expectedwww.dajiba.com

RESEARCHARTICLEOpenAccessGenome-wideassociationmulti-locusandmulti-variatelinearmixedmodelsrevealtwolinkedlociwithmajoreffectsonpartialresistanceofapricottobacterialcankerMariemOmrani1,2,3,MorganeRoth1,5,GuillaumeRoch1,4,AlainBlanc1,CindyE.
Morris2andJean-MarcAudergon1*AbstractBackground:DiseasescausedbyPseudomonassyringae(Ps)arerecognizedasthemostdamagingfactorsinfruittreeswithasignificanteconomicandsanitaryimpactoncrops.
Amongthem,bacterialcankerofapricotisexceedinglydifficulttocontrolduetoalackofefficientprophylacticmeasures.
Severalsourcesofpartialresistancehavebeenidentifiedamonggeneticresourcesbuttheunderlyinggeneticpatternhasnotbeenelucidatedthusfar.
Inthisstudy,wephenotypedbacterialcankersusceptibilityinanapricotcore-collectionof73accessionsover4yearsbymeasuringcankerandsuperficialbrowninglengthsissuedfromartificialinoculationsintheorchard.
Inordertoinvestigatethegeneticarchitectureofpartialresistance,weperformedagenome-wideassociationstudyusingbestlinearunbiasedpredictorsongenetic(G)andgeneticxyear(G*Y)interactioneffectsextractedfromlinearmixedmodels.
Usingasetof63,236single-nucleotidepolymorphismmarkersgenotypedinthegermplasmoverthewholegenome,multi-locusandmulti-variatemixedmodelsaimedatmappingtheresistancewhilecontrollingforrelatednessbetweenindividuals.
Results:Wedetected11significantassociationsover7candidatelocilinkedtodiseaseresistanceunderthetwomostsevereyears.
ColocalizationsbetweenGandG*Ytermsindicatedamodulationonalleliceffectdependingonenvironmentalconditions.
Amongthecandidateloci,twolocionchromosomes5and6hadahighimpactonbothcankerlengthandsuperficialbrowning,explaining41and26%ofthetotalphenotypicvariance,respectively.
Wefoundunexpectedlong-rangelinkagedisequilibrium(LD)betweenthesetwomarkersrevealinganinter-chromosomalLDblocklinkingthetwounderlyinggenes.
Thisresultsupportsthehypothesisofaco-adaptationeffectduetoselectionthroughpopulationdemography.
Candidategenesannotationssuggestafunctionalpathwayinvolvingabscisicacid,ahormonemainlyknownformediatingabioticstressresponsesbutalsoreportedasapotentialfactorinplant-pathogeninteractions.
Conclusions:OurstudycontributedtothefirstdetailedcharacterizationofthegeneticdeterminantsofpartialresistancetobacterialcankerinaRosaceaespecies.
Itprovidedtoolsforfruittreebreedingbyidentifyingprogenitorswithfavorablehaplotypesandbyprovidingmajor-effectmarkersforamarker-assistedselectionstrategy.
Keywords:GWAS,Prunus,Partialresistance,Bacterialcanker,Pseudomonassyringae,Candidategenes,Apricot,Multi-locusmixedmodel,Multi-variatemixedmodel,Linkagedisequilibrium*Correspondence:jean-marc.
audergon@inra.
fr1INRA,UR1052GénétiqueetAméliorationdesFruitsetLégumes,CentredeRecherchePACA,Montfavet,FranceFulllistofauthorinformationisavailableattheendofthearticleTheAuthor(s).
2019OpenAccessThisarticleisdistributedunderthetermsoftheCreativeCommonsAttribution4.
0InternationalLicense(http://creativecommons.
org/licenses/by/4.
0/),whichpermitsunrestricteduse,distribution,andreproductioninanymedium,providedyougiveappropriatecredittotheoriginalauthor(s)andthesource,providealinktotheCreativeCommonslicense,andindicateifchangesweremade.
TheCreativeCommonsPublicDomainDedicationwaiver(http://creativecommons.
org/publicdomain/zero/1.
0/)appliestothedatamadeavailableinthisarticle,unlessotherwisestated.
Omranietal.
BMCPlantBiology(2019)19:31https://doi.
org/10.
1186/s12870-019-1631-3BackgroundFordecades,breedingprogramshavenotablyfocusedontheintrogressionofmajorresistantgenesintocropcul-tivars.
Howevermultipleepisodesofresistancebreak-downhaveledresearcheffortstotargetpolygenic(quantitativeorpartial)resistances[1].
Althoughbreed-ingstrategiesrelyingonpyramidingquantitativetraitloci(QTLs)ontopofmonogenicgenesmightbecostlyandtime-consuming,thelong-termefficiencyandsus-tainabilityofQTLshavebeenassumediftheyarede-ployedparsimoniouslyinbothspaceandtime[2,3].
Resistancesustainabilityhasaparticularinterestforper-ennialwoodycropsconsideringtheirlonggenerationtimeandlife-spaninorchards,andoverallthelong-termcommitmenttotheirbreedingschemes.
WithinthePrunusgenus,apricot(PrunusarmeniacaL.
)isoneofthemostpopularandtypicalcropsoftheMediterraneanBasinwhichprovides49%ofthetotalproductiontotheworldmarket[4].
Amongthebioticstressesaffectingapricotcropdurabilityandinabroaderwaystonefruitspecies,bacterialcanker,causedbyubi-quitousbacteriainthespeciescomplexPseudomonassyringae(Ps),isoneofthemostdamaging.
Thisdiseasecouldpotentiallyleadtothedeathoftreesintheor-chard,especiallyyoungtreeswithintheirfivefirstyearsafterplanting.
Threemaingroups-Pspv.
syringae,Psviridiflava[5]andPspv.
morsprunorum[6],associatedwithphylogroups2,7and3,respectively[7],havebeenhistoricallyassociatedwiththediseaseonapricot.
Thesymptomsaffectaerialorgans,resultinginlesionsandshotholesinleaves,budandblossomdieback,andap-icalnecrosis,potentiallyleadingtomoreseveresymp-tomswhenbacteriagettothevesselsandspreadthroughthevascularsystem[8,9].
Whentheinfectionissystemic,cankerscorrespondingtonecrosisandflatten-ingofexternaltissueslinkedtoadissymmetricgrowthofthecambiuminthespring,arevisibleonbranchesandscaffoldlimbs[10].
Severalabioticfactorsrelyingonsoilandclimaticconditionscanfavorbacterialcankerseverity.
Amongthem,coldwintertemperaturesandes-peciallyhighfrequencyoffrost-defrostepisodeshavebeenhighlightedasmajorfactorsfavoringbacterialcan-ker[11,12].
Integratedmanagementpracticessuchastheuseofresistantandsoil-adaptedrootstockmaterialandgraftingatatallheightcouldlowerbacterialcankerincidenceinorchards[13,14].
Theseculturalrecom-mendationsarenevertheless,technicallychallengingfororchardmanagementandprovideonlyapartialprotec-tioninorchards.
Therefore,thedevelopmentofpartiallyresistantculti-varsseemstobeapromisingmeasureinadditiontopre-ventivepracticestoassureorcharddurability.
Inthiscontext,researcheffortshavefocusedonscreeningapricotgeneticresourcesbothundernaturalinfectionsorcontrolledinoculations.
Severalgeneticbackgroundswithpartialresistancetobacterialcankerhavebeenidentified:'Bakour'[14],'HtifColomer','Luizet','Palsteyn'[15]and'Orangered'[16].
Moreglobally,similarresearchinitiativeshavebeenconductedtophenotypicallycharacterizediffer-entialsusceptibilitiestobacterialcankerinPrunusroot-stocks[17],sweetcherry[17–19]andplum[20],buttodatethequestionofthegeneticdeterminantsofpartialre-sistancetobacterialcankerfromthenaturaldiversityinRosaceaefruittreeshasneverbeeninvestigated.
Recentprogressofhigh-throughput-sequencingtech-nologiesfosteringthediscoveryofthousandsandevenmillionsofsinglenucleotidepolymorphisms(SNPs)overwholegenomeshasopenedupnewopportunitiestoad-dresstheunderlyingarchitectureofquantitativetraits[21].
Genome-WideAssociationStudies(GWAS)havebeenextensivelyandsuccessfullyusedinplantbreedingthankstotheadvancementofthesenovelgenomics-basedapproaches[22].
GWASenablemappingofQTLsandgenesaffectingtraitvariationinawidecollectionofmostlyunrelatedindividualsusuallysampledfromwildrelativepopulations,breedingcultivarsandlandraces.
Thisisinstarkcontrasttotraditionallinkagemappingmethodsthatrequiretheestablishmentofsegregatingpopulationsbeforehand[23,24].
Associationstudiesbenefitfromthenumerousrecombinationevents,whichhaveoccurredthroughspeciesdemographichistory,andrelyonlinkagedisequilibrium(LD)causedinpartbyselectionandpopu-lationstructure,toexploitnaturalallelicdiversityandidentifylinksbetweenmarkersandcausallociunderlyingthetraitofinterest[25–27].
GWASturnsouttobeaparticularlysuitableapproachforplantsandespeciallyforperennialcropssincecostsas-sociatedwithmakingandmaintaininglargeprogeniesintheorchardcanbespared.
Therefore,theuseofGWAShasgraduallybecomemorewidespreadfordeterminingthegeneticbasisofvariationincomplextraitsinRosaceaefruitspecies[28–30].
Moreparticularly,theapricotgen-omedisplaysmanyadvantagesforGWASapplicationsdueto:(i)itsdiploidy(2n=16)andsmallsize(294Mb/n)[31],(ii)itshighlevelofheterozygosityresultingfromageneraloutcrossingmatingsystem[32]and(iii)itshighnucleotidediversityrelatedbothtotheearlyseedpropagation[33]andgeographicallybroaddistributionofthegermplasmwithadiversificationfromCentralAsia[34,35].
Moreover,theapricotgenomehasbeencharacterizedwithaveryfastLDdecaywithin100basepairs(bp)inalargegeneticdi-versitypanel[30],allowingassociationmappingwithaverypreciseresolutionconditionallyupontheuseofahighdensityofmarkers[36].
Inthepresentstudy,weinvestigatetheunderlyinggeneticarchitectureofpartialresistancetobacterialcan-kerinapricotusinganassociationapproach.
Thespe-cificobjectivesofthisresearchareto(i)identifytheOmranietal.
BMCPlantBiology(2019)19:31Page2of18geneticandenvironmentalcomponentsofbacterialcan-kersusceptibilityusingacore-collectionof73apricotaccessionsthathasbeenphenotypedintheorchardovera4-yearcampaign,(ii)developanappropriategenome-wideassociationmethodologytakingintoaccountthespeciesgeneticarchitectureandthestratificationofourpopulationset,and(iii)identifycandidategenesandde-cipherthemolecularbasisofpolygenicresistancetobacterialcanker.
ResultsPhenotypicvariancedecompositionandheritabilityofpartialresistancetobacterialcankerConsideringdifferencesbetweenallcontrolsandinocu-latedshootsregardlessofthegenotypeeffect,ahighlysignificanteffectofbacterialinoculationwasobserved(Wilcoxonunilateralt.
testp0.
05).
IndividualeffectsofLG6_15273858,LG5_5394803andallotherdetectedlociontheirrespectiveGorG*YBLUPphenotypesareshowninAdditionalfile7.
Interestingly,arelativelyhighLDsquarecorrelationvalue(r2vs=9.
76E-02),higherthanthe99thpercentileofr2vsinter-chromosomaldistribution,wasrevealedmorespecificallybetweenthetwomaincandidatesLG6_15273858andLG5_5394803(Additionalfile8).
ThiscorrelationbetweenassociatedlociprovedtobewithinalongerLDblock(Fig.
2)revealinganinter-chromosomallinkagebetweentheunderlyingcan-didategenespp000004m.
g(chromosome5)andpp018341m.
g(chromosome6).
Likewise,thecorrelationbetweenLG5_5394803andLG5_4842835(r2vs=2.
94E-01)detectedbothonchromosome5appearedtobesignifi-cantlyhigherthanthe99thpercentileofr2vsintra-chromosomaldistribution(Additionalfile8).
ThegenotypegroupdistributionfromalldetectedSNPsfromthemulti-yearmodelBLUPswastestedonthewithin-yearmodelBLUPswithheterogeneoussignificancelevelsaccordingtotheyearandthemarkerPVE.
MarkerLG7_18047191wasonlysignifi-cantforbs2013andbs2015.
Significanceofthetwomainlociwasconfirmedindividuallyforlgc2013,lgc2015,lgc2016andbs2015withLG6_15273858andforbs2013andbs2015consideringLG5_5394803(uniandmulti-factorialANOVAandpost-hoctestsinAdditionalfile9).
Strict-senseheritabilityestimatesrangedfrom10.
85%(lgc2014)to64.
46%(bs2015)withnon-negligiblevalueseveninthenon-favorableyears2014and2016.
Onlybs2016ledtonon-significantre-sultsforalldetectedmarkers(Additionalfile9).
HaplotypesassociatedwithpartialresistanceWefoundthreegroupsofhaplotypesamongthecore-collectionhavingseveralSNPsassociatedtocankerresistance;18,45and9haplotypesshared3,5and2markersassociatedtolgcG,bsGandbs(G*Y)2015BLUPs,respectively(Additionalfile10).
MeancomparisonwasperformedonBLUPvaluesfrommulti-yearandwithin-yearmodelsafterdiscardinglowfrequencyclassesinordertolimitfalse-positivefindings(typeIerror).
Fivehaplotypeswereidentifiedwithfavorableallelesandstabil-ityoftheresistancethroughyears(1withlgcG,3withbsG,and1withbs(G*Y)2015BLUP,seeAdditionalfile10).
AccessionsA1814andHarlayneshowedtwoandallofthethreedefinedfavorablehaplotypes,respectively.
Fig.
3GandG*YBLUPdistributionsaccordingtothehaplotypeconstitutedbyLG6_15273858andLG5_5394803.
Statisticaldifferencesbetweengroupsareindicated–HSDpost-hoctestsoneachG&G*Yterm.
ForbsGand(G*Y)2015BLUPs,testswereperformedonresidualsfromanANOVAmodelaccountingtheotherassociatedlocifrommulti-locusmodel(LG3_10897844andLG52_22535054forbsGLG3_4322444andLG5_4842835forbs(G*Y)2015).
Y-axisvaluesareindicatedinln(x+1)scale.
Haplotypeclass'AGAA'(1individual)wasdiscardedintheboxplotOmranietal.
BMCPlantBiology(2019)19:31Page7of18A0008andA1314wereinthesamewaynoticeablefortheirpartialresistance(seeAdditionalfile2C)buttheircorrespondinghaplotypeswereeithernotlocatedinthemostresistantgroupsor,iftheywere,theyappearedtobeunstablebetweenyears(seeAdditionalfile10).
Themeancomparisontestwasnon-significantforthewithin-yearmodellgc2015BLUPforthefirstgroupofhaplotypes(relatedtothe3candidatelociassociatedtolgcGBLUP)whichmightbeduetoalossofpowerwhenhaplotypenumber(18)ishighcomparedtopopu-lationsize(ANOVAtestp=0.
1322).
Noneofthecom-parisontestsbetweenhaplotypesweresignificantforyears2014and2016,whichmightbeexplainedbythelowgeneticvarianceintheseyearswheretherewaslowdiseaseseverity.
CandidategenesforresistancetobacterialcankerPutativecandidategeneannotationsforallmulti-locusandmvLMMassociatedSNPswerereportedinTable1.
Todate,noneoftheassociatedSNPswerereportedfordiseaseresistanceinaRosaceaespeciesorcolocalizedwithpreviouslyknownQTL.
Itisnoteworthythat,overallthe7detectedloci,only2candidatesLG3_10897844andLG3_4322444targetedpolymorphismsincodingse-quences(CDS)ofppa017602m.
gandppa008956m.
g,re-spectively.
Mutationswerebothnon-synonymouswithonesubstitution(K49Minppa017602m.
g)leadingtoadeleteriouspredictiveeffectonthecorrespondingprotein,aglutamyl-tRNA(Gln)amidotransferasesubunitCpro-tein.
AllotherlocicomprisingthetwomostsignificantlociLG6_15273858(ppa018341m.
g)andLG5_5394803(ppa000004m.
g)werelocalizedinintrons.
TheassociatedfunctionalannotationsofthetwomaincandidategeneswererelatedtoHarbingertransposase-derivednucleaseandPleckstrinhomology-like(PH)domains,correspondingtoasuperfamilyoftransposonsfoundinplantsandanimals[57]andaconserveddomainparticularlyabundantinproteinsinvolvedinsignaltransductionpathways[58],respectively.
DiscussionWeaimedatinvestigatingthegeneticpatternofpartialresistanceofapricottobacterialcankercausedbyP.
syr-ingaeusingabroadgeneticbasisinthespecies.
Weshowedthat,althoughbeingpolygenic,bacterialcankerpartialresistanceappearedtorestuponalimitednum-berofadditivecomponents:5SNPsonchromosomes2,3,5,6and7partiallyexplaining45and62%ofthetotalphenotypicvarianceofcanker(lgc)andsuperficialbrowning(bs)lengths,respectively.
Alargepartofthegenotypicvariancewasduetoawintereffectallowingthedetectionoftwoadditionallocionchromosomes3and5intheyearmostfavorablefortheexpressionofcankersymptoms.
Amongthe7candidateloci,twomainSNPsonchromosome5and6displayedadditivecontri-butionsreaching41and26%ofthevariationoflgcandbs.
Furthermore,thesetwomarkersshowedalong-rangeinter-chromosomalLDrevealingamulti-locus-linkedse-lectioneventthroughpopulationhistory.
DuetotheshortdecayofLDinapricot,candidatelociwereidenti-fiedwithahighresolutionleadingtotheidentificationofpromisingcandidategenesinvolvingapotentialsig-nalingcross-talkbetweenabscisic(ABA),jasmonic(JA),andsalicylic(SA)acids,moleculesknowntomediatelong-distancesignalinginplant-pathogeninteractions.
Overall,thisstudycontributedtotheveryfirstcharacterizationofthegeneticvariationandgenomicdeterminantsofpartialresistancetobacterialcankerinafruittreespecies.
DetectionofgenomicregionscontrollingvariationofpartialresistancetobacterialcankerisdependentonwinterfrostintensityAshighlightedbytheheterogeneityofgeneticvariancesacross4years,geneticvariationwashighlydependentonwinter-frostseveritywithlowgeneticvariationsin2014and2016whilethehighestwereregisteredin2013and2015.
Consideringthisenvironmentaldependency,thescreeningandcharacterizationofcultivarsforbacterialcankerresistancewasadifficulttask.
Otherlimitsofbac-terialcankerphenotypingconcerntime-consumingeffortsintheorchard,averylongperiodofsymptomdevelop-ment,theneedtocalibratetheobservationsbetweenop-eratorsandnoguaranteeofgettingsignificantgeneticvariationbetweenindividuals.
Marker-assistedselection(MAS)programswouldbeinthatcaseanattractive,powerfulandcost-efficientalternativetophenotyping.
Amongthetwomostsevereyears,broad-senseherit-abilityestimatesrangedfrom35%(lgcin2013)to78%(bsin2015)showingaconsiderablelevelofvariationofsusceptibilityinthecore-collection.
Previousstudiesonotherpatho-systemsinfruitandforesttreeshavere-portedmoderatetohighheritabilityestimatesrangingfrom30to40%forpitchcankerresistanceinloblollypine[59],from70to89%inpeachxPrunusdavidiana[60]and60to87%inapple[61]forpowderymildewre-sistance.
Importantly,thesestudiesreportedthecrucialneedforcontrollingGxEinteractionsinordertoesti-mateaccurategeneticeffectsinthefaceofenvironmen-talnoiseinpluri-annualdata.
Thetwophenotypeslgcandbsdisplayedahighcorrel-ation(Pearsoncorrelationr=0.
57±0.
10,p=1.
48E-07***).
Regardingtheimportantcontributionofthegeneticfactortothetotalvarianceofthephenotypes,thiscorrelationsuggeststhatacommongeneticpatternmaycontrolbothphenotypes.
Weidentifiedatotalof11associations,over7candi-dateSNPsonchromosomes2,3,5,6and7,linkedtoOmranietal.
BMCPlantBiology(2019)19:31Page8of18thevarianceoflgcandbsGandG*YBLUPs.
Amongthestableassociations,twomainSNPsweredetectedonchromosomes5(LG5_5394803)and6(LG6_15273858)andexplained41and26%ofthevariationoflgcandbs,respectively.
Inthecaseoflgc,auniquelocus(LG6_15273858)wasdetectedbothforG*YinteractionBLUPsandtheoverallGtermwithamodulationofthealleliceffectandthePVEaccordingtotheyearduetotheheterogeneityofwintertemperatures.
Moreover,newassociationsforbsweredetectedin2015whichwasthemostsevereyearintermsofdiseaseexpression,suggest-ingthatdataonspecificyearsandclimaticconditionshavethepotentialtorevealadditionalregionslinkedtodiseaseresistance.
SimilarresultshavebeenobtainedinthecaseofQTLmappingontreearchitecture[62]andphenologicaltraits[63]inappleunderlyingtheuseful-nessoffocusingonGandG*Ytermstotargetbothstableandenvironment-specificgeneticdeterminants.
CandidategenescanbefoundbyexploitingthequickLDdecayinapricotMappingresolutioninGWASmostlydependsonthedecayofLD:themoretheLDdecreases,thesharpertheresolutionappearsaroundthedetectedlocus[64].
Con-sideringtheoriginalreproductivecharacteristicsofthefamily,mostRosaceaespeciesarecross-pollinatingbe-causeoftheirself-incompatiblysystem.
ThisresultsinahighnumberofeffectiverecombinationeventsandarapidexpectedLDdecayoverthegenome[65].
Inourstudy,globalmeanLDoverallchromosomesdecayedoververyshortdistances(100to200bp)consideringbothcorrectionsforpopulationstratificationanduncor-rectedestimates.
Moreover,LDinthecaseofourpopu-lationisnotlikelytorelyonconfoundingstructureeffectsrelatedtodifferencesinallelefrequenciesbe-tweengroupsresultingfromnon-randommating[66].
Theseresultsarecongruentwiththosereportedinthepreviousassociationstudyonapricot[30]basedonapopulationcomprisingthemajorpartofthematerialusedinourwork.
Regardingalltheseobservations,GWASisaverysuit-ableapproachtomapthegeneticregionslinkedtobac-terialcankerpartialresistanceinourdiversitypanelwithaverypreciseresolution.
Todate,noneoftheputativecandidategenesordo-mainsassociatedwiththeSNPsidentifiedherehavebeenreportedforplantresistancefunctionintheRosaceaefam-ily.
ThetwomainSNPsLG6_15273858andLG5_5394803werelocatedingeneswithsignificanthomologiestoHar-bingertransposase-derivednuclease(ppa018341m.
g)andPleckstrinHomologyPH(ppa000004m.
g)domains,re-spectively.
Thesemotifsareconserveddomainsfoundinawiderangeofuncharacterizedproteinswithfunctionsrespectivelyin,chromatinremodeling[67]andsignaltransductionthroughinteractionwithmembranephos-phoinositides[68,69].
Aroleforphosphoinositides(bind-ingtopleckstrinhomologydomain-ppa000004m.
g)inregulatingplantnuclearfunctionsandpossiblytranscrip-tionalactivitiesthroughchromatinremodeling(Harbingertransposase-derivednuclease-ppa018341m.
g)wasre-vealedasacommonresponsetoalargerangeofbothabi-oticandbioticstresses[70].
Itisnoteworthythattheproteinencodedbythecandidategeneppa023961m.
glo-catedonchromosome2(SNP:LG2_22535054)belongstoalargefamilyofproteases:thesubtilisin-likeproteases(orsubtilases)whoseinvolvementinplantdefenseresponseshasbeenmoreandmoredescribedrecently[71].
TwocloselyrelatedmembersoftheP69subtilasefamily(P69BandP69C)wereshowntobetranscriptionallyactivatedafterPsDC3000infectionwithanelicitationfromSAandJA[72,73].
AlthoughthelinkbetweenSBT4.
6encodedbyppa023961m.
gandplantresistanceremainstobeinvesti-gated,transcriptsfromacloserelativeSBT4.
14(alsoknownasAtXSP1)wereevidencedascontributingtoxylemdifferentiationinA.
thaliana[74].
Moreover,SBT4.
6subtilasewaspredictedtobeintheextracellularspace(Uniprotdatabase)suggestingaroleinrecognitionofPsbeforeentryandcolonizationthroughthexylem,inthecaseofacompatibleinteraction.
Interestingly,theGproteinβWD-40repeat,oneofthedomainsofthepro-teinencodedbyanothercandidategene(ppa008956m.
gforbs(G*Y)2015association),hadbeenpreviouslyshowntobindin-vitrowithPHdomains(ppa000004m.
g)[68].
IthasbeenshowninA.
thalianathatmyo-inositolpoly-phosphate5-phosphatases,alargefamilyenglobingIP5P2(ppa020388m.
gincludingdetectedlociLG5_4842835),hydrolyzeawiderangeofphosphoinositidephosphatesubstratesandisinvolvedinstressresponsesthroughtheabscisicacid(ABA)signalingpathway[75].
Inaddition,re-searchthatdeployedtranscriptionalapproachesdemon-stratedthattheABApathwaywasoneofthemaintargetsofeffectorssecretedbyPs[76].
Regardingalltheirfunc-tions,mostofthecandidategenesordomainsseemtosupportaroleinaregulatorynetworkinvolvingsignaltransductionthroughmembranephosphoinositidesinacrosstalkinvolvingABA,withSAandJA,potentially.
ABAmostlyactsasnegativeregulatorofdiseaseresist-ancewithanantagonisticeffectonSAandJA[77,78].
Inaddition,overallthesevencandidateloci,onlytwosinglepolymorphismstargetedCDSwithnon-synonym-oussubstitutions.
Thisobservationemphasizestheim-portanceofintronswhichcanaffectgeneexpressionlevelorinducealternatesplicingwithanimpactonthephenotype.
Moreover,thegenotypedataweusedintheassociationmappingwasrestrictedtovariantslocatedingene-spaceregions.
Inturn,wemayhavemissedinter-genicallelicdiversityandregulatoryvariantspo-tentiallycontributingtophenotypevariation.
TheroleofOmranietal.
BMCPlantBiology(2019)19:31Page9of18non-codingDNAsuchaspromotersor/andenhancersinthevariabilityofsusceptibilitytoplantdiseaseresist-ancehasbeendemonstratedinseveralpatho-systemsin-cludingrice/Xanthomonasoryzaepv.
oryzae[79]andmaize/maizeroughdwarfvirus[80]andwouldbeworthfurtherinvestigationinthecaseofbacterialcankerpar-tialresistance.
TheseresultsopenthedoorforsubsequentvalidationstoconfirmthecandidatelocipolymorphismsoverapoolofgeneticresourcesandthecorrelationsbetweentheSNPgenotypeandthesusceptibilityleveltothedisease.
Toachievethisgoal,pyro-sequencingcouldbeperformedinordertoconfirmthehaplotypes.
Moreover,inordertotakeintoaccountthesignificanceofnon-codingpoly-morphismcandidatesfromourstudy,validationofthecandidategenescouldbeperformedwithqRT-PCRacrossapooloffavorableandunfavorablehaplotypesidentifiedinourstudy.
Combininggenome-widemulti-locusandmulti-variateassociationmodelsisapowerfulmethodtodecipherthegenomicpatternofpartialresistancetobacterialcankerTherehavebeenveryfewassociationstudiesusingbothgenome-widemulti-locusandmulti-variatemixedmodels[81,82].
Ourapproachcombiningbothmethodsprovedtobepowerfulinlimitingthenumberofcandi-datesbycontrollingforLD,capturingalargepartofbroad-senseheritabilityanddetectingcommongeneticvariantsimpactingthetwophenotypes.
ThetwomainSNPsLG5_5394803andLG6_15273858wereassociatedindependentlywithmulti-locusmodelsonlgc(chromosome6)andbs(chromosome5)andco-detectedonbothphenotypeswiththemulti-variatemodel.
ThehighercorrelationbetweenLG5_5394803andLG6_15273858(r2vs=9.
76E-02)comparedtoback-groundpairwiselocicouldexplainthelackofco-localizationwiththemulti-locusmodel.
Indeed,themulti-locusmodelcorrectsforpotentialspuriousassoci-ationsduetointraorinter-chromosomicLDbyre-estimatinggeneticvarianceatthestepwiseinclusionofeachassociatedlocusasregressorinthemodel[50].
Thedrawbackofsuchcorrectionisthatpleiotropicgeneticvariantscannotbedetected,whichinturnemphasizestheadvantageofusingmulti-variateass-ociationmodelsasacomplementaryapproach.
Noteworthy,multi-variatemixedmodels,bytakingad-vantageofcorrelationsbetweenphenotypescanpoten-tiallyalsocapturevariantswithsmallereffectsthanthosedetectedwithtraditionaluni-variateanalyses[83].
Asanexample,thisallowedustodetectthesmall,butsignificanteffectofLG7_18047191(chromosome7)onbothlgcandbsGBLUP(withPVEreaching3and12%,respectively).
Despitethesmallsizeofourcore-collection(73acces-sions),wehavesuccessfullyshownthatourmethodhasenoughpowertodetectmarker-traitassociationsforbacterialcankerresistance.
Nevertheless,ourpopulationmightnothavebeenlargeenoughtocaptureSNPswithminoreffectsduetoalackofstatisticalpowerwithtypeII(false-negative)errors.
Aninterestingnextanalyticperspectivewouldbetheuseofgenomicselection(GS)modelsonahighernumberofindividuals[84].
Thiswouldassesstheeffectofallmarkersandinparticularlow-effectvariantsinthephenotypicvarianceregardlessoftheirfrequencyinthepopulation,allowingtovalidatethecandidateswhiledetectingnewlociandthusgivinganoverallbetterestimationofthegeneticeffects[85].
Inourcase,duetogenotypicsamplingbias,thesegeneticeffectscouldhavebeeneitheroverestimatedforthedetectedloci[86]orunderestimatedforrarevariantspossiblylinkedtoresistance.
Forinstance,theculti-varsA0008andA1314noticedforparticularpheno-typeofpartialresistancewerenothighlightedbytheanalysisofhaplotypesinourstudy,buttheymightberelevantmaterialtocrossinprogenies.
GSmodelsorlinkageanalysisusingbi-parental,multi-parentalorinterconnectedprogenies[63,87–89]derivedfromthismaterialcouldthusmapadditionalrarevariantswhichcouldhavebeenremovedinourstudy(filterMAF0.
05).
AnalysesofVariance(ANOVA)wereusedtoassesssignificanceoffixedeffects:year,operator,genotype,genotype-by-yearinteractionandthefixedcovariatesdiameterandheightofthebranchonlgcandbsaccord-ingtothefollowingmodel:PijkμYiY=hiY=iYi=OjGkGkYieijkwherePijkisthephenotypicvalueofaccessionknotedbythejthoperatorinyeari,μtheoverallmean,Yitheef-fectofyeari,(Y/h)iand(Y/)ithenestedeffectsoftheshootdiameterandheight,respectivelywithintheyeari,Yi/Ojthenestedeffectofthejthoperatorwithintheyeari,Gktheeffectoftheaccessionk,Gk*Yitheinteractiontermbetweenyeariandgenotypek,andeijktherandomindependentandidenticallydistributedresidualterm.
Twomainapproacheshavebeenconductedforchar-acterizingtheinfluenceoftheinoculationintotheob-servedaccessions:amulti-yearsandwithin-yearmodelsthroughtheuseoflinearmixedmodels.
Themulti-yearsmodelintegratedtheeffectoftheYear(Y)(asfixedfactor)andboth–Genotype(G)andGeno-typexYear(GxY)interactions(asrandomfactors).
Thismodelwasusedtomodelcorrelationsofthesamepheno-typebetweenyears.
Preliminaryanalysisrevealedthatgen-eticvariancesdifferedsignificantlybetweenyears.
Thus,ahomogenouscorrelation(corh)variance-covariance(vcov)matrixwasfittedontheinteractionterminthemodel.
Theresidualerrorwasmodeledeitherforallyearsorforeachyearindependently.
Inordertoselecttheoptimummodelforeachphenotype,agoodness-of-fitcomparisonwasmadebasedonRestrictedMaximumLikelihood(REML)statisticsusingtheAkaikeInformationCriterion(AIC).
Duetotheexpectedannualeffectonphenotypesandthemissingdatainourdesign(Additionalfile11),aFig.
4Photographoftypicalbacterialcankersymptomsonbranchesaftercontrolledinoculation.
Cankerlengthlgc(bluearrows)andsuperficialbrowningbs(redarrows)asobserved6monthsafterinoculationOmranietal.
BMCPlantBiology(2019)19:31Page12of18within-yearmodelwasperformedforeachyearaccord-ingtothefollowinglinearmixedmodel:PjkμhOjGkejkMoreover,byconsideringawithin-yearmodelforeachphenotype,weensuredthebiologicalrelevanceofmiss-ingdatareplacementinthemulti-yearmodelallowedbytheimputationduetothecorhvcovstructure.
Bestunbiasedlinearpredictors(BLUPs)weresubse-quentlyassessedfrombothwithin-yearandmulti-yearmodelsonGandG*Yrandomeffectsforthepheno-typeslgcandbs.
Thesepredictorsextractedfromthemulti-yearmodelwerethenusedasphenotypicinputdataforGWAS.
Avalidationoftheresultswasper-formedusingthewithin-yearBLUPs.
InthecaseofG*YBLUPs,imputeddatawereremovedformissingindi-vidualsineachyearinordertoavoidstatisticalredun-dancywiththeGBLUPprediction.
REMLestimatesofgenetic(σ2G)andresidual(σ2e)varianceswerealsocom-putedfromthemulti-yearmodel.
Broad-senseheritabil-ityH2foreachphenotypeperyearwascalculatedas:H2σ2Gσ2Gσ2enwheren=3isthenumberofreplicatesperaccession.
GenotypingandSNPmarkerdatafilteringGenotypingandSNPsalignmentwerepreviouslyde-scribedin[30].
Briefly,thesetofplantsforGWASwasgenotypedusingNGS-IlluminaHiSeq2000/2500se-quencingwithestimateddepthsbetween15to25foldsdependingontheaccession.
SNPsalignmentwasper-formedusingthePrunuspersica(P.
persica)v1.
0refer-encegenomeandconsideringgene-spaceregionsovertheeightchromosome-levelscaffoldassembliescovering99%ofthegenome[41].
Genotypeassignmentwasper-formedaccordingtoaninferencemethodbasedonmax-imumlikelihoodforestimatingallelefrequenciesbyusingbasecountsateachposition[42].
Forourstudy,aminorallelefrequency(MAF)thresh-oldof5%(accordingto[43])wasappliedontheinitialsetofSNPsinordertoremoverarevariantsandthusavoidfalse-positiveassociations.
Thenafilteronphysicalmapposition,discardingoneofapairofconsecutiveSNPswhosepairwisedistancewaslessthan10bp,wasperformedtolimitthenumberofmultiplestatisticaltests.
Afterconductingallbioinformaticfilters,aselec-tionof63,236SNPswaskeptforassociationstudy.
Markerswerenamedaccordingtotheirphysicalposi-tionsonthegenome,'LG1_72881'foraSNPlocatedonchromosome1at72,881bp,forexample.
PopulationstratificationanalysisAnadditionalfilterbasedonLDpairwisepruning,thatinvolveddiscardingmarkersforwhichsquaredcorrel-ationr2>0.
2(window50-stepwise5SNPs),wasexe-cutedinordertokeepasetof21,942independentmarkers(undertheassumptionoflinkageequilibrium)forinferringpopulationstructure.
TheAdmixtureprogram[44]wasusedforcomputingmaximumlikelihoodestimationsofindividualancestryfractionswhiletestingscenariifromk=2tok=10ances-tralgroups.
Themodelchoicecriteriawasbasedonminimizationofthecross-validationerror.
Inordertoinsurethereliabilityoftheoptimalchoiceofk,acomple-mentaryPCAanalysiswasperformedonthegenotypematrix.
Inferenceofrelatednessbetweenindividualsre-sultedinthecalculationofanidentity-by-state(IBS)allelesharingmatrixfromthesamesubsetof21,942SNPsusingtheemmax-kinfunctionimplementedintheEMMAX(EfficientMixed-ModelAssociationeXpedited)program[45].
Boththeancestralproportionfractionmatrix(Q)calculatedaccordingtotheoptimummodelfromAdmix-tureandtheIBSpairwisematrixbetweenindividuals(K)wereusedinGWAStocorrectinflationofp-valuesduetostratificationartefactsinthepopulation.
LinkagedisequilibriumestimationPairwiseLDfromasamplingof1000markersperchromosomewascomputedusingr2.
Acorrectedpro-cedurecompactedintheLDcorSVRpackage[46]wasusedallowingremovalofpopulationstratificationbiasonLD.
Consideringasetofmarkersoneachchromo-some,bothinitialandcorrectedr2estimates(r2vs)werethenplottedagainstphysicaldistancesinordertoinves-tigateintra-chromosomalLDdecayaccordingtothefol-lowingmodelassumingdrift-recombinationequilibrium[47]r2114bdewherer2isthesquareoflocicorrel-ationbetweenamarkerpair,disthepairwisephysicaldistancebetweenthetwomarkers,bisadecaycoeffi-cientcalculatedwithleastsquaresestimatesinanon-linearregression(nlsfunctioninRsoftware)andereferstoaresidualestimate.
Genome-wideassociationanalysisCumulativeeffectsofKandQmatricesovergenotypicdatawerefirsttestedwithEMMAX[45]asacovariancegeneticmatrixandfixedancestralcovariatesimpactingexpressionofphenotypes.
Quantile-quantile(Q-Q)plotswererealizedtoselectthebestpredictivemodelbetweenK,QandK+QforeachphenotypeusingtheqqmanRpackage[48].
Thusaccordingtothebestmodelchoice,twosupplementarymixedmodelswereperformed:are-centlyimplementedmulti-locusmixed-model[49]fromtheMLMMalgorithm[50](onbothGandG*YBLUPsOmranietal.
BMCPlantBiology(2019)19:31Page13of18fromthemulti-yearmodel)andamulti-variatelinearmixed-model(mvLMM)algorithm(betweenlgcandbsGBLUPsfromthemulti-yearmodel)developedinGEMMAsoftware[51].
Multi-locusmodelsconsistofforwardstepwisemixed-modelregressionswithanin-creasingnumberofincludedmarkers(regressors)whilere-estimatinggeneticandresidualvariancecomponentsateachstep.
TheimplementationfromMLMMallowsamoresuitablehandlingoftheso-called"highdimensionissue"resultingfromalowpopulationsizeandahighnumberofpossibleregressorsintroducedintothemodel[49].
Insteadofusingaclassicalp-valuethresholdthatconsidersonlythenumberofmultipletests(numberofSNPs),themodelselectioncriterionisbasedonthecal-culationofamorepermissiveextendedBayesianInfor-mationCriterioneBIC[52].
P-valuesofmarker-traitassociationsaregivenconsideringboththeoptimalmodel(withpossiblyseveralregressors)andeachassoci-atedSNPasuniqueregressoravoidingp-valueinflationduetoforwardregressorinclusion.
Themulti-traitmodeltakesadvantageofthecorrelationstructurebetweenmultiplephenotypestoincreasepowertodetectnotonlypleiotropicgeneticvariantsbutalsospecificvariantsaffectingonlyoneofthecorrelatedphe-notypes[51].
Multi-testingcorrectiononoutputp-valueswasperformedconsideringFalseDiscoveryRate(FDR)[53]controlperchromosome(5%significancelevel).
Forallsignificantassociations,wecomputedthealleliceffectαas(Minorallelemean–Majorallelemean)/2andtheindividualpercentageofphenotypicvarianceex-plained(PVE).
Furthermore,foreachtrait,strictsenseheritabilitywascalculatedastheratioh2σ2PSNPσ2PSNPσ2Ewhereσ2PSNPistheadditivegeneticvarianceexplainedbyallassociatedlociandσ2Eistheremainingvariancelinkedtobothdominance,epistasisgeneticeffectsandnon-geneticresidualerror.
Inordertoensuretheirreli-ability,alldetectedassociationsweresubjectedtofurthervalidations(ANOVAtests)onBLUPsfromthewithin-yearmodel.
Estimatesofr2vsaroundmarker-trait-associatedlociwerecomputedtogiveagraphicalrepresentationofthepair-wiseLDusingamodifiedversionfromthesnp.
plotterRpackage[54].
Moreover,performinggenome-widesam-plingof1000markers,r2vsdistributionswereprospectedconsideringbothinter-chromosomalandcandidatelociSNPpairs,andcomparedtointra-chromosomalscaleinordertodetectanyspecificlinkages.
HaplotypeanalysisGroupsofmarkerswereconstructedbyconsideringalldetectedlociforeachGandG*YBLUPphenotype.
Markerhaplotypeswereidentifiedamongalltheacces-sionsaccordingtothegenotypesonthecandidatelociforeachgroupofmarkers.
Meanphenotypicdistribu-tionsofthedifferenthaplotypescomprisingatleasttwoaccessionswerecomparedusingHSDpost-hoctests(α=5%)fortheassociatedGBLUPphenotypesconsideringbothmulti-yearandwithin-yearmodels.
Forhaplotypesrepresentedbyonlyoneindividual,meanvaluewascomparedwiththesignificantlydifferinggroupsfromtheHSDtests.
Favorable(andunfavorable)haplotypesweredefinedforeachmulti-yearandwithin-yearmodelGBLUPphenotypeaccordingtothefollowingcriteria:(i)showingasignificantlylower(orupper)groupmeanvaluethanallotherhaplotypesgroupsand(ii)withstableperformancesamongyears.
CandidategeneidentificationForeachdetectedloci,candidategeneswereidentifiedandlocalizedonthefirstandsecondversionofthepeachgenome(P.
persicav1.
0andv2.
1,publiclyavail-ableathttps://www.
rosaceae.
org/species/prunus/all).
Pu-tativehomologproteinannotationswereobtainedfollowingBlastpsearchesupontheArabidopsisthaliana(A.
thaliana)genomeusingtheTAIR10database(https://www.
arabidopsis.
org/).
SNPgenelocalizationandeffectontheproteinsequenceweredeterminedusingtheJBrowsetool(https://www.
rosaceae.
org/tools/jbrowse)ontheP.
persicav1.
0genomeandORFfinder(https://www.
ncbi.
nlm.
nih.
gov/orffinder/),respectively.
Then,thepredictedimpactofnon-synonymousSNPonthebiologicalfunctionoftheproteinwasevaluatedwithProveanv1.
1.
3web-interfacesoftware(http://provean.
jcvi.
org/index.
php)[55].
AdditionalfilesAdditionalfile1:Timeseriesofdegree-daysoverwinterperiodinl'Amarine.
2013to2016annualdatafromNovember15thtoMarch31st.
Februarymonthisdelimitedbytheredarrows.
(PDF294kb)Additionalfile2:Genetic(G)andgeneticxyear(G*Y)distributionsoflgcandbsBLUPs.
A.
andB.
lgcandbsdensityplotsofG(red),(G*Y)2013(green)and(G*Y)2015(blue)adjustedBLUPs.
C.
ScatterplotshowingtheregressionlinebetweenlgcandbsGBLUPsincluding95%confidenceinterval.
AllBLUPvaluesarerepresentedonaln(x+1)scale.
(TIFF347kb)Additionalfile3:PopulationStructure(Q)oftheapricotcore-collectionbasedondataof21,942independentSNPs.
Genotypicdataprunedinordertokeepmarkerswithpairwiser2values2.
Greenandredcases:extremegroupvaluesrelatedtopartiallyresistantandsusceptibleaccessionsforeachphenotype.
Forhaplotypescarriedbyonlyoneindividual,assessmentwasperformedincomparisonwithsignificantlydifferentgroupsfromHSDpost-hoctests.
Boldvaluesshowoutliersinindividualhaplotypesfromtheextremegroups(valuesoverthemean±standarddeviationoftheextremesignificantgroupsissuedfromHSDpost-hoctests).
(XLSX19kb)Additionalfile11:Informationofplantmaterialusedinthestudy:geographicalorigin,nameandnumberofyearrepetitionsthrough2013to2016phenotypingperiod.
(XLSX11kb)Additionalfile12:Qancestralcovariates,G,G*YBLUPsfromthemulti-yearandGBLUPsfromthewithin-yearmodelofphenotypicdataintheapricotcore-collection.
NA:missingvalues.
Q1,Q2andQ3covariatesarerelativetoancestralfractionsfrom(i)Central&EasternAsia,(ii)ContinentalEuropeand(iii)Irano-Caucasia&MediterraneanBasin,respectively.
(XLSX23kb)Additionalfile13:Genotypicdataofthe63,236SNPsusedforGWAS.
MAF:MinorAlleleFrequency.
Genotypesarecodedaccordingtominoralleledose.
0/2=homozygousmajor/minorallele.
1=heterozygous.
MarkersnamesareindicatedasLG[chromosomenumber]_[physicalpositioninbp].
(XLSX17724kb)Abbreviationsh2:Strict-senseheritability;H2:Broad-senseheritability;ABA:Abscisicacid;AIC:Akaikeinformationcriterion;ANOVAq:Analysisofvariance;BLUP:Bestlinearunbiasedpredictor;bp:Basepair;bs:Superficialbrowning;CDS:Codingsequences;corh:Correlation;eBIC:Extendedbayesianinformationcriterion;FDR:Falsediscoveryrate;G:Genetic;glm:Generalizedlinearmodel;GWAS:Genome-wideassociationstudy;JA:Jasmonicacid;K:Kinship;LD:Linkagedisequilibrium;LG:Linkagegroup;lgc:Cankerlength;MAF:Minor-allelefrequency;MAS:Marker-assistedselection;MLE:Maximumlikelihoodestimation;mvLMM:Multivariatelinearmixedmodel;NA:Non-applicable;PCA:Principalcomponentanalysis;Ps:Pseudomonassyringae;PVE:Percentageofvariationexplained;Q:Populationstructure;QQ-plots:Quantile-quantileplots;QTL:Quantitativetraitloci;REML:Restrictedmaximumlikelihood;SA:Salicylicacid;SNP:Single-nucleotidepolymorphism;vcorr:Variance-covariance;Y:YearAcknowledgementsWeacknowledgetheexperimentalteamsofUR-GAFLandAmarineExperimentalfarmfortheirhelpinimplementingtheexperimentations.
WethanksApricotrepositoryasapartofPrunusBiologicalResourceCenterinAmarineexperimentalfarmfortheavailabilityoftheapricotgermplasmandthedatasetsissuedfromprimarydescriptors.
WethanksAlbertAbbottandVéroniqueDecroocqforthegenotypingdatasetsissuedfromABRIWGANRproject.
WeparticularlythankCharlotteChandeysson,JeanLeonetti,EricMartin,TimmyDefert,CarlosGilandAnne-MarieFerreolfortheirhighlyvaluablecontributioninfieldexperimentandphenotyping.
ManythankstoMounaHadjBrahimforherworkasMastersstudent.
WearealsogratefultoBrigitteManginforsharingscriptandprovidingvaluablescientificadvice.
WefinallythankChristopherSauvage,BénédicteQuilot-Turion,BrigitteManginandLucieTamisierforreviewingthemanuscript.
FundingTheResibacCASDAR(2013–2016)andABRIWG(2012–2014)projectssupportedthiswork.
MOwassupportedbyafellowshipfromtheFrenchMinistryofAgricultureandAgriFoodandhostedasaPhDstudent.
AvailabilityofdataandmaterialsThedatasetsgeneratedand/oranalyzedduringthecurrentstudyareincludedinAdditionalfiles12and13.
Authors'contributionsMO,MR,GR,ABconductedtheexperimentaltrialsandphenotyping.
MOandMRconductedthestatisticalanalysesforphenotypicstudyandassociationmapping.
MOwroteandreviewedtheredactionofthemanuscript.
J-MAandCMEcoordinatedtheproject,constructedtheexperimentaldesign,orientedtheanalysisandmonitoredtheredactionofthearticle.
Allauthorsread,approvedthefinalmanuscriptanddeclaredthattheyhavenoconflictofinterestinthepublicationofthisdocument.
EthicsapprovalandconsenttoparticipateNotapplicable.
ConsentforpublicationNotapplicable.
Omranietal.
BMCPlantBiology(2019)19:31Page15of18CompetinginterestsTheauthorsdeclarethattheyhavenocompetinginterests.
Publisher'sNoteSpringerNatureremainsneutralwithregardtojurisdictionalclaimsinpublishedmapsandinstitutionalaffiliations.
Authordetails1INRA,UR1052GénétiqueetAméliorationdesFruitsetLégumes,CentredeRecherchePACA,Montfavet,France.
2INRA,UR407PathologieVégétale,CentredeRecherchePACA,Montfavet,France.
3ENGREF,AgroParisTech,Paris,France.
4CEPInnovation,Lyon,France.
5PresentAddress:Agroscope,ResearchDivisionPlantBreeding,Wdenswil,Switzerland.
Received:25May2018Accepted:4January2019References1.
StClairDA.
Quantitativediseaseresistanceandquantitativeresistancelociinbreeding.
AnnuRevPhytopathol.
2010;48(1):247–68.
2.
MundtCC,CowgerC,GarrettKA.
Relevanceofintegrateddiseasemanagementtoresistancedurability.
Euphytica.
2002;124(2):245–52.
3.
MundtCC.
Durableresistance:akeytosustainablemanagementofpathogensandpests.
InfectGenetEvol.
2014;27:446–55.
4.
Cropproductioninformation:Apricots[http://www.
fao.
org/faostat/en/#compare].
Accessed20Apr2018.
5.
GardanL,PrunierJP,LuisettiJ,BezelguesJJ.
ResponsabilitédediversPseudomonasdansledépérissementbactériendel'abricotierenFrance.
RevuedeZoologieAgricoleetdePathologieVégétale.
1973;4:112–20.
6.
CrosseJE.
Bacterialdiseasesofstone-fruittreesinBritain:IX.
Bacteriosisofapricot.
TransBrMycolSoc.
1953;36(1):38–45.
7.
BergeO,MonteilCL,BartoliC,ChandeyssonC,GuilbaudC,SandsDC,MorrisCE.
Auser'sguidetoadatabaseofthediversityofPseudomonassyringaeanditsapplicationtoclassifyingstrainsinthisphylogeneticcomplex.
PLoSOne.
2014;9(9):e105547.
8.
ScortichiniM.
Epidemiologyandpredisposingfactorsofsomemajorbacterialdiseasesofstoneandnutfruittreesspecies.
JPlantPathol.
2010;92(1):S73–8.
9.
LamichhaneJR,VarvaroL,ParisiL,AudergonJM,MorrisCE.
DiseaseandfrostdamageofwoodyplantscausedbyPseudomonassyringae:seeingtheforestforthetrees.
AdvAgron.
2014;126:235–95.
10.
KennellyMM,CazorlaFM,deVicenteA,RamosC,SundinGW.
Pseudomonassyringaediseasesoffruittrees:progresstowardunderstandingandcontrol.
PlantDis.
2007;91(1):4–17.
11.
VigourouxA.
IngressandspreadofPseudomonasinstemsofpeachandapricotpromotedbyfrost-relatedwater-soakingoftissues.
PlantDis.
1989;73(10):854–5.
12.
KlementZ,RozsnyayDS,BálóE,PánczélM,PrileszkyG.
TheeffectofcoldondevelopmentofbacterialcankerinapricottreesinfectedwithPseudomonassyringaepv.
syringae.
PhysiolPlantPathol.
1984;24(2):237–46.
13.
YoungJM.
Orchardmanagementandbacterialdiseasesofstonefruit.
NZJExpAgric.
1987;15(2):257–66.
14.
PrunierJP,BordjibaO.
Effectoffrostonbacterialnecrosisofapricotbuds.
ActaHortic.
1991;293:495–502.
15.
PrunierJP,AudergonJM,CottaJ.
Etudedelasensibilitévariétaledel'abricotierauchancrebacterien.
In:ProgrammederechercheAgrimed:Deuxièmesrencontressurl'abricotier:1993.
Avignon:Officedespublicationsofficiellesdescommunautéseuropéennes;1993.
p.
111–2.
16.
BrunL,WarlopF,MercierV,BroquaireJ,ClauzelG,GomezC,ParveaudCE,AudergonJM.
Quellesélectionfruitièrepouruneproductiondurable,àfaibleniveaud'intrantsMéthodologiepourunréseaudesélectionvariétaledécentralisée.
InnovationsAgronomiques.
2011;15:105–15.
17.
GarrettCME.
ScreeningPrunusrootstocksforresistancetobacterialcanker,causedbyPseudomonasmorsprunorum.
JHorticSci.
1979;54(3):189–93.
18.
Mgbechi-EzeriJ,PorterL,JohnsonKB,OraguzieN.
Assessmentofsweetcherry(PrunusaviumL.
)genotypesforresponsetobacterialcankerdisease.
Euphytica.
2017;213(7):145.
19.
BedfordKE,SholbergPL,KappelF.
Useofadetachedleafbioassayforscreeningsweetcherrycultivarsforbacterialcankerresistance.
In:XXVIInternationalHorticulturalCongress:GeneticsandBreedingofTreeFruitsandNuts:2003.
Leuven:InternationalSocietyforHorticulturalScience(ISHS);2003.
p.
365–8.
20.
LiB,HulinMT,BrainP,MansfieldJW,JacksonRW,HarrisonRJ.
Rapid,automateddetectionofstemcankersymptomsinwoodyperennialsusingartificialneuralnetworkanalysis.
PlantMethods.
2015;11:57.
21.
DaveyJW,HohenlohePA,EtterPD,BooneJQ,CatchenJM,BlaxterML.
Genome-widegeneticmarkerdiscoveryandgenotypingusingnext-generationsequencing.
NatRevGenet.
2011;12(7):499–510.
22.
ZhuC,GoreM,BucklerES,YuJ.
Statusandprospectsofassociationmappinginplants.
PlantGenomeJ.
2008;1(1):5.
23.
KhanMA,KorbanSS.
Associationmappinginforesttreesandfruitcrops.
JExpBot.
2012;63(11):4045–60.
24.
OguraT,BuschW.
Fromphenotypestocausalsequences:usinggenomewideassociationstudiestodissectthesequencebasisforvariationofplantdevelopment.
CurrOpinPlantBiol.
2015;23:98–108.
25.
NordborgM,WeigelD.
Next-generationgeneticsinplants.
Nature.
2008;456(7223):720–3.
26.
IngvarssonPK,StreetNR.
Associationgeneticsofcomplextraitsinplants.
NewPhytol.
2011;189(4):909–22.
27.
HuangX,HanB.
Naturalvariationsandgenome-wideassociationstudiesincropplants.
AnnuRevPlantBiol.
2014;65(1):531–51.
28.
FontiForcadaC,OraguzieN,Reyes-Chin-WoS,EspiauMT,SociasiCompanyR,FernándeziMartíA.
Identificationofgeneticlociassociatedwithqualitytraitsinalmondviaassociationmapping.
PLoSOne.
2015;10(6):e0127656.
29.
CaoK,ZhouZK,WangQ,GuoJ,ZhaoP,ZhuGR,FangWC,ChenCW,WangXW,WangXL,etal.
Genome-wideassociationstudyof12agronomictraitsinpeach.
NatCommun.
2016;7:13246.
30.
MarietteS,WongJunTaiF,RochG,BarreA,ChagueA,DecroocqS,GroppiA,YhL,LambertP,TriconD,etal.
Genome-wideassociationlinkscandidategenestoresistancetoplumpoxvirusinapricot(Prunusarmeniaca).
NewPhytol.
2016;209(2):773–84.
31.
ArumuganathanK,EarleED.
NuclearDNAcontentofsomeimportantplantspecies.
PlantMolBiolReport.
1991;9(3):208–18.
32.
FoltaKM,GardinerSE.
Genomics-basedopportunitiesinapricot.
In:FoltaKM,GardinerSE,editors.
GeneticsandGenomicsofRosaceae.
NewYork:SpringerNewYork;2009.
p.
315–35.
33.
ZaurovD,MolnarT,EisenmanS,FordTM,MavlyanovaR,CapikJM,ReedFunkC,GoffredaJC.
Geneticresourcesofapricots(PrunusarmeniacaL.
)inCentralAsia.
HortScience.
2013;48(6):681–91.
34.
YilmazKU,GurcanK.
Geneticdiversityinapricot.
In:IntechOpen,editor.
GeneticDiversityinPlantsMahmutCaliskan;2012.
p.
249–70.
35.
BourguibaH,AudergonJ-M,KrichenL,Trifi-FarahN,MamouniA,TrabelsiS,D'OnofrioC,AsmaBM,SantoniS,KhadariB.
LossofgeneticdiversityasasignatureofapricotdomesticationanddiffusionintotheMediterraneanBasin.
BMCPlantBiol.
2012;12:49.
36.
RikkerinkEHA,OraguzieNC,GardinerSE.
Prospectsofassociationmappinginperennialhorticulturalcrops.
In:OraguzieNC,RikkerinkEHA,GardinerSE,DeSilvaHN,editors.
AssociationMappinginPlants.
NewYork:Springer;2007.
p.
249–69.
37.
BartoliC,CarrereS,LamichhaneJR,VarvaroL,MorrisCE.
Whole-genomesequencingof10Pseudomonassyringaestrainsrepresentingdifferenthostrangespectra.
GenomeAnnounc.
2015;3(2):e00379–15.
38.
KingEO,WardMK,RaneyDE.
Twosimplemediaforthedemonstrationofpyocyaninandfluorescin.
JLabClinMed.
1954;44(2):301–7.
39.
CaoT,SaylerRJ,DeJongTM,KirkpatrickBC,BostockRM,ShackelKA.
Influenceofstemdiameter,watercontent,andfreezing-thawingonbacterialcankerdevelopmentinexcisedstemsofdormantstonefruit.
Phytopathology.
1999;89(10):962–6.
40.
GilmourAR,ThompsonR,CullisBR.
AverageinformationREML:anefficientalgorithmforvarianceparameterestimationinlinearmixedmodels.
Biometrics.
1995;51(4):1440–50.
41.
VerdeI,AbbottAG,ScalabrinS,JungS,ShuS,MarroniF,ZhebentyayevaT,DettoriMT,GrimwoodJ,CattonaroF,etal.
Thehigh-qualitydraftgenomeofpeach(Prunuspersica)identifiesuniquepatternsofgeneticdiversity,domesticationandgenomeevolution.
NatGenet.
2013;45(5):487–94.
42.
KimSY,LohmuellerKE,AlbrechtsenA,LiY,KorneliussenT,TianG,GrarupN,JiangT,AndersenG,WitteD,etal.
Estimationofallelefrequencyandassociationmappingusingnext-generationsequencingdata.
BMCBioinformatics.
2011;12(1):231.
43.
AulchenkoYS,RipkeS,IsaacsA,vanDuijnCM.
GenABEL:anRlibraryforgenome-wideassociationanalysis.
Bioinformatics.
2007;23(10):1294–6.
Omranietal.
BMCPlantBiology(2019)19:31Page16of1844.
AlexanderDH,NovembreJ,LangeK.
Fastmodel-basedestimationofancestryinunrelatedindividuals.
GenomeRes.
2009;19(9):1655–64.
45.
KangHM,SulJH,ServiceSK,ZaitlenNA,KongS-y,FreimerNB,SabattiC,EskinE.
Variancecomponentmodeltoaccountforsamplestructureingenome-wideassociationstudies.
NatGenet.
2010;42(4):348–57.
46.
ManginB,SiberchicotA,NicolasS,DoligezA,ThisP,Cierco-AyrollesC.
Novelmeasuresoflinkagedisequilibriumthatcorrectthebiasduetopopulationstructureandrelatedness.
Heredity.
2011;108:285–91.
47.
HeifetzEM,FultonJE,O'SullivanN,ZhaoH,DekkersJCM,SollerM.
Extentandconsistencyacrossgenerationsoflinkagedisequilibriumincommerciallayerchickenbreedingpopulations.
Genetics.
2005;171(3):1173–81.
48.
TurnerSD.
qqman:anRpackageforvisualizingGWASresultsusingQ-Qandmanhattanplots:biorXiv;2014.
49.
BonnafousF,FievetG,BlanchetN,BonifaceM-C,CarrèreS,GouzyJ,LegrandL,MarageG,Bret-MestriesE,MunosS,etal.
ComparisonofGWASmodelstoidentifynon-additivegeneticcontroloffloweringtimeinsunflowerhybrids.
TheorApplGenet.
2018;131(2):319–32.
50.
SeguraV,VilhjálmssonBJ,PlattA,KorteA,Seren,LongQ,NordborgM.
Anefficientmulti-locusmixedmodelapproachforgenome-wideassociationstudiesinstructuredpopulations.
NatGenet.
2012;44(7):825–30.
51.
ZhouX,StephensM.
Efficientmultivariatelinearmixedmodelalgorithmsforgenome-wideassociationstudies.
NatMethods.
2014;11(4):407–9.
52.
ChenJ,ChenZ.
Extendedbayesianinformationcriteriaformodelselectionwithlargemodelspaces.
Biometrika.
2008;95(3):759–71.
53.
BenjaminiY,HochbergY.
Controllingthefalsediscoveryrate:apracticalandpowerfulapproachtomultipletesting.
JRStatSocSerBMethodol.
1995;57(1):289–300.
54.
LunaA,NicodemusKK.
snp.
plotter:anR-basedSNP/haplotypeassociationandlinkagedisequilibriumplottingpackage.
Bioinformatics.
2007;23(6):774–6.
55.
ChoiY,ChanAP.
PROVEANwebserver:atooltopredictthefunctionaleffectofaminoacidsubstitutionsandindels.
Bioinformatics.
2015;31(16):2745–7.
56.
MuirW,NyquistWE,XuS.
Alternativepartitioningofthegenotype-by-environmentinteraction.
TheorApplGenet.
1992;84(1–2):193–200.
57.
KapitonovVV,JurkaJ.
HarbingertransposonsandanancientHARBI1genederivedfromatransposase.
DNACellBiol.
2004;23(5):311–24.
58.
MusacchioA,GibsonT,RiceP,ThompsonJ,SarasteM.
ThePHdomain:acommonpieceinthestructuralpatchworkofsignallingproteins.
TrendsBiochemSci.
1993;18(9):343–8.
59.
QuesadaT,GopalV,CumbieWP,EckertAJ,WegrzynJL,NealeDB,GoldfarbB,HuberDA,CasellaG,DavisJM.
Associationmappingofquantitativediseaseresistanceinanaturalpopulationofloblollypine(PinustaedaL.
).
Genetics.
2010;186(2):677–86.
60.
FoulongneM,PascalT,PfeifferF,KervellaJ.
QTLsforpowderymildewresistanceinpeach*Prunusdavidianacrosses:consistencyacrossgenerationsandenvironments.
MolBreed.
2003;12(1):33–50.
61.
CalengeF,DurelCE.
BothstableandunstableQTLsforresistancetopowderymildewaredetectedinappleafterfouryearsoffieldassessments.
MolBreed.
2006;17(4):329–39.
62.
SeguraV,DurelC-E,CostesE.
Dissectingappletreearchitectureintogenetic,ontogeneticandenvironmentaleffects:QTLmapping.
TreeGenetGenomes.
2009;5(1):165–79.
63.
AllardA,BinkMCAM,MartinezS,KelnerJ-J,LegaveJ-M,diGuardoM,DiPierroEA,LaurensF,vandeWegEW,CostesE.
DetectingQTLsandputativecandidategenesinvolvedinbudbreakandfloweringtimeinanapplemultiparentalpopulation.
JExpBot.
2016;67(9):2875–88.
64.
BrachiB,MorrisGP,BorevitzJO.
Genome-wideassociationstudiesinplants:themissingheritabilityisinthefield.
GenomeBiol.
2011;12:232.
65.
ArúsP,GardinerS.
GenomicsforimprovementofRosaceaetemperatetreefruit.
In:VarshneyRK,TuberosaR,editors.
Genomics-AssistedCropImprovement:Vol2:GenomicsApplicationsinCrops.
Dordrecht:SpringerNetherlands;2007.
p.
357–97.
66.
ZhangW,CollinsA,GibsonJ,TapperWJ,HuntS,DeloukasP,BentleyDR,MortonNE.
Impactofpopulationstructure,effectivebottlenecktime,andallelefrequencyonlinkagedisequilibriummaps.
ProcNatlAcadSciUSA.
2004;101(52):18075–80.
67.
DuanC-G,WangX,XieS,PanL,MikiD,TangK,HsuC-C,LeiM,ZhongY,HouY-J,etal.
Apairoftransposon-derivedproteinsfunctioninahistoneacetyltransferasecomplexforactiveDNAdemethylation.
CellRes.
2016;27(2):226–40.
68.
RebecchiMJ,ScarlataS.
Pleckstrinhomologydomains:acommonfoldwithdiversefunctions.
AnnuRevBiophysBiomolStruct.
1998;27(1):503–28.
69.
vanLeeuwenW,krészL,BgreL,MunnikT.
Learningthelipidlanguageofplantsignalling.
TrendsPlantSci.
2004;9(8):378–84.
70.
DieckCB,BossWF,PereraIY.
Aroleforphosphoinositidesinregulatingplantnuclearfunctions.
FrontPlantSci.
2012;3:50.
71.
FigueiredoJ,SousaSilvaM,FigueiredoA.
Subtilisin-likeproteasesinplantdefence:thepast,thepresentandbeyond.
MolPlantPathol.
2018;19(4):1017–28.
72.
JordáL,CoegoA,ConejeroV,VeraP.
Agenomicclustercontainingfourdifferentiallyregulatedsubtilisin-likeprocessingproteasegenesisintomatoplants.
JBiolChem.
1999;274(4):2360–5.
73.
ZhaoY,ThilmonyR,BenderCL,SchallerA,HeSY,HoweGA.
VirulencesystemsofPseudomonassyringaepv.
tomatopromotebacterialspeckdiseaseintomatobytargetingthejasmonatesignalingpathway.
PlantJ.
2003;36(4):485–99.
74.
ZhaoC,JohnsonBJ,KositsupB,BeersEP.
ExploitingsecondarygrowthinArabidopsis.
ConstructionofxylemandbarkcDNAlibrariesandcloningofthreexylemendopeptidases.
PlantPhysiol.
2000;123(3):1185–96.
75.
GunesekeraB,TorabinejadJ,RobinsonJ,GillaspyGE.
Inositolpolyphosphate5-phosphatases1and2arerequiredforregulatingseedlinggrowth.
PlantPhysiol.
2007;143(3):1408–17.
76.
deTorres-ZabalaM,TrumanW,BennettMH,LafforgueG,MansfieldJW,RodriguezEgeaP,BgreL,GrantM.
Pseudomonassyringaepv.
tomatohijackstheArabidopsisabscisicacidsignallingpathwaytocausedisease.
EMBOJ.
2007;26(5):1434–43.
77.
LievensL,PollierJ,GoossensA,BeyaertR,StaalJ.
Abscisicacidaspathogeneffectorandimmuneregulator.
FrontPlantSci.
2017;8:587.
78.
CheckerVG,KushwahaHR,KumariP,YadavS.
Roleofphytohormonesinplantdefense:signalingandcrosstalk.
In:SinghA,SinghIK,editors.
MolecularAspectsofPlant-PathogenInteraction.
Singapore:SpringerSingapore;2018.
p.
159–84.
79.
CuiH,WangC,QinT,XuF,TangY,GaoY,ZhaoK.
PromotervariantsofXa23allelesaffectbacterialblightresistanceandevolutionarypattern.
PLoSOne.
2017;12(10):e0185925.
80.
ShiL,WengJ,LiuC,SongX,MiaoH,HaoZ,XieC,LiM,ZhangD,BaiL,etal.
IdentificationofpromotermotifsregulatingZmeIF4Eexpressionlevelinvolvedinmaizeroughdwarfdiseaseresistanceinmaize(ZeaMaysL.
).
MolGenGenomics.
2013;288(3):89–99.
81.
AlbertE,SeguraV,GricourtJ,BonnefoiJ,DerivotL,CausseM.
Associationmappingrevealsthegeneticarchitectureoftomatoresponsetowaterdeficit:focusonmajorfruitqualitytraits.
JExpBot.
2016;67(22):6413–30.
82.
JaiswalV,GahlautV,MeherPK,MirRR,JaiswalJP,RaoAR,BalyanHS,GuptaPK.
Genomewidesinglelocussingletrait,multi-locusandmulti-traitassociationmappingforsomeimportantagronomictraitsincommonwheat(T.
aestivumL.
).
PlosOne.
2016;11(7):e0159343.
83.
YangJJ,WilliamsLK,BuuA.
Identifyingpleiotropicgenesingenome-wideassociationstudiesformultivariatephenotypeswithmixedmeasurementscales.
PLoSOne.
2017;12(1):e0169893.
84.
MeuwissenTHE,HayesBJ,GoddardME.
Predictionoftotalgeneticvalueusinggenome-widedensemarkermaps.
Genetics.
2001;157(4):1819–29.
85.
CrossaJ,Pérez-RodríguezP,CuevasJ,Montesinos-LópezO,JarquínD,delosCamposG,BurgueoJ,González-CamachoJM,Pérez-ElizaldeS,BeyeneY,etal.
Genomicselectioninplantbreeding:methods,models,andperspectives.
TrendsPlantSci.
2017;22(11):961–75.
86.
UtzHF,MelchingerAE,SchnCC.
Biasandsamplingerroroftheestimatedproportionofgenotypicvarianceexplainedbyquantitativetraitlocideterminedfromexperimentaldatainmaizeusingcrossvalidationandvalidationwithindependentsamples.
Genetics.
2000;154(4):1839–49.
87.
DiPierroEA,GianfranceschiL,DiGuardoM,Koehorst-vanPuttenHJJ,KruisselbrinkJW,LonghiS,TroggioM,BiancoL,MurantyH,PagliaraniG,etal.
Ahigh-density,multi-parentalSNPgeneticmaponapplevalidatesanewmappingapproachforoutcrossingspecies.
HorticRes.
2016;3:16057.
88.
BillotteN,JourjonMF,MarseillacN,BergerA,FloriA,AsmadyH,AdonB,SinghR,NouyB,PotierF,etal.
QTLdetectionbymulti-parentlinkagemappinginoilpalm(ElaeisguineensisJacq.
).
TheorApplGenet.
2010;120(8):1673–87.
89.
TisnéS,PomièsV,RiouV,SyahputraI,CochardB,DenisM.
Identificationofganodermadiseaseresistancelociusingnaturalfieldinfectionofanoilpalmmultiparentalpopulation.
G3.
2017;7(6):1683–92.
90.
SkellyDA,MagwenePM,StoneEA.
Sporadic,globallinkagedisequilibriumbetweenunlinkedsegregatingsites.
Genetics.
2016;202(2):427–37.
91.
BrachiB,MeyerCG,VilloutreixR,PlattA,MortonTC,RouxF,BergelsonJ.
CoselectedgenesdetermineadaptivevariationinherbivoreresistanceOmranietal.
BMCPlantBiology(2019)19:31Page17of18throughoutthenativerangeofArabidopsisthaliana.
ProcNatlAcadSci.
2015;112(13):4032–7.
92.
Pilet-NayelM-L,MouryB,CaffierV,MontarryJ,KerlanM-C,FournetS,DurelC-E,DelourmeR.
Quantitativeresistancetoplantpathogensinpyramidingstrategiesfordurablecropprotection.
FrontPlantSci.
2017;8:1838.
93.
CalengeF,FaureA,GoerreM,GebhardtC,VandeWegWE,ParisiL,DurelCE.
Quantitativetraitloci(QTL)analysisrevealsbothbroad-spectrumandisolate-specificQTLforscabresistanceinanappleprogenychallengedwitheightisolatesofVenturiainaequalis.
Phytopathology.
2004;94(4):370–9.
94.
DirlewangerE,GrazianoE,JoobeurT,Garriga-CaldereF,CossonP,HowadW,ArusP.
Comparativemappingandmarker-assistedselectioninRosaceaefruitcrops.
ProcNatlAcadSci.
2004;101(26):9891–6.
Omranietal.
BMCPlantBiology(2019)19:31Page18of18