introduction to marker-assisted selection4 Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fish Summary This chapter provides

Section i

introduction to marker-assisted selection

Chapter 1

marker-assisted selection as a tool for genetic improvement of

crops, livestock, forestry and fish in developing countries:

an overview of the issues

John Ruane and Andrea Sonnino

Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fish4

SummaryThischapterprovidesanoverviewofthetechniques,currentstatusandissuesinvolvedinusing marker-assisted selection (MAS) for genetic improvement in developing countries.Molecular marker maps, the necessary framework for any MAS programme, have beenconstructed for the majority of agriculturally important species, although the densityof the maps varies considerably among species. Despite the considerable resources thathavebeeninvestedinthisfieldanddespitetheenormouspotentialitstillrepresents,withfewexceptions,MAShasnotyetdelivered its expectedbenefits in commercialbreedingprogrammesforcrops,livestock,foresttreesorfarmedfishinthedevelopedworld.Whenevaluating the potential merits of applying MAS as a tool for genetic improvement indevelopingcountries,someoftheissuesthatshouldbeconsideredareitseconomiccostsandbenefits,itspotentialbenefitscomparedwithconventionalbreedingorwithapplicationofotherbiotechnologies,andthepotentialimpactofintellectualpropertyrights(IPRs)onthedevelopmentandapplicationofMAS.

Chapter 1 – An overview of the issues 5

introduCtionThe potential benefits of using markerslinked to genes of interest in breedingprogrammes,thusmovingfromphenotype-based towards genotype-based selection,have been obvious for many decades.However, realization of this potential hasbeen limited by the lack of markers. WiththeadventofDNA-basedgeneticmarkersin the late 1970s, the situation changedand researchers could, for the first time,begintoidentifylargenumbersofmarkersdispersedthroughoutthegeneticmaterialofanyspeciesofinterestandusethemarkerstodetectassociationswithtraitsofinterest,thus allowing MAS finally to become areality. This led to a whole new field ofacademic research, including the milestonepaperbyPatersonet al.(1988).Thisshowedthat with the availability of large numbersof genetic markers for their species ofinterest (tomato), the effects and locationofmarker-linkedgeneshavinganimpactona number of quantitative traits (fruit traitsin their case) could be estimated using anapproach that could be applied to dissectthe genetic make-up of any physiological,morphological and behavioural trait inplantsandanimals.

Most of the traits considered in animalandplantgeneticimprovementprogrammesarequantitative, i.e. theyarecontrolledbymany genes together with environmentalfactors,andtheunderlyinggeneshavesmalleffects on the phenotype observed. Milkyield and growth rate in animals or yieldandseedsizeinplantsaretypicalexamplesof quantitative traits. In classical geneticimprovementprogrammes,selectioniscar-ried out based on observable phenotypes

ofthecandidatesforselectionand/ortheirrelativesbutwithoutknowingwhichgenesare actually being selected. The develop-ment of molecular markers was thereforegreetedwithgreatenthusiasmasitwasseenasamajorbreakthroughpromisingtoover-comethiskey limitation.AsYoung(1999)wrote:“BeforetheadventofDNAmarkertechnology, the idea of rapidly uncoveringthe loci controlling complex, multigenictraitsseemedlikeadream.Suddenly,itwasdifficult to open a plant genetics journalwithoutfindingdozensofpapersseekingtopinpoint many, if not most, agriculturallyrelevantgenes.”However,despitethecon-siderableresourcesthathavebeeninvestedinthisfieldanddespitetheenormouspoten-tial it still represents, with few exceptions,MAShasnotyetdelivereditsexpectedben-efits in commercial breeding programmesfor crops, livestock, forest trees or farmedfish in thedevelopedworld. Indevelopingcountries, where investments in molecularmarkers have been far smaller, delivery ofbenefitshaslaggedevenfurtherbehind.

Thefocusofthischapterisontheuseofmolecularmarkersforgeneticimprovementof populations through MAS, includingmarker-assisted introgression. Its aim is toprovide an easily understandable overviewof the techniques, applications and issuesinvolved in the use of DNA markers inMASforgeneticimprovementofdomesticplantandanimalpopulationsindevelopingcountries.Inthenextsectionofthechapter,a brief description of the technical aspectsofmolecularmarkersandMASisprovided.The current status of the application ofMAS in crops, forestry, livestock and fishisthensummarized,whilethefinalsection

Note: This chapter is based on the Background Document to Conference 10 (on molecular marker-assistedselection as a potential tool for genetic improvement of crops, forest trees, livestock and fish in developingcountries) of the FAO Biotechnology Forum, 17November–14 December 2003 (available at www.fao.org/biotech/C10doc.htm).


highlights issues that might be importantto applications of MAS in developingcountries.Althoughmolecularmarkersmaybeusedforawiderangeofdifferenttasks,such as to quantify the genetic diversityand relationships within and betweenagricultural populations (e.g.livestockbreeds), to investigate biological processes(suchasmatingsystems,pollenmovementor seed dispersal in plants) or to identifyspecificgenotypes(e.g.clonedforesttrees),theseapplicationsarenotconsideredhere.

BaCkground to maSmolecular markersAll living organisms are made up of cellsthat are programmed by genetic materialcalled DNA. This molecule is made up ofa long chain of nitrogen-containing bases(therearefourdifferentbases–adenine[A],cytosine[C],guanine[G]andthymine[T]).OnlyasmallfractionoftheDNAsequencetypicallymakesupgenes, i.e.thatcodeforproteins, while the remaining and majorshare of the DNA represents non-cod-ingsequences, theroleofwhich isnotyetclearlyunderstood.Thegeneticmaterial isorganized into sets of chromosomes (e.g.fivepairsinArabidopsis thaliana;30pairsinBos taurus[cow]),andtheentiresetiscalledthe genome. In a diploid individual (i.e.wherechromosomesareorganizedinpairs),there are two alleles of every gene– onefromeachparent.

Molecular markers should not be con-sideredasnormalgenesastheyusuallydonothaveanybiologicaleffect.Instead,theycan be thought of as constant landmarksinthegenome.Theyare identifiableDNAsequences,foundatspecificlocationsofthegenome, and transmitted by the standardlaws of inheritance from one generationto the next. They rely on a DNA assay,in contrast to morphological markers that

arebasedonvisibletraits,andbiochemicalmarkersthatarebasedonproteinsproducedbygenes.

Different kinds of molecular markersexist, such as restriction fragment lengthpolymorphisms (RFLPs), random ampli-fiedpolymorphicDNA(RAPDs)markers,amplified fragment length polymorphisms(AFLPs), microsatellites and single nucle-otide polymorphisms (SNPs). They maydiffer in a variety of ways – such as theirtechnical requirements (e.g. whether theycan be automated or require use of radi-oactivity); the amount of time, moneyand labour needed; the number of geneticmarkers that can be detected throughoutthe genome; and the amount of geneticvariation found at each marker in a givenpopulation. The information provided tothebreederbythemarkersvariesdependingonthetypeofmarkersystemused.Eachhasitsadvantagesanddisadvantagesand,inthefuture,othersystemsarelikelytobedevel-oped.MoredetailsontheindividualmarkersystemsareprovidedinChapter3.

from markers to maSThe molecular marker systems describedabove allow high-density DNA markermaps(i.e.withmanymarkersofknownloca-tion,interspersedatrelativelyshortintervalsthroughout thegenome) tobeconstructedfor a range of economically importantagricultural species, thus providing theframeworkneededforeventualapplicationsofMAS.

Using the marker map, putative genesaffecting traits of interest can then be de-tectedby testing for statistical associationsbetween marker variants and any trait ofinterest. These traits might be geneticallysimple–forexample,manytraitsfordiseaseresistanceinplantsarecontrolledbyoneora few genes (Young, 1999). Alternatively,


they could be genetically complex quan-titative traits, involving many genes (i.e.so-called quantitative trait loci [QTL])and environmental effects. Most economi-cally important agronomic traits tend tofall into this latter category. For example,using 280molecular markers (comprising134RFLPs, 131AFLPs and 15microsat-ellites) and recording populations of ricelines for various plant water stress indica-tors, phenology, plant biomass, yield andyieldcomponentsunderirrigatedandwaterstressconditions,Babuet al.(2003)detecteda number of putative QTL for droughtresistancetraits.

Having identified markers physicallylocated beside or even within genes ofinterest, in thenextstep it isnowpossibletocarryoutMAS,i.e.toselectidentifiablemarker variants (alleles) in order to selectfor non-identifiable favourable variants ofthegenesofinterest.Forexample,considerahypothetical situationwhereamolecularmarker M (with two alleles M1 and M2),identified using a DNA assay, is knownto be located on a chromosome close toa gene of interest Q (with a variant Q1that increases yield and a variant Q2 thatdecreases yield), that is, as yet, unknown.If a given individual in the population hastheallelesM1andQ1ononechromosomeandM2andQ2ontheotherchromosome,then any of its progeny receiving the M1allelewillhaveahighprobability(howhighdependsonhowcloseMandQaretoeachotheronthechromosome)ofalsocarryingthe favourable Q1 allele, and thus wouldbepreferredforselectionpurposes.Ontheotherhand,thosethatinherittheM2allelewilltendtohaveinheritedtheunfavourableQ2 allele, and so would not be preferredfor selection. With conventional selectionwhichreliesonphenotypicvalues,itisnotpossibletousethiskindofinformation.

The success of MAS is influenced bythe relationship between the markersand the genes of interest. Dekkers (2004)distinguishedthreekindsofrelationship:• Themolecularmarker is locatedwithin

thegeneofinterest(i.e.withinthegeneQ, using the example above). In thissituation,onecanrefer togene-assistedselection (GAS). This is the mostfavourable situation for MAS since, byfollowing inheritance of the M alleles,inheritance of the Qalleles is followeddirectly.Ontheotherhand,thesekindsofmarkersarethemostuncommonandarethusthemostdifficulttofind.

• Themarkeris inlinkagedisequilibrium(LD) with Q throughout the wholepopulation.LDisthetendencyofcertaincombinationsofalleles(e.g.M1andQ1)to be inherited together. Population-wide LD can be found when markersand genes of interest are physicallyvery close to each other and/or whenlines or breeds have been crossed inrecentgenerations.SelectionusingthesemarkerscanbecalledLD-MAS.

• The marker is not in linkage disequi-librium (i.e. it is in linkage equilibrium[LE])withQthroughoutthewholepop-ulation. Selection using these markerscanbecalledLE-MAS.ThisisthemostdifficultsituationforapplyingMAS.TheuniversalnatureofDNA,molecular

markers and genes means that MAS can,in theory, be applied to any agriculturallyimportant species. Indeed, active researchprogrammeshavebeendevotedtobuildingmolecularmarkermapsanddetectingQTLsforpotentialuse inMASprogrammes inawhole range of crop, livestock, forest treeand fish species. In addition, MAS can beapplied to support existing conventionalbreeding programmes. These programmesusestrategiessuchas:recurrentselection(i.e.


usingwithin-breedorwithin-lineselection,important in livestock); development ofcrossbreds or hybrids (by crossing severalimprovedlinesorbreeds)andintrogression(where a target gene is introduced from,for example, a low-productive lineor breed (donor) into a productive line(recipient) that lacks the target gene (astrategy especially important in plants).SeeDekkersandHospital (2002) formoredetails.MAScanbe incorporated intoanyone of these strategies (e.g. for marker-assisted introgression by using markers toaccelerate introduction of the target gene).Alternatively,novelbreedingstrategiescanbedevelopedtoharnessthenewpossibilitiesthatMASraises.

Current StatuS of appliCationS of maS in agriCultureBelow is a brief summary of the currentstatus regarding application of MAS inthedifferentagriculturalsectors.Formoredetails, a number of case studies for cropsarepresentedinSectionIIofthebookandfor livestock, forestry and fish in SectionsIII,IVandV,respectively.

CropsThe promise of MAS has possibly beengreetedwiththemostenthusiasmandexpec-tation in this particular agricultural sector,stimulating tremendous investments in thedevelopment of molecular marker mapsandresearchtodetectassociationsbetweenphenotypesandmarkers.Molecularmarkermaps have been constructed for a widerangeofcropspecies.Informationonmajorplantprojects(suchasthesequencingoftheentire rice genome) can be found at www.ncbi.nlm.nih.gov/genomes/PLANTS/PlantList.html.

Inarecentreview,however,DekkersandHospital (2002) noted that “as theoretical

andexperimentalresultsofQTLdetectionhave accumulated, the initial enthusiasmfor the potential genetic gains allowed bymolecular genetics has been tempered byevidence for limits to the precision of theestimatesofQTLeffects”,andthat“overall,there are still few reports of successfulMAS experiments or applications.” Theyreportedthatmarker-assisted introgressionofknowngeneswaswidelyusedinplants,particularly by private breeding compa-nies,whereasmarker-assistedintrogressionofunknowngeneshadoftenproved tobeless useful in practice than expected. AsYoung(1999)wrote:“eventhoughmarker-assisted selection now plays a prominentrole in the field of plant breeding, exam-ples of successful, practical outcomes arerare. It is clear that DNA markers holdgreat promise, but realizing that promiseremainselusive.”

There is also considerable divergencewith respect to the applications of MASamongdifferentcropspecies.Forexample,Koebner (2003) highlighted the relativelyfastuptakeofMASinmaizecomparedwithwheat and barley, arguing that this largelyreflected the breeding structure. Thus,whereas maize breeding is dominated inindustrializedcountriesbyasmallnumberof large private companies that produceF1 hybrids, a system allowing protectionfrom farm-saved seed and competitor use,breedingfor theothermajorcerealspeciesisprimarilybypublic sectororganizationsandmostvarietiesareinbredpurebreedinglines,asystemallowinglessprotectionoverthe released varieties. Progress in arablecrops is nevertheless quite advanced com-pared with horticultural crop species suchasapplesandpears,wheredevelopmentofmolecularmarkermapshasbeenslowandonlyfewQTLhavebeendetected(Tartarini,2003),evenifMAScanpotentiallybevery


useful for genetic improvement of suchlong-cycleplants.

livestockAgain, much effort has been put into thedevelopment of molecular marker mapsin this sector. The first reported map inlivestock was for chicken in 1992, whichwasquicklyfollowedbythepublicationofmapsforcattle,pigsandsheep.Sincethen,the search for useful markers has contin-uedandfurtherspecieshavebeentargeted,including goat, horse, rabbit and turkey(see www.thearkdb.org/ for the currentstatus regarding somemajor livestock spe-cies). Microsatellite markers have been ofmajorimportance.

Dekkers(2004)recentlyreviewedcom-mercial applications of MAS in livestockand noted that several gene or markertests are available on a commercial basisindifferentspeciesandfordifferenttraits,andthatthemajorityofusesinvolveGAS,where an important gene (e.g. responsibleforacongenitaldefect)hasbeenidentifiedor,toalesserdegree,LD-MAS.Hepointedout that documentation is poor since,althoughseveralgenetictestsareavailable,theextent towhichtheyareused incom-mercial applications is unclear, as is themannerinwhichtheyareusedandwhethertheiruseleadstogreaterresponsestoselec-tion.Heconcludedthat“opportunitiesfortheapplicationofMASexist, inparticularfor GAS and LD-MAS and, to a lesserdegree, for LE-MAS because of greaterimplementation requirements. Regardlessof the strategy, successful application ofMAS requires a comprehensive integratedapproachwithcontinuedemphasisonphe-notypic recording programmes to enableQTL detection, estimation and confirma-tion of effects, and use of estimates inselection.Althoughinitialexpectationsfor

theuseofMASwerehigh,thecurrentatti-tudeisoneofcautiousoptimism.”

forestryAs for crops, extensive efforts have beendevotedtoconstructionofmolecularmarkermaps for the major commercial genera,suchaseucalypts,pinesandacacia.RFLPs,RAPDs, microsatellites and AFLPs havebeenextensivelyused.TheWebsitehttp://dendrome.ucdavis.edu/index.php providesupdatedinformationonthestatusregardingmolecularmarkermapsinforestry.

Molecularmapshavebeenusedtolocatemarkers associated with variation in for-estry traits of commercial interest, such asgrowth, frost tolerance, wood properties,vegetativepropagation,leafoilcompositionand disease resistance. Since MAS allowsearly selection before traits of interest(e.g.wood quality) are expressed, a majorincentive for using molecular techniquesin tree breeding is to improve the rate ofgenetic gain by reducing the long gen-eration interval However, Butcher (2003)notedthat“MAShasyettobeincorporatedin operational breeding programmes forplantation species” and she referred to thehigh costs of genotyping, the large familysizes required to detect QTL and the lackof knowledge of QTL interactions withgenetic background, tree age and environ-mentasexplanatoryfactors.

In a recent review of biotechnology inforestry,Yanchuk(2002)alsohighlightedthepotentialadvantageofearlyselectionusingMAS,butagainpointedoutthatMASisnotyetbeingappliedroutinelyintreebreedingprogrammes,largely“becauseofeconomicconstraints(i.e.theadditionalgeneticgainsaregenerallynotlargeenoughtooffsetthecostsofapplyingthetechnology).Thusitislikely thatMASwillonlybeapplied forahandfulofspeciesandsituations,e.g.afew


of the major commercially used pine andEucalyptus species. Molecular markers arethereforeprimarilyaninformationtoolandareusedtolocateDNA/genesthatcanbeofinterestforgenetictransformation,orinfor-mation on population structure, matingsystemsandpedigreeconfirmation.”

fishMolecular marker maps have been con-structed for a number of aquaculture spe-cies, e.g. tilapia, catfish, giant tiger prawn,kuruma prawn, Japanese flounder andAtlantic salmon, although their density isgenerallylow.Densityishighfortherain-bowtrout,wherethemappublishedin2003has over 1300markers spread throughoutthegenome–thevastmajorityareAFLPsbut it also includes over 200microsatellitemarkers(Nicholset al.,2003).SomeQTLsofinteresthavebeendetected(e.g.forcoldandsalinitytoleranceintilapiaandforspe-cificdiseasesinrainbowtroutandsalmon).InarecentreviewofMASinfishbreedingschemes, Sonesson (2003) suggested thatMASwouldbeespeciallyvaluablefortraitsthatareimpossibletorecordonthecandi-datesforselectionsuchasdiseaseresistance,fillet quality, feed efficiency and sexualmaturation,andconcludedthatMASisnotused in fish breeding schemes today andthatthelackofdensemolecularmapsisthelimitingfactor.

ConclusionsMolecular marker maps, the necessaryframework for any MAS programme,have been constructed for the majorityof agriculturally important species but thedensity of the maps varies considerablyamong species. Currently, MAS does notplay a major role in genetic improvementprogrammes in any of the agriculturalsectors. Enthusiasm and optimism remain

concerning the potential contributionsthat MAS offers for genetic improvement.However, this seems to be tempered bytherealizationthatitmaybemoredifficultand therefore take longer than originallythought before genetic improvement ofquantitativetraitsusingMASisrealized.TheconclusionsfromthereviewbyDekkersandHospital(2002)areagoodreflectionofthis:“Furtheradvancesinmoleculartechnologyandgenomeprogrammeswillsooncreateawealthofinformationthatcanbeexploitedfor the genetic improvement of plants andanimals. High-throughput genotyping,for example, will allow direct selection onmarker information based on population-wide LD. Methods to effectively analyseandusethisinformationinselectionarestillto be developed. The eventual applicationof these technologies in practical breedingprogrammeswillbeonthebasisofeconomicgrounds, which, along with cost-effectivetechnology,willrequirefurtherevidenceofpredictableandsustainablegeneticadvancesusing MAS. Until complex traits can befully dissected, the application of MASwill be limited to genes of moderate-to-large effect and to applications that donotendanger the response toconventionalselection.Untilthen,observablephenotypewill remain an important component ofgenetic improvementprogrammes,becauseittakesaccountofthecollectiveeffectofallgenes.”

Some faCtorS relevant to applying maS in developing CountrieSInthedebateontheroleorvalueofMASasapotentialtoolforgeneticimprovementin developing countries, some of thepotential factors thatshouldbeconsideredare described briefly below, as they mayinfluenceapplicationsofthetechnology.


economic factors As with any new technology promisingincreased benefits, the costs of applica-tionmustalsobeconsidered.AccordingtoDekkers and Hospital (2002), “economicsisthekeydeterminantfortheapplicationofmolecular genetics in genetic improvementprogrammes.Theuseofmarkersinselectionincursthecoststhatareinherenttomolecu-lartechniques.ApartfromthecostofQTLdetection, which can be substantial, costsforMAS include thecostsofDNAcollec-tion,genotypingandanalysis.”Forexample,Koebner (2003) suggested that the currentcosts of MAS would need to fall consider-ablybeforeitwouldbeusedwidelyinwheatand barley breeding. In practice, therefore,althoughMASmayleadtoincreasedgeneticresponses,decision-makersneedtoconsiderwhetheritmaybecost-effectiveorwhetherthemoneyandresourcesspentondevelop-ing and applying MAS might instead bemore efficiently used on improving exist-ing conventional breeding programmes oradoptingothernewtechnologies.

Little consideration has been givento this issue. Some results have, how-ever, been published recently from studiesat the International Maize and WheatImprovementCenter(CIMMYT)inMexicoon the relative cost-effectiveness of con-ventional selection and MAS for differentmaize breeding applications. One applica-tionconsideredbyMorriset al.(2003)wasthetransferofanelitealleleatasingledom-inantgenefromadonorlinetoarecipientline. Here, conventional breeding is lessexpensive but MAS is quicker. For situa-tions like this, where the choice betweenconventional breeding and MAS involvesa trade-off between time and money, theysuggested that the cost-effectiveness ofusingMASdependsonfourparameters:therelative cost of phenotypic versus marker

screening;thetimesavedbyMAS;thesizeandtemporaldistributionofbenefitsasso-ciatedwithacceleratedreleaseofimprovedgermplasm and, finally, the availability tothebreedingprogrammeofoperatingcap-ital. They conclude that “all four of theseparameters can vary significantly betweenbreeding projects, suggesting that detailedeconomicanalysismaybeneededtopredictinadvancewhichselectiontechnologywillbeoptimalforagivenbreedingproject.”

In the applications considered byCIMMYT, the costs of developing molec-ular markers associated with the trait ofinterest were not considered, as it wasassumed that they were already available.Thereisadistinctionbetweendevelopmentcosts(e.g.identifyingmolecularmarkersonthegenome,detectingassociationsbetweenmarkers and the traits of interest) andrunning costs (typing individuals for theappropriate markers in the selection pro-gramme) of MAS. Development costs canbe considerable, so developing countriesneed to consider whether to develop theirowntechnologyor,alternatively,toimportthe technology developed elsewhere, ifavailable.

AnotheraspecttobeconsideredishowtoevaluatetheeconomicbenefitsofMAS.Fora publicly-funded breeding programme, itshouldincludeeconomicbenefitstofarmersfromgeneticimprovementoftheirplantsoranimals.Forprivatecompaniesontheotherhand, the impacts of using MAS on theirmarket share, and not on rates of geneticimprovement,wouldbeofgreatestinterest.

The economics of MAS are consideredin more detail later, in particular inChapter19.

maS versus conventional methodsAlthough conventional breeding pro-grammes that rely on phenotypic records


have their limitations, they have shownovertimethattheycanbehighlysuccessful.Application of MAS will not occur in avacuumandthepotentialbenefits(genetic,economic, etc.) of using MAS need to becomparedwith thoseachievedorexpectedfrom any existing conventional breedingprogrammes.

Inthedifferentagriculturalsectors,thisquestionhasreceivedmuchattentionfromresearchers.Thereseemstobegeneralcon-sensus that the relative success of MAScomparedwithconventionalbreedingmaydependonthekindoftrait(ortraits)tobegenetically improved. If the trait is diffi-culttorecordorisnotroutinelyrecordedin conventional programmes, MAS willoffermoreadvantagesthanifitisroutinelyrecorded.Similarly,ifthetraitissex-limitedor can only be measured late in life thenMAS is favoured, as marker informationcanbeusedinbothsexesandatanyage.

InconsideringthemeritsofMASversusconventional breeding, it is also impor-tant to keep in mind that the existence ofa strong breeding programme is a pre-requisite for the application of advancedmolecular technologies such as MAS. Insituations where the infrastructure andcapacity are insufficient to support a suc-cessful conventional breeding programme,MASwillnotprovideashortcuttogeneticimprovement.

maS versus other biotechnologies for genetic improvement The relative costs and benefits of apply-ing MAS should be compared not onlywith conventional breeding but also withtheuseofothernewtechnologiesthatcanpotentially improve agricultural popula-tions genetically. These include tissue cul-tureincropsandforesttrees,reproductivetechnologies(e.g.embryotransferorclon-

ing)inlivestockandtriploidizationorsex-reversal in farmed fish. They also includegeneticmodification,atechnologythatcanbe applied to all sectors. Compared withgenetic modification, regulation of MAS,be it at the level of research and develop-ment, field testing, commercial release orimport/export of developed products, ismore relaxed; in addition, public accept-anceofthetechnologyisnotanissue.

intellectual property rights issuesAsdiscussed inConference6of theFAOBiotechnology Forum (FAO, 2001), theissueof intellectualproperty rights (IPRs)is playing an ever greater role in foodand agriculture in developing countries.Participants in that conference, inter alia,suggestedthatthisissuewashavingagen-erally negative influence on the quality ofagriculturalresearchcarriedoutandonthenature of research collaborations betweenthe public and private sector and betweendevelopinganddevelopedcountries.

It is therefore obvious that IPRs mayalsohaveanimpactonthedevelopmentandapplicationofMASindevelopingcountries.For example, theAFLPmolecularmarkermapping technique is patented. Molecularmarkerscanbepatented,althoughthiscanoftenbeovercomebyusingothermarkersnear the gene of interest. Individual genescanalsobepatented.WithIPRs,however,there is nevertheless public disclosure ofthe inventionor information.Non-disclo-sureofinformation,wherepatentsarenotsoughtbut the informationonmarkersordetected QTL is nevertheless kept secret,canalsohavenegativeimpacts,bydenyingdeveloping countries access to potentiallyusefulinformation.

Moredetailson IPRs andMAScanbefoundinChapter20.


referenCeSBabu, R.C., Nguyen, B.D., Chamarerk, V., Shanmugasundaram, P., Chezhian, P., Jeyaprakash, P.,

Ganesh, S.K., Palchamy, A., Sadasivam, S., Sarkarung, S., Wade, L.J. & Nguyen, H.T. 2003.Geneticanalysisofdroughtresistanceinricebymolecularmarkers:associationbetweensecondarytraitsandfieldperformance.Crop Sci.43:1457–1469.

Butcher, P.A.2003.Molecularbreedingoftropicaltrees.In A.Rimbawanto&M.Susanta,eds. Proc. Int. Conf. Adv. in Genetic Improvement of Tropical Tree Species,1–3October2002.Centre forForestBiotechnologyandTreeImprovement,Yogyakarta,Indonesia.

Dekkers, J.C.M. 2004. Commercial application of marker- and gene-assisted selection inlivestock: strategies and lessons. J. Anim. Sci. 82: E313–E328 (available at http://jas.fass.org/cgi/reprint/82/13_suppl/E313)

Dekkers, J.C.M. & Hospital, F.2002.Theuseofmoleculargeneticsintheimprovementofagricul-turalpopulations.Nature Revs. Genet.3:22–32.

FAO.2001.Agricultural biotechnology for developing countries – results of an electronic forum,byJ.Ruane&M.Zimmermann.FAOResearchandTechnologyPaperNo.8.Rome(alsoavailableatwww.fao.org/docrep/004/Y2729E/Y2729E00.htm).

Koebner, R.2003.MASincereals:greenformaize,amberforrice,stillredforwheatandbarley.In Proc. Int. Workshop on Marker-Assisted Selection: A Fast Track to Increase Genetic Gain in Plant and Animal Breeding?(availableatwww.fao.org/biotech/docs/Koebner.pdf).

Morris, M., Dreher, K., Ribaut, J-M. & Khairallah, M. 2003. Money matters (II): costs ofmaizeinbredlineconversionschemesatCIMMYTusingconventionalandmarker-assistedselection. Mol. Breed.11:235–247.

Nichols, K.M., Young, W.P., Danzmann, R.G., Robison, B.D., Rexroad, C., Noakes, M., Phillips, B., Bentzen, P., Spies, I., Knudsen, K., Allendorf, F.W., Cunningham, B.M., Brunelli, J., Zhang, H., Ristow, S., Drew, R., Brown, K.H., Wheeler, P.A. & Thorgaard, G.H. 2003. A consolidatedlinkagemapforrainbowtrout(Oncorhynchus mykiss).Anim. Genet.34:102–115.

Paterson A.H., Lander, E.S., Hewitt, J.D., Peterson, S., Lincoln, S.E. & Tanksley, S.D. 1988.ResolutionofquantitativetraitsintoMendelianfactorsbyusingacompletelinkagemapofrestric-tionfragmentlengthpolymorphisms.Nature335:721–726.

Sonesson, A.K.2003.Possibilitiesformarker-assistedselectioninfishbreedingschemes.InProc. Int. Workshop on Marker-Assisted Selection: A Fast Track to Increase Genetic Gain in Plant and Animal Breeding?(availableatwww.fao.org/biotech/docs/Sonesson.pdf).

Tartarini, S. 2003. Marker-assisted selection in pome fruit breeding. In Proc. Int. Workshop on Marker-Assisted Selection: A Fast Track to Increase Genetic Gain in Plant and Animal Breeding?(availableatwww.fao.org/biotech/docs/Tartarini.pdf).

Yanchuk, A.2002.Theroleandimplicationsofbiotechnologyinforestry.Forest Genetic Resources30:18–22(availableatwww.fao.org/docrep/005/Y4341E/Y4341E06.htm).

Young, N.D. 1999. A cautiously optimistic vision for marker-assisted breeding. Mol. Breed. 5:505–510.

Chapter 2

an assessment of the use of molecular markers in developing countries

Andrea Sonnino, Marcelo J. Carena, Elcio P. Guimarães, Roswitha Baumung, Dafydd Pilling and Barbara Rischkowsky


SummaryFourdifferentsourcesofinformationwereanalysedtoassessthecurrentusesofmolecularmarkersincrops,foresttreesandlivestockindevelopingcountries:theFAOBiotechnologyinDevelopingCountries(FAO-BioDeC)databaseofbiotechnologyindevelopingcoun-tries; country reports evaluating the current status of applied plant breeding and relatedbiotechnologies; country reports on animal genetic resources management for preparingthe First Report on the State of the World’s Animal Genetic Resources (SoW-AnGR);andtheresultsofaquestionnairesurveyonanimalgeneticdiversitystudies.Evenifstilllargelyincomplete,thecurrentdatashowthatmolecularmarkersarewidelyusedforplantbreedinginthedevelopingworldandmostprobablytheirusewillincreaseinthefuture.In the animal sector the use of molecular markers seems less developed and limited orabsent in most developing countries. Major differences exist among and within regionsregardingtheapplicationofmolecularmarkertechniquesinplantandanimalbreedingandgenetics.These canbe explainedby the relativelyhigh investments in infrastructure andhumanresourcesnecessarytoundertakeresearchinthesefields.Thespectrumofapplica-tionofmolecularmarkers incropplants isquitewide,coveringmanyplants relevant totheenhancementoffoodsecurity,butotherimportantplantspeciesarestillneglected.Thepracticalresultsofmarker-assistedselection(MAS)inthefieldaredisappointinglymodest,possiblydue to: low levelsof investment; limitedcoordinationbetweenbiotechnologistsandpracticalbreeders;instable,non-focusedorill-addressedresearchprojects;andthelackoflinkagesbetweenresearchandfarmers.Partnershipsbetweendevelopedanddevelopingcountriesmaybeameansofbetterrealizingthepotentialofmolecularmarkertechniquesforimprovingbothanimalandcropproduction.

Chapter 2 – An assessment of the use of molecular markers in developing countries 17

introduCtionAssessmentsrelatingtotheuseofmolecularmarkers in crop plants are based on twosources of information: (i) FAO-BioDeC,a searchable database of biotechnologyproducts and techniques inuse and in thepipelineindevelopingandtransitioncoun-tries (available at www.fao.org/biotech/inventory_admin/dep/default.asp); and(ii)FAO country reports produced bynational agricultural research systems(NARS)aspartofasurveyofcountryinfor-mationandtrendsinresourcesallocatedforappliedplantbreedingandrelatedbiotech-nology,with theaimof raisingawareness,evaluating opportunities for investmentand designing national, regional and/orglobalstrategiestostrengthenthecapacityof national plant breeding programmes(Guimarães,KuenemanandCarena,2006).

AstheFAO-BioDeCdatabasecontainslittle information on the use of molecularmarkers in relation to animals, it is evenmore difficult to give a comprehensiveoverview of the situation with respect tolivestockindevelopingcountriesthanit isforcrops.However,informationontheuseofmolecularmarkerswasdrawnfromthecountryreportsonanimalgeneticresources(AnGR) management submitted to FAOas part of the preparation of the FirstReportontheStateoftheWorld’sAnimalGeneticResources(SoW-AnGR)andfromaquestionnairesurveyongeneticdiversitystudies.Thecountryreportscoveredawidevariety of aspects of AnGR managementandcontainonlyquitegeneralinformationabouttheroleofmoleculartechniques.Thequestionnairesurveylookedspecificallyatthe use of molecular markers in livestockgenetic diversity studies and was directedtoresearchersinvolvedinsuchstudies.Assuch,itgivesanindicationofwheregeneticdiversitystudiesarebeingundertakenand

which markers are primarily used, but itdoesnotprovideacompletepicture.

While this book focuses on the useof markers to assist in genetic selection(MAS),itisoftendifficulttoobtainspecificinformationontheextenttowhichmarkersare used for this purpose in developingcountries.Forthisreason,someofthedatapresented in this chapter cover theoveralluse of molecular markers in developingcountries anddonot allowdiscriminationbetween molecular markers used forselection from uses for other purposes,such as the descriptive studies of geneticdiversity within populations or geneticdistance between populations. Other datapresentedheredescribetheuseofmolecularmarkers for measuring genetic diversityonly. In this case, the information can beconsidered as an indicator of the humancapacity and infrastructure available foruseofmarkersinMAS.Forthesereasons,anddue to the incompletenatureof someofthe informationavailable, thisoverviewshouldbeconsideredpreliminary,butstillmeaningful.

fao-BiodeCAt the time of writing (September 2006),FAO-BioDeC includes 2336 entries re-lated to crops and 829 entries related toforest trees.Thedatabasecurrentlycovers74developing countries, including coun-trieswitheconomiesintransition.

Noquantitativeinformationisavailableconcerningthehumancapacityorfundinginvolvedinanyresearchinitiative.Activitiescarried out in developed countries or atinternationalresearchcentres,suchasthosethat are part of the Consultative Groupon International Agricultural Research(CGIAR),arenotconsidered.

To compile the data in FAO-BioDeC,several sources of information were


consulted (for a complete description seeFAO, 2005). In particular, information onplantbiotechnologyproductsandtechniqueswas gathered from a survey undertakenin Latin America by the InternationalServiceforNationalAgriculturalResearch(ISNAR)andfromcountrybiotechnologystatus assessment reports prepared forFAO in South and Southeast Asia, AfricaandtransitioncountriesinEasternEurope.Other information was obtained fromcountryreportsandpublishedliterature.

The initial biotechnology applicationdataobtainedwasclassifiedonacountry/regional/continentalbasis,byspecies, traitanalysedortechniqueused,andbywhethertheapplicationwasintheresearchorfieldtestingphasesorwasalreadycommerciallyreleased.

FAO-BioDeC currently contains677entries related to the use of molec-ular marker techniques, 489 of which areassociated with crop plants and 188 with

forest trees. Table1 suggests that earlygenerationDNA-basedmolecularmarkerssuch as randomly amplified polymorphicDNAs(RAPDs)aremorewidelyusedthanthe more recently developed markers, e.g.amplified fragment length polymorphisms(AFLPs), while isozymes are still largelyusedintheforestrysector.

Onlyinfivecaseshavetheresearchini-tiatives reported reached the final stage ofdevelopment,givingrisetocommercializedproducts(Table2).TheseareonevarietyofanunspecifiedornamentalplantreleasedinBrazil; one variety of rice commercializedinIndonesia;onestrainofRhizobium etli,the soil bacterium inducing the formationofnitrogen-fixingnoduleson the rootsofa common bean obtained in Mexico; onericevarietycontainingpyramidedgenesforbacterial leaf blight resistance obtained inthe Netherlands Antilles; and one varietyofanunspecifiedforesttreeinBurundi.In115cases(107inthecropsectorandeight

table 1number of research initiatives utilizing genetic markers in the crop and forestry sectors sorted by type of markers

markers Crop forestry total

RFlP 61 9 70

RaPD 158 15 173

SSRs/Microsatellites 68 19 87

aFlP 65 3 68

isozymes 2 50 52

chloroplast Dna markers 0 11 11

rDna (ribosomal Dna sequences) 0 4 4

other or not specified 135 77 212

total 489 188 677

table 2number of research initiatives utilizing genetic markers in the crop and forestry sectors according to the development stage of the technique or product

phase Crop forestry total

experimental phase 344 179 523

Field tests 107 8 115

commercial phase 4 1 5

Unspecified 34 0 34

total 489 188 677


for forest trees), the research initiativeshave reached the field test stage, while in523 cases (344 of which are related to thecropsector),theyareatearlierstages.

The use of molecular markers is wide-spreadinLatinAmericaandtheCaribbeanwith molecular research being reportedfrom ten countries. Special emphasis ison the crop sector and includes applica-tions on Andean local roots and tubers,sugar cane, rice, cocoa, banana, bean andmaize (Table 3). In the Asia and Pacificregion, research activities with molecularmarkers focus on forest trees, sugar cane,rice, jute,banana,coconutandwheat.TheFAO-BioDeC database shows that, whileresearch involving molecular markers inAfrica is under way in only a few coun-tries including Ethiopia, Nigeria, SouthAfrica and Zimbabwe, the crops understudy range from traditional commoditiestotropicalfruits.MolecularresearchintheNear East and North Africa is reportedforonlysixcountriesand focusesondatepalm,durumandbreadwheat,rice,barleyand olive trees. In transition countries ofEastern Europe, molecular markers targetseveralcropplantsincludingwheat,maize,pulses,vegetablesandtobaccoacrosssevencountries.

Table 4 shows that most attentionfocuses on cereals, especially durumand bread wheat, barley, maize and rice.Other important cereal or pseudo-cerealspecies such as sorghum, amaranthus and

buckwheat receive less attention and noresearch initiatives are reported for teffor millets. Among the pulses, molecularresearch projects are reported for beans(18),chickpea(5),cowpea(9)andsoybean(7) and little or no attention is dedicatedto lentil, pigeon pea, faba bean and otherlocallyimportantleguminousplantssuchasbambaragroundnut.Amongrootandtubercrops, potato, sweet potato and cassavaattract the most research effort involvingmolecularmarkers,butsomeresearchisalsoundertaken on Andean roots and tubers.Fewornorecordsareavailableforrootandtuber species important for food securityinmanydevelopingcountriessuchasyam,taro(ordasheen),cocoyamandotheraroids.Researchonfruittreesinvolvingmolecularmarkersincludestropicalfruittreessuchasbanana,cocoa,coconutandpapaya,aswellasplantsmoretypicaloftemperateclimatessuch as strawberry and apple, while less

table 3number of research initiatives utilizing genetic markers in the crop and forestry sectors according to geographical origin

region Crop forestry total

africa 52 17 69asia and Pacific 98 103 201europe (transition countries) 42 13 55latin america and caribbean 249 55 304near east and north africa 48 0 48total 489 188 677

table 4number of research initiatives utilizing genetic markers according to the crop of application

Crop group number of projects

cereals and pseudo-cereals 134Pulses 54Root and tubers 51Fruit trees 53Vegetables 29industrial crops 74Fodder crops 16aromatics 5other or not specified 73total 489


research was reported for citrus, mango,pineappleandmanyotherfruittreeslargelycultivated in developing countries. Severalresearch initiatives are applying molecularmarkers to industrialcropspecies, suchassugarcane,cotton,rubber,jute,coffee,flaxandoilpalm.

fao plant Breeding and related BioteChnology CapaCity aSSeSSmentIn2002,adraftquestionnairewasdesignedtogathercountryinformationonresourceallocationtrendsinplantbreedingandbio-technology related activities. Later in thesame year, a group of experts includingrepresentatives from CGIAR centres, thepublic and private sectors and non-gov-ernmental organizations (NGOs), met atFAO headquarters to discuss the natureof the information tobecollectedand theprocedure for its collection. This resultedin a questionnaire being developed andsenttoallpublicandprivateappliedplantbreedingprogrammesaswellastobiotech-nologylaboratoriesindevelopingcountriesand countries in transition. Among otherissues, the survey gathered informationon the number of full-time equivalentplant breeders and biotechnologists avail-ableduringeachfive-yearperiodbeginningfrom1985.Thequestionnairealsorequestedinformationconcerningtrendsofresourcesallocated to biotechnology as well as togermplasm improvement (pre-breeding),line development and line evaluation.One of the objectives of the survey wasto assess the gap between biotechnologytools and their successful deployment inappliedbreedingprogrammes(Guimarães,KuenemanandCarena, 2006).The surveythereforealsoconcentratedonprioritiesforbreeding,potentialinternationalsupporttostrengthen national breeding programmes,

the number of varieties released and thefactors that are most likely to limit thesuccess of applied plant breeding pro-grammes, including the current status ofbiotechnology. The work of gatheringthe information and preparing a technicalreport on the current status of nationalplant breeding and related biotechnologywasassignedtoawell-knownandrespectednational plant breeding scientist. This hasbeenthekeytoidentifyinggapsinordertodevelopstrategiesforstrengtheningeffortsdirected at the sustainable use of plantgenetic resources for food and agriculture(PGRFA)innationalprogrammes.

For the purposes of this chapter, bio-technology data were gathered from25countries to complement the prelimi-nary assessments based on FAO-BioDeCon theuseofmolecularmarkers indevel-opingcountries(Table5).Thedatagatheredindicate that tissue culture is the mostcommonbiotechnologytechniqueasitwasused in 88percent of all cases, followedby MAS (44percent), the double-hap-loid technique (32percent), interspecificcrosses (28percent), molecular characteri-zation(24percent)andgeneticengineering(12percent).

Applications of molecular markersinclude a number of categories withinbiotechnology such as MAS, molecularcharacterization, facilitating genetic engi-neeringandtrackingdesirablechromosomesegments when making wide crosses (e.g.interspecificcrosses).TheresultsinTable5suggest that molecular markers might bean integral part of developing countryagricultural efforts. MAS seems to be thesecond most utilized biotechnology toolapplied after tissue culture, implying thatemphasisshouldbegiventothedevelopmentof molecular markers to make selectionmoreefficient.However,rapidandefficient


advancementofplantbreedingeffortsmightnot be achieved through MAS because ofthe complexity encountered in multitraitand multistage selection for economicallyimportant traits. Consequently, today inthe developed world, molecular markersdo not have a prominent role in breedingprogrammes(Hallauer,1999).

uSe of moleCular teChniqueS in angr managementFAO invited 188 countries to participatein the preparation of the First Reporton the SoW-AnGR. One hundred andsixty-ninecountryreports(CR)onAnGRwere submitted (available at www.fao.org/dad-is/).

The countries were offered guidelinesforthepreparationofthecountryreports,onesectionofwhichwastobedevotedtoreviewing the state of national capacitiesand assessing future capacity buildingrequirements(FAO,2001).Countrieswereassigned to seven regions on the basisof the regional classification establishedby FAO for the purpose of preparingthe SoW-AnGR. This analysis considered148countryreportsavailablebyJuly2005,of which 42 were from Africa, 25 fromAsia, 39 from Europe and the Caucasus,22fromLatinAmericaandtheCaribbean,7 from the Near and Middle East, 2 fromNorthAmericaand11fromtheSouthwestPacific(Pillinget al.,2007).

table 5Biotechnology applications in plant genetic resources for food and agriculture in use in 25 developing countries

Country tC maS iC dh mC ge

algeria X1 X X X n2 nangola X n n n n narmenia X n X X X ncameroon X X n n n ncosta Rica X n X n X XDominican Republic n n n n n nethiopia X X n X n nGeorgia X X X n X nGhana X X n n n nMali X n n n n nKenya X X X X n XMalawi X n n n n nMoldova X n n n n nMozambique n n n n n nnicaragua X X n n X nniger X X n n n nnigeria X X X X X nSenegal X n X X n nSierra leone X n n n n nSri lanka n n n n n nSudan X n n n n ntunisia X X n X n nUzbekistan X n n X X nZambia X n n n n nZimbabwe X X n n n X

1 one or more institutions in the country are using the tool. However, this does not measure its impact. 2 not in use.tc = tissue culture; MaS = marker-assisted selection; ic = interspecific crosses; DH = double-haploid technology; Mc = molecular characterization; Ge = genetic engineering


Not surprisingly, the information pro-videdbythecountryreportsindicatesthatthereisalargegapbetweendevelopedanddevelopingcountriesintermsofcapacitytoutilizemolecularmarkersforthestudyandmanagementofAnGR(Table6).Comparedwith other developing regions, a higherpercentage of countries from Asia andLatinAmericaandtheCaribbeanreportedtheir use. In Africa, the Southwest Pacific(excludingAustralia),theNearandMiddleEast,andEasternEuropeandtheCaucasus,very fewcountries report theuseof thesetechnologies, the prominent exception inthe last case being Ukraine which hascarriedoutmolecularcharacterizationandgenetic distance studies on a number oflivestockspecies(CRUkraine,2004).

In Africa, only four countries describetheexistenceofcharacterizationorgeneticdistancestudiesbasedontheuseofmolecularmarkersandinallcasesthestudiesrelatetolocal breeds. One country report indicatesthat local breeds of goat, pig and chickenarethesubjectofmolecularcharacterizationcarriedoutabroad.InnocaseistheuseofMASreportedfromthisregion.

Excluding Japan, seven Asian coun-tries (out of 15 providing information onwhether or not the technologies are used)reportmolecularmarkerstudies,ofwhichfivespecifygeneticdistancestudiesandonementionsresearchintoMAS(CRMalaysia,

2003). A range of species are the subjectof molecular characterization, the mostcommonbeingcattle,chickens,sheep,goatsand pigs; however, some studies involvingbuffaloes, ducks, horses, camels or deerare also reported. Systematic studies ofAsian breeds are being conducted by theSociety for Research on Native LivestockinJapan,includinganalysisofgeneticrela-tionships based on mitochondrial DNApolymorphisms and other DNA markers(CRJapan,2003).

In Latin America and the Caribbean,11 countries out of the 15 that providedinformationindicatesomeuseofmolecularmarkers. Among nine countries providinginformation on the species involved inmolecular characterization studies, sevenmentioned cattle while smaller numbersmention sheep, pigs, chickens, horses,goats, buffaloes, llamas, alpacas, vicuñasorguanacos.Severalcountries indicatetheinclusionoflocallyadaptedbreedsinsuchstudies,but therewas little indicationthatmolecularmarkershavebeenincorporatedwithin breeding programmes. However,the report from Colombia (2003) notedthe potential significance of MAS pro-grammes for utilizing the genes of theBlanco Orejinegro cattle breed, which isreported to show resistance to brucellosisand which has been the subject of molec-ularcharacterization.

table 6use of molecular markers reported in country reports on angr management

region number providing

information

reporting use of molecular

markers %

number with information on

species

reporting use of molecular markers

in cattle %

in other species %

europe 29 83 18 89 100africa 29 14 3 100 33asia 16 50 7 86 100latin america and the caribbean 15 73 9 78 89Southwest Pacific 9 11 0 - -north america 2 100 1 100 100near and Middle east 5 40 2 0 100


Apart from Australia, no countries intheSouthwestPacificregionreporttheuseofmolecularmarkers.

IntheNearandMiddleEastonereport(CR Jordan, 2003) refers to molecularcharacterization and genetic distancestudies in indigenous goats, while another(CREgypt,2003) notes that molecularstudies of buffalo, sheep and goats hadrecently been initiated with the aid ofregionalandinternationalorganizations.

Survey on the uSe of moleCular markerS in genetiC diStanCe StudieS in liveStoCkMorespecificanddetailed informationonthe use of molecular markers in AnGRresearch was obtained from a question-nairestudylaunchedin2003.Onehundredand thirty-two questionnaires were sentout via e-mail to research teams that hadbeen involved in genetic distance studiesduring the past ten years. The researcherswere identified througha literature searchand enquiry via several Internet discus-sion groups. The points covered in thesurveywere:numberofbreedsandsamplesizes; number and type of markers used;additional breed information such as phe-notypic traits or geographic spread; andthe mathematical and statistical methodschosenformeasuringgeneticdistance.Thestudy also aimed to verify the degree offamiliarityandacceptanceofmeasurementofdomesticanimaldiversity(MoDAD)rec-ommendations, which had been proposedas standards for genetic diversity studiesby the International Society for AnimalGenetics(ISAG)andFAOabouttenyearsearlier (FAO, 1998a; b). Compliance withthe recommendations was seen as impor-tant as it would enable the compilationof results from different genetic distancestudies.

Informationon87geneticdistancestud-ies was obtained from 57 researchers. Thestudiescoveredbreedsfrom13mammalianandavian species and investigated samplesfrom 93 countries; the largest number ofcountrieswasinEurope,followedbythosein Asia and the Pacific (Table 7). Most ofthestudiesfocusedonruminants.Thesizeof the projects varied between one and120breedsoriginatingfromupto33coun-tries.However,a largenumberofnationalprojectsfocusedonbreedswithinaspecificcountry or region. There were also a fewlargeinternationalprojectsinvolvingcattleand goats (Table8). A smaller number ofpigandchickenprojectswereimplemented.Nofeedbackwasreceivedregardingbreedsofllamas,ducks,turkeysorgeese.

Withregardtocompliancewiththerec-ommendationsoftheFAO/ISAGadvisorygroup, 95percent of all projects aimed tofulfiltheminimumrequirementofsampling25animalsperbreed.Althoughmicrosatel-litemarkerswereusedin90percentofthestudies,inonly23percentwereallmarkerstaken from the recommended marker list.In about 57percent of studies some rec-ommended microsatellites were used. Thedegree of acceptance of the recommen-dations was highest in pigs and lowest inchickens.MoredetailedinformationontheresultsisgivenbyBaumung,SimianerandHoffmann(2004)andFAO(2004).

table 7number of countries where samples were collected for angr genetic distance studies

fao region number of countries

africa 13

asia and the Pacific 19

europe 37

latin america and the caribbean 10

near east 9

north america 2

total 93


ConCluSionSEvenifstilllargelyincomplete,thecurrentdata allow some general conclusions tobe drawn regarding the use of molecularmarkers in agricultural research anddevelopmentindevelopingcountries.

Molecular markers are widely utilizedintheplantproductionsectorofthedevel-oping world even if the present uptake ofmolecular marker technologies does notreflecttheiractualpotential.Itmightthere-forebespeculatedthatasignificantincreaseintheirutilizationmightbeexpectedinthenear future. However, it is recommendedthateachtechniqueiscarefullyassessedforits actual potential for improving the effi-ciency of plant breeding and germplasmcharacterization.Untilthisisdemonstrated,the use of molecular markers would bea costly investment with limited returns.Publishingallmarkerresearchthathasnotbeensuccessfulisalsostronglyencouragedin order to avoid potential failures and/orimportinginappropriatetechnologiesfromdevelopedcountries.

Majordifferencesexistbetweenregions(andwithinregions)regardingtheapplica-

tionofmolecularmarkertechniquesinplantbreedingandgenetics.Whilesomecountrieshave developed quite extensive researchprogrammes, vast geographical areas, par-ticularly in Africa, remain excluded fromthese technological advancements or cancountonlyonminimalactivities.Thiscanbeexplainedbytherelativelyhighinvestmentsininfrastructureandhumanresourcesnec-essary to undertake research in this field.Highcostscanalsobeindicatedasacauseof the low technological level of geneticmarker research inmanycountries,whichfocus on isozymes or on restriction frag-ment length polymorphisms (RFLPs) andhave not yet adopted the more advancedpolymerase chain reaction (PCR)-basedmarkers. However, the life span of PCR-basedmarkersisveryshortanditmightbebettertowaituntilimprovedmarkerssuchassinglenucleotidepolymorphisms(SNPs)becomeavailable.Thespectrumofapplica-tion of molecular markers to crop plantsin developing countries is quite wide andcoversmanyplantspeciesthatarerelevantfor the enhancement of food security orfor the improvement of farmers’ incomes

table 8number of projects and countries in which samples were collected according to animal species and fao regions

Species number of projects

number of countries

fao region

buffalo 3 9 africa, asia and the Pacific, europe, latin america and the caribbean cattle 24 40 africa, asia and the Pacific, europe, latin america and the caribbeanGoat 11 28 africa, asia and the Pacific, europe, latin america and the caribbeanSheep 19 56 africa, asia and the Pacific, europe, latin america and the caribbeanPig 6 19 africa, asia and the Pacific, europe ass 1 1 europe Horse 5 25 africa, asia and the Pacific, europe, latin america and the caribbean,

north america bactrian camel 1 2 asia and the Pacific Dromedary 2 7 africa, near east alpaca 3 2 near east, latin america and the caribbean Rabbit 1 19 africa, asia and the Pacific, europe chicken 8 34 africa, asia and the Pacific, europe, latin america and the caribbean,

near east Yak 2 8 asia and the Pacific, europe, near east


in tropical areas. However, other impor-tantplantspeciesarestillneglectedbytheongoingresearchinitiatives.

AccordingtothedatareportedinFAO-BioDeC, only five products obtainedthroughtheuseofmolecularmarkershavebeencommerciallyreleasedtodateindevel-opingcountries.Even ifmorecommercialproductshavebeenreleasedbutaremissingfromthedatabase,suchasthosereportedbyToenniessen, O’Toole and DeVries (2003)or others obtained by the internationalagricultural research centres or the pri-vate sector, the totalityofpractical resultsobtained from using molecular markersis disappointingly modest compared withthe declared potential of the approach.The reasons for the poor results to dateare multiple and include: the low level ofinvestmentsinbothbiotechnologyresearchand applied plant breeding; the limitedcoordination between biotechnology lab-oratories and plant breeding programmes;managerial and political frailties leadingto instable, unfocused or ill-addressedresearch projects; legal, infrastructural ortechnical weaknesses of the seed produc-tion and commercialization systems; andthe lack of linkages between research andpractical application of research productsbyfarmers.

Appliedplantbreedingshouldcontinuetobethefoundationfortheapplicationofmolecularmarkers.Focusingusefulmolec-ulartechniquesontherighttraitswillbuilda strong linkage between genomics andplantbreedinginordertoproducenewandbettercultivars.Therefore,morethanever,thereistheneedforbettercommunicationand cooperation among scientists in plantbreeding and biotechnology. Public plantbreeding and biotechnology programmesin developing countries are being seri-ously eroded through lack of funding.

Thislossofpublicsupportaffectsbreedingcontinuity and objectivity and, equallyimportantly, the training of future plantbreedersandbiotechnologistsand theuti-lization and improvementofplantgeneticresourcescurrentlyavailable.Thefactthatpoor farmers rely on public and privatebreedinginstitutionsforsolvinglong-termchallenges should influence policy-makersto reverse the trend of reduced funding.Cooperation between industry and publicinstitutions is a promising approach tofollow. Ensuring strong applied breedingprogrammes incorporating the applicationof molecular markers will be essential inensuring the sustainable use and enhance-mentofplantgeneticresources.

AnGR management shows a similarpatterntotheuseofMASinplantbreedingmanagementintermsofthedifferencesthatexistamongregionsintheuseofmolecularmarker techniques. Within several regionstherearealsodifferencesbetweenmoreandless developed countries. The reasons aresimilar to those mentioned above, namelya lack of financial, human and technicalresources. In particular, human capacitiesin animal genetics and breeding are muchsmaller than those existing in the cropsector.Consequently,theuseofmoleculartechniquestoevaluategeneticresources,toplanconservationefforts,ortofacilitatetheachievementofdesiredbreedingobjectivesis limited or absent in most developingcountries.

Nevertheless,countryreportsexpressedastrongdesiretodevelopgreatercapacityto carry out molecular studies of nationalAnGR, and the responses to the FAOquestionnairealsoindicatedahighlevelofinterest in doing so. For the near future,microsatellite loci will remain the mostusefultypeofgeneticmarkerforgeneticdis-tancestudiesandforgenetic improvement


programmes but SNPs were singled outaspromisingmarkers for the future.Withpartnershipsbetweendevelopedanddevel-oping countries within or across regions,geneticdiversitystudiesmaybeameansofrealizingthepotentialofmolecularmarkertechniquestoimprovedecision-makingonbreed development and the prioritizationofbreedsforconservationprogrammes.

The successful application of MAS inanimal breeding necessitates a high levelof expenditure in terms of establishmentandmaintenancecostsandrequiresskilledhuman resources, equipment, laboratoriesandsupportiveinfrastructure.Assuch,thecost-effectivenessofthesestrategieshastobe carefully evaluated before promotingtheminresource-poorenvironments.

referenCeSBaumung, R., Simianer, H. & Hoffmann, I. 2004. Genetic diversity studies in farm animals – a

survey.J. Anim. Breed. Genet.121:361-373.FAO.Countryreportsonthestateofanimalgeneticresources(availableatwww.fao.org/dad-is/).FAO. 1998a. Secondary guidelines for development of National Farm Animal Genetic Resources

Management Plans. Measurement of domestic animal diversity (MoDAD): original Working Group report.Rome.

FAO. 1998b. Secondary guidelines for development of National Farm Animal Genetic Resources Management Plans. Measurement of domestic animal diversity (MoDAD): recommended microsatellite markers.Rome.

FAO. 2001. Preparation of the first report on the state of the world’s animal genetic resources.GuidelinesforthepreparationofCRs.Rome.

FAO.2004.Measurement of domestic animal diversity – a review of recent diversity studies. Documentprepared for the third session of the Intergovernmental Working Group on Animal GeneticResources. Rome (available at www.fao.org/ag/againfo/programmes/en/genetics/documents/ITWG3_Inf3.pdf).

FAO. 2005.Status of research and application of crop biotechnologies in developing countries: prelimi-nary assessment, byZ.Dhlamini,C.Spillane,J.P.Moss,J.Ruane,N. Urquia&A.Sonnino.Rome(availableatwww.fao.org/docrep/008/y5800e/y5800e00.htm).

Guimarães, E.P., Kueneman, E. & Carena, M.J.2006.AssessmentofnationalplantbreedingandbiotechnologycapacityinAfricaandrecommendationsforfuturecapacitybuilding.Hort. Sci.41:50–52.

Hallauer, A.R.1999.Temperatemaizeandheterosis.In J.G.Coors&S.Pandey,eds.The genetics and exploitation of heterosis in crops, pp.353-361.Proc.Internat.Symp.ofHeterosisinCrops.MexicoCity,18–22August1997.ASA,CSSAandSSSA,Madison,WI,USA.

Pilling, D., Cardellino, R., Zjalic, M., Rischkowsky, B., Tempelman, K.A. & Hoffmann, I. 2007.Theuseofreproductiveandmolecularbiotechnologyinanimalgeneticresourcesmanagement–aglobaloverview.Anim. Genet. Resources Information.FAO,Rome(inpress).

Toenniessen G.P., O’Toole, J.C. & DeVries, J. 2003.Advancesinplantbiotechnologyanditsadop-tionindevelopingcountries.Curr. Opin. in Plant Biol.6:191–198.

Section ii

marker-assisted selection in crops – case studies

Chapter 3

molecular markers for use in plant molecular breeding and

germplasm evaluation

Jeremy D. Edwards and Susan R. McCouch


SummaryAnumberofmolecularmarkertechnologiesexist,eachwithdifferentadvantagesanddisad-vantages.Whenavailable,genomesequenceallowsforthedevelopmentofgreaternumbersandhigherqualitymolecularmarkers.Whengenomesequenceislimitedintheorganismofinterest,relatedspeciesmayserveassourcesofmolecularmarkers.Somemolecularmarkertechnologies combine the discovery and assay of DNA sequence variations, and there-forecanbeusedinspecieswithouttheneedforpriorsequenceinformationandup-frontinvestmentinmarkerdevelopment.Asaprerequisiteformarker-assistedselection(MAS),theremustbeaknownassociationbetweengeneticmarkersandgenesaffectingthephe-notypetobemodified.Comparativedatabasescanfacilitatethetransferofknowledgeofgeneticmarker-phenotypeassociationacrossspeciessothatdiscoveriesinonespeciesmaybeappliedtomanyothers.FurthergenomicsresearchandreductionsinthecostsassociatedwithmolecularmarkerswillcontinuetoprovidenewopportunitiestoemployMAS.

Chapter 3 – Molecular markers for use in plant molecular breeding and germplasm evaluation 31

introduCtionMolecular markers are valuable tools forthe classification of germplasm and inMAS. The purpose of this chapter is toprovide guidance in selecting appropriatemolecular marker systems based on theavailability of technological resources invariousspeciesandtoprovidesomeexam-plesofMASapplications.OneofthemanybenefitsoftheincreasingamountofDNAsequenceinformationinmanyorganismsistheexpandingopportunityforthedevelop-mentofnewmolecularmarkers.Asthefullgenome sequence will not be available formostspeciesof interest in thenear future,it is importanttofindstrategiesfordevel-oping and using molecular markers whensequenceresourcesarelimited.Thischapterdescribes several technologies that existfordevelopingmolecularmarkerswithoutDNAsequence information. Italsodrawsonsomeexamplesfromrice(Oryza sativaL.) to illustrate how molecular markerdevelopment was influenced by the addi-tionofeachlayerofsequenceinformation,culminatinginthepresentstatusofriceasthefirstcropwithnearlycompletegenomesequenceinformation.

moleCular marker teChnologieSrestriction fragment length polymorphismsRestrictionfragmentlengthpolymorphisms(RFLPs)werethefirstDNA-basedmolec-ular markers. An application of Southernanalysis (Southern, 1975), RFLPs exploittheabilityofsinglestrandedDNAtobind(hybridize)toDNAwithacomplementarysequence. RFLP markers detect variationin DNA sequences at the same loci indifferent individuals or accessions. Tech-nically, RFLP technology involves thehybridizationofclonedDNAtorestrictionfragments of differing molecular weights

fromrestrictionenzyme-digestedgenomicDNA. The digested DNA fragments aresize-separated on agarose gels by electro-phoresis and transferred as denatured(single stranded) arrays of fragments tofilters through capillary action. The filtersare then incubated with specific labelledprobes (genes or anonymous fragmentsof single stranded DNA), washedand exposed to x-ray film. To identifypolymorphisms between individuals oraccessions, the genomic DNA extractedfromeachindividualisdigestedwithaseriesof restriction enzymes to find enzymesthatproduce fragments (bands) thatdifferin molecular weight between accessionsandcanbedistinguishedbyhybridizationwithagivenprobe.Toensure thatprobeshybridizetosinglefragmentsonagel, theDNA used as a probe should be from asingleor lowcopy (non-repetitive) regionofthegenome.Probesmayrepresentgenes(i.e. derived from complementary DNA[cDNA])ortheymayrepresentanonymoussequences derived from genomic DNA.Genomicprobesaregeneratedbyshearingor digesting DNA and cloning the frag-mentsintoaplasmidvectorthatallowsforamplification of the cloned fragment in asuitablehost.Toincreasethefrequencyoflow copy clones in a genomic library, theDNAmaybedigestedwithamethylation-sensitive enzyme, such as PstI. Therepetitiveregionsofagenomearetypicallyheavily methylated and thus producefragments >25 kb when digested with amethylation-sensitiveenzyme.Asaresult,these fragments do not clone efficientlyinto plasmid vectors and consequentlyare effectively filtered out of the analysis.Thus,useofmethylation-sensitiveenzymesincreasestherepresentationofunmethylatedand typically low copy gene sequences inRFLP analysis. Sharing of anonymous,


unsequenced RFLP markers amongresearchers requires an infrastructure forthemaintenanceanddistributionofclonedprobes for use by multiple researchers.However, if end-sequence or full-clonesequenceinformationisavailable,theprobescan be amplified readily from genomicDNA via the polymerase chain reaction(PCR), and the cumbersome aspects ofclone maintenance and distribution areavoided. The polymorphisms detected byRFLPsmayresultfromsinglebasechangescausingalossofrestrictionsitesoragainofnewrestrictionsites,orfrominsertionsanddeletions (indels) between restriction sites(McCouch et al., 1988; Edwards, Lee andMcCouch,2004).

pCr-based markersManyadvances inmolecularmarker tech-nologyhavecomethroughapplicationsofthe PCR method (Mullis et al., 1986). InPCR, a thermo-stable DNA polymeraseenzymemakescopiesofa target sequencebeginning from two small pieces of syn-thetically produced DNA (primers) thatare complementary to sequences brack-eting the target. Through iterations of theprocesswithheatingtoseparatethedoublestranded DNA molecules and cooling toallow the primers to re-anneal, the targetsequence is exponentially amplified.Polymerase chain reaction-based markersrequire much less DNA per assay thanRFLPsandaremorecompatiblewithauto-matedhigh-throughputgenotyping(i.e.theabilitytoprocesslargenumbersofsamplesquicklyandefficiently).

randomly amplified polymorphic dna markersRandomly amplified polymorphic DNAmarkers (RAPDs) use PCR to amplifystretches of DNA between single primers

ofarbitrarysequence(Williamset al.,1990;WelshandMcClelland,1990).Amplificationoccurs only where sequences complemen-tary to the primers are in close enoughproximity for successful PCR. The typ-ical oligonucleotide used for RAPDs istenbases longandwill amplifymany locisimultaneously, allowing multiple markersto be assayed in a single PCR reactionandasinglelaneonanagarosegel.Astheprimers are arbitrary, RAPD technologycanbeapplieddirectlytoanyspecieswithno prior sequence knowledge. This tech-nology is particularly useful when thereis a need to assay loci across the entiregenome. The polymorphisms are detectedonly as thepresenceor absenceof abandof a particular molecular weight, and itis not possible to differentiate betweenhomozygous and heterozygous markers.RAPDsarenotoriouslyunreliablebecause,asidefromsequencedifferences,theampli-fication or failure of amplification of anyband may be sensitive to any number offactors, including DNA template quality,PCRconditions,reagentsandequipment.

amplified fragment length polymorphisms Amplifiedfragmentlengthpolymorphisms(AFLPs) are molecular markers derivedfrom the selective amplificationof restric-tionfragments(Voset al.,1995).GenomicDNAisdigestedwithapairofrestrictionenzymes and oligonucleotide adaptors areligatedtotheendsofeachrestrictionfrag-ment. The fragments are amplified usingprimersthatannealtotheadaptorsequenceand extend into the restriction fragment.Only a portion of restriction fragmentswillbewithintherangeofsizesthancanbeamplifiedbyPCRandvisualizedonpolya-cylamidegels(between50and350bp).Forlarge genomes, additional selective bases


canbeadded to theprimers toreduce thenumberofco-amplifiedbands.AFLPshavemanyoftheadvantagesofRAPDs,buthavemuch better reproducibility. AFLP tech-nologyrequiresgreatertechnicalskillthanRAPDsand,becauseAFLPsrunonpoly-acrylamidegelsinsteadofagarose,theyalsorequire a larger investment in equipmentthan RAPDs. Using manual gels, AFLPbands are detectable using silver stain, orby labelling of the primers with a radio-active isotope. Alternatively, for higherthroughput,AFLPscanbedetectedwithanautomatedDNAsequencerbyusingfluo-rescentlylabelledprimers.

Diversity array technology (DArT) isa modification of the AFLP procedureusingamicroarrayplatform(Jaccoudet al.,2001)thatgreatlyincreasesthroughput.InDArT, DNA fragments from one sampleare arrayed and used to detect polymor-phisms for the fragments inothersamplesby differential hybridization (Wenzl et al.,2004).

developing moleCular markerS with dna SequenCe informationWhentheDNAsequenceisavailable,it ispossibletodesignprimerstoamplifyacrossa specific locus. However not all loci willbe polymorphic. Targeting highly variablesequence features increases the likelihoodof detecting polymorphism. These highlyvariable features include tandem repeatssuchasmicrosatellites,anddispersedcom-plexrepeatssuchastransposableelements.

microsatellitesSimple sequence length polymorphisms(SSLPs), also known as simple sequencerepeats(SSRs),ormicrosatellites,consistoftandemly repeated di-, tri- or tetra-nucle-otide motifs and are a common feature ofmost eukaryotic genomes. The number of

repeats is highly variable because slippedstrandmis-pairing causes frequentgainorloss of repeat units. With their high levelofallelicdiversity,microsatellitesarevalu-ableasmolecularmarkers,particularlyforstudiesofcloselyrelatedindividuals.

PCR-based markers are designed toamplify fragments that contain a micros-atellite using primers complementary tounique sequences surrounding the repeatmotif (Weber and May, 1989). DifferencesinthenumberoftandemrepeatsarereadilyassayedbymeasuringthemolecularweightoftheresultingPCRfragments.Asthedif-ferencesmaybeassmallastwobasepairs,the fragments are separated by electro-phoresis on polyacrylamide gels or usingcapillary DNA sequencers that providesufficientresolution.

Without prior sequence knowledge,microsatellitescanbediscoveredbyscreeninglibraries of clones. Clones containing therepeat motif must be sequenced to finduniquesitesforprimerdesignflankingtherepeats.Microsatellitemarkerdevelopmentfrompre-existingsequenceisfarmoredirect.Good reviews of microsatellite markerdevelopment include those of McCouchet al. (1997) and Zane, Bargelloni andAtarnello(2002).Microsatellitesdiscoveredinnon-codingsequenceoftenhaveahigherrate of polymorphism than microsatellitesdiscovered in genes. However, in somespecies such as spruce (Picea spp.) withhighly repetitive genomes, SSR markersdevelopedfromgenesequenceshavefewerinstancesofnullalleles,i.e.failureofPCRamplification(Runguset al.,2004).

Microsatellite markers have severaladvantages.Theyareco-dominant;thehet-erozygousstatecanbediscernedfromthehomozygous state. The markers are easilyautomated using florescent primers on anautomated sequencer and it is possible


to multiplex (combine) several markerswith non-overlapping size ranges on asingle electrophoresis run. The results arehighly reproducible, and the markers areeasilysharedamongresearcherssimplybydistributing primer sequences. AlthoughSSRs are abundant in most eukaryoticgenomes, their genomic distribution mayvary. Uneven distributions of microsatel-liteslimittheirusefulnessinsomespecies.

Inter-SSRs (ISSRs) are another type ofmolecularmarkerthatmakesuseofmicro-satellitesequences.ISSRsusePCRprimersanchored in the termini of the repeatsextendingintotheflankingsequencebysev-eral nucleotides (Zietkietkiewicz, RafalskiandLabuda,1994).PCRproductsarepro-duced for each pair of microsatellites thatare in sufficient proximity for PCR tooccur, or may be generated by anchoringone primer in the SSR motif and using asecond “universal” primer correspondingto a sequence that has been ligated ontotheendsofrestrictionfragments(asintheAFLP technique described above, wheregenomic DNA is first digested with arestriction enzyme and oligonucleotideadaptors are ligated to the ends of eachrestrictionfragment,exceptthatoneprimerresides in an SSR motif that is brack-etedby the restrictionsites) (Guptaet al.,1994; Goodwin, Aitken and Smith, 1997).Markersatmultiplelociareassayedasthepresenceorabsenceofbandsofparticularsizes. ISSRs can be visualized on agarosegels, on silver stained polyacrylamide gelsorfluorescentlylabelledfordetectionwithanautomatedDNAsequencer.

transposable element-based markersTransposable elements (TEs) are anotherrapidly changing feature of the genomethat canbe exploited as a sourceofvaria-bilityformolecularmarkers.Discoveryof

TEsequencesisaprerequisitefortheiruseasmarkers.WhileTEsmaybediscoveredas mutations in alleles of genes conferringmutant phenotypes, they have also beendiscovered directly in genomic sequence(reviewedbyFeschotte,JiangandWessler,2002). Transposon display is a modifiedAFLP procedure that differs only in thatoneof the twoprimers isdesignedwithinthe consensus sequence of a TE family sothat amplification depends on the pres-ence of a TE insertion within a restrictionfragment (Casa et al., 2000). Using thisapproach, thepresenceorabsenceofaTEcan be assayed simultaneously at manyloci throughout the genome. To assay fora TE insertion at a specific locus, singlecopy “anchor markers” can be designedwith primers located in unique sequencesflankingtheregionofinterest.Asizepoly-morphismindicatesthepresenceorabsenceoftheTEinthatparticularlocation.Anchormarkersareadvantageousbecausetheyareco-dominant,canberunonasimpleagarosegel system and are biologically informa-tive in that they provide evidence of bothcomplete,or incomplete, insertionorexci-sionevents.Thismethodologycanalsobeappliedtoanyknownindelfeatureregard-lessofwhetherornot it isderived fromaTE.

Single nucleotide polymorphismsSingle nucleotide polymorphisms (SNPs)are an abundant source of sequence vari-ants that can be targeted for molecularmarker development. Of all the molec-ular marker technologies available today,SNPs provide the greatest marker den-sity. SNPs are often the only option forfinding markers very near or within agene of interest, and can even be used todetect aknown functionalnucleotidepol-ymorphism (FNP). Discovery of SNPs


requiresobtaininganinitialDNAsequenceinareferenceindividualfollowedbysomeform of re-sequencing in other varietiesto find variable base pairs. In addition todirectsequencing,SNPscanbediscoveredthroughecotillingwiththeCELIenzyme(Comaiet al.,2004)orbydenaturinghighpressureliquidchromatography(DHPLC)to measure small conformational differ-ences when PCR amplified sequences arehybridizedtoareferencesequence(Kwok,2001). Inaddition toSNPdiscovery,bothDHPLCandecotillingareviabletechnolo-giesforSNPdetection.Thereisamyriadofother SNP assay technologies in develop-mentand todatenosinglemethodstandsout as superior to the others. Table1 listssomeexamplesofSNPallelediscriminationmethodsanddetectionsystemsthatcanbecombined invariousways (seereviewsbyKwok,2001andGut,2001).ThebenefitsofSNPassaysincludeincreasedspeedofgen-otyping,lowercostandtheparallelassayofmultipleSNP.

Single feature polymorphisms and microarray-based genotypingIndelpolymorphisms,alsoknownassinglefeature polymorphisms (SFPs), are par-ticularly amenable to microarray-basedgenotyping.Theseassaysaredonebylabel-linggenomicDNA(target)andhybridizingto arrayedoligonucleotideprobes that arecomplementary to indel loci. Each SFPis scored by the presence or absence of ahybridizationsignalwithitscorrespondingoligonucleotide probe on the array. Both

spotted oligonucleotides (Barrett et al.,2004)andAffymetrix-typearrays(Borevitzet al.,2003)havebeenusedintheseassays.The SFPs can be discovered throughsequence alignments or by hybridizationof genomic DNA with whole genomemicroarrays. The advantage of microarrayplatforms for genotyping is that they arehighlyparallel,andtheyarewellsuitedforapplications such as quantitative trait loci(QTL)analysis,wherewholegenomecov-eragewithmanymarkersisdesirable.

SpeCial ConSiderationS for diverSity StudieS and germplaSm evaluationTheinterpretationofmolecularmarkerdatafor germplasm classification and diversitycanbeconfoundedbyuncertaintyabouttheunderlying sources of the polymorphismsand by homoplasy (false homology). ForRFLPs in rice, indels can account for asmuch or more of the polymorphism aschanges in the restriction sites themselves(Edwards,LeeandMcCouch, 2004).AFLPsand RAPDs can also be sensitive to bothindelsandbasechanges.Theratioofindelsto base changes is important for diversitystudies because, when molecular markersareusedtoestimatenucleotidedivergence,the divergence will be overestimated ifindel-derivedpolymorphismsarecommon(Upholt,1977;NeiandMiller,1990;Innanet al., 1999). The greatest certainty of theunderlying polymorphism comes fromSNP technologies that directly assay forsinglebasechanges.

ForSSRmarkersamongcloselyrelatedindividuals, most polymorphism shouldbe caused by expansion or contractionof the number of repeat units. However,as genetic distance between the varietiesincreases, there is an increasing chancethat indel eventswill causeadditional size

table 1Snp technologies

allele discrimination detection methods

• Hybridization• Primer extension• ligation• invasive cleavage

• Gel separation• arrays• Mass spectrometry• Plate readers


polymorphism(Chen,ChoandMcCouch,2002). Thus, the use of stepwise SSRmutationmodelswouldbeinappropriateforhighlydivergedpopulations.HomoplasyisalsoaprobleminSSRmarkersbecausethehyper-variabilityleadstosomesharedallelesizesthroughparallelism,convergenceandreversion (Doyleet al.,1998).Homoplasyfrom reversions can affect transposon-based markers or any markers withpolymorphisms potentially derived fromClassIIDNAtransposableelements.ThisclassofTEshasacutandpastemechanismoftransposition,soaTEmayinsertontoalocusandlaterexcise.

In RAPDs, ISSRs and AFLPs, homo-plasy can occur when two or more lociproducePCR fragmentsof similarmolec-ularweight.Althoughitisdesirabletohavehigh numbers of bands to maximize theamountof informationper lane, thismustbe balanced against the increasing risk ofhomoplasyasmorelociarerepresented.

SpeCial ConSiderationS for marker-aSSiSted SeleCtionQuality markers for use in MAS shouldbe reliable and easily shared amongresearchers.Co-dominantmarkersarepre-ferredtoavoidtheneedforprogenytesting.SometimeslessdesirablemarkersforMASsuch as RAPDs, ISSRs and AFLPs areuseful for finding markers linked to thedesiredallele.Oncesuchamarkerisfound,it is possible to extract and sequence thecorresponding band. This sequence canbe used to develop co-dominant markerssuch as cleaved amplified polymor-phic sequences (CAPS) (Konieczny andAusubel, 1993) or to sequence character-izedpolymorphicregions(SCARs)(ParanandMichelmore, 1993). SCARandCAPSmarkersareco-dominantandsimplify thescreening of large numbers of individuals.

When a genetic map exists, markers canbepositionedonthemapandotherlinkedmarkerscanbesubstituted.Theadditionalmarkers are useful for higher resolutionmapping to find markers more closelylinkedtothedesiredalleleorultimatelyforpositionalcloningoftheunderlyinggene.

reproducibility of molecular marker dataFororphanspecies,clearlythere isahugevaluetotheanonymousprimerapproaches(AFLP, DArTs, ISSRs and RAPDs) thatdo not require sequence information ormuch up-front investment. However,the data can be difficult to score, andreproducibility requires a lot of technicalskill. Technologies that depend on thepresence or absence of PCR amplifiedbands are susceptible to changes in PCRconditionsandthequalityofsampleDNA,and the data from separate experimentsmay differ. Further, in any method thatdepends on accurate measurement ofmolecularweightdifferencesbetweenbands(e.g. SSRs), the exact mole-cular weightsassigned to each allele may be differentin each analysis because of differencesin labelling of PCR products, roundingof allele molecular weight estimates andbinning of alleles. Without controls foreach allele encountered, it is difficult orimpossible to merge separate sets of data.Despite discrepancies in the exact dataderivedfrommolecularmarkers,theresultsandconclusionsshouldbeconsistentwithinindependent experiments. For reliabilityin making inferences across independentdata-sets,SNPmarkersarepreferred.SNPdata-sets can be easily integrated basedon sequence, and SNPs have properties(such as a low mutation rate) that areparticularly valuable for evolutionaryinference(Nielsen,2000).


ChooSing a moleCular marker teChnologyClearly there is no single best choice ofmolecularmarkerforallsituations.Factorsinfluencing the decision may include theobjectives of the study, availability oforganism specific sequences, equipmentand technical resources, and biologicalfeatures of the species. Several importantadvantages/disadvantages for each type ofmolecular marker discussed are summa-rized in Table2 (see review by Powell et al.,1996).

If available, microsatellite or SNPmarkersareoftenthebestchoice.Therateof adoption of SSR markers can be facili-tated, and thecosts reduced,bypreparing“kits” of selected SSR markers for certainspeciestoprovideareliablesetofmarkerswith good amplification, reasonable pol-ymorphism and good genome coverage.ThiswasdoneintheearlydaysofthericeSSReffortandSSRkitsweredistributedatvery low cost through Research Genetics(called Rice-Pairs; McCouch et al., 1997).Similarly, for SNPs, there is a need todevelop useful sets of markers that arewidelyavailableandcanbemass-produced(at reduced cost) for distribution to theinternational community. SNP kits wouldalso have a clear benefit for databasing

andanalysingdatasetsobtainedfrommul-tiple laboratories. In addition to kits ofmarkers, there is a need to distribute setsof“controlgenotypes”as samples,partic-ularlytoaddresstheproblemsurroundingthedifficulties in integratingSSRdatasets.WhenSNPsorSSRsarenotavailable,itissometimes possible to transfer molecularmarkersfromcloselyrelatedspecies(Guptaet al.,2003;LaRotaet al.,2005;Zhanget al., 2005). When financial resources arerestricted, RAPDs, AFLPs and ISSRs canprovide large numbers of markers with alimitedinvestment.AFLPs,SSRsandISSRscanprovidehighthroughputusinganauto-matedsequencer,whileRAPDsandISSRscan be run on agarose gels with minimalinvestmentinequipment.Theeffectivenessofeachmethodmayvarybyspeciesandbyapplication. Therefore, it is reasonable totrytousemorethanonemethod,particu-larlyattheearlystagesofresearch.

impaCt of the riCe genome SequenCe: a CaSe hiStoryDNA sequence information greatlyaccelerates the development of molecularmarkers. This is evident in the historyof rice microsatellite marker proliferationcoinciding with the release of data fromrice genome sequencing projects. Figure 1

table 2key features of common molecular marker technologies

marker type

pCr-based

uses restriction enzymes

poly-morphism

abun-dance

Co-dominant

auto-mation

loci per assay Specialized equipment

RFlP no yes moderate moderate yes no 1 to few Radioactive isotopeRaPD yes no moderate moderate no yes many agarose gelsaFlP yes no moderate moderate no yes many Polyacrylamide gels/capillaryiSSR yes no moderate moderate no yes many agarose/polyacrylamide gelsDart yes yes moderate moderate no yes many MicroarraycaPS yes yes variable moderate yes yes single agarose gelsScaR yes no low moderate yes yes single agarose gelsSSR yes no low moderate yes yes 1 to about 20 Polyacrylamide gels/capillaryte-anchor yes no variable variable yes yes single agarose gelsSnP yes no variable highest yes yes 1 to thousands Variable


tracksthepublicationofricemicrosatellitemarkersderivedfromscreeninglibrariesofclones and from the various categories ofsequencesdepositedinpublicdatabases.Theearliestmethodofdevelopingmicrosatellitemarkersinricewasbyusingmicrosatellitesequencesasprobesto isolateclonesfromgenomiclibraries(ZhaoandKochert,1993;WuandTanksley,1993;Panaud,ChenandMcCouch, 1996; Akagi et al., 1996; Chenet al.,1997;Temnykhet al.,2000).In1996,Akagi et al. used microsatellite repeatsfound in rice sequences from databasesearches to develop 35 new markers andin2000,Temnykhet al.published91newmicrosatellite markers developed fromexpressed sequence tag (EST) sequences.Temnykh et al. (2001) developed 200 newmarkers,mostlyfromendsequencesofricebacterial artificial chromosomes (BACs).However, the most dramatic increase inmicrosatellitemarkers(2240newmarkersin 2002 and 25 000 in 2004) was madepossibleprimarilythoughtheuseofwholegenomeshotgunsequences(McCouchet al.,2002;G.Wilson,personalcommunication).

Complete genome sequence provides anadditional advantage in electronicallydetermining the position of new markerson genetic and physical maps. However,fullgenomicsequenceisnotarequirementformicrosatellitemarkerdevelopment,andthereareanumberofmicrosatellitemarkersthathavebeendevelopedforawidearrayofcropspecies(Table3)withoutthebenefitoffullgenomicsequence.

marker-aSSiSted SeleCtion StrategieS and exampleS MASinabreedingcontextinvolvesscoringindirectly for the presence or absenceof a desired phenotype or phenotypiccomponent based on the sequences orbanding patterns of molecular markerslocatedinornearthegenescontrollingthephenotype. The sequence polymorphismorbandingpatternofthemolecularmarkerisindicativeofthepresenceorabsenceofaspecificgeneorchromosomalsegmentthatisknowntocarryadesiredallele.

DNA markers can increase screeningefficiency in breeding programmes in a

FiGURe 1progress of microsatellite marker development in rice

0 500 1000 1500 2000

Zhao and Kochert, 1993

Wu and Tanksley, 1993

Panaud, Chen andMcCouch, 1996

Akagi et al., 1996

Chen et al., 1997

Temnykh et al., 2000

Temnykh et al., 2001

McCouch et al., 2002

Screened libraries Genes/ESTs BAC-ends Shotgun sequences Finished BACs

Number of microsatellite markers


number of ways. For example, theyprovide:• theabilitytoscreeninthejuvenilestage

fortraitsthatareexpressedlateinthelifeoftheorganism(i.e.grainorfruitquality,malesterility,photoperiodsensitivity);

• the ability to screen for traits that areextremely difficult, expensive or timeconsuming to score phenotypically (i.e.quantitatively inherited or environ-mentally sensitive traits such as rootmorphology, resistance to quarantinedpestsor to specific racesorbiotypesofdiseases or insects, tolerance to certainabioticstressessuchasdrought,saltandmineraldeficienciesortoxicities);

• the ability to distinguish the homozy-gous from the heterozygous conditionof many loci in a single generationwithouttheneedforprogenytesting(asmolecularmarkersareco-dominant);

• the ability to perform simultaneousMAS for several characters at one time

(ortocombineMASwithphenotypicorbiochemicalevaluation).This section provides examples of

how molecular markers are being used inbreedingandgermplasmevaluation.Whiletheseexamplesaredrawnmostlyfromrice,they illustrate applications of MAS tech-niquesthatareusedinotherspecies.

Before molecular markers can be usedfor selection purposes, their associationwith genes or traits of interest must befirmly established. While the number ofeconomically important genetic loci thathave been cloned or tagged via linkage tomolecular markers is still limited in mostspecies, work towards this end is acceler-ating rapidly. This is particularly true inrice, due to the availability of completegenomesequenceinformation.

Nonetheless, a great deal of time andeffort is required to identify the geneticloci and specific allelic variants that areresponsible for the tremendous array of

table 3examples of SSr markers available across different plant species

Common name Species number of SSrs

reference

Rice Oryza sativa 2240 Mccouch et al., 2002Maize Zea mays 1669 MapPairs (mp.invitrogen.com)Soybean Glycine max 597 MapPairs (mp.invitrogen.com)cassava Manihot esculenta 318 MapPairs (mp.invitrogen.com)arabidopsis Arabidopsis thaliana 290 MapPairs (mp.invitrogen.com)cotton Gossypium spp. 217 MapPairs (mp.invitrogen.com)Sugar cane Saccharum spp. 200 www.intl-pag.org/pag/9/abstracts/W30_04.htmlWheat Triticum aestivum 193 MapPairs (mp.invitrogen.com)Grape Vitis vinifera 152 noGroundnut Arachis hypogaea 110 Ferguson et al., 2004cucumber Cucumis sativus 110 Fazio, Staub and chung, 2002Peach Prunus persica 109 aranzana et al., 2004Kiwifruit Actinidia spp. 105 testolin et al., 2001barley Hordeum vulgare 44 MapPairs (mp.invitrogen.com)Potato Solanum tuberosum 31 Ghislain et al., 2004Pine trees Pinus spp. 28 MapPairs (mp.invitrogen.com)banana Musa spp. 28 MapPairs (mp.invitrogen.com)Sweet potato Ipomoea batatas 26 MapPairs (mp.invitrogen.com)Sugar beet Beta vulgaris 25 www.intl-pag.org/pag/10/abstracts/PaGX_W306.htmleggplant Solanum melongena 23 www.intl-pag.org/pag/11/abstracts/P3b_P181_Xi.html

From: thomson, Septiningsih and Sutrisno, 2003 (reprinted with permission of author)


characters that breeders are concernedaboutinpopulationorvarietyimprovementprogrammes. Given the complexity ofquantitative traits, many different lines orcrosses must be carefully analysed overdifferentyearsandenvironmentstounravelimportantcomponentsofgeneinteraction.In a breeding context, understanding thegenetic basis of genotype by genotypeinteraction (GxG) and genotype byenvironment interaction (GxE) is criticalas the basis for predicting how QTL arelikely tobehave.Informationfroma largenumberofstudiesaddressingeachofthesepoints must then be assembled into adatabasethatofferseasyaccesstousersandallows many different kinds of data to beintegratedwithasimplequery.

The Gramene database represents abeginning in the quest to serve this usercommunity. Gramene is a comparativegenomedatabase forgrassesandcurrentlyoffersacompleteinventoryofallpublishedQTLthathavebeenidentifiedinrice(www.gramene.org/qtl/index.html),allowinguserstofindinformationaboutwherealongthechromosomeaQTL is located,whatphe-notypeisassociatedwiththeQTL,howitwas measured, what germplasm was used,whatmolecularmarkersresidenearby,whatthecorrespondingpositionisonacompar-ativemapofanothergrassspeciesandwithwhat statistical significance the QTL wasdetected.Thedatabasealsoprovidesalinkto the published article so that users canreadilyfindmore informationonthesub-ject. Similar inventories and databases arebeingassembledforotherfamiliesofplantsand are critical to the implementation ofeffectivemolecularbreedingstrategies.

Comparative genome methods takeadvantage of the fact that some specieshavemoredevelopedgeneticsystemsthanothers. Examples of well studied “model”

organismswithavailablegenomicsequenceinclude species such as Arabidopsis andrice for plants, Populus (Taylor, 2002) andEucalyptus (Poke et al., 2005) specificallyfor forestry, and Fugu (Aparicio et al.,2002) and zebrafish (Guryev et al., 2006)for fisheries. Relying heavily on the useof comparative maps and comparativesequenceanalysis,genomedatabasesallowresearchers to make predictions about thelocation and phenotypic consequences ofhomologousgenesinrelatedspecies.Thus,understandinghowageneorQTLbehavesin one species can potentially shortcutthe process of identifying a related geneor QTL in the genetic system of anotherspecies. This approach underscores thesearch for QTL associated with abioticstress tolerance in cereals. A global effortto identify loci associated with droughttolerancehasrecentlybeeninitiatedunderthe umbrella of the Generation ChallengeProgramme(www.generationcp.org).

Markers associated with tolerance fora variety of environmental stresses rank asimportant targets for molecular MAS incerealbreedingbecausethesecomplextraitsare often prohibitively difficult to screenusing classical selection techniques. EffortstoidentifyQTLassociatedwithtolerancetodrought,saltandmineraldeficienciesortox-icities(Champouxet al.,1995;Flowerset al.,2000;Nguyenet al.,2002;Kamoshitaet al.,2002; Price et al., 2002; Gregorio, 2002) inanumberofgeneticbackgroundsrepresentan important first step towards achievingthis goal. Additional studies have specifi-callyaddressedtheproblemsassociatedwithGxGandGxE (Zhenget al., 2000;Liet al.,2003;Hittalmaniet al.,2003).

Intheareaofbioticstress,severalgeneshave been cloned and characterized forresistance to major diseases such as bac-terial blight and blast (Song et al., 1995;


Yoshimura et al., 1998; Wang et al., 1999;Bryan et al., 2000; Sun et al., 2004) andmany other genes for disease resistancehavebeentaggedwithlinkedmarkers.Thisopens thedoor for targetedapproaches toMAS(Valentet al.,2001).Whilethediseaseresistanceliteratureistoovasttosummarizehere,itisimportanttonotethatadvancesinthis area are having an impact on varietalimprovementprogrammes(www.syix.com/rrb/98rpt/MarkerAssist.htm). Pyramidingof resistance genes into a single varietyandtheconstructionofmultilinevarieties,eachwithoneormoreRgenes(resistancegenes) that can be used in various combi-nations, are under way to develop moredurable forms of disease and insect resist-ance(Yoshimuraet al.,1992;Yoshimuraet al.,1995;Hittalmaniet al.,1995;BlairandMcCouch, 1997; Ndjiondjop et al., 1999;Davierwala et al., 2001; Su et al., 2002;Conaway-Bormanset al.,2003;Lorieuxet al., 2003;Hayashiet al.,2004).

Marker-based selection is also helpfulin attempts to transfer genes from exoticgermplasm into cultivated lines. In rice,several workers have used RFLP andSSR markers to monitor introgression ofbrown planthopper resistance from O. officinalis (Kochert, Jena andZhao,1990),bacterial blight resistance from O. longis-taminata (Ronald et al., 1992), aluminumtoleranceoryieldandquality-relatedtraitsfrom O. rufipogon (Nguyen et al., 2002;Thomson et al., 2003; Septiningsih et al.,2003a,b)orfromotherwildspeciessuchasO. glumaepatula (Brondaniet al.,2002)orO. glaberrima (Jones et al., 1997; Lorieuxet al.,2003)intocultivatedO. sativaback-grounds. Marker-assisted introgressionstrategieshavealsobeenusedinanumberof livestock breeding programmes but,becauseof longergeneration intervals and

lower reproductive rates, this is generally

feasible forgenesof largeeffect (Dekkers,2004; Chapter 10). Identifying the recom-binants with the least amount of donorDNA flanking the genes of interest isenhancedbytheuseofmolecularmarkers(Monnaet al., 2002;Takeuchiet al., 2003;Blair, Panaud and McCouch, 2003). Inthese examples, MAS offers a powerfulstrategy for making efficient use of thewealthofusefulgeneticvariationthatexistsin the early landraces and wild speciesof cultivated food crops (Tanksley andMcCouch,1997).

Asthiskindofinformationaccumulates,MAS permits rapid identification of indi-viduals that may contain only one geneticcomponentofacomplextrait.Onceiden-tified, such an individual can be crossedwithanotherindividual inabreedingpro-gramme so that multiple, complementarygenesarecombinedtooptimizeaquantita-tivelyinheritedtrait.Individualscontainingonlyonegeneof interestoftendefyaccu-rate phenotypic identification wherepolygenictraitsareconcernedbecausevar-ioustypesofepistasis,orgeneinteraction,mayberequiredtogeneratethephenotypeof interest (Yamamoto et al., 2000; Zhenget al.,2000).

Linkage disequilibrium (LD) mappingis another marker-assisted approachthat provides important informationthat is immediately relevant to breedingprogrammes (Remington, Ungererand Purugganan, 2001; Flint-Garcia,Thornsberry and Buckler, 2003). Usingcollections of distantly related germplasmaccessionsrather thanpopulationsderivedfrombi-parentalcrossesallowsresearchersto explore the relationship betweenphenotype and genotype in materials thathave been amply tested over years andenvironments, often as part of an appliedbreeding programme. This provides


critical information about how specificcombinations of genes and alleles interactinrelevantvarietalbackgroundsandallowsbreeders tocomparethephenotypiceffectofgenesorchromosomalsegmentsthathavebeeninheritedfromacommonancestorandselectedinmultiple-crosscombinations.

InadditiontotheuseofMASintradi-tional crossing and selection programmes,breeders also have opportunities to adjustparticular traits or phenotypes via theintroduction of genes using a transgenicapproach (Ye et al., 2000; James, 2003;NuffieldCouncilonBioethics,2004).Onceintroducedintothegenepool,atransgenecan be tracked with the aid of molecularmarkers (designed to tag the transgenesequenceitself)throughsubsequentcrosses,justaswouldbedoneforanyothergeneofinterestinabreedingprogramme.

Another use of molecular markers invariety improvement involves marker-assisted germplasm evaluation (Xu, IshiiandMcCouch,2003).Populationstructureanalysisoffers insightabouthowdiversityis partitioned within a species and canhelpdefine clusters,or subpopulations,ofgermplasm that are likely to contain highfrequencies of particular alleles (Garris,McCouchandKresovich,2003).Thistypeof analysis can also guide allele miningefforts aimed at identifying valuableaccessions in a germplasm collection foruse as parents in a breeding programme.Suchapproacheshavethepotentialtomakeparentalselectionmoreefficient,toexpandthe gene pool of modern cultivars andultimatelytospeedupthedevelopmentofproductivenewvarieties.Asinformationisgeneratedaboutwhichgenesandallelesareassociated with phenotypic characters ofagronomicimportance,andasthecomplexinteractionsamonggenesareenumeratedinthe context of specific gene pools and the

environments to which they are adapted,breeders are increasingly empowered tomake predictions about how to combinediverseallelesproductively.

Toexploitmolecularbreedingstrategiesfully,informationresourcesmustbedevel-oped so that theoverwhelming amountofinformationaboutgenes,allelesandnaturalgeneticvariationcanbefunnelledintoausefultool for breeding applications. This willinvolve averydifferent approach to infor-mation resources than currently employedby the large genome databases, which areoriented towards genomics researchersand molecular biologists rather than thebreeding community. Nonetheless, a fewexamplesofferbeaconsofinspirationinthisarea, including the emerging InternationalRice Information System (IRIS) data-base (Bruskiewich et al., 2003; www.icis.cgiar.org/), the GeneFlow database (www.geneflow.com),themarker-assistedselectionwheat (MASwheat) database (http://mas-wheat.ucdavis.edu/) and software such asReal Time QTL (http://zamir.sgn.cornell.edu/Qtl/Html/home.htm).

Inconclusion,genomicsresearchisgen-eratinginformationaboutthelocationandphenotypicconsequencesofspecificgenesandallelesinawiderangeofspecies.Thisinformationcanbetranslatedintotoolsforbreeders.Molecularmarkertechnologycanbenefitbreedingobjectivesbyincreasingtheefficiencyandreliabilityofselectionandbyprovidingessentialinsightsintohowgenesbehave in different environments and indifferentgeneticbackgrounds.Oncegenesand QTL are identified, markers allowinteresting alleles to be traced throughthe pedigrees of breeding programmes orminedoutofgermplasmcollectionstoserveas the basis for future varietal improve-ment.Usingmarkers in combinationwithbothQTLandassociationapproaches,the


effect of specific alleles on a phenotypecan be monitored with relative precision.As all this information is assembled andorganized in databases that provide easy

querying capabilities for plant breeders,breeders will take advantage of the powerthatcomesfromtheapplicationofgenomebasedstrategiesforplantimprovement.

referenCeSAkagi, H., Yokozeki, Y., Inagaki, A. & Fujimura, T.Y. 1996.MicrosatelliteDNAmarkersforrice

chromosomes.Theor. Appl. Genet.93:1071–1077.Aparicio, S., Chapman, J., Stupka, E., Putman, N., Chia, J.M., Dehal, P., Christoffels, A., Rash,

S., Hoon, S., Smit, A., Gelpke, M.D.S., Roach, J., Oh, T., Ho, I.Y., Wong, M., Detter, C., Verhoef, F., Predki, P., Tay, A., Lucas, S., Richardson, P., Smith, S.F., Clark, M.S., Edwards, Y.J.K., Doggett, N., Zharkikh, A., Tavtigian, S.V., Pruss, D., Barnstead, M., Evans, C., Baden, H., Powell, J., Glusman, G., Rowen, L., Hood, L., Tan, Y.H., Elgar, G., Hawkins, T., Venkatesh, B., Rokhsar, D. & Brenner, S.2002.Whole-genomeshotgunassemblyandanalysisofthegenomeofFugu rubripes.Science297:1301–1310.

Aranzana, M.J., Pineda, A., Cosson, P., Dirlewanger, E., Ascasibar, J., Cipriani, G., Ryder, C.D., Testolin, R., Abbott, A., King, G.J., Iezzoni, A.F. & Arus, P. 2003.Asetofsimple-sequencerepeat(SSR)markerscoveringthePrunusgenome.Theor. Appl. Genet.106:819–825.

Barrett, M.T., Scheffer, A., Ben-Dor, A., Sampas, N., Lipson, D., Kincaid, R., Tsang, P., Curry, B., Baird, K., Meltzer, P.S., Yakhini, Z., Bruhn, L. & Laderman, S. 2004.Comparativegenomichybridization using oligonucleotide microarrays and total genomic DNA. Proc. Nat. Acad. Sci. USA101:17765–17770.

Blair, M.W. & McCouch, S.R.1997.Microsatelliteandsequence-taggedsitemarkersdiagnosticforthericebacterialleafblightresistancegenexa-5.Theor. Appl. Genet.95:174-184.

Blair, M.W., Panaud O. & McCouch, S.R. 1999. Inter-simple sequence repeat (ISSR) amplification foranalysis of microsatellite motif frequency and fingerprinting in rice (Oryza sativa L.). Theor. Appl. Genet.98:780–792.

Borevitz, J. O., Liang, D., Plouffe, D., Chang, H.S., Zhu, T., Weigel, D., Berry, C.C., Winzeler, E. & Chory, J.2003.Large-scaleidentificationofsingle-featurepolymorphismsincomplexgenomes.Genome Res.13:513–523.

Brondani, C., Rangel, N., Brondani V. & Ferreira, E. 2002. QTL mapping and introgression ofyield-relatedtraitsfromOryza glumaepatulatocultivatedrice(Oryza sativa)usingmicrosatellitemarkers.Theor. Appl. Genet.104:1192–1203.

Bruskiewich, R.M., Cosico, A.B., Eusebio, W., Portugal, A.M., Ramos, L.M., Reyes, M.T., Sallan, M.A., Ulat, V.J., Wang, X., McNally, K.L., Sackville Hamilton, R. & McLaren, C.G. 2003.Linkinggenotypetophenotype:theInternationalRiceInformationSystem(IRIS).Bioinformatics 19 Suppl 1:i63–i65.

Bryan, G.T., Wu, K.S., Farrall, L., Jia, Y., Hershey, H.P., McAdams, S.A., Faulk, K.N., Donaldson, G.K., Tarchini, R. & Valent, B.2000.AsingleaminoaciddifferencedistinguishesresistantandsusceptibleallelesofthericeblastresistancegenePi-ta.The Plant Cell12:2033–2046.

Casa, A.M., Brouwer, C., Nagel, A., Wang, L., Zhang Q., Kresovich, S. & Wessler, S.R. 2000.Inaugural article: the MITE family heartbreaker (Hbr): molecular markers in maize. Proc. Nat. Acad. Sci. USA97:10083–10089.


Champoux, M.C., Sarkarung, G.W.S., Mackill, D.J., O’Toole, J.C., Huang, N. & McCouch, S.R.1995. Locating genes associated with root morphology and drought avoidance via linkage tomolecularmarkers.Theor. Appl. Genet.90:969–981.

Chen, X., Cho Y.G. & McCouch, S.R.2002.SequencedivergenceofricemicrosatellitesinOryzaandotherplantspecies.Mol. Genet. Genomics268:331–343.

Chen, X., Temnykh, S., Xu, Y., Cho, X.G. & McCouch, S.R. 1997.Developmentofamicrosatelliteframeworkmapprovidinggenome-widecoverage inrice (Oryza sativaL.).Theor. Appl. Genet.95:553–567.

Comai, L., Young, K., Till, B.J., Reynolds, S.H., Greene, E.A., Codomo, C.A., Enns, L.C., Johnson, J.E., Burtner, C., Odden, A.R. & Henikoff, S. 2004.EfficientdiscoveryofDNApolymorphismsinnaturalpopulationsbyEcotilling.Plant J.37:778–786.

Conaway-Bormans, C.A., Marchetti, M.A., Johnson, C.W., McClung, A.M. & Park, W.D. 2003. Molecular markers linked to the blast resistance gene Pi-z in rice for use inmarker-assistedselection.Theor. Appl. Genet.107:1014–1020.

Davierwala, A.P., Reddy, A.P., Lagu, M.D., Ranjekar, P.K. & Gupta, V.S. 2001. Marker assistedselectionofbacterialblightresistancegenesinrice.Biochem. Genet.39:261–278.

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

Doyle, J.J., Morgante, M., Tingey S.V. & Powell, W. 1998.Sizehomoplasy inchloroplastmicro-satellitesofwildperennial relativesof soybean (Glycine subgenusGlycine).Mol. Biol. Evol. 15:215–218.

Edwards, J.D., Lee, V.M. & McCouch, S.R. 2004. Sources and predictors of resolvable indelpolymorphismassessedusingriceasamodel.Mol. Genet. Genomics271:298–307.

Fazio, G., Staub, J.E. & Chung, S.M. 2002.DevelopmentandcharacterizationofPCRmarkersincucumber.J. Am. Soc. Hortic. Sci.127:545–557.

Ferguson, M.E., Burow, M.D., Schulze, S.R., Bramel, P.J., Paterson, A.H., Kresovich, S. & Mitchell, S. 2004.Microsatelliteidentificationandcharacterizationinpeanut(A. hypogaea L.).Theor. Appl. Genet.108:1064–1070.

Feschotte, C., Jiang N. & Wessler, S.R. 2002. Plant transposable elements: where genetics meetsgenomics.Nature Revs. Genet.3:329–341.

Flint-Garcia, S.A., Thornsberry, J.M. & Buckler, E.S.T.2003.Structureoflinkagedisequilibriuminplants.Ann. Rev. Plant Biol. 54:357–374.

Flowers, T.J., Koyama, M.L., Flowers, S.A., Sudhakar, C., Singh, K.P. & Yeo, A.R. 2000.QTL:theirplaceinengineeringtoleranceofricetosalinity.J. Exp. Bot.51:99–106.

Garris A., McCouch, S. & Kresovich, S.2003.Populationstructureanditseffectonhaplotypediver-sityandlinkagedisequilibriumsurroundingthexa5locusofrice(Oryza sativaL.).Genetics 165:759–769.

Ghislain, M., Spooner, D.M., Rodriguez, F., Villamon, F., Nunez, J., Vasquez, C., Waugh, R. & Bonierbale, M.2004.Selectionofhighlyinformativeanduserfriendlymicrosatellites(SSRs)forgenotypingofcultivatedpotato.Theor. Appl. Genet.108:881–890.

Goodwin, I.D., Aitken, E.A.B. & Smith, L.W. 1997. Application of inter-simple sequence repeat(ISSR)markerstoplantgenetics.Electrophoresis18:1524–1528.

Gregorio, G.B. 2002. Progress in breeding for trace minerals in staple crops. J. Nutr. 132:500S–502S.


Gupta, M., Chyi, Y.S., Romero-Severson, J. & Owen, J.L. 1994.AmplificationofDNAmarkersfromevolutionarilydiversegenomesusingsingleprimersofsimple-sequencerepeats.Theor. Appl. Genet. 89:998–1006.

Gupta, P.K., Rustgi, S., Sharma, S., Singh, R., Kumar, N. & Balyan, H.S. 2003.TransferableEST-SSRmarkersforthestudyofpolymorphismandgeneticdiversity inbreadwheat.Mol. Genet. Genomics270:315–323.

Guryev, V., Koudijs, M.J., Berezikov, E., Johnson, S.L., Plasterk, R.H., van Eeden, F.J. & Cuppen, E.2006.Geneticvariationinthezebrafish.Genome Res.16:491–497.

Gut, I.G. 2001.Automationingenotypingofsinglenucleotidepolymorphisms.Hum. Mutat. 17:475–492.

Hayashi, K., Hashimoto, N., Daigen, M. & Ashikawa, I.2004.DevelopmentofPCR-basedSNPmarkersforriceblastresistancegenesatthePizlocus.Theor. Appl. Genet.108:1212–1220.

Hittalmani, S., Foolad, M.R., Mew, T., Rodriguez, R.L. & Huang, N. 1995. Development of aPCR-based marker to identify rice blast resistance gene, Pi-2(T), in a segregating population.Theor. Appl. Genet.91:9–14.

Hittalmani, S., Huang, N., Courtois, B., Venuprasad, R., Shashidhar, H.E., Zhuang, J.Y., Zheng, K.L., Liu, G.F., Wang, G.C., Sidhu, J.S., Srivantaneeyakul, S., Singh, V.P., Bagali, P.G., Prasanna, H.C., McLaren, G. & Khush, G.S. 2003.IdentificationofQTLforgrowth-andgrainyield-relatedtraitsinriceacrossninelocationsofAsia.Theor. Appl. Genet.107:679–690.

Innan, H., Terauchi, R., Kahl, G. & Tajima, F.1999.AmethodforestimatingnucleotidediversityfromAFLPdata.Genetics151:1157-1164.

Jaccoud, D., Peng, K., Feinstein, D. & Kilian, A.2001.Diversityarrays:asolidstate technologyforsequenceinformationindependentgenotyping.Nucl. Acids Res.29:E25.

James, C. 2003.Globalreviewofcommercializedtransgeniccrops:2002Feature:Bt Maize.ISAAA Briefs.

Jones, M.P., Dingkuhn, M. Aluko, G.K. & Semon, M.1997.InterspecificO. sativaL.xO. glaber-rimaSteud.Progeniesinuplandriceimprovement.Euphytica92:237–246.

Kamoshita, A., Wade, J., Ali, L., Pathan, S., Zhang, J., Sarkarung, S. & Nguyen, T. 2002.MappingQTLsforrootmorphologyofa ricepopulationadapted torain-fed lowlandconditions.Theor. Appl. Genet. 104:880–893.

Kochert, G., Jena, K. & Zhao, X.P. 1990.Genomecharacterizationand introgression.In4th Ann. Mtg. Rockefeller Foundation’s Internat. Program on Rice Biotech.LosBaños,Philippines,IRRI.

Konieczny, A. & Ausubel, F.M. 1993.Aprocedure formappingArabidopsismutationsusingco-dominantecotype-specificPCR-basedmarkers.Plant J.4:403–410.

Kwok, P.Y. 2001. Methods for genotyping single nucleotide polymorphisms. Ann. Rev. Genomics Hum. Genet.2:235–258.

La Rota, M., Kantety, R.V., Yu, J.K. & Sorrells, M.E.2005.Non-randomdistributionandfrequen-ciesofgenomicandEST-derivedmicrosatellitemarkersinrice,wheat,andbarley.BMC Genomics6:23.

Li, Z.K., Yu, S.B., Lafitte, H.R., Huang, N., Courtois, B., Hittalmani, S., Vijayakumar, C.H., Liu, G.F., Wang, G.C., Shashidhar, H.E., Zhuang, J.Y., Zheng, K.L., Singh, V.P., Sidhu, J.S., Srivantaneeyakul, S. & Khush, G.S.2003.QTLxenvironmentinteractionsinrice.I.Headingdateandplantheight.Theor. Appl. Genet.108:141–153.


Lorieux, M., Reversat, G., Garcia Diaz, S.X., Denance, C., Jouvenet, N., Orieux, Y., Bourger, N., Pando-Bahuon, A. & Ghesquiere, A. 2003.LinkagemappingofHsa-1(Og),aresistancegeneofAfricanricetothecystnematode,Heterodera sacchari.Theor. Appl. Genet.107:691–696.

McCouch, S.R., Kochert, G., Yu, Z.H., Wang, Z.Y., Khush, G.S., Coffman, W.R. & Tanksley, S.D. 1988.Molecularmappingofricechromosomes.Theor. Appl. Genet. 76:815–829.

McCouch, S.R., Chen, X., Panaud, O., Temnykh, S., Xu, Y., Cho, Y.G., Huang, N., Ishii, T. & Blair, M.1997.Microsatellitemarkerdevelopment,mappingandapplicationsinricegeneticsandbreeding.Plant Mol. Biol.35:89–99.

McCouch, S.R., Teytelman, L., Xu, Y., Lobos, K.B., Clare, K., Walton, M., Fu, B., Maghirang, R., Li, Z., Xing, Y., Zhang, Q., Kono, I., Yano, M., Fjellstrom, R., DeClerck, G., Schneider, D., Cartinhour, S., Ware, D. & Stein, L. 2002.Developmentandmappingof2240newSSRmarkersforrice(Oryza sativa L.)(supplement).DNA Res.9:257–279.

Monna, L., Lin, X., Kojima, S., Sasaki, T. & Yano, M. 2002.Geneticdissectionofagenomicregionforaquantitativetraitlocus,Hd3,intotwoloci,Hd3aandHd3b,controllingheadingdateinrice.Theor. Appl. Genet.104:772–778.

Mullis, K., Faloona, F., Scharf, S., Saiki, R., Horn, G. & Erlich, H.1986.Specificenzymaticampli-ficationofDNAin vitro:thepolymerasechainreaction.Cold Spring Harb. Symp. Quant. Biol.51Pt1:263–273.

Ndjiondjop, M.N., Albar, L., Fargette, D., Fauquet, C. & Ghesquiere, A. 1999.Thegeneticbasisofhighresistancetoriceyellowmottlevirus(RYMV)incultivarsoftwocultivatedricespecies.Plant Disease83:931–935.

Nei, M. & Miller, J.C. 1990.Asimplemethodforestimatingaveragenumberofnucleotidesubstitu-tionswithinandbetweenpopulationsfromrestrictiondata.Genetics 125:873–879.

Nguyen, B.D., Brar, D.S., Bui, B.C., Nguyen, T.V., Pham, L.N. & Nguyen, H.T. 2002.Identificationand mapping of the QTL for aluminum tolerance introgressed from the new source, Oryza rufipogonGriff.,intoindicarice(Oryza sativa L.).Theor. Appl. Genet.106:583–593.

Nielsen, R.2000.Estimationofpopulationparametersandrecombinationrates fromsinglenucle-otidepolymorphisms.Genetics154:931–942.

Nuffield Council on Bioethics.2004.The use of genetically modified crops in developing countries.London,UK.

Panaud, O., Chen, X. & McCouch, S.R.1996.Developmentofmicrosatellitemarkersandcharac-terizationof simple sequence lengthpolymorphism(SSLP) in rice (Oryza sativa L.).Mol. Gen. Genet.252:597–607.

Paran, I. & Michelmore, R.W.1993.DevelopmentofreliablePCR-basedmarkerslinkedtodownymildewresistancegenesinlettuce.Genetics 85:985–993.

Poke, F.S., Vaillancourt, R.E., Potts, B.M. & Reid, J.B. 2005. Genomic research in Eucalyptus.Genetica125:79–101.

Powell, W., Morgante, M., Andre, C., Hanafey, M., Vogel, J., Tingey, S. & Rafalski, A.1996.ThecomparisonofRFLP,RAPD,AFLPandSSR(microsatellite)markersforgermplasmanalysis.Mol. Breed.2:225–238.

Price, A.H., Townend, J., Jones, M.P., Audebert, A. & Courtois, B. 2002.MappingQTLsassoci-atedwithdroughtavoidanceinuplandricegrowninthePhilippinesandWestAfrica.Plant Mol. Biol.48:683–695.

Remington, D.L., Ungerer, M.C. & Purugganan, M.D. 2001.Map-basedcloningofquantitativetraitloci:progressandprospects.Genet. Res.78:213–218.


Ronald, P.C., Salmeron, J.M., Carland, F.M. & Staskawicz, B.J. 1992.TheclonedavirulencegeneavrPtoinducesdiseaseresistanceintomatocultivarscontainingthePtoresistancegene.J. Bacteriol. 174:1604–1611.

Rungus, D., Berube, Y., Zhang, J., Ralph, S., Ritland, C.E., Ellis, B.E., Douglas, C., Bohlmann, J. & Ritland, K. 2004.Robustsimplesequencerepeatmarkersforspruce(Piceaspp.)fromexpressedsequencetags.Theor. Appl. Genet.109:1283–1294.

Septiningsih, E.M., Prasetiyono, J., Lubis, E., Tai, T.H., Tjubaryat, T., Moeljopawiro, S. & McCouch, S.R. 2003a.IdentificationofquantitativetraitlociforyieldandyieldcomponentsinanadvancedbackcrosspopulationderivedfromtheOryza sativavarietyIR64andthewildrelativeO. rufipogon.Theor. Appl. Genet.107:1419–1432.

Septiningsih, E.M., Trijatmiko, K.R., Moeljopawiro, S. & McCouch, S.R. 2003b.IdentificationofquantitativetraitlociforgrainqualityinanadvancedbackcrosspopulationderivedfromtheOryza sativavarietyIR64andthewildrelativeO. rufipogon.Theor. Appl. Genet.107:1433–1441.

Song, W.Y., Wang, G.L., Chen, L.L., Kim, H.S., Pi, L.Y., Holsten, T., Gardner, J., Wang, B., Zhai, W.X., Zhu, L.H., Fauquet, C. & Ronald, P. 1995.Areceptorkinase-likeproteinencodedbythericediseaseresistancegene,Xa21.Science270:1804–1806.

Southern, E.M. 1975.DetectionofspecificsequencesamongDNAfragmentsseparatedbygelelec-trophoresis.J. Mol. Biol.98:503–510.

Su, C.C., Cheng, X.N., Zhai, H.Q. & Wan, J.M. 2002.DetectionandanalysisofQTLforresist-ancetothebrownplanthopper,Nilaparvata lugens(Stal),inrice(Oryza sativaL.),usingbackcrossinbredlines].Yi Chuan Xue Bao29:332–338.

Sun, X., Cao, Y., Yang, Z., Xu, C., Li, X., Wang, S. & Zhang, Q.2004.Xa26,ageneconferringresistancetoXanthomonas oryzaepv.oryzaeinrice,encodesanLRRreceptorkinase-likeprotein.Plant J.37:517–527

Takeuchi, Y., Lin, S.Y., Sasaki, T. & Yano, M. 2003. Fine linkage mapping enables dissection ofclosely linkedquantitative trait loci forseeddormancyandheading inrice.Theor. Appl. Genet.107:1174–1180.

Tanksley, S.D. & McCouch, S.R. 1997.Seedbanksandmolecularmaps:unlockinggeneticpotentialfromthewild.Science277:1063–1066.

Taylor, G., 2002.Populus:arabidopsisforforestry.Doweneedamodeltree?Ann. Bot. (London)90:681–689.

Temnykh, S.P., Park, W.D., Ayres, N., Cartinhour, S., Hauck, N., Lipovich, L., Cho,Y-G., Ishii, T. & McCouch, S.R. 2000.Mappingandgenomeorganizationofmicrosatellitesequencesinrice(Oryza sativa L.).Theor. Appl. Genet.100:697–712.

Temnykh, S., DeClerck, G., Lukashova, A., Lipovich, L., Cartinhour, S. & McCouch, S. 2001.Computational andexperimental analysisofmicrosatellites in rice (Oryza sativa L.): frequency,length variation, transposon associations, and genetic marker potential. Genome Res. 11:1441–1452.

Testolin, R., Huang, W.G., Lain, O., Messina, R., Vecchione, A. & Cipriani, G. 2001.Akiwifruit(Actinidia spp.) linkagemapbasedonmicrosatellitesandintegratedwithAFLPmarkers.Theor. Appl. Genet.103:30–36.

Thomson M., Septiningsih E. and Sutrisno. 2003. WorkshoponIPRinrelationtoPlantBreeding,Cipanas,Indonesia,10–18November.

Thomson, M.J., Tai, T.H., McClung, A.M., Lai, X.H., Hinga, M.E., Lobos, K.B., Xu, Y., Martinez, C.P. & McCouch, S.R. 2003. Mapping quantitative trait loci for yield, yield components and


morphologicaltraitsinanadvancedbackcrosspopulationbetweenOryza rufipogonandtheOryza sativacultivarJefferson.Theor. Appl. Genet. 107:479–493.

Upholt, W.B. 1977.EstimationofDNAsequencedivergencefromcomparisonofrestrictionendonu-cleasedigests.Nucl. Acids Res.4:1257–1265.

Valent, B., Bryan, G.T., Jia, Y., Farrall, L., McAdams, S.A., Faulk, K.N. & Levy, M.2001.Enhancingdeploymentofriceblastresistancegenes:opportunitiesfromcloningaresistancegene/avirulencegenepair.In G.J.Khush,D.S.Brar&B.Hardy,eds.Rice genetics IV, pp.309-322.Enfield,NH,USA,SciencePublishers.

Vos, P., Hogers, R., Bleeker, M., Reijans, M., van de Lee, T., Hornes, M., Frijters, A., Pot, J., Peleman, J. & Kuiper, M. 1995.AFLP:anewtechniqueforDNAfingerprinting.Nucl. Acids Res.23:4407–4414.

Wang Z.X., Yano, M., Yamanouchi, U., Iwamoto, M.I., Monna, L., Hayasaka, H., Katayose, Y. & Sasaki, T. 1999.ThePib geneforriceblastresistancebelongstothenucleotidebindingandleucine-richrepeatclassofplantdiseaseresistancegenes.Plant J. 19:55–64.

Weber, J.L. & May, P.E. 1989.AbundantclassofhumanDNApolymorphismswhichcanbetypedusingthepolymerasechainreaction.Am. J. Hum. Genet.44:388–396.

Welsh, J. & McClelland, M. 1990.Finger-printinggenomesusingPCRwitharbitraryprimers.Nucl. Acids Res.18:7213–7218.

Wenzl, P., Carling, J., Kudrna, D., Jaccoud, D., Huttner, E., Kleinhofs, A. & Kilian, A. 2004.Diversityarraystechnology(DArT)forwhole-genomeprofilingofbarley.Proc Natl Acad Sci USA101:9915–9920.

Williams, J.G., Kubelik, A.R., Livak, K.J., Rafalski J.A. & Tingey, S.V. 1990.DNApolymorphismsamplifiedbyarbitraryprimersareusefulasgeneticmarkers.Nucl. Acids Res.18:6531–6535.

Wu, K.S. & Tanksley, S.D.1993.Abundance,polymorphismandgeneticmappingofmicrosatellitesinrice.Mol. Gen. Genet.241:225–235.

Xu, Y., Ishii, T. & McCouch, S.R. 2003.Marker-assistedevaluationofgermplasmresourcesforplantbreeding,InT.W.Mew,D.S.Brar,S.Peng,D.Dawe&B.Hardy,eds.Rice science: innovations and impact for livelihoods. Proc. Internat. Rice Res. Conf.,pp.213-230. Beijing,China,IRRI,ChineseAcademyofEngineeringandChineseAcademyofAgriculturalSciences.

Yamamoto, T., Lin, H., Sasaki, T. & Yano, M.2000.IdentificationofheadingdatequantitativetraitlocusHd6andcharacterizationofitsepistaticinteractionswithHd2inriceusingadvancedback-crossprogeny.Genetics 154:885–891.

Ye, X., Al-Babili, S., Kloti, A., Zhang, J., Lucca, P., Beyer, P. & Potrykus, I. 2000.EngineeringtheprovitaminA(beta-carotene)biosyntheticpathwayinto(carotenoid-free)riceendosperm.Science287:303–305.

Yoshimura, S., Yoshimura, A., Saito, A., Kishimoto, N., Kawase, M., Yano, M., Nakagahara, M., Ogawa, T. & Iwata, N.1992.RFLPanalysisofintrogressedchromosomalsegmentsinthreenear-isogeniclinesofriceforbacterialblightresistancegenes,Xa-1,Xa-3andXa-4.Jpn. J. Genet.67:29–37.

Yoshimura. S., Iwata, N.A.Y., McCouch, S.R., Abenes, M.L., Baraoidan, M.R., Mew, T.W. & Nelson, R.J. 1995.TaggingandcombiningbacterialblightresistancegenesusingRAPDandRFLPmarkers.Mol. Breed.1:375–387.


Yoshimura, S., Yamanouchi, U., Katayose, Y., Toki, S., Wang, Z.X., Kono, I., Kurata, N., Yano, M., Iwata, N. & Sasaki, T.1998.ExpressionofXa1,abacterialblight-resistanceinrice,isinducedbybac-terialinoculation.Proc. Nat. Acad. Sci. USA95:1663–1668.

Zane, L., Bargelloni, L. & Atarnello, T.P. 2002.Strategiesformicrosatelliteisolation:areview.Mol. Ecol.11:1–16.

Zhao, X. & Kochert, G. 1993. Phylogenetic distribution and genetic mapping of a (GGC)nmicrosatellitefromrice(Oryza sativa L.).Plant Mol. Biol.21:607–614.

Zhang, L.Y., Bernard, M., Leroy, P., Feuillet, C. & Sourdille, P. 2005.HightransferabilityofbreadwheatEST-derivedSSRstoothercereals.Theor. Appl. Genet.111:677–687.

Zheng, H.G., Babu, R.C., Pathan, M.S., Ali, L., Huang, N., Courtois, B. & Nguyen, H.T.2000.Quantitativetraitlociforroot-penetrationabilityandrootthicknessinrice:comparisonofgeneticbackgrounds.Genome43:53–61.

Zietkietkiewicz, E.A.R., Rafalski, A. & Labuda, D. 1994.Genomefingerprintingbysimplesequencerepeat(SSR)-anchoredpolymerasechainreactionamplification.Genomics 20:176–183.

Chapter 4

marker-assisted selection in wheat: evolution, not revolution

Robert Koebner and Richard Summers


SummaryThischapterreviewstheuptakeofmarker-assistedselection(MAS)inwheatinaEuropeancontext.Althoughlessintensethanthescaleofitsapplicationinmaize,reflectingthefactthatmaizevarietiesarepredominantlyF1hybrids,theuseofMASinwheathasgrownoverthelastfewyears.Thisgrowthhasbeenencouragedbyanincreaseinthenumberofame-nabletargettraits,butmoresignificantlybyacombinationoftechnologicalimprovements,particularlyintheareasofDNAacquisition,laboratorymanagementsystemsandintegra-tionintothebreedingcycle,whichtogetherhaveservedtoreducetheperunitcostofeachdatapoint.Microsatellites(simplesequencerepeats[SSRs])are,andwilllikelyremainforsometime,themarkerofchoicebecauseoftheirflexibilityandtheknowledgebaseassoci-atedwiththem.SomecurrentexamplesareprovidedoftheuseofMASinamajorUnitedKingdomcommercialbreedingprogramme.

Chapter 4 – Marker-assisted selection in wheat: evolution, not revolution 53

introduCtionWheat is a very important world staplecrop. The 2005 United States Departmentof Agriculture (USDA) estimates for theglobal production of wheat (both breadand durum) and maize are, respectively,627milliontonnesand708milliontonnes.In Europe, bread wheat is without doubtthe most important broad-acre crop, witha production in the extended EuropeanUnion of 25 states of 115 million tonnes(maize48milliontonnes).Thelargestpro-duction and highest productivity of breadwheat are achieved in northwest Europe.Historically, wheat has been bred largelyby government-sponsored national andregional programmes, but the introduc-tionofplantvariety rights intoEurope inthe 1960s encouraged participation by theprivate sector. Currently, wheat breedingin northwest Europe is almost exclusivelycarried out by private companies, withsomeresearchunderpinningby thepublicsector. Breeders continue to be successfulintheproductionofhigh-yielding,disease-resistant,high-qualityvarieties and, in theUnitedKingdomatleast,geneticadvancesforyieldhavebeenrunningatbetween0.5to1percentperannumformanyyears.

Wheat is a naturally inbreeding spe-cies, and although a level of heterosis canbe demonstrated, difficulties in enforcingcross-pollinationinareliableandcost-effec-tivewayhavehinderedthedevelopmentofany significant contribution of F1 hybridsto the variety pool. Most varietal develop-ment programmes are therefore based onversions of the long-established pedigreebreedingsystem,wherelargeF2populationsare generated and conventional phenotypicselection is carried out in early genera-tions for highly heritable, qualitative traits(suchasdiseaseresistance)andinlateronesfor quantitative traits (primarily yield and

quality). Thus, most varieties are bred andgrown as inbred, pure breeding lines. As aresult, the unitvalueofseedandeconomicmargins for breeders are low. By contrast,maizeisanaturallyout-crossingspeciesthatshowshighlysignificant levelsofheterosis.This has resulted in the majority of maizebreeding being geared to the productionof F1 hybrids. In industrialized countries,maize hybrid breeding has for some timebeendominatedbyasmallnumberoflargeprivatesectorcompaniesthatareabletosus-tainprofitabilitythroughtheircontroloverthegenotypeoftheirvarieties.Norevenueislostasaresultoftheuseoffarm-savedseed,and the inbred components of a successfulhybrid are not available to competitors touseasparentalmaterial for theirownvari-etal improvement programmes. This hasfar-reaching implications on the feasibilityof MAS in maize, and largely explains thelead that maize enjoys over wheat in thedeploymentofMAStechnology.

The continuing development of molec-ularmarkertechnologyoverthelastdecadehas been a happy by-product of “bigbiology” genomics research. As recentlyas1996,thedefinitionof5000SSRlociinthehumangenomemeritedamajorpubli-cationinNature(Dibet al.,1996),butthenumberofknownhumansinglenucleotidepolymorphisms(SNPs)nowrunsintomil-lions. Thus, although marker availability,potentially at least, is no longer limitingin crops, and the clear potential benefitsof marker deployment to plant breedingare undisputed, only relatively recentlyhasitbeguntomakemorethanamarginalimpact on breeding methodology. Eveninmaize,where the levelofDNAmarkerpolymorphism is high, large-scale deploy-mentofMASdidnotgatheranysignificantmomentumuntilmore than15years afterthepublicationofthefirstrestrictionfrag-


ment length polymorphism (RFLP)-basedmaize genetic map. In the less geneti-cally variable cereals, prominently wheat,the level of polymorphism is not now inpractice likelytorepresentthemajorcon-strainttoMASuptake,althoughinthepastitwasarguedthat thiswas thecase.Whathas changed in recent times is that cur-rent marker technology, and systems ofDNAacquisition, laboratorymanagementand integration into the breeding cycle,havealldevelopedtotheextentwherethebenefits of MAS can be increasingly real-ized in actual practice. As many of theseimprovements are incremental rather thansudden, we argue that the trends in MASapplication in wheat are characteristicallyevolutionaryratherthanrevolutionary.

target traitS for maS in northweSt european winter wheat BreedingThe use of MAS to date has a history ofabout20years,anduntilrecentlyinvolvedthe exploitation of just two non-DNA-based assays. The first, which has beenretained with only slight modificationssince its inception, exploits a correlationbetween bread-making quality and allelicstatus at the Glu-1 (endosperm storageproteinsubunit)loci.Ituseselectrophoreticprofiles obtained by the straightforward,robustandcheapproceduresodiumdodecylsulphatepolyacrylamidegelelectrophoresis(SDS-PAGE) from crude seed proteinextracts, which have been shown to bepartiallypredictiveofend-usequality.Thesecondispredictiveforthepresenceofthegene Pch1, which confers a high level ofresistance to eyespot, a stem base diseasethatisdifficulttoscreenusingconventionalpathologymethods.Boththesetargetshavein the meanwhile become assayable bypolymerase chain reaction (PCR)-based

assays, although SDS-PAGE remains inroutineusethankstoitsflexibilityandcosteffectiveness. In recent years, the numberof loci for which DNA-based assays havebeen generated has increased dramatically,the majority using PCR as a technologyplatform. Over 50 of these are described(specifically inaUnitedStatesofAmericacontext) at http://maswheat.ucdavis.edu/,which reports the output of an ongoingUnited States Department of Agriculture(USDA)-funded programme. The focusis heavily on disease and pest resistance,reflecting the generally simple inheritanceofgenesconferringthesetraits.

Someoftheabovetraitsareofsufficientrelevance to the United Kingdom contextthat identical or equivalent assays havebeenincorporatedinanumberofbreedingprogrammes,wheretheyareusedasguidestoparentalselectionand/orinearlygener-ationselection.ProminentamongthesearemarkersforthegenesRht-1(responsibleforthe “Green Revolution” semi-dwarfism),Pinb (grain texture), Pch1, Lr37/Yr17 (agenecomplexconferringresistancetotwoof the most important leaf fungal path-ogens) and the wheat/rye translocation1B/1R(whichisassociatedwithhighlevelsofyield).EmergingMAStargetsareneces-sarilyprogramme-dependent,butthebroadfocus is on quantitative trait locus (QTL)targets that could have a major impact onbreedingefficiency.IntheUnitedKingdom,aselsewhereworldwide,currentfocusisonresistance to the diseases Fusarium headblight (FHB), Septoria tritici blotch (STB)andbarleyyellowdwarfvirus(BYDV),andondurableresistancetoyellowrust.Othercurrenttargets,morespecifictotheUnitedKingdom and northwest European con-text,butinroutineuse,areresistancetotheinsectpestorangeblossommidge (OBM),andsoil-bornemosaicvirus(SBMV).


fhBTheimportanceofFHBislessinitseffectonyieldreduction,butratheronthepoten-tially damaging reduction in grain qualityassociated with infected grain, which canbeheavilycontaminatedby the fungal tri-cothecin toxins. An important source ofFHBresistanceoriginatesfromtheChinesevarietySumai3,andamajorcomponentofthis resistance (up to 50percent) has beenassociated with a single QTL (Waldron et al.,1999;Andersonet al.,2001;Buerstmayret al., 2002). While this QTL is largelyeffective in preventing the spread of thepathogenfollowinginfection,afurtherQTLthatgivesasignificantdegreeofprotectionagainstinitialinfectionhasbeenmappedtoadifferentchromosome(Buerstmayret al.,2003).SelectionforFHBresistancebycon-ventionalmeansiscomplicatedbothbythequantitative nature of the Sumai 3 resist-ance and by difficulties in ensuring evenandreliableartificialinfectionsinbreedingnurseries. However, SSR-based MAS pro-tocolshavebeendeveloped forbothQTL(see http://maswheat.ucdavis.edu/ andBuerstmayr et al., 2003), and the urgencyofbreedingforresistancehasensuredthatincreasinguseisbeingmadeofsuchassays.BoththeseQTLinconcertdonotexplainallthegeneticresistanceofSumai3toFHB,buttheremainderappearstobedeterminedbyQTLofminoreffectsand/orpleiotropiceffectsassociatedwithanearmorphology,which is inconsistent with a northwestEuropeanwinterwheatideotype.

StBSTB of wheat is caused by the fungusMycosphaerella graminicola (syn. Septoria tritici),andinrecentyearshasbecomethemajorleafdiseaseofwheatinmanyregionsof theworld. Inpastyears,good levelsofcontrolwereachievedbytheapplicationof

strobilurin fungicides, but their heavy usehasledtotheemergenceofpathogenstrainsthatcannotbesoeasilycontrolledbychem-icalmeans.Anumberbothofmajorgenesgiving near-complete resistance to specificraces of the pathogen and of quantitativerace non-specific resistances with poly-genic inheritance have been defined, andoneof theformer,Stb6,whichmapscloseto the SSR locus Xgwm369 on chromo-some 3A (Chartrain, Brading and Brown,2004),iscommoninmanygenepools.Thisensures that the gene has been retained inelitematerials,anditsknownmappositionhas made it relatively straightforward touseamarkerassaytotrackitspresence inbreedingpopulations.

Bydv SignificantgrainyieldlossesareattributabletonaturalinfectionsofBYDV,andnomajorsource of resistance has been identified todate in wheat. Control is achieved in theabsenceofgeneticresistancebyinsecticidalspray, which is associated with both aneconomic and an environmental cost.However,apotentresistanceispresentintherelated species Thinopyrum intermedium.It is possible to generate sexual hybridsbetween wheat and this grass, but the F1plants are self-sterile and either have tobe rescued by chromosome doubling orback-crossed to wheat. By this route, adistal segment of the grass chromosomethat carries the BYDV resistance geneBdv2 has been introduced into wheat. Asthis introgression comprises a significantlength of non-wheat chromosome, it hasbeenrelativelystraightforwardtogeneratemarkerssuitableforMASuse(Ayalaet al.,2001a;Zhanget al.,2004).AMASapproachforscreening isattractivebecauseartificialinoculation involves the propagation ofvirus-bearingaphids,whilenaturalinfections


are unreliable. Interestingly, unlike theexperience with many alien introgressionsegments,noobviousnegativeeffectsofitspresence on agronomic performance haveyet been detected either in InternationalMaize and Wheat Improvement Center(CIMMYT)trials(Ayalaet al.,2001b)oratRAGTSeeds(Cambridge,UK).

durable resistance to yellow rustYellowrustishistoricallythemostdamagingof the leaf fungal pathogens in temperateEurope. Control has been achieved in thepast largely by a combination of fungi-cide application and of combinations ofmajor seedling resistance genes, of whicha significant number have been describedin the literature.However, likemost race-specific resistances, most of these majorgeneshavelosttheireffectiveness,andthishas led to a renewed effort in the defini-tionofpartialoradultplantresistances tothis disease. The French variety Cappelle-Desprez dominated the wheat crop acrossFranceandtheUnitedKingdomduringthe1960s and 1970s, and maintained its levelof adult resistance to yellow rust over thewhole of this period. A major part of thegeneticbasisforthisdurableresistancewaslocated to a translocated wheat chromo-some(LawandWorland,1997),andthishasbeenconfirmedbyarigorousQTLanalysis(Mallardet al.,2005),whichhasprovidedanumberofinformativeSSRmarkersforthiseffect.Other independent sourcesofadultresistance have been identified in FrenchandEasternEuropeangermplasmatRAGTSeeds,andthemajorQTLresponsiblehavebeendefinedandmarked.

oBm OBMlarvae feedondevelopinggrainandheavy infestations result in a significantreductioningrainqualityandsomelossin

yield.As formanysporadicpests,pheno-typicscreeningisunreliableandanindirectmeans of selection would be valuable.The gene Sm1 confers resistance to OBM(Sitodiplosis mosellana)bytheexpressionofanantibioticthatkillsorslowsthedevelop-mentoflarvae.Thomaset al.(2005)definedthe map position of Sm1 and proposed aclose linkage with an SSR locus Xbarc35.This linkage remains to be validated inUnited Kingdom breeding populations, asit remains unclear whether the antibioticeffect shown by a few United KingdomwheatvarietiesisconferredbySm1.

SBmvSBMVisoneoftwoknownviralpathogenstransmitted by the soil fungus Polymyxa graminis(anotheronebeingyellowmosaicvirus[YMV]),andcanbeanimportantagentofyieldlossinsomeareas.Chemicalcontrolisnot feasible,andoncesoil is infectedbythe virus-bearing host, the only solutionspossible are to abandon wheat culture orto use resistant varieties. Phenotyping isparticularly difficult as plant infection isenvironmentallysensitive,andthedetectionofinfectionislaboriousandpronetoerror.AproprietaryassayforresistancetoSBMVoriginating from European germplasm hasbeen in routine use at RAGT Seeds since2000 with a very high level of marker/phenotype association. More recently, abulk segregant analysis along with a QTLapproach has allowed the definition of aresistance locus to YMV from Chinesegermplasm, and a number of linked SSRmarkers have been identified (Liu et al.,2005).

high-throughput infraStruCtureSTechnical considerations of DNA acquisi-tion, laboratory information management


system(LIMS),laboratoryautomationanddata capture and analysis are generic foranyMASset-up,andthesearewellcoveredelsewhere in this volume. The limitationsaffecting MAS deployment in wheat flowfrom the restricted revenue generated bybreeding a self-pollinated, homozygous,non-hybridproduct.Asaresult,thevolumeofcapitalinvestmentaffordableinmaizeisnotavailabletoawheatMASprogramme.Financial constraints also affect the devel-opment of marker platforms. It is wellknownthatthepredictiveabilityofalinkedmarkerwillbedisruptedbyrecombination,and therefore that “perfect” markers aremoredesirablethanlinkedones.However,the development of genome-wide gene-basedmarkers,pre-eminentlySNPs,whichare particularly suited to high-throughputgenotypingonautomatedplatforms,isstillsome way off. At present, an insufficientnumberofsuchassayshasbeenestablished(grain hardness, semi-dwarfness and graintexture) to consider adjusting the presentmajor genotyping methodology, which isfoundedonSSRs.DoubtshavebeenraisedthatSNPfrequency inexonsequencewillbe high enough to generate informativeassays for many critical genes, but earlyexperiencesuggeststhatsequencepolymor-phismismorethanadequateinintronsandotheruntranslatedregionsofwheatgenes.At present, the consensus is that there isplenty of mileage left in SSR technology,andwheatmapscontinuetoberefinedbytheadditionofnewSSRloci.

ConCluSion In 1999, Young set out his “cautiouslyoptimisticvision”forMAS.Sevenyearson,the situation continues to crystallize. Thetechnologyitselfisnolongerlimiting.With

respecttomarkeravailability,SSRsremainuseful and SSR-based genetic maps arebecoming increasingly densely populated,while SNPs may eventually represent asourceofplentifulperfectmarkersforgenesof defined function. The “big biology”spawned by the genomics revolution hasbroughtminiaturizationandautomationtobiologicalassayssothatlevelsofthroughputrelevant to the wheat breeding processare becoming attainable. The issue thatremains unresolved is the affordability oflarge-scaleMAS.Aswheat isabroad-acrecommodity product, its value is low, andthis impedes the abilityof the industry toinvest inMAS infrastructure to theextentthat is possible for crops such as maizewherethegenerationofF1hybridseedisaviableproposition.However,aseconomiesof scale and improvements in technologycontinue to drive down assay price, thepenetrationofMASintocommercialwheatbreeding will surely grow. This growthshouldprogressivelyallowawideningintherangeofpossibleMAStargets,inparticularextendingtocriticalonessuchasQTLforyield and its components (mean kernelsize, kernel number per ear and numberof fertile tillers per unit area). These arealreadywidelyexploitedinmaizebreedingandtheirdefinitionandvalidationinwheatrepresent a significant research theme inboththepublicandprivatesectors.Inthemeantime,muchMASusewillbedirectedtowardsspecificpurposessuchasacceleratedselection of a few traits that are difficultto manage by conventional phenotyping,for the maintenance of recessive allelesin backcrossing programmes, for thepyramidingofdiseaseresistancegenesandforguidingthechoiceofparentstobeusedincrossingprogrammes.


referenCeSAnderson, J.A., Stack, RW., Liu, S., Waldron B.L., Fjeld, A.D., Coyne, C., Moreno-Sevilla, B.,

Mitchell Fetch, J., Song, Q.J., Cregan, P.B. & Frohberg, R.C.2001.DNAmarkersforFusariumheadblightresistanceQTLsintwowheatpopulations.Theor. Appl. Genet.102:1164–1168.

Ayala, L., Henry, M., Gonzalez-de-Leon, D., van Ginkel, M., Mujeeb-Kazi, A., Keller, B. & Khairallah, M.A. 2001a. Diagnostic molecular marker allowing the study of Th. intermedium-derived resistance to BYDV in bread wheat segregating populations. Theor. Appl. Genet. 102:942–949.

Ayala, L., van Ginkel, M., Khairallah, M., Keller, B. & Henry, M.2001b.ExpressionofThinopyrumintermedium-derived barley yellow dwarf virus resistance in elite bread wheat backgrounds.Phytopath.91:55–62.

Buerstmayr, H., Lemmens, M., Hartl, L., Doldi, L., Steiner, B., Stierschneider, M. & Ruckenbauer, P. 2002. Molecular mapping of QTLs for Fusarium head blight resistance in spring wheat. I.Resistancetofungalspread(TypeIIresistance).Theor. Appl. Genet.104:84–91.

Buerstmayr, H., Steiner, B., Hartl, L., Griesser, M., Angerer, N., Lengauer, D., Miedaner, T., Schneider, B. & Lemmens, M.2003.MolecularmappingofQTLsforFusarium headblightresist-ance in spring wheat. II. Resistance to fungal penetration and spread. Theor. Appl. Genet. 107:503–508.

Chartrain, L., Brading, P.A. & Brown, J.K.M. 2004. Presence of the Stb6 gene for resistance toSeptoria tritici blotch (Mycosphaerella graminicola) in cultivars used in wheat-breeding pro-grammesworldwide.Plant Pathol.54:134–143.

Dib, C., Faure, S., Fizames, C., Samson, D., Drouot, N., Vignal, A., Millasseau, P., Marc, S., Hazan, J., Seboun, E., Lathrop, M., Gyapay, G., Morissette, J. & Weissenbach, J.1996.Acompre-hensivegeneticmapofthehumangenomebasedon5264microsatellites.Nature 380:152–154.

Law, C.N. & Worland, A.J.1997.Thecontrolofadultplantresistancetoyellowrustbythetranslo-catedchromosome5BS-7BSofbreadwheat.Plant Breed.116:59–63.

Liu, W.H., Nie, H., Wang, S.B., Li, X., He, Z.T., Han, C.G., Wang, J.R., Chen, X.L., Li, L.H. & Yu, J.L.2005.MappingaresistancegeneinwheatcultivarYangfu9311toyellowmosaicvirus,usingmicrosatellitemarkers.Theor. Appl. Genet.111:651–657.

Mallard, S., Gaudet, D., Aldeia, A., Abelard, C., Besnard, A.L., Sourdille, P. & Dedryver, F.2005.Genetic analysis of durable resistance to yellow rust in bread wheat. Theor. Appl. Genet. 110:1401–1409.

Thomas, J., Fineberg, N., Penner, G., McCartney, C., Aung, T., Wise, I. & McCallum, B. 2005.ChromosomelocationandmarkersofSm1:ageneofwheatthatconditionsantibioticresistancetoorangewheatblossommidge.Mol. Breed.15:183–192.

Waldron, B.L., Moreno-Sevilla, B., Anderson, J.A., Stack, R.W. & Frohberg, R.C. 1999. RFLPmappingofQTLsforFusariumheadblightresistanceinwheat.Crop Sci.39:805–811.

Young, N.D. 1999. A cautiously optimistic vision for marker-assisted selection. Mol. Breed. 5:505–510.

Zhang, Z.Y., Xu, J.S., Xu, Q.J., Larkin, P. & Xin, Z.Y.2004.DevelopmentofnovelPCRmarkerslinked to the BYDV resistance gene Bdv2 useful in wheat for marker-assisted selection. Theor. Appl. Genet.109:433–439.

Chapter 5

marker-assisted selection for improving quantitative traits

of forage crops

Oene Dolstra, Christel Denneboom, Ab L.F. de Vos and E.N. van Loo


SummaryThischapterprovidesanexampleofusingmarker-assistedselection(MAS)forbreedingperennial ryegrass (Lolium perenne), a pasture species. A mapping study had shownthepresenceofquantitative trait loci (QTL)forsevencomponent traitsofnitrogenuseefficiency(NUE).TheNUE-relatedQTLclusteredinfivechromosomalregions.TheseQTLwerevalidatedthroughdivergentmarkerselectioninanF2population.Thecriterionused for plant selection was a summation index based on the number of positive QTLalleles. The evaluation studies showed a strong indirect response of marker selectiononNUE.Marker selectionusinga summation index suchas appliedhereproved tobeveryeffective fordifficultandcomplexquantitative traits suchasNUE.Thestrategy iseasily applicable in outbreeding crops to raise the frequency of several desirable allelessimultaneously.

Chapter 5 – Marker-assisted selection for improving quantitative traits of forage crops 61

introduCtionMostagronomicalcharacteristicsofforagecrops have a quantitative, polygenic andmostlycomplexnature.Forthesereasons,genetic improvement of such traits islaborious and time consuming. Improvingnitrogenuseefficiency(NUE)inperennialryegrass (Lolium perenne, 2n = 14), themajorgrass species innorthernEurope, isin this respect a good example. The highinput of nitrogen needed to attain highforage yields for animal husbandry hascaused severe water pollution (van Loo et al.,2003),andtherefore loweringnitrogeninputs throughimprovingnitrogenusebybreedingisofutmostimportance.

SelectionforNUE,however,isnoteasilyimplemented in conventional grass breed-ing based on field evaluations. Adequatetestingrequiresseparateandlong-termtri-alswithgoodcontrolof theNstress,andsuch experiments tend to be rather inac-curate. To circumvent the disadvantagesof field testing,ahydroponics systemwasused in this study in which the crop situ-ation is simulated with growth-dependentN application (van Loo et al., 1992), theaim being to grow plants having an equalsuboptimal N content. The set-up has acapacitytotestabout1600plantsinparal-lelandenablesallplantstoexperiencemoreor less thesameNstrain.Criteriausedtomeasure NUE are several plant growthcharacteristics, such as tillering, and shootandrootgrowth.Eachtestusuallyrequiresfourtofivecuts.Thetraitisvigour-relatedandcomplex,andisextremelyimportantinrelationtoregrowthaftercutting.Together,all theseaspectsmakeNUEaveryattrac-tivetraitforMAS.

analySiS of genetiC variationThe genetic variation for NUE presentin an F1 plant originating from a cross

between two contrasting genotypes forNUE was first analysed by crossing theF1 with a doubled haploid. The resultingtest cross progeny was then used to pro-duceamolecularmarkermapandanalysethe variation. This approach was chosento avoid inbreeding effects and to be ableto use dominant molecular markers. TheperformanceofthemappingpopulationforNUE-relatedtraitswasstudiedonhydro-ponicswiththesystemsetatamoderatelylow nitrogen deficiency (3.6percent N ofleafdryweight).Theoutcomeofthemap-ping study was a genetic map with sevenlinkagegroups.

Putative genes (quantitative trait loci[QTL])forthecomponentsofNUEwerefoundonfourlinkagegroups.ThelocationoftheselectionmarkersforQTLisdepict-edinFigure1.Themapshowsfivegenomicsites with 1-5 QTL. In total, 13 QTL forsevenNUErelatedtraitswerefound.ThreesitescontainmorethanoneQTL.

The findings of the current study aretypical forgeneticanalysesofquantitativetraits in forage crops and also indicativeof the problems associated with exploita-tionofQTLinformationthroughmarker-assisted breeding. These included uncer-taintieswith respect toeffect and locationof QTL, the fairly large number of QTLoften found in genetic analyses, the co-segregation of QTL and the weighing ofthedifferentcomponenttraitsofNUEandNUE-QTL.Belowisadescriptionofhowthese breeding problems were solved orcircumvented in a divergent marker selec-tionstudytovalidatetheQTLfoundinthemappingstudy.

divergent marker SeleCtionThe plant materials used in the validationstudy were an F2 generation obtained byselfing of the heterozygous F1 genotype


used to generate the mapping populationmentionedabove(vanLooet al.,2003).Intotal,about200genotypesweregenotypedfor five amplified fragment length poly-morphism(AFLP)selectionmarkersusingthefluorescentAFLPtechniquedevelopedby Applied BioSystems (Figure 1). Themarkers were co-dominantly scored usingtheheightsof the fluorescencepeaks rela-tive to thoseofhomozygous fragments asacriterion.

The genotyping data were used sub-sequently as a basis for a divergent massselection programme. The selection strat-egy is outlined in Figure 2. The selectioncriterion was a genotype-specific selec-tion index, being the summation of allpositiveQTLalleles (orchromosomeseg-ments)overthefiveQTLsitesconsidered(Figures1and2).

appliCation of marker SeleCtionThe AFLP technique is usually not themarker technology of choice for selectionpurposesbecauseofitsdominantnatureandhigh costs per selection marker. However,co-dominant scoring of the five selectionmarkers was quite adequate. The trimo-dalfrequencydistributionsallowedproperclassificationofplants,althoughsomemis-classificationcannotbefullyexcluded.Theadvantages of co-dominant AFLP scoringfromaselectionpointofviewaresolargethat a small number of genotyping errorsareacceptable.

Thedecisiontouseasummation indexasthecriterionforselectionwasmadepri-marilybecauseofthedifficultyofweightingthe individual NUE related traits and theco-localization of QTL. The designationof the positive QTL alleles (chromosome

FiGURe 1map position of the selection markers (in blue) for qtl for seven components of

nitrogen use efficiency

E32M49-164 E37M49-403 K10F08-141 K10F08-144

E32M52-02R E32M49-219 E32M54-05R E38M48-04R E38M48-130 E32M52-352 E32M48-303

E41M48-239 E32M52-01R E36M48-220 E41M47-425 E32M52-04R E41M61-362

LG1

E36M48-180 E36M50-094 Uni-001-147 K07H08-165

E40M47-380 E38M51-285 E38M49-223 E36M49-213 E37M49-343 E37M49-118 E35M48-252 E38M49-309 E41M48-479 E38M51-429 K14F12-113 K14F12-123

E32M54-03R E38M49-305 E35M48-339 E38M51-01R E33M49-233 E32M51-313

E37M49-201 E37M48-265 E41M48-205 E38M49-242

LG3

Rye35-117 K04D01-157 K04D01-239 E35M48-328 E41M48-268 E37M48-347 E36M50-233 E38M48-11R

ADH-1000 Rye12-148 Rye26-147

E32M51-303 E36M49-133 E38M48-416 E32M51-328 E41M48-296 E38M51-247 E33M49-277 E36M49-217 E38M48-08R E37M48-331 E38M51-14R E41M48-299 E32M51-212 E38M47-334 E38M47-255 E38M47-348

E35M47-135 E41M48-208 E38M48-093 E38M51-291 E38M51-04R E32M51-304 E33M51-03R E38M52-262 E38M52-260 E38M51-12R E38M51-267 Meth300-250 E36M48-108 E37M47-441 KXX104-108

LG4

Rye14-221 Rye14-223

E38M52-289 E36M49-297 E36M50-173 E32M52-169 E38M51-13R E38M48-10R E38M48-05R E38M48-06R E32M52-08R E38M48-110 E38M51-264 E32M52-126 E32M52-06R E36M48-163

E35M48-141 E38M47-116 E32M54-06R E37M47-327 E32M52-137 E32M52-07R E35M47-278

LG6

0 510 15 20 25 30 35 40 45 50556065707580859095100105110115120125130135140145150155160165170

LW1 LAE3

LWR1

TN1 DWS1 LAE1 DWT1

TN2 DWS2 LAE2 DWT2 DWR1

TN: Tiller number LAE: Leaf area expansion LW: Leaf width LWR: Leaf weight ratio DWR: Dry weight roots DWS: DWT: Total dry weight

LWR2

Dry weight shoots


fragments) turned out to be straightfor-ward.Figure3showstheF2frequencydis-tribution for thenumberof“plusalleles”.The population mean is somewhat belowthe expected number of five owing to thefactthattheAFLPmarkeronLG1showeda skewed segregation. This is likely duetogametophytic selection in favourof thenegativeQTLallele,perhapsduetolinkagewithanincompatibilitylocus.

The intensities of selection were setat about 25 percent, representing about50genotypesper selection (Figure3).Theselection pressure was kept fairly lowbecauseoftheneedtohavesufficientseedsfor measuring selection responses. In thisway,theinfluenceofgeneticdriftaccompa-nyingmarkerselectionwasminimized.Thecut-off point for the top selection was sixpositive alleles and three for the oppositeselection (Figure 3). The frequency of theplus alleles was on average 0.66 and 0.27,respectively. Selection showed a positiveresponse for all NUE loci. However, thebetween-selection difference in allele fre-

quencyofthelocirangedfrom0.18to0.77,showingthatindexselectiondidnotaffectallNUElocitothesamedegree.Thediffer-enceswereprobablymainlyduetochance.

indireCt reSponSe to marker SeleCtionThe selections were then multiplied usingapolycross scheme (aftervegetativeprop-agation) to obtain sufficient seeds forevaluationonhydroponicsandundervar-iousfieldconditions.ThemarkerselectionswereevaluatedforNUEinareplicatedtrialwith two cuts on hydroponics at two Nlevels,being2.5and5percentNinleaves(vanLooet al.,2003).Thesamesetofplantcharacteristics as in the original mappingstudieswasmonitoredaftereachcut.Leafareaexpansionrate, leaf lengthandwidth,as well as tiller number, were determinedone week after cutting. The determina-tionofshootandrootdryweightfollowedthree weeks later. The indirect responsesto marker selection are summarized inFigure4. At low N supply, the NUEplus

FiGURe 2Strategy applied for marker selection

F2

Co-dominant scoring of markers associated with QTLs

QTL

1

QTL

2

QLT

3

QTL

4

QTL

5

Sum

1 2 1 0 0 2 52 1 1 0 0 2 43 0 2 2 1 1 64 2 2 2 1 1 85 2 2 1 1 1 7. . .

195 1 1 2 2 2 8196 1 1 2 0 2 6197 0 2 0 1 2 5198 0 2 0 1 1 4199 2 2 1 1 1 7200 2 1 2 2 1 8

Number of plus alleles

Genoty

peSelection criterion


FiGURe 3frequency distribution for number of plus alleles found for five qtls

0

10

20

30

40

50

60

70

0 1 2 3 4 5 6 7 8 9 10

Number of plus alleles

Nu

mb

er o

f p

lan

ts

selection showed a remarkable 40percenthigher tillering rate and dry matter pro-duction than the NUEmin selection. The40percenthighertilleringrateisassociated

with a 40percent higher leaf area increaseafterdefoliation(datanotshown).Relativerootgrowth(expressedastheratioofroottototalgrowth)andleaflengthwerehardly

FiGURe 4performance of the marker selections, nuemin (white) and nueplus (blue) on hydroponics at low n-supply (2.5%n in leaves) and high n-supply (5%n in leaves). the performance is

expressed as a percentage of the mean of the two selections

60 80 100 120

Tiller number

Dry weight roots

Dry weight shoots

Leaf length

Root weight ratio

60 80 100 120

Low N - supply High N - supply

the two contrasting marker selections, i.e. nUeplus (right) and nUemin (left), are marked in blue.


changedthroughmarkerselection.AthighN supply, the performances of NUEplusandNUEminwerefairlysimilar(Figure4).Theselectionsalsoshowedstrikingdiffer-ences in field trials in Germany, Englandand the Netherlands. At suboptimal N,the NUEplus selection significantly out-performed its counterpart inyieldsofdrymatter and water soluble carbohydrates,whiletotalNuptakewasslightlylower.

ConCluSionSDivergent mass selection has shown thatmarker selectionusinga summation indexcan be very effective for difficult andcomplexquantitativetraitssuchasNUE.Acollateraladvantageofsuchanapproachis

thatitoffersatruevalidationoftheputativegenes (QTL)for thetraitsof interest.Theassociated response to marker selectiondistinctively indicates thepresenceof truegenes affecting NUE, particularly in thevicinity of markers, which were stronglyaffected by the selection imposed. Theresults also indicate that recurrent massselectiontoincreasethenumberofpositiveallelesisworthwhile.Thestrategyiseasilyapplicableinoutbreedingcrops.

aCknowledgementSThe researchworkwascarriedoutwithinthe framework of the EU-FAIR projectNIMGRASS(CT98-4063).

referenCeSvan Loo, E.N., Schapendonk, A.H.C.M. & de Vos, A.L.F. 1992. Effects of nitrogen supply on

tilleringdynamicsandregrowthofperennialryegrasspopulations.Netherlands J. Agric. Sci.40:401–419.

van Loo, E.N., Dolstra, O., Humphreys, M.O., Wolters, L., Luessink, W., de Riek, W. & Bark N.2003.Lowernitrogenlossesthroughmarkerassistedselectionfornitrogenuseefficiencyandfeedingvalue(NIMGRASS).Vorträge Pflanzenzüchtung59:270–279.

Chapter 6

targeted introgression of cotton fibre quality quantitative trait loci

using molecular markers

Jean-Marc Lacape, Trung-Bieu Nguyen, Bernad Hau and Marc Giband


SummaryWithin the framework of a cotton breeding programme, molecular markers are used toimprovetheefficiencyoftheintrogressionoffibrequalitytraitsofGossypium barbadenseinto G. hirsutum. A saturated genetic map was developed based on genotyping dataobtainedfromtheBC1(75plants)andBC2(200plants)generations.Phenotypicmeasure-mentsconductedoverthreegenerations(BC1,BC2andBC2S1)allowed80quantitativetraitloci (QTL) tobedetected for fibre length,uniformity, strength,elongation, finenessandcolour.PositiveQTL,i.e.thoseforwhichfavourableallelescamefromtheG. barbadenseparent, were harboured by 19 QTL-rich regions on 15 “carrier” chromosomes. In sub-sequentgenerations(BC3andBC4),markersframingtheQTL-richregionswereusedtoselectabout10percentofover400plantsanalysedineachgeneration.AlthoughBCplantsselectedthroughthemarker-assistedselection(MAS)processshowpromisingfibrequality,onlytheirfullfieldevaluationwillallowvalidationoftheprocedure.

Chapter 6 – Targeted introgression of cotton fibre quality quantitative trait loci using molecular markers 69

introduCtionAmongthefourspeciesofGossypium that produce seedswith spinnable fibres calledcotton,Gossypiumhirsutumdominatestheworld’scottonfibreproduction,accountingforapproximately90percentoftotalworldproduction.Thesecondmostcultivatedspe-cies,G. barbadense,includessuperiorextralong, strong and fine cottons. However,comparedwithG. hirsutum,themarketingadvantageof“highquality”G. barbadense cottons is offset by their lower produc-tivityandanarroweradaptabilitytoharshenvironments.Breedingapproacheswithinthesetwospecieshaveessentiallyreliedonhybridizationandselectionmethods (sub-sequent to simple or complex crosses, apedigreesystem,sometimescombinedwithrecurrent selection, is applied). AlthoughG. hirsutum and G. barbadense displaycomplementarycharacteristics,attemptstoutilizedeliberateinterspecificG. hirsutum/G. barbadense recombinations throughconventional breeding have had limitedimpactoncultivardevelopment.

In thepast10–15years,DNAmarkersforanalysesofQTLandMAShavereceivedconsiderableattentionbyplantandanimalbreeders (Dekkers and Hospital, 2002).However,followinganinitialkeeninterestandpromisesformolecular-basedbreedingapproaches, the successful application ofthistechnologyhasbeenshowntodependonthereliabilityandaccuracyoftheQTLanalyses,whichinturnarestronglyaffectedbybothpopulationsizeandenvironmentalfactors (Schön et al., 2004). Examples ofapplied MAS in breeding programmes arestill scarce, particularly when complextraits (yield components, product quality)areunderconsideration.

Inthecaseofcotton,itisonlyrecentlythat the results of efforts to gain a betterunderstanding of the genome and the

molecular basis of fibre quality have beenpublished. Most of the earlier efforts incotton molecular breeding concentratedon interspecific hybridization, due to thefact that, intraspecifically, the major spe-cies G. hirsutum displayed a very lowlevel of molecular variability (Brubakerand Wendel, 2001). Based on studies ofinterspecific G. hirsutum x G. barbadensepopulations,publishedreportsrelate(i) totheconstructionofhigh-resolutiongeneticmaps (Lacape et al., 2003; Rong et al.,2004);and(ii)totheidentificationoffibrequality-related QTL (Jiang et al., 1998;Kohel et al., 2001; Paterson et al., 2003;Lacape et al., 2005). In parallel, data haveaccumulated describing the cotton fibre transcriptome (reviewed by Wilkins andArpat,2005).Thesestudiesconfirmedthatkeyfibrequalityproperties,suchaslength,finenessandstrength,arecontrolledquan-titatively, thus complicating conventionalbreedingforfibreimprovement.

Withintheframeworkofamarker-assistedbackcross introgression scheme aimed attransferringfibrequalitytraitsfromalow-productivitylineofG. barbadense(donor)intoaproductivelineofG. hirsutum(recip-ient), a saturatedgeneticmapof tetraploidcotton was first developed (Lacape et al.,2003).ThischapterdescribeshowmolecularmarkerswereusedintheearlyBC1andBC2generations to identify QTL-rich regionsinvolved in determining fibre quality, asrecently reported by Lacape et al. (2005),andhowMASwasactuallyimplementedinthelaterBC3andBC4generations.

methodologyThe major milestones (Figure 1) in themarker-assistedbackcrossselectionprocessincluded the construction of two geneticmaps from the BC1 and BC2 populations,the detection of fibre quality QTL from


threephenotypingdatasets(BC1,BC2andBC2S1)andtheactualmarker-basedselectionin the BC3 and BC4generations, followedbytheanalysisofmarker-traitassociationsintheBC3andBC4generations.

plant materialThe initial interspecific cross involved theG. hirsutum variety Guazuncho 2 andthe G.barbadense variety VH8-4602.Guazuncho 2 is a modern pure lineG. hirsutumvarietycreatedinArgentinaandwaschosenasa recipient in thebackcrossgenerationsforitsgoodoverallagronomicperformance. VH8-4602, a G. barbadensevariety of the Sea Island type, was thedonor parent for superior fibre quality, inparticularforlength(+9to+12mmascom-paredwithGuazuncho2),strength(+12to+16g/tex)andfineness(-30to-50millitex);

conversely its fibrecolour indices (reflect-anceandyellowness)areoflowervalue.1

The plant material used in the multi-generation QTL analyses included threepopulations: BC1, BC2 and BC2S1 (Lacapeet al., 2005). The first backcross genera-tion (BC1), consisted of 75 plants grownin a greenhouse in Montpellier (France)during the summer of 1999; these servedasfemaleparentsforthesecondbackcrossto Guazuncho 2. Two hundred individualfield-grown BC2 plants that had shown asatisfactory production of BC3 seeds andoriginating from 53 different BC1 plantswere used in 2000. Open pollinated seedsharvested from BC2 plants were grown as200 BC2S1 progenies in 2001 under fieldconditions in Brazil. Each BC2S1 line was

FiGURe 1Schematic representation of the marker-assisted backcross selection scheme

G. hirsutum x G. barbadense (var. Guazuncho 2) (var. VH8 -4602)

F1

BC 1

G.h

G.h

BC 2 G.h

BC 2S1

BC 3 G.h

BC 4

x

x

x

x

BC 1generation:1160 loci (RFLP + SSR + AFLP)

phenotypic data and QTL mappingNo selection

G.h

BC 5

x

BC4S1

BC 2 generation: 200 plants 514 loci (SSR + AFLP)

200 BC 2 + 200 BC 2S1 phenotypic data and QTL mappingNo selection

BC 3generation: 411 plants with 35 SSR loci and43 plants with 340 AFLP loci

Marker selection

BC 4 generation:450 plants with 50 SSR loci and 37 plants with 340 AFLP loci

Marker and phenotypic selection

Pyramiding QTL - NILs

-

:

11tex=1gram/kilometre


plantedintworeplications,eachplot(onerow) measuring 5 m. The next BC3 andBC4 generations were grown under field(411BC3in2002)orgreenhouse(450BC4in 2003) conditions in Montpellier. EveryplantineachBC1-4generationwasusedforDNA extraction from young fresh leavesusing different methods described else-where (Lacapeet al., 2003;Nguyen et al.,2004).IneachBC3andBC4generation,anearly genotyping was conducted (beforefloweringofBC3plantsandattheseedlingstageforBC4plants),toreducethenumberof plants to be manipulated and raised tofloweringforselfingandbackcrossing.

Fromeachgeneration(75BC1,200BC2,400BC2S1,43selectedBC3and37selectedBC4), the cotton seed harvest was ginned(separationofthefibrefromtheseed)onalaboratoryrollerginandthefibrewassam-pled for analyses at the Fibre TechnologyLaboratory of the French AgriculturalResearch Centre for International Devel-opment(CIRAD).

fibre analysesAll fibre quality measurements (11 traits)were conducted at CIRAD, Montpellier,onahighvolumeinstrumentline(ZellwegerUster900,UsterTechnologies,Switzerland).Theseincludedlength,uniformity,strength,elongationandcolour.AFMT3maturimeter(ShirleyDevLtd.,UK)wasusedtodeterminemicronairevalue,maturityandfineness.

molecular analysesThe different types of markers displayingpolymorphism between G. hirsutum andG. barbadenseincludedrestrictionfragmentlengthpolymorphisms(RFLPs)(usedonlyin the BC1 generation), simple sequencerepeats(SSRs)andamplifiedfragmentlengthpolymorphisms (AFLPs). Details of themarkers and protocols used are provided

in Lacape et al. (2003) and Nguyen et al.(2004).TheAFLPmarkerswereallderivedfrom combinations of EcoRI/MseI primerpairs(64pairsintheBC1,45intheBC2and30 in the BC3 and BC4 generations). Thecotton microsatellites were derived essen-tiallyfromtwopubliclibraries,BrookhavenNational Laboratory (BNL) and CIRAD(CIR).IntheBC1generation,themicrosat-ellitesusedincluded188polymorphicBNLmarkers out of the 216 available (Lacapeet al., 2003) and 204 CIR markers out of392developed(Nguyenet al.,2004).Fromthe results of the combined QTL analysesoftheBC1/BC2/BC2S1generations(Lacapeet al., 2005), QTL-rich regions were iden-tified on “carrier” chromosomes, and SSRloci present within or in the vicinity ofthese regions were assembled for consti-tutinggroupsofthreeSSRs(onegroupperregion) to be tested as multiplexes, takinginto account both annealing temperatureandcompatibilityofsizesofamplifiedfrag-ments.Asubsetof60SSR(20region-specifictriplexes)wasusedforearlygenotypingofall411BC3and450BC4plants(seeexamplesinFigure2).The individualplants selectedfromBC3andBC4(43and37plantsrespec-tively)were further analysedusingknownAFLPstoprovidebroadgenomecoverage.Inthecontextofourmarker-assistedintro-gression programme, the SSR markerstarget theQTL-richregions, i.e. those lociof the “foreground genome” expected tohave been introgressed, while the AFLPmarkersessentiallyservetocovertherestofthegenome,i.e.the“backgroundgenome”,aimedatreturningtotherecipientgenomecomposition.

Construction of genetic map The BC1 (75 individuals) and BC2 (200individuals) maps were constructed sepa-rately using the MapMaker 3.0 software


(Lander et al., 1987). The MapMaker“group” (using a logarithm of the oddsratio [LOD] of 5.0 and 30 as a maximalrecombination frequency), “order” and“sequence” commands were used in eachcase.AfteraligningtheBC1andBC2mapsusing common loci, a consensus frame-work BC1/BC2 map was constructed bysimpleextrapolationofthepositionsoftheadditional BC2 loci on the BC1 map usedas a backbone map. The allelic constitu-tionthroughoutthe26chromosomesofallBC1-4individualswasdisplayedgraphicallyusing Graphical GenoTyping software (R.vanBerloo,LaboratoryofPlantBreeding,Wageningen,Netherlands)andrepresentedalongtheconsensusBC1mapdata.

qtl analysesThecombinedmarkerandphenotypicdatathenservedforthree(BC1,BC2andBC2S1)

separate QTL analyses of fibre qualitycomponents. The association betweenphenotypeandmarkergenotypewasinves-tigated through simple marker analysis(SMA), interval mapping (IM) and com-posite interval mapping (CIM) using thecomputersoftwareQTLCartographer1.13(Basten,WeirandBeng,1999)asdescribedin Lacape et al. (2005). In each data set(trait, generation), permutation-basedthresholds were considered at a 5percentriskat thegenome level. Intervalmethodsrelied on the positions of the loci on theconsensus BC1 map. Molecular data offurther generations (BC3 and BC4) werealso combined with phenotypic measure-ments for conducting the SMA option ofQTLCartographer.Cottonfibrepropertieswere considered from a product trans-formation perspective, meaning thatdecreasesthe infibrefinenessandyellow-

FiGURe 2examples of autoradiograms showing the segregation of 2 SSr triplexes observed among BC4 plants and used for the targeted genotyping of specific regions along chromosomes c5

and c25 (a subset of around 15 BC4 plants from the 450 plants analysed is shown)

CIR034

BNL3995

CIR301

CIR034

BNL3995

CIR301

c5

CIR109

BNL1047a

CIR280a

c25

CIR109

BNL1047a

CIR280a

CIR034

BNL3995

CIR301

CIR034

BNL3995

CIR301

c5

CIR109

BNL1047a

CIR280a

c25

CIR109

BNL1047a

CIR280a


ness index, for example, were positivelyconsidered.

Details of the plant material used andthe types of analyses undertaken duringthedifferentstepsoftheMASprocessaregiveninFigure1.

reSultSphenotypic variationThetwoparentswerecharacterizedbytheircontrasting fibre properties (Table1) withsignificantadvantagesfortheG. barbadenseparent in terms of length (+9.7mm onaverageoveralldatasets),strength(+15.9g/tex) and fineness (-38mtex). By contrast,theG. hirsutumparentdisplayedbetteryel-lownessindex/colourreflectance.ForeachBC population, it was observed that thedatafittednormaldistributions,thattrans-gressive segregants were regularly in thelowerrangeofphenotypicvaluesandthat,although progeny values rarely reachedthose of G. barbadense, high phenotypicvalueswereobserved,includingwithinthemostadvancedBC4generation(Table1).

genetic mappingThe first step in the programme involvedthe construction of two genetic maps of

tetraploidcottonbycombiningRFLP,SSRand AFLP markers generated separatelyfrom the first two backcross generations(BC1andBC2).The initialBC1mapcom-prising 888 loci grouped in 37linkagegroups and spanning 4400cM (Lacape et al.,2003),benefitedfromthedevelopmentand integration of new additional micro-satellitemarkers(Nguyenet al.,2004).ThisupdatedsaturatedBC1mapspans5500cMandcomprisesatotalof1160lociorderedalong 26chromosomes or linkage groups(Nguyenet al.,2004).On theotherhand,theBC2mapconstructedusingAFLPandSSRmarkershad514lociintotal.Thetwomapsagreedperfectlyfor lociorder.Theyhad 373 loci in common (between sevenand26perchromosomethroughoutthe26chromosomes), thusallowingtheirmergerintoacombinedconsensusmap.Thecon-sensus framework map comprises 1 306loci and spans 5597cM, with an averagemarkerintervalof4.3cM.

qtl detectionThe QTL analyses, conducted throughcomposite interval mapping, used twomolecular data sets (BC1 and BC2) andthreesetsoffibremeasurements(perplant

table 1range of parental (G.hirsutum[Gh]and G.barbadense[Gb]) values over the five sets of data

Gh Gb BC1 n=75

BC2 n=200

BC2S1

n=200BC3 n=43

BC4 n=37

length (mm)* 27.5–31.8 39.2–43.7 33.8 (27.8–38.2)

28.6 (22.9–35.3)

30.7 (26.9–36.1)

30.0 (25.0–33.9)

31.7 (28.1–37.1)

length uniformity 81.3–85.5 83.9–87.1 85.0 (82.0–88.2)

81.3 (73.3–86.7)

83.3 (80.6–85.1)

81.9 (77.6–86.4)

86.1 (82.5–89.5)

Strength (g/tex) 26.5–32.5 41.4–46.7 35.7 (29.7–41.6)

28.3 (17.8–43.7)

29.0 (23.7–34.5)

24.5 (16.8–32.8)

33.8 (29.7–39.1)

elongation 5.1–6.4 5.5–6.0 6.3 (5.7–7.4)

5.5 (3.9–7.6)

6.3 (5.4–7.4)

5.7 (4.5–7.0)

6.3 (4.9–7.4)

Fineness (mtex)** 207–243 178–191 218 (177–308)

224 (165–379)

225 (176–285)

128*** (117–148)

243 (192–283)

colour reflectance 71.2–77.7 74.6–75.6 74.3 (65.9–81.1)

72.2 (56.8–81.3)

74.1 (69.8–77.5)

71.5 (64.7–76.6)

75.1 (67.1–82.0)

* length is upper half mean length (UHMl), ** standard fineness, *** low fibre fineness values in bc3 generation because of poor maturitiesnote: Mean values and range (in brackets) observed in each bc1–4 generation (number of plants, n, indicated) of fibre technological parameters.


basis for BC1 and BC2 and per-line basiswithtworeplicatesforBC2S1).Thegenera-tions BC1 and BC2 were conducted withno selection, except for choosing thoseplants that produced backcrossed seeds.The fibre measurements, which initiallyincludedeleven traits,were reduced to sixgroupsafterconsideringthestrongcorrela-tionsthatexistedbetweensometraits.Thefibre characteristics that were retained formeasurement included length, length uni-formity, strength, elongation, fineness ormaturity,andcolour.

For the six fibre quality componentsstudied, 50 QTL were identified thatmet permutation-based LOD thresholds(ranging between 3.2 and 4.0 for mostof the traits). Thirty additional suggestiveQTL(havingaLODvaluebelowthresholdbut above 2.5) were also taken into con-sideration after comparing the resultsbetween the threepopulationsorbetweenthe present results and those reported intheliterature(Jianget al.,1998;Kohelet al.,2001;Patersonet al.,2003;Meiet al.,2004).Table2 summarizes the data generatedfromtheQTLanalysesforthesixtraitsofinterest and the phenotypic effects of thedetected QTL. In general, the contribu-tionofeachQTL,measuredasapercentageofexplainedvariationofagiventrait,was

variable and in most cases fairly low. Forexample,fortraitsofeconomicimportance,individualcontributionsvariedfrom4.8to14.8percentinthecaseoffibrelength,4.4to21.3percentforfibrestrengthand4.6to29.1percentforcolourreflectance.

Overall, it was observed that these80QTL partitioned as expected from thephenotypic values of the G. hirsutum andG. barbadenseparents:amajorityofposi-tive alleles for length (12 of the 15 QTL),strength (8 of the 12 QTL) and fineness(13 of the 21 QTL) derived from theG. barbadense parent, while a majority ofpositive alleles for fibre colour (13 of the16QTL) derived from the G. hirsutumparent (Table2). Furthermore, the QTLdetected for the various traits often co-localizedwithinQTL-richregions(Lacapeet al., 2005). In some cases, QTL detec-tion and mapping were in agreementbetween generations (BC1 and BC2) and,very interestingly, in 26 cases (33percentofthe80QTL)theyconfirmedtheresultsreported in the literature, both for theposition of a QTL and for the sign of itsphenotypic effect. The most prominentcasesofQTLconsistentlydetected in thisstudy as well as in those of Paterson et al. (2003) and Kohel et al. (2001), i.e. indifferent crosses/populations, were found

table 2number of qtl for each trait and range of observed phenotypic effects conferred by theG.barbadense alleles (either positive, “Gb+”, or negative, “Gb–”) detected over the three populations (BC1, BC2 and BC2S1)

qtl Gb+

range phenotypic effects qtl Gb–

range phenotypic effects

length (mm)* 12 +0.7 to +2.1 3 –1.6 to –1.8

length uniformity 3 +0.5 to +1.5 3 –1.1 to –3.3

Strength (g/tex) 8 +0.8 to +2.8 4 –0.9 to –3.4

elongation 6 +0.2 to +0.5 4 –0.3 to –0.6

Fineness (mtex)** 13 –10 to –20 8 +9 to +40

colour reflectance 3 +1.8 to +2.5 13 –0.9 to –3.5

total 45 35

* length is upper half mean length (UHMl), ** standard fineness.


along chromosome 3 for QTL for fibrestrengthandfineness,andchromosome23forQTLforfibrestrengthandlength.

The chromosome regions carryingco-localized QTL (corresponding to asingle or to several traits measured on asingle or on several populations) whosepositive alleles derived from the G. bar-badense donor genome, were reduced to19QTL-rich regions that were carriedby 15different “carrier” chromosomes(Table3).Altogether, theconfidence inter-vals(oneLOD)oftheinvolvedQTL-richregionsdelimited a total lengthof636cM(20percent of the carrier genome), or11.5percentof the totalgenome(Table3).Eleven non-carrier chromosomes weredevoidofpositiveQTL,orharbouredneg-ative (positive alleles derived from the G. hirsutumalleles)QTL.

maS in the BC3 and BC4 generations and allelic transmission throughout generationsTheearlyselectionofBC3andBC4plantsusingSSRmarkers that framed the19tar-geted regions of interest made it possibleto choose those plants that showed anallelic constitution with as many intro-gressed loci within the targeted regions aspossible.Intotal,43BC3plantsoutof411(11.4percent)and37BC4plantsoutof450(8.2percent)wereretainedbasedupontheinformation provided by the markers, i.e.without any phenotypic selection at thisstage.Theseplantswerebackcrossedtotherecurrentparent(andself-pollinatedinthecaseoftheBC4plants).

Theallelictransmissionobservedinthefour groups of BC4 derived from fourdifferent BC1 plants is given in Table 4.

table 3identification of the 19 targeted regions mapped on 15 different chromosomes and comprising one or several co–localized fibre quality qtl from G.barbadensefor introgression into a G.hirsutum genetic background

Carrier chromosome Chromosome length (cm)

target interval (cm)

target size (cm)

trait

c14 197 28–57 29 lengthc3 153 32–67 35 length, fineness

90–138 48 length, strength, finenessc4 190 102–118 16 Finenessc22 139 112–139 27 Finenessc5 360 78–101 23 Strengthc6 296 137–144 7 length, finenessc25 183 44–73 29 length, strengthc16 168 65–117 52 Strength, fineness, colourc23 173 45–66 21 Strength (elongation –, colour –)

113–135 22 length, strengthc10 192 0–21 21 Fineness

78–120 42 length, fineness, colourc20 268 88–161 73 elongation, finenessc26 195 67–143 76 length (colour –)a01 233 16–54 38 length

171–209 38 Strengthc18 158 32–46 14 Fineness

a03 271 209–234 25 Strength, uniformitytotal 3176 total 636

note: all targeted Qtl show a positive contribution from the G.barbadenseallele, except for a few negative cases indicated in brackets. the target region is defined as situated between the two loci flanking the Qtl peak loD value at a one loD confidence interval.


Moderate deviations were observed fromtheoretical transmission values (62, 26, 14and 8percent compared with 50, 25, 12.5and6.25percentattheBC1,BC2,BC3andBC4 stages, respectively), with a bias infavour of a higher rate of G. barbadenseallele transmission. This bias was prob-ablyduetotheselectionpressureimposedat least in the BC3 and BC4 generations.Throughout the BC1 and BC2 generationsthat have undergone no deliberate selec-tion, the introgression of G. barbadensealleles (at the heterozygous state) cov-ered the complete genome fairly well, i.e.introgressed segments were found on allof the26chromosomes (not shown).ThisresultcontradictsthefindingsofJianget al.(2000)whodetectedimportantdeficienciesin donor (G. barbadense) allele transmis-sion in a population of 3 662 BC3 plantsoriginatingfrom21BC1plants.

After combining the SSR and AFLPmarkerdata,itwasobservedthattheintro-gression rate differed between target andnon-target regions. When averaged overthe37BC4plants,thepercentageofintro-gressed loci (8percent genome-wide) wasmuch lower in the non-target regions(5percent) than that reachedwithin targetregions(21percent)(Table4).ThedifferentBC4plantsintrogressedbetweenthreeandsix QTL-rich target regions in different

combinations. As an illustration of theselectionpressure applied through theuseof molecular markers, Figure 3 shows thegraphicalgenotypeoftwoBC4individualsas well as that of the BC1 plant (No. 16)fromwhichtheseindividualswerederived.The two BC4 plants had a common BC1ancestorbutoriginatedfromtwodifferentBC2 plants. In this particular example,starting from a common BC1 plant (No.16) which harboured 13 out of 19 pos-sibleQTL-richregions,thetwoBC4plants(Nos.104and419)derivedfromitpartlyorcompletely retained respectively five (c16,c23top,c23bot,c25andA03)andfour(c6,c25,c26andA01bot)genomicregionscar-rying favourablealleles.TheotherregionscarryingQTLonc3,c4,c23,c20,A01andA03, which had been introgressed andwere heterozygous in the BC1 plant, hadreturnedtothehomozygousG. hirsutum/G. hirsutumstate.Thepercentagesofintro-gressedlociintargetandnon-targetregionsinthesetwoexampleswere29and10per-cent,andof29and5percentinthetwoBC4plants(Nos.104and419)respectively.

This example shows that, at least insome cases, the process used was effi-cientinselectingforchromosomalregionsof interest (foreground selection), whileallowing the rest of the genome to returntowardsthatoftherecurrentparent.

table 4percentage of introgressed loci (i%), at the heterozygous state, of 37 BC4 plants that were derived from four different BC1 plants (nos. 3, 11, 16, and 27)

BC1 BC2 BC3 BC4

plant number

number of loci

i% number of

plants

number of loci

i% number of plants

number of loci

i% number of

plants

number of loci

i% global

i% target/non-target

no. 3 646 55 1 479 14 1 467 8 9 456 5 10/4

no. 11 654 64 1 446 28 1 403 13 1 408 10 29/6

no. 16 681 67 2 464 31 3 420 15 21 428 9 25/6

no. 27 668 63 1 471 26 1 433 15 6 439 10 25/6

Mean 62 26 14 8 21/5

note: the number of plants and of loci analysed at each generation are given. at the bc4 generation, the percentage of introgression is also differentiated between target and non-target (as defined in table 3) regions.


fibre characteristics of BC3 and BC4 generation plantsOwing to the limited number of indi-viduals and the unbalanced frequenciesof genotypic classes in the BC3 and BC4material, significant marker-trait associa-tions were less frequent than observedfrom the BC1and BC2 data. For example,markers mapped along five, nine and sixchromosomeregionscontributed(P=0.01),respectively, to length, strength or fine-nessvariationusingBC4marker-traitdata,as compared with 15, 12 and 21 from the

BC1 and BC2 data (Table 2). However,the majority of significant associations,particularly those determined in the BC4generation, were observed within previ-ouslydetectedregions(notshown).Usingfibrestrengthasanexample,outoftheeightstrengthQTL-harbouringregionsonchro-mosomes c3bot, c5, c16, c23sup, c23bot,c25,A01andA03identifiedfromthecom-bined BC1 and BC2 data (Table 3), theBC4dataconfirmedsignificantmarker-traitassociationsinfiveoftheseregions,i.e.formarkers mapped on chromosomes c3bot,

FiGURe 3graphical genotypes (26 chromosomes) of a BC1 plant (no. BC1/16) (upper panel) and of two

selected BC4 plant (nos. BC4/104 and BC4/419) (lower panel) derived from it

BC4 /104 BC4 /419

BC1/16

c3top

c4

c3bot

c6

c25 c16 c23top

c23bot

A03

c26 A01bot

c25 c16 c23top

c23bot

c6

c25

A01top

c26 A01bot

c20

A03

BC /104 BC /419

BC /16

c3top

c4

c3bot

c6

c25 c16 c23top

c23bot

A03

c26 A01bot

c25 c16 c23top

c23bot

c6

c25

A01top

c26 A01bot

c20

A03

the two possible allelic forms, homozygous Gh/Gh and heterozygous Gh/Gb, are denoted in dark grey and black respectively. Regions in black are introgressed with G. barbadense alleles. light grey areas indicate portions of unknown allelic composition. boxed areas represent the localization of Qtl-rich regions localized on 15 carrier chromosomes shown to the left (11 non-carrier chromosomes are shown to the right). arrows indicate the regions totally or partially introgressed.


c16, c23bot, c25 and A03. Furthermore,it is worth noting that the BC4plant No.104presentedinFigure3,whichhadintro-gressedallthesefiveregions,alsodisplayedthehighestfibrestrengthvalueofitsgener-ation(39.1g/tex,comparedwith33.1g/texfortheGuazuncho2parent).Theconcom-itantintrogressionofG. barbadenseallelesdisplayingpositivemarker-traitassociationsforotherfibrepropertiessuchaslengthorfinenesswasalsoobserved.Thistranslatedinto the development of different highlyvaluable BC progenies. These preliminaryresults suggest that the improvement ofG. hirsutum fibre properties through theintrogressionofG. barbadense fibreQTLappearsfeasible.

diSCuSSionIn an attempt to overcome the limitationsof conventional breeding for improvingcotton fibre quality through the use ofinterspecific hybridization, molecularmarkers were used in a MAS scheme toimprove the efficiency of introgressingfibre quality traits. The advanced back-cross-QTL (AB-QTL) strategy (TanksleyandNelson,1996)wasusedasthisallowedconcomitantdevelopmentofageneticmapofthecottongenomeandanalysisoffibrequality QTL, and attempts to introgressfavourable alleles in an adequate recipientgeneticbackground(Figure1).

In contrast to monogenic characteris-tics such as disease and insect resistance,many important traits including yield andquality show continuous phenotypic vari-ation and are governed by a number ofQTL. Cotton fibre quality is a complexconceptthatinvolvesanumberoftraitsorcharacteristics. Each of these is under theinfluence of numerous QTL, indicatinga complex genetic determinism. Indeed,from the present results, at least six QTL

governfibreuniformityandupto21QTLinfluence fibre fineness. When consid-ering six traits that can account for fibrequality, a total of 80 QTL were detected(Table2).Thisfigurefallswithinthesamerange as that found by Paterson et al.(2003).AssomeoftheseQTLco-localizedwithin the same chromosome region, bychoosing thoseQTLwhosepositive allelederivedfromthedonorparentandhadthestrongesteffectoneconomicallyimportantfibre characteristics, the number of targetregions to be introgressed was reduced to19 (Table3).Nevertheless, thisnumberofQTLremainstoohighto identifyasingleplantthatwouldcarrythemall.Indeed,inthe authors’ experience, at the BC3 stage,single plants carried a maximum of fiveregions of interest (eight if consideringregions only partially introgressed), whileattheBC4stage,thisnumberwasreducedto four (seven if considering regions onlypartiallyintrogressed).

At this stage of the MAS process, tworoutes areunderway (Figure1).The firstinvolves identifying the best BC4 plants,i.e. those showing the highest amount offavourable QTL introgression, and thenfixingthefavourableallelebyself-pollina-tion.SuchBC4S1plantshavebeencrossedwithotherBC4S1plantsofdifferentascentinordertopyramidasmanyQTLaspos-sible (each contributing todifferent traits)within the same genome. Similarly, BC4S1plantswereusedtopyramidvariousQTLresponsible for a given trait (“selectivepyramiding”). This latter strategy couldespecially apply to traits of commercialimportance,suchas fibrestrengthor fine-ness.ThesecondavenueinvolvesrepeatingthebackcrossingprocessuntilnearisogeniclinesdifferingonlyatagivenQTL(QTL-NILs) are developed. Such plant materialcould prove useful not only to study the


effect of a single given QTL on the phe-notypicvalueofaplantharbouringit,butalsoincasetheintrogressedQTLisproventocontributesignificantlytotheimprove-mentofagiventrait(Bernacchiet al.,1998).Also, QTL-NILs could be used as donormaterial for QTL pyramiding (PelemanandvanderVoort,2003).Finally,anintro-gression library, i.e. a collection of NILs,willtypicallyserveasprimaryplantmate-

rial for QTL fine mapping and eventualQTLcloning(SalviandTuberosa,2005).

Howeversuccessfulmarker-aidedintro-gression of genomic regions of interestmaybe,onlyphenotypicanalysisofplantmaterialstemmingfromtheMASprocess,includingtheassessmentofitsadaptabilitytoanygivensetoflocalagronomicandeco-logical conditions,will allowvalidationofthisprocedure.

referenCeSBasten, C., Weir, B. & Beng, Z-B. 1999.QTL cartographer, version 1.13.Dept.ofStatistics,North

CarolinaStateUniv.,Raleigh,NC,USA.Bernacchi, D., Beck-Bunn, T., Emmatty, D., Eshed, Y., Inai, S., Petiard, V., Sayama, H., Uhlig, J.,

Zamir, D. & Tanksley, S.D. 1998.AdvancedbackcrossQTLanalysisintomato.II.Evaluationofnearisogeniclinescarryingsingle-donorintrogressionsfordesirablewildQTL-allelesderivesfromLycopersicon hirsutumandL. pimpinellifolium.Theor. Appl. Genet.97:170–180.

Brubaker, C.L. & Wendel, J.F.2001.RFLPdiversityincotton.InJ.N.Jenkins&S.Saha,eds.Genetic improvement of cotton, emerging technologies,pp81-101.Enfield,NH,USA,SciencePublisherInc.

Dekkers, J.C.M. & Hospital, F.2002.Theuseofmoleculargeneticsintheimprovementofagricul-turalpopulations.Nature Genet.3:22–32.

Jiang, C., Chee, P., Draye, X., Morrell, P., Smith, C.W. & Paterson, A.H.2000.Multilocusinter-actions restrict gene introgression in interspecific populations of polyploid Gossypium (cotton).Evolution54:798–814.

Jiang, C., Wright, R., El-Zik, K.M. & Paterson, A.H. 1998. Polyploid formation created uniquevenuesforresponsetoselectioninGossypium(cotton).Proc. Nat. Acad. Sc. USA95:4419–4424.

Kohel, R.J., Yu, J., Park, Y.-H. & Lazo, G.2001.Molecularmappingandcharacterizationoftraitscontrollingfiberqualityincotton.Euphytica121:163–172

Lacape, J.-M., Nguyen, T.-B., Thibivilliers, S., Courtois, B., Bojinov, B.M., Cantrell, R.G., Burr, B. & Hau, B. 2003.AcombinedRFLP-SSR-AFLPmapoftetraploidcottonbasedonaGossypium hirsutumxGossypium barbadense backcrosspopulation.Genome46:612–626.

Lacape, J.-M., Nguyen, T.-B., Courtois, B., Belot, J.-L., Giband, M., Gourlot, J.-P., Gawryziak, G., Roques, S. & Hau, B.2005.QTLanalysisofcottonfiberqualityusingmultipleG. hirsutumxG. barbadense backcrossgenerations.Crop Sci.45:123–140.

Lander, E., Green, P., Abrahamson, J., Barlow, A., Daly, M.J., Lincoln, S. & Newburg, L.1987.MapMaker: an interactive computer package for constructing primary linkage maps of experi-mentalandnaturalpopulations.Genomics1:174–181.

Mei, M., Syed, N.H., Gao, W., Thaxton, P.M., Smith, C.W., Stelly, D.M. & Chen, Z.J. 2004.GeneticmappingandQTLanalysisoffiber-relatedtraitsincotton(Gossypium).Theor. Appl. Genet. 108:280–291.

Nguyen, T.-B., Giband, M., Brottier, P., Risterucci, A.-M. & Lacape, J.-M.2004.Widecoverageoftetraploidcottongenomeusingnewlydevelopedmicrosatellitemarkers.Theor. Appl. Genet.109:167–175.


Paterson, A.H., Saranga, Y., Menz, M., Jiang, C. & Wright, R.J.2003.QTLanalysisofgenotypexenvironmentinteractionsaffectingcottonfiberquality.Theor. Appl. Genet.106:384–396.

Peleman, J.D. & van der Voort, J.R.2003.Breedingbydesign.Trends Plant Sci.7:330–334.Rong, J., Abbey, C., Bowers, J.E., Brubaker, C.L., Chang, C., Chee, P., Delmonte, T.A., Ding, X.,

Garza, J.J., Marler, B.S., Park, C.-H., Pierce, G.J., Rainey, K.M., Rastogi, V.K., Schulze, S.R., Trolinder, N., Wendel, J.F., Wilkins, T.A., Williams-Coplin, T.D., Wing, R.A., Wright, R.J., Zhao, X., Zhu, L. & Paterson, A.H.2004.A3347locusgeneticrecombinationmapofsequencetaggedsitesrevealsfeaturesofgenomeorganization,transmissionandevolutionofcotton(Gossypium).Genetics166:389–417.

Salvi, S. & Tuberosa, R.2005.TocloneornottocloneplantQTLs:presentandfuturechallenges.Trends in Plant Sci.10:297–304.

Schön, C., Utz, H.F., Groh, S., Truberg, B., Openshaw, S. & Melchinger, A.E.2004.Quantitativetraitlocusmappingbasedonresamplinginavastmaizetestcrossexperimentanditsrelevancetoquantitativegeneticsforcomplextraits.Genetics 167:485–498.

Tanksley, S.D. & Nelson, J.C.1996.AdvancedbackcrossQTLanalysis:amethodforthesimulta-neousdiscoveryandtransferofvaluableQTLsfromunadaptedgermplasmintoelitebreedinglines.Theor. Appl. Genet.92:191–203.

Wilkins, T.A. & Arpat, A. 2005. The cotton fiber transcriptome. Physiologia plantarum 124:295–300.

Documents

introduction to marker-assisted selection4 Marker-assisted selection – Current status and future perspectives in crops, livestock, forestry and fish Summary This chapter provides