fat, and protein production, and fat and protein concentration.2 Somatic cell concentration


foreachanimal,includinganimalsthatdidnot have production records (Westall andvan Vleck, 1987). Genetic evaluations fortheseanimalsarederivedviathenumeratorrelationshipmatrix,whichisincludedinthemodel.Inaddition,a“permanentenviron-mental”effectiscomputedforeachanimalwith records to account for similaritiesamong multiple records of the same cowthatarenotduetoadditivegeneticeffects.As noted previously, analysis of QTLeffectshasgenerallybeenbasedonanalysisof genetic evaluations or DYD, which aretheadjustedmeansofthedaughterrecordsof a bull but which, unlike genetic evalu-ations, are not regressed. However, thestatistical properties of DYD are not wellunderstood,andQTLeffectsderivedfromanalysis of DYD are still biased (Israeland Weller, 1998). Theoretically, it shouldbe possible to derive unbiased QTL esti-matesiftheseeffectsareincorporatedintoageneticevaluationschemebasedonanal-ysisoftheactualrecords,suchastheanimalmodel. In practice, the inclusion of QTLeffects into genetic evaluation models iscomplicatedbythreemainfactors:• actual QTL location is unknown, and

there is only partial linkage betweengeneticmarkersandQTL;

• linkage phase between genetic markersandQTLdiffersamongindividuals,andisgenerallyunknown;

• onlyasmallfractionofthepopulationisgenotyped.An analysis including only genotyped

individuals isnotaviableoptionas itwillgenerallynotbepossibletoderiveaccuratefixed effects, such as herd-year-seasons,fromthissample.

Fernando and Grossman (1989) pro-posed modifying the individual animalmodel described above to a “gametic”model that assumes the two QTL alleles

ofeachindividualarerandomeffectssam-pled from a distribution with a knownvariance.Theydevelopedamethodtoesti-mate breeding 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 markerareknown.Thismodel is suitable foranypopulationstructureandcanalsoincorpo-ratenon-linkedpolygeniceffectsandother“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-assistedBLUP”or“MA-BLUP”.

Each individual with unknown ances-tors isassumedtohave twouniqueallelesfortheQTL,whichare“sampled”fromaninfinite population of alleles. For animalsthat are not genotyped, the probability ofreceivingeitherallelefromeitherparentwillbeequal.However, ifboth theparentandprogenyaregenotypedforalinkedgeneticmarker,thentheprobabilityofreceivingaspecificparentalalleleforaQTLlinkedtothegeneticmarkerwillbeafunctionoftheprogenymarkergenotypeandrecombina-tionfrequency.Basedontheseprobabilities,Fernando and Grossman (1989) demon-strated how a variance-co-variance matrixcouldbeconstructedfortheQTLgameticeffects. They further described a simplealgorithm to invert this matrix analo-goustoHenderson'smethodforinvertingthe numerator relationship matrix. Thismethodhasbeenextended tohandlemul-tiple markers and traits (Goddard, 1992).CantetandSmith(1991)demonstratedthatthe number of equations could be signifi-cantly reduced by analysis of the reducedanimalmodel.


Thedisadvantagesofthismodelarethatit assumes that both recombination fre-quencyandthevarianceduetotheQTLareknown a priori. Studies on simulated datahavedemonstratedthatalthoughrestrictedmaximum likelihood methodology can beused to estimate these parameters, theyare completely confounded for a singlemarker locus (van Arendonket al., 1994).Methods to estimate the variance contrib-utedbyQTLwithmultiplemarkersweredevelopedbyGrignola,HoescheleandTier(1996). Furthermore, as each individualwithunknownparents isassumedtohavetwo unique alleles, the prediction errorvariances of the effects for any individualwillbequitelargeand,therefore,notveryinformative. Finally, the assumption of anormaldistributionofpossibleQTLalleleeffectsmaynotberealistic.

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.TheQTLeffectisthenincludedin the complete animal model analysis asa fixed effect. For individuals that arenot genotyped, probabilities of receivingeither allele are included as regressionconstants.Theseprobabilitiescanbereadilycomputed 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)extendedthemethodtoQTLanalysis based on flanking marker, usingthe method of Whittaker, Thompsom andVisscher (1996) to estimate QTL effectsand locationfromtheregressionestimatesofflankingmarkers.Thismethodhasbeentestedextensivelyonsimulatedpopulations,

and was able to derive unbiased estimatesof QTL effect and location. It has alsobeenappliedtoactualdatafromtheIsraeliHolstein population for a segregatingQTL on chromosome 14 that affectedmilkproductiontraits(Welleret al.,2003).However, in this case the QTL effectwas underestimated. Further research isrequired to determine the reason for thisdiscrepancy.

methodS for qtl deteCtion and maS in developing CountrieSAs noted previously, dairy cattle breedingintropicalandsubtropicalcountriesisgen-erallybasedoncrossbreedingbetweenhighproduction breeds adapted to temperateclimates, and tropical strains which areadaptedtothelocalenvironment,includingresistancetolocaldiseases.Inotheranimalspecies, synthetic strains have been pro-duced by selecting those individuals thatretain the positive characteristics frombothstrains.Forexample,theAssafsheepbreedwasproducedfromacrossbetweenthe Middle East Awassi breed and theEast Friesian breed (www.sheep101.info/breedsA.html).Indairycattle,theproblemof an appropriate strategy for future gen-erations has not been adequately solved,for reasons considered previously. If theeconomicallyimportantgeneswereidenti-fied, then the time andeffort required forproductionofthedesiredsyntheticstrainscouldbereduced.

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. MarkersflankingtheQTLwillberequiredinordertoselectbackcrossprogenythatreceivedthedonorQTLallele.As therewillbeuncer-tainty with respect to the QTL location,


the flanking markers must be sufficientlyclose to the QTL so that it will be pos-sible to determine with relative certaintythat the QTL is in fact located betweenthe flanking markers. Although marker-assisted introgression does decrease thenumber of generations required to obtainfixation of the desired allele, it increasestwo key cost elements. First, with tradi-tional introgression, half of the progenywill carry the donor allele for the intro-gressed gene, and all of these can be usedasparentsinthenextgeneration.However,if only a small fraction of the progeny isselected based on genetic markers, thenmany more individuals must be producedeach generation. Second, genotyping costsforalargenumberofmarkersateachgen-erationwillalsobesignificant.

Crosses between cattle breeds can alsobe used for QTL detection and they havebeenusedindevelopingcountries.Inmostplant species, the parental lines are com-pletely inbred, and there will be completeLD in the F2 or backcross generation.However, cattle are outbreeders and incrossesbetweenbreedstherewillthereforeonly be partial LD between segregatingQTL and linked genetic markers. Song,SollerandGenizi(1999)proposedthefull-sib intercross line (FSIL) design for QTLdetectionandmappingforcrossesbetweenstrainsofoutcrossingspecies.Theyassumedthatthetwoparentalstrainsdifferinallelicfrequencies, but were not at fixation foralternativeQTLalleles.

For given statistical power, the FSILdesign requires only slightly more indi-viduals thananF2designderived fromaninbredlinecross,butsix-toten-foldfewerthan a half-sib or full-sib design. In addi-tion, as the population is maintained bycontinuedintercrossing,DNAsamplesandphenotypicinformationcanbeaccumulated

acrossgenerations.Continuedintercrossingin future generations also leads to mapexpansion, and thus to increased map-ping accuracy in the later generations. AnFSIL can therefore be used for fine map-pingofQTLandthis isconsideredbelowindetail.

AlthoughthesemethodshavenotasyetbeenappliedtodetectQTLrelatedtomilkproduction,theyhavebeenappliedtoQTLfor disease resistance. Trypanosomosis(sleeping sickness) is a major constrainton livestock productivity in sub-SaharanAfrica. Hanotte et al. (2003) mappedQTLaffecting trypanotolerance inacrossbetweenthe“tolerant”N’DamabreedandthesusceptibleBoranbreed.PutativeQTLaffecting 16 traits associated with diseasesusceptibility were mapped tentatively to18 autosomes. Excluding chromosomeswith ambiguous effects, the allele associ-ated with resistance was derived from theN’DamastrainfornineQTLandfromtheBoranstrainforfiveQTL.Theseresultsareconsistent with many plant crossbreedingexperimentsinwhichthestrainwithoverallphenotypic inferiority for the quantitativetraitneverthelessharboursQTLallelesthatare superior to the alleles present in thephenotypicallysuperiorstrain(e.g.Weller,SollerandBrody,1988).

from qtl to qtn – theoryAs noted by Darvasi and Soller (1997),withasaturatedgeneticmap,theresolvingpower for QTL will be a function of theexperimentaldesign,numberofindividualsgenotyped and QTL effect. Weller andSoller(2004)computedthatthe95percentconfidenceinterval(CI) inpercentrecom-binationforhalf-sibdesigns,includingthedaughter and granddaughter designs, was3073/d2N,wheredistheQTLsubstitutioneffectinunitsofthestandarddeviation,and


Nisthesamplesize.Inthecaseofagrand-daughterdesign,theunitsforthestandarddeviation will be either units of the bulls’DYDorgeneticevaluations.Forexample,ifdis0.5andNis400,theCIwillbe31per-cent recombination, or approximately 35cM.Thus,exceptforthelargestQTL,CIswill generally include several tens of cM.ConsideringthateachcattlecMincludes~8genesandonemillionbp,detectionoftheactual polymorphism responsible for theobservedQTLeffects(thequantitativetraitnucleotide,QTN)appearsatfirstglancetobea“missionimpossible”.

Various strategies have been pro-posed to reduce the CI based on multiplecrosses, but most are not applicable todairycattle(e.g.Darvasi,1998).MeuwissenandGoddard (2000)proposed thatCI forQTL location could be reduced to indi-vidual cMbyapplicationofLDmapping.If a QTL polymorphism is due to a rela-tively recent mutation or to a relativelyrecent introduction from another popula-tion, then it should be possible to detectpopulation-wide LD between the QTLand closely linked genetic markers. Thecloser themarker to theQTL, thegreaterwill be the extent of LD. They developedamethodtoestimateQTLlocationandCIbasedonLDbetweenaQTLanda seriesof closely linkedmarkers. TheCIcanbefurtherreducedbycombining linkageandLDmapping(Meuwissenet al.,2002),andby a multitrait analysis (Meuwissen andGoddard,2004).However,unlesstheQTLeffect isverylarge,theCIwillstillextendoverseveralcM.

In order to determine the actual generesponsiblefortheQTL,moststudieshaveused the “candidate gene” approach, i.e.todeterminea likelycandidateamongthegeneswithintheCI,basedonknowngenefunction,orspecificgeneexpressioninthe

organ of interest. Examples are given inthe following section. However, even if apolymorphismisdetectedinthecandidategene and the polymorphism has a majorLDeffectontheQTL,howdoesoneprovethat this polymorphism is not merely inLDwiththeactualQTN?

Mackay (2001) proposed two alterna-tives for proof positive that a candidatepolymorphismisinfacttheQTN,namely,co-segregation of intragenic recombinantgenotypes in a candidate gene with theQTL phenotype, and functional comple-mentation where the trait phenotype is“rescued”inatransgenicorganism.NeitheroftheseisapplicabletoQTLindairycattle.Inthiscase,Mackay(2001)postulatedthattheonlyoptiontoachievethestandardofrigorousproof for identificationof ageneunderlying a QTL in commercial animalpopulations is to collect “multiple piecesofevidence,nosingleoneofwhichiscon-vincing, but which together consistentlypointtoacandidategene”.EvidencecanbeprovidedbyconcordanceofpolymorphismwithdeducedQTLgenotype,quantitativedifferences of gene expression in physi-ologically relevant 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 ances-tral gene) and alteration of gene proteinin bovine cell lines by “short interferingRNA” (siRNA) technology. (The siRNAmoleculesbindwithproteinstoformaunitcalled the “RNA-induced silencing com-plex”thatsuppressestheexpressionofthegene to which it corresponds in the viralgenome,silencingthegenefromwhichthesiRNAisderived.)

Fordairycattle,todate,themostcom-pellingevidenceis“concordance”,i.e.that


thededucedQTLgenotypesofasampleofindividualscorrespondcompletelytotheirgenotypesfortheputativeQTN.Allindi-viduals heterozygous for the QTL shouldbe heterozygous for the putative QTN,with the sameQTNallele associatedwiththesameQTLalleleinallindividuals,andall individuals homozygous for the QTLshouldalsobehomozygousfortheQTN.Theoretically, the sample of individualsanalysed should be large enough to rejectstatistically the hypothesis that concord-ance was obtained by chance. However,in dairy cattle, the only individuals forwhichQTLgenotypecanbederivedwithany level of reliability are sires that havebeen analysed by either a daughter orgranddaughter design, and the numberof these individuals will always be lim-ited. Furthermore, there is at present noaccepted theory to compute concordanceprobabilities by chance, considering thatanypolymorphismveryclosetotheQTNwilldisplaysignificantLD.Severalstudieshaveaddressedtheproblem(Cohen-Zinderet al.,2005;Schnabelet al.,2005).Thecasefor identification of the QTN is clearlymorecompellingifconcordanceisobtainedintwodifferentpopulations.

from qtl to qtn – reSultSTo date, the QTN has been determinedintwocasesindairycattle,onBTA6andBTA 14. In both cases the QTL chieflyaffected fat andprotein concentration andthe QTL effect was large enough that theconfidence interval for QTL location was<10cM.AQTLonBTA14nearthecen-tromere that chiefly affected fat quantityand both fat and protein concentration inboththeUnitedStatesandIsraeliHolsteinpopulations was first detected by Ron et al. (1998), and further studies were ableto map the QTL to a region of approxi-

mately10cM(Coppieterset al., 1999). In2002, two studies independently showedthatamis-sensemutation,causingreplace-ment of a lysine residue with alanine inexon VIII of the gene acylCoA:diacyg-lycerol acyltransferase (DGAT1), is theQTN (Grisart et al., 2002; Winter et al.,2002). Discovery was aided by the factthatDGAT1wasanobviousphysiologicalcandidate. In addition to mapping to theputative QTL region, DGAT1 encodesa microsomal enzyme that catalyses thefinalstepoftriglyceridesynthesisandmicelacking both copies of DGAT1 are com-pletelydevoidofmilksecretion.Completeconcordance between this polymorphismandtheQTLwas found in threedifferentdairybreeds.

The QTL near the middle of BTA 6affecting protein concentration was firstdetected by Georges et al. (1995) in theUnited States Holstein population. ThisQTL was then detected in several otherHolstein populations, including FinnishAyrshire cattle (Velmala et al., 1999) andNorwegiancattle(Olsenet al.,2002).Ronet al. (2001) reduced the CI to 4 cMcentred on microsatellite BM143. Olsenet al. (2002) used physical mapping andcombined linkage and LD mapping todetermine that thisQTL is locatedwithina 420 000 bp region between the genesABCG2andLAP3.

In 2005, two research groups claimedto have found the QTN in two differentgenes.Schnabelet al.(2005)claimedthattheQTNislocatedinapoly-Asequenceinthepromoter region of the osteopontin gene,alsodenotedSPP1,whileCohen-Zinderet al. (2005) claimed that the QTN is a mis-sense mutation in exon 14 of the ABCG2gene. Both studies based their claim ongene function and concordance of bullswithknowngenotypes.Bothgenesaredif-


ferentiallyexpressedinthemammaryglandduringlactation,ascomparedwiththeliver.Furthermore, anti-sense SPP1 transgenicmice displayed abnormal mammary glanddifferentiation and milk secretion (Nemiret al.,2000).

Schnabel et al. (2005) found concord-ancebasedonfourheterozygousandfourhomozygous sires for the United StatesHolstein population, as determined by agranddaughterdesign,whileCohen-Zinderet al. (2005) found concordance for threeheterozygous and 15homozygous siresfrom both the United States and IsraeliHolsteinpopulations.Cohen-Zinderet al.(2005) also analysed the site proposed bySchnabelet al. (2005), and found that thissite was hyper-variable in that at leastfoursinglenucleotidechangeswerefoundwithin the 20 bp region centred on thepoly-A sequence. Eight of nine Israelisiresanalysedbythedaughterdesignwereheterozygous for at least one of thesepolymorphisms.

Many studies have found a QTLaffectingallfivemilkproductiontraitsandSCSnearthemiddleofBTA20.Blottet al.(2003) claimed that a mis-sense mutationin the bovine growth hormone receptorwas responsible for the QTL affectingmilk yield and composition on BTA20,butdidnotfindconcordanceforthebullsheterozygousfortheQTL.Thus,thispol-ymorphism may be responsible for onlypartof theobservedeffectonBTA20,ormaybeaphysiologicallyneutralmutationinLDwiththeQTN.

For both the QTL on BTA 6 and 14,thepolymorphismsanalysedapparentlydonotaccountfortheentireeffectobservedinthese chromosomal regions (Bennewitz et al.,2004a;Kuhnet al.,2004;Cohen-Zinderet al.,2005).Theeffectassociatedwiththemis-sensemutationinABCG2explainsthe

entireeffectobservedonmilkyieldandfatand protein concentration, but does notexplain the effects associated with fat andprotein yield. It is likely that in the nearfuture additional QTN will be resolved.Asnoted,themeta-analysis(Khatkaret al., 2004)foundsignificanteffectsonBTA1,3,9and10,inadditiontotheeffectsdescribedonBTA6,14and20.

methodS and theory for maS in dairy CattleConsidering the long generation interval,thehighvalueofeach individual, theverylimited female fertility and the fact thatnearly all economic traits are expressedonly in females, it would seem that dairycattle should be a nearly ideal species forapplicationofMAS.However,most theo-retical studieshavebeenratherpessimisticwithrespecttotheexpectedgainsthatcanbe obtained by MAS. As noted by Weller(2001),MAScanpotentiallyincreaseannualgenetic gain by increasing the accuracy ofevaluation, increasing the selection inten-sityanddecreasingthegenerationinterval.

The following dairy cattle breedingschemes that incorporate MAS have beenproposed:• a standard PT system, with informa-

tionfromgeneticmarkersbeingusedtoincrease the accuracy of sire evaluationsin addition to phenotypic informationfrom daughter records (Meuwissen andvanArendonk,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 (MeuwissenandvanArendonk,1992);

• PT schemes in which information ongeneticmarkersisusedtopreselectyoungsires for entrance into the PT (Kashi,


Hallerman and Soller, 1990; MackinnonandGeorges,1998);

• selectionofbullsireswithoutaPT,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 genet-ic markers (Spelman, Garrick and vanArendonk,1999);

• use of genetic markers to reduce errorsin parentage determination (Israel andWeller,2000).Meuwissen and van Arendonk (1992)

foundthatinclusionofmarkerinformationto increase the accuracy of sire evalua-tions increased the rate of genetic gain byonly5percentwhenthemarkersexplained25percent of the genetic variance. ThisresultisnotsurprisingconsideringthattheaccuracyofsireevaluationsbasedonaPTof50to100daughtersisalreadyquitehigh.In “open” and “closed” nucleus breedingschemes,ratesofgeneticgainwereincreasedby 26and 22percent, respectively. Theadvantage of MAS in this case is greater,becauseyoungsiresarenotprogenytested,andtheirreliabilitiesbasedonlyonhalf-sibinformationaremuchlower.

MackinnonandGeorges(1998)proposed“top-down”and“bottom-up”strategies toapply the third scheme listed above, pre-selectionofyoungsirespriortoPT.Inthe“top-down” strategy, QTL genotypes aredetermined for the elite sires used as bullsiresbyagranddaughterdesign. If adensemarkermapisavailable,itwillthenbepos-sibletodeterminewhichQTLalleleispassedto each son. Elite bulls from among thesesons are then selected as bull sires for thenextgeneration.Iftheoriginalsirewashet-erozygousforaQTL,itcanbedeterminedwhich of his sons received the favourableallele.Sonsofthesesiresarethengenotyped

andselectedbasedonwhethertheyreceivedthe favourable grandpaternal QTL alleles.Itisassumedthatthedamsofthecandidatesiresarealsogenotyped,andthatthesecowswillbeprogenyof the sires evaluatedbyagranddaughter design. Thus, grandpaternalallelesinheritedviathecandidates’damscanalsobetraced.Adisadvantageofthisschemeisthatonlythegrandpaternalallelesarefol-lowed.Someofthesonsoftheoriginalsiresthat were evaluated by a granddaughterdesignwillalsohavereceivedthefavourableQTLallelefromtheirdams,butnotviathegenotypedgrandsires.However,youngsireswillbeselectedbasedonlyonthegrandpa-ternalhaplotypes.

Inthe“bottom-up”scheme,QTLgen-otypes of elite sires are determined bya daughter design. These sires are thenused as bull sires. The candidate bullsare then pre-selected for those QTL het-erozygous in their sires, based on whichpaternal haplotype they received. As theQTLphaseisevaluatedonthesiresofthebull 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 hundreddaughtersarerequiredtoevaluateeachbullsire.Onthenegativeside,moredaughtersthansonsmustbegenotypedtodetermineQTL genotype. Mackinnon and Georges(1998)assumedthatineitherschemeitwillnotbenecessary to increasemeangenera-tionintervalabovethatofatraditionalPTprogramme, although this will probablynotbethecase(Weller,2001).

Kashi, Hallerman and Soller (1990),MackinnonandGeorges(1998),andIsraelandWeller(2004)alladdressedtheproblemthat QTL determination will be subjectto error. Deciding that a specific sire ishomozygous for the QTL when in fact


the sire is heterozygous will be denotedthe“typeI”error.DecidingthattheQTLis heterozygous in a specific sire, whilethe sire is in reality homozygous will bedenoted the “type II” error. In the firstcase,segregatingQTLwillbemissedwhile,inthesecondcase,selectionforthepositiveQTL allele will be applied to no advan-tage. All three studies found that geneticgains will be maximized with a relativelylargeproportionof typeIerrors,between5 and 20percent. This is due to the factthatastypeIerrorincreases,typeIIerrordecreases, 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 thetypeIerrorratewassetat20percent.

Spelman, Garrick and van Arendonk(1999) considered three different breedingschemesbydeterministicsimulation:• astandardPTwiththeinclusionofQTL

data;• thesameschemeexceptthatyoungbulls

without PT could also be used as bullsiresbasedonQTLinformation;

• ascheme inwhichyoungsirescouldbeused as both bull sires and cow sires inthe general population, based on QTLinformation.It was assumed that only bulls were

genotyped but that, once genotyped, theinformation on QTL genotype and effectwasknownwithouterror.Itwasthenpos-sibletoconductacompletelydeterministicanalysis. They varied the fraction of thegeneticvariancecontrolledbyknownQTLfrom zero to 100percent. Even withoutMAS,aslightgainwasobtainedbyallowing

young sires tobeused asbull sires, and agenetic gain of 9percent was obtained ifyoungsireswithsuperiorevaluationswerealsouseddirectlyasbothsiresofsiresandingeneralservice.Asnotedpreviously,thegenetic gain was limited where MAS wasusedonlytoincreasetheaccuracyofyoungbull evaluations for a standardPT schemebecause the accuracy of the bull evalua-tionswasalreadyhigh.Thus,evenifallthegeneticvariancewasaccountedforbyQTL,the genetic gain was less than 25percent.However, if young sires are selected forgeneral servicebasedonknownQTL, therateofgeneticprogresscanbedoubled.Themaximum rate of genetic gain that can beobtainedinthethirdscheme,the“allbulls”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 obtainedwasgreaterthanthatpossiblewithnucleusbreedingschemes.

The final scheme, with use of geneticmarkers to reduce parentage errors, is themost certain to produce gains, as it doesnotrelyonQTLgenotypedetermination,which may be erroneous. Weller et al.(2004) genotyped 6 040 Israeli Holsteincows from 181 Kibbutz herds for 104microsatellites. The frequency of rejectedpaternitywas11.7percent,andmosterrorswere due to inseminator mistakes. Mostadvanced breeding schemes already usegenetic markers to confirm parentage ofyoungsires.IsraelandWeller(2002)foundbysimulationsthatiftheparentageofbulldamsandthetestdaughtersofyoungsiresare also verified, genetic gain increasedby 4.3percent compared with a breedingprogramme with 10percent incorrectpaternity. This scheme is economicallyjustified ifgenotypingcostsper individualarenomorethanUS$15.


Current StatuS of maS in dairy CattleTwo ongoing MAS programmes in dairycattle have been reported to date, inFrench and German Holsteins (Boichardet al.,2002,2006;Bennewitzet al.,2004b).Currently in the German programme,markers on three chromosomes are used.The MA-BLUP evaluations (FernandoandGrossman,1989) are computedat theVIT-computing centre in Verden, and aredistributed to Holstein breeders who canuse these evaluations for selection of bulldamsandpreselectionofsiresforprogenytesting. The MA-BLUP algorithm onlyincludesequationsforbullsandbulldams,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 theGermangranddaughterdesignandthecurrentgen-erationofbulls,3600bullsweregenotypedin2002.Asthen,about800bullshavebeenevaluated each year (N.Reinsch, personalcommunication).Onlybullsandbulldamsaregenotypedastissuesamplesarealreadycollected for paternity testing. Thus addi-tionalcostsduetoMASare lowandevenaverymodestgeneticgaincanbeeconomi-callyjustified.Thisschemeissimilartothe“top-down” scheme of Mackinnon andGeorges (1998) in that evaluation of thesonsisusedtodeterminewhichgrandsiresare heterozygous for the QTL and theirlinkage phase. This information is thenused to select grandsons based on whichhaplotype was passed from their sires. ItdiffersfromtheschemeofMackinnonandGeorges (1998) in that the grandsons arepreselected for PT based on MA-BLUPevaluations,whichincludegeneralpedigreeinformationinadditiontogenotypes.

The French MAS programme includes

elements of both the “top-down” and“bottom-up” MAS designs. Similar to theGerman programme, genetic evaluationsincluding marker information were com-puted by a variant of MA-BLUP, andonly genotyped animals and non-geno-typed connecting ancestors were includedin thealgorithm.Genotypedfemaleswerecharacterizedbytheiraverageperformancebased 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. RegionswithputativeQTLaffectingmilkproduc-tionorcompositionarelocatedonBTA3,6, 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, 7and21.EachregionwasfoundtoaffectonetofourtraitsandonaveragethreeregionswithsegregatingQTLwerefoundforeachtrait. Each region is monitored by two tofourevenlyspacedmicrosatellites,andeachanimalincludedintheMASprogrammeisgenotypedforatleast43markers.Siresanddams of candidates for selection, all maleAIancestors,upto60AIunclesofcandi-dates,andsamplingdaughtersofbullsiresandtheirdamsaregenotyped.Thenumberofgenotypedanimalswas8000in2001and50 000 in 2006. An additional 10000ani-mals are genotyped per year, with equalproportionsofcandidatesforselectionandhistoricalanimals.

future proSpeCtive for maS in dairy CattleAlthough the first large experiment inQTL detection in dairy cattle was pub-lished in 1961 by Neimann-Sørensen andRobertson, in 1985 it still looked as if


MAS was a long way off for commercialanimalpopulationsastherewereveryfewknown genetic markers and methodologywasrudimentary.Inthelast20yearstherehave been huge advances in both DNAtechnology and statistical methodology,anditcannowbestatedwithnearcertaintythat the technology is available to detectand map accurately segregating QTL indairy cattle. Furthermore, although manyeffects reported in the literature are “falsepositives”,thereisawealthofevidencethatseveralQTLareinfactrealasanumberofeffectshavebeenrepeatedacrossnumerousexperiments, and the actual QTN havebeenidentifiedforatleasttwoQTL.

The main limitation at this point todetecting and mapping more QTL is thesamplesizesavailable,especiallythenumberofprogenytestedbullsperfamily.TomapQTL of smaller magnitude accurately, itwill be necessary to combine data acrossexperiments (e.g. Khatkar et al., 2004) orsignificantlyincreasesamplesizes.Thiscanonly be done by genotyping cows, eventhough power per individual genotypedwillbelower.

The fact that only two countries haveactually started MAS programmes high-lights the current limitations to practicalapplicationofMAS.Todate,veryfewseg-regatingQTLwith economic impacthave

been identified in commercial dairy cattlepopulations. Of the two QTNs that havebeendetected,eachhasdisadvantageswithrespect to application in MAS. The alleleof DGAT1 that increases fat productionand decreases water content in the milk,bothdesirable,alsodecreasesproteinyield,which is undesirable (Weller et al., 2003).The allele of ABCG2 that decreases milkproduction and increases protein percentisclearlythefavourableallele innearlyallcurrent selection indices, but this allele isalreadyataveryhighfrequencyinallmajordairycattlepopulations(Ronet al.,2006).

In addition to the limitation of defini-tivelyidentifiedQTLwitheconomicvalue,suitable software for genetic evaluationincluding QTL effects is also a limitingfactor.Atpresent, thosecountries thatareapplying MAS are using two-step proce-dures,i.e.apreliminaryanalysistocomputegeneticevaluationsbasedonlyonpedigreeand 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 popula-tionincludingtheeffectsofknownQTL.

aCknowledgementSIthankM.Ron,E.SeroussiandM.Ashwellfortheirinput.

referenCeSAshwell, M.S. & Van Tassell, C.P.1999.Detectionofputativelociaffectingmilk,health,andtype

traitsinaUSHolsteinpopulationusing70microsatellitemarkersinagenomescan.J. Dairy Sci.82:2497–2502.

Ashwell, M.S., Van Tassell, C.P. & Sonstegard, T.S.2001.Agenomescanto identifyquantitativetrait 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.MappingeconomictraitlociforsomaticcellscoreinHolsteincattleusingmicrosatellitemarkersandselectivegenotyping.Anim. Genet.27:235–242.


Ashwell, M.S., Rexroad, C.E., Miller, R.H., VanRaden, P.M. & Da, Y. 1997. Detection of lociaffectingmilkproductionandhealthtraitsinaneliteUSHolsteinpopulationusingmicrosatellitemarkers.Anim. Genet.28:216–222.

Ashwell, M.S., Da, Y., VanRaden, P.M., Rexroad, C.E. & Miller, R.H.1998a.DetectionofputativelociaffectingconformationaltypetraitsinanelitepopulationofUnitedStatesHolsteinsusingmic-rosatellitemarkers.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.Detectionofputativelociaffectingmilkproductionandcomposition,health,andtypetraitsinaUnitedStatesHolsteinpopulation.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.Detectionofquantitativetraitlociaffectingmilkproduction,health,andreproductivetraitsinHolsteincattle.J. Dairy Sci.87:468–475.

Bennewitz, J., Reinsch, N. & Kalm, E. 2003. Comparison of several bootstrap methods for biasreductionofQTLeffectestimates.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.Combinedanalysisofdatafromtwogranddaughterdesigns:AsimplestrategyforQTLconfirmationandincreasingexperimentalpowerindairycattle.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.Markerassistedselection inGermanHolsteindairycattlebreeding:outlineoftheprogramandmarkerassistedbreedingvalueestimation.Ann. Meet. Europ. Ass. Anim. Prod.54thSessionG1.9. 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.TheDGAT1K232Amutationisnotsolelyresponsibleforthemilkproductionquantitativetraitlocusonthebovinechromosome14.J. Dairy Sci.87:431–442.

Bennewitz, J., Reinsch, N., Reinhardt, F., Liu, Z. & Kalm, E.2004b.Topdownpreselectionusingmarkerassistedestimatesofbreedingvaluesindairycattle.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.Moleculardissectionofaquantitativetrait locus:Apheny-lalanine-to-tyrosine substitution in the transmembrane domain of the bovine growth hormonereceptorisassociatedwithamajoreffectonmilkyieldandcomposition.Genetics163:253–266.

Boichard, D., Fritz, S., Rossignol, M.N., Boscher, M.Y., Malafosse, A. & Colleau, J.J. 2002.Implementationofmarker-assistedselectioninFrenchdairycattle.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.Implementationofmarker-assistedselection:practical lessons fromdairycattle.Proc. 8th World Cong. Genet. Appl. Livest. Prod.22:11.13–18August2006,BeloHorizonte,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 threeFrenchdairycattlebreeds.Genet. Sel. Evol.35:77–101.


Brascamp, E.W., van Arendonk J.A.M. & Groen A.F.1993.Economicappraisaloftheutilizationofgeneticmarkersindairycattlebreeding.J. Dairy Sci.76:1204–1213.

Cantet, R.J.C. & Smith, C. 1991.Reducedanimalmodel formarker assisted selectionusingbestlinearunbiasedprediction.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.IdentificationofamissensemutationinthebovineABCG2genewithamajoreffectontheQTLonchromosome6affectingmilkyieldandcompositioninHolsteincattle.Genome Research15:936–944.

Coppieters, W., Kvasz, A., Arranz, J.J., Grisart, B., Riquet, J., Farnir, F. & Georges, M.1999.Thegreat-grand-daughterdesign:asimplestrategytoincreasethepowerofagrand-daughterdesignforQTLmapping.Genetical Research74:189–199.

Cunningham, E.P.1989.Formulationofbreedingplansfordairyanddual-purposecattle.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.AsimplemethodtocalculateresolvingpowerandconfidenceintervalofQTLmaplocation.Behavior. Genet.27:125–132.

Darvasi, A., Vinreb, A., Minke, V., Weller, J.I. & Soller, M.1993.Detectingmarker-QTLlinkageand 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.Mappingofmultiplequantitativetrait locibysimpleregressioninhalf-sibdesigns.J. Anim. Sci.79:616–622.

de Vaccaro, L.1990.SurvivalofEuropeandairybreedsandtheircrosseswithZebusinthetropics.Anim. Breed. Abstr.58:475–494.

Fernando R.L. & Grossman, M.1989.Markerassistedselectionusingbestlinearunbiasedpredic-tion. 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.Extensivegenome-widelinkagedisequilibriumincattle.Genome Research10:220–227.

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.Mappingquantitativetraitlocicontrollingmilkproductionindairycattlebyexploitingprogenytesting.Genetics139:907–920.

Goddard, M.E.1992.Amixedmodelforanalysesofdataonmultiplegeneticmarkers.Theor. Appl. Genet. 83:878–886.

Grignola, F.E., Hoeschele, I. & Tier, B.1996.Mappingquantitativetraitlociinoutcrosspopulationsviaresidualmaximumlikelihood.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.PositionalcandidatecloningofaQTLindairycattle:IdentificationofamissensemutationinthebovineDGAT1genewithmajoreffectonmilkyieldandcomposition.Genome Research12:222–231.


Haley, C.S. & Knott, S.A.1992.Asimpleregressionmethodformappingquantitativetraitlociinlinecrossesusingflankingmarkers.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.MappingofquantitativetraitlocicontrollingtrypanotoleranceinacrossoftolerantWestAfricanN’DamaandsusceptibleEastAfricanBorancattle.Proc. Nat. Acad. Sci. USA100:7443–7448.

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.AgenomescanforQTLinfluencingmilkproductionandhealthtraitsindairycattle.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.MappingofQTLforbodyconformationandbehaviorincattle.J. Heredity94:496–506.

Hill, W.G.1971.Investmentappraisalfornationalbreedingprogrammes.Anim. Prodn.13:37–50.Holmberg, M. & Andersson-Eklund, L. 2004. Quantitative trait loci affecting health traits in

Swedishdairycattle.J. Dairy Sci.87:2653–2659.International Human Genome Sequencing Consortium.2004.Finishingtheeuchromaticsequence

ofthehumangenome.Nature431:931–945.Israel, C. & Weller, J.I.1998.Estimationofcandidategeneeffectsindairycattlepopulations.J. Dairy

Sci.81:1653–1662.Israel, C. & Weller, J.I.2000.Effectofmisidentificationongeneticgainandestimationofbreeding

valueindairycattlepopulations.J. Dairy Sci.83:181–187.Israel, C. & Weller, J.I.2002.Estimationofquantitativetraitlocieffectsindairycattlepopulations.

J. Dairy Sci.85:1285–1297.Israel, C. & Weller, J.I.2004.EffectoftypeIerrorthresholdonmarker-assistedselectionindairy

cattle.Livest. Prod. Sci.85:189–199.Kashi, Y., Hallerman, E. & Soller, M.1990.Marker-assistedselectionofcandidatebullsforprogeny

testingprogrammes.Anim. Prod.51:63–74.Kerr, R. J. & Kinghorn, B.P. 1996.Anefficientalgorithmforsegregatinganalysisinlargepopula-

tions.J. Anim. Breed. Genet.113:457–469.Khatkar, M.S., Thomson, P.C., Tammen, I. & Raadsma, H.W.2004.Quantitativetraitlocimapping

indairycattle:reviewandmeta-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.Quantitativetraitlociaffectingclinicalmastitisandsomaticcellcountindairycattle.Mamm. Genome12:837–842.

Knott, S.A., Elsen J.M. & Haley, C.S.1996.Methodsformultiple-markermappingofquantitativetraitlociinhalf-sibpopulations.Theor. Appl. Genet.93:71–80.

Kosgey, I. S., Kahi, K.A. & van Arendonk J.A.M.2005.Evaluationofclosedadultnucleusmultipleovulationandembryotransferandconventionalprogenytestingbreedingschemesformilkpro-ductionintropicalcrossbredcattle.J. Dairy Sci.88:1582–1594.


Kuhn, 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.QuantitativetraitlocimappingoffunctionaltraitsintheGermanHolsteincattlepopulation.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 betterexplainsaquantitativetiptraitlocuswithmajoreffectonmilkfatcontentincattle.Genetics 167:1873–1881.

Lander, E.S. & Botstein, D.1989.MappingMendelianfactorsunderlyingquantitative traitsusingRFLPlinkagemaps.Genetics 121:185–199.

Mackay, T.F.2001.Thegeneticarchitectureofquantitativetraits.Ann. Rev. Genet.35:303–339.Mackinnon, M.J. & Georges, M.A.J.1998.Marker-assistedpreselectionofyoungdairysirespriorto

progeny-testing.Livest. Prod. Sci.54:229–250.Mackinnon, M.J. & Weller, J.I.1995.Methodologyandaccuracyofestimationofquantitativetrait

lociparametersinahalf-sibdesignusingmaximumlikelihood.Genetics141:755–770.Madalena, F.E.1993.Asimpleschemetoutilizeheterosisintropicaldairycattle.World Anim. Rev.

74/75:17–25.Martinez, O. & Curnow, R.N.1992.Estimatingthelocationsandthesizesoftheeffectsofquantita-

tivetraitlociusingflankingmarkers.Theor. Appl. Genet.85:480–488.Meuwissen, T.H.E. & Goddard, M.E.2000.Finemappingofquantitativetraitlociusinglinkagedis-

equilibriawithcloselylinkedmarkerloci.Genetics 155:421–430.Meuwissen, T.H.E. & Goddard, M.E.2004.MappingmultipleQTLusinglinkagedisequilibriumand

linkageanalysisinformationandmultitraitdata.Genet. Sel. Evol36:261–279.Meuwissen, T.H.E. & van Arendonk, J.A.M.1992.Potentialimprovementsinrateofgeneticgain

frommarkerassistedselectionindairycattlebreedingschemes.J. Dairy Sci.75:1651–1659.Meuwissen, T.H.E., Karlsen, A., Lien, S., Olsaker, I. & Goddard, M.E. 2002.Finemappingof a

quantitativetraitlocusfortwinningrateusingcombinedlinkageandlinkagedisequilibriummap-ping.Genetics 16:373–379.

Mosig, M.O., Lipkin, E., Khutoreskaya, G., Tchourzyna, E., Soller, M. & Friedmann, A.2001.AwholegenomescanforquantitativetraitlociaffectingmilkproteinpercentageinIsraeli-Holsteincattle,bymeansofselectivemilkDNApoolinginadaughterdesign,usinganadjustedfalsedis-coveryratecriterion.Genetics 157:1683–1698.

Nadesalingam, J., Plante, Y. & Gibson, J.P.2001.DetectionofQTLformilkproductiononchro-mosomes1and6ofHolsteincattle.Mammal. Genome12:27–31.

Neimann-Sørensen, A. & Robertson, A.1961.TheassociationbetweenbloodgroupsandseveralproductioncharactersinthreeDanishcattlebreeds.Acta. Agr. Scand.11:163–196.

Nemir, M., Bhattacharyya, D., Li, X., Singh, K., Mukherjee, A.B. & Mukherjee, B.B.2000.Targetedinhibitionofosteopontinexpressioninthemammaryglandcausesabnormalmorphogenesisandlactationdeficiency.J. Biol. Chem.275:969–976.

Nicholas, F.W. & Smith C.1983.Increasedratesofgeneticchangeindairycattlebyembryotransferandsplitting.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.


2002.AgenomescanforquantitativetraitlociaffectingmilkproductioninNorwegiandairycattle.J. Dairy Sci. 85:3124–3130.

Plante, Y., Gibson, J.P., Nadesalingam, J., Mehrabani-Yeganeh, H., Lefebvre, S., Vandervoort, G. & Jansen, G.B.2001.Detectionofquantitativetraitlociaffectingmilkproductiontraitson10chromosomesinHolsteincattle.J. Dairy Sci.84:1516–1524.

Powell, R.L. & Norman, H.D.1988.Accuracyofcowindexesaccordingtorepeatability,evaluation,herdyield,andregistrystatus.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.DetectionandanalysisofalocusaffectingmilkconcentrationintheUSandIsraelidairycattlepopulations.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.Multiplequantitativetraitlocusanalysisofbovinechromosome6intheIsraeliHolsteinpopulationbyadaughterdesign.Genetics159:727–735.

Ron, M., Cohen-Zinder, M., Peter, C., Weller, J.I. & Erhardt, G.2006.ABCG2polymorphisminBos indicusandBos tauruscattlebreeds.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.AcompletegenomescanoftheIsraeliHolsteinpopu-lationforquantitativetraitlocibyadaughterdesign.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-mappingmilkproductionquantitativetraitlocionBTA6:Analysisofthebovineosteopontingene.Proc. Nat. Acad. Sci. USA102:6896–6901.

Schrooten, C., Bovenhuis, H., Coppieters, W. & van Arendonk, J.A.M.2000.Wholegenomescantodetectquantitativetraitlociforconformationandfunctionaltraitsindairycattle.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.QuantitativetraitlociforhealthtraitsinFinnishAyrshirecattle.J. Dairy Sci.87:443–449.

Smith, C.1988.Geneticimprovementoflivestock,usingnucleusbreedingunits.World Anim. Rev.65:2–10.

Song, J-Z., Soller, M. & Genizi, A.1999.Thefull-sibintercrossline(FSIL):aQTLmappingdesignforoutcrossingspecies.Genet. Research73:61–73.

Spelman, R.J., Garrick, D.J. & van Arendonk, J.A.M. 1999. Utilization of genetic variation bymarkerassistedselectionincommercialdiarycattlepopulations.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-Friesianpopulation.Genetics 144:1799–1808.

Spelman, R.J., Huisman, A.E., Singireddy, S.R., Coppieters, W., Arranz, J., Georges, M. & Garrick, D.J.1999.Quantitativetraitlocianalysison17nonproductiontraitsintheNewZealanddairypopulation.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 devia-tionsandderegressedproofswithinawholegenomescanforQTL.J. Anim. Breed. Genet.118:357–370.


van Arendonk, J.A.M., Bovehuis, H., van der Beek, S. & Groen, A.F. 1994.Detectionandexploi-tationofmarkerslinkedtoquantitativetraitsinfarmanimals.Proc. 5th World Cong. Genet. Appl. Livest. Prod.21:193–200.Guelph,Canada.

VanRaden, P.M. & Wiggans, G.R.1991.Derivation,calculationanduseofnationalanimalmodelinformation.J. Dairy Sci.74:2737–2746.

Velmala, R.J., Vilkki, H.J., Elo, K.T., de Koning, D.J. & Maki-Tanila, A.V. 1999.Asearchforquan-titative trait 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.QuantitativetraitlociaffectingmilkproductiontraitsinFinnishAyrshiredairycattle.J. Dairy Sci.86:1828–1836.

Vilkki, H.J., de Koning, D.J., Elo, K.T., Velmala, R. & Maki-Tanila, A.1997.Multiplemarkermap-pingofquantitativetraitlociofFinnishdairycattlebyregression,J. Dairy Sci.80:198–204.

Visscher, P.M., Haley, C.S. & Thompson, R. 1996. Marker-assisted introgression in backcrossbreedingprograms.Genetics144:1923–1932.

Visscher, P.M., Thompson, R. & Haley, C.S.1996.ConfidenceintervalsinQTLmappingbyboot-strapping.Genetics 143:1013–1020.

Weller, J.I.1986.Maximumlikelihoodtechniquesforthemappingandanalysisofquantitativetraitlociwiththeaidofgeneticmarkers.Biometrics42:627–640.

Weller, J.I.1994.Economic aspects of animal breeding.London,Chapman&Hall.244pp.Weller, J.I.2001.Quantitative trait loci analysis in animals.London,CABIPublishing.287pp.Weller, J.I. & Soller, M.2004.Ananalytical formula toestimateconfidence intervalofQTLloca-

tionwithasaturatedgeneticmapasafunctionofexperimentaldesign.Theor. Appl. Genet.109:1224–1229.

Weller, J. I., Kashi, Y. & Soller, M.1990.Powerof“daughter”and“granddaughter”designsforgeneticmappingofquantitativetraitsindairycattleusinggeneticmarkers.J. Dairy Sci.73:2525–2537.

Weller, J.I., Shlezinger, M. & Ron, M.2005.Correctingforbiasinestimationofquantitativetraitlocieffects.Genet. Sel. Evol.37:501–522.

Weller, J.I., Soller, M. & Brody, T. 1988. Linkage analysis of quantitative traits in an interspecificcrossoftomato(L. esculentumxL. pimpinellifolium)bymeansofgeneticmarkers.Genetics118:329–339.

Weller, J.I., Song, J.-Z., Heyen, D.W., Lewin, H.A. & Ron M.1998.Anewapproachtotheproblemofmultiplecomparisonsinthegeneticdissectionofcomplextraits.Genetics150:1699–1706.

Weller, J.I., Weller, H., Kliger, D. & Ron, M.2002.Estimationofquantitativetraitlocusallelefre-quencyviaamodifiedgranddaughterdesign.Genetics162:841–849.

Weller, J.I., Golik, M., Seroussi, E., Ezra, E. & Ron, M.2003.Population-wideanalysisofaQTLaffectingmilk-fatproductionintheIsraeliHolsteinpopulation.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.FactorsaffectingincorrectpaternityassignmentintheIsraeliHolsteinpopulation.J. Dairy Sci.87:2627–2640.

Westell, R.A. & van Vleck, L.D.1987.Simultaneousgeneticevaluationofsiresandcowsforalargepopulationofdairycattle.J. Dairy Sci.70:1106–1017.

Whittaker, J.C., Thompsom, R. & Visscher, P.M.1996.OnthemappingofQTLbyregressionofphenotypeonmarker-type.Heredity 77:23–32.


Winter, A., Kramer, W., Werner, F.A.O., Kollers, S., Kata, S., Durstewitz, G., Buitkamp, J., Womack, J.E., Thaller, G. & Fries, R.2002.Associationofalysine-232/alaninepolymorphisminabovinegeneencodingacyl-CoA:diacylglycerolacyltransferase(DGAT1)withvariationataquan-titativetraitlocusformilkfatcontent.Proc. Nat. Acad. Sci. USA99:9300–9305.

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.Mappingquantitativetraitlociformilkproductionandhealthofdairycattleinalargeoutbredpedigree.Genetics149:1959–1973.

marker-assisted selection in livestock – case studiesused to enhance selection, have raised high...

Documents