Genetic diversity and accession structure in European Cynara cardunculus collections

23 Pages·2017·16.37 MB·English

RESEARCHARTICLE
Genetic diversity and accession structure in
European Cynara cardunculus collections
MarioA.Pagnotta1,2*,JuanA.Ferna´ndez3,4,GabriellaSonnante5,CatalinaEgea-
Gilabert2,4
1 DipartimentodiScienzeAgrarieeForestali(DAFNE),TusciaUniversity,Viterbo,Italy,2 Cienciay
Tecnolog´ıaAgraria,E.T.S.Ingenier´ıaAgro´noma,UniversidadPolite´cnicadeCartagena,Cartagena(Murcia),
Spain,3 Produccio´nvegetal,E.T.S.Ingenier´ıaAgro´noma,UniversidadPolite´cnicadeCartagena,Cartagena
(Murcia),Spain,4 InstitutodeBiotecnolog´ıaVegetal(IBV),UniversidadPolite´cnicadeCartagena,Cartagena
(Murcia),Spain,5 InstituteofBiosciencesandBioresources,NationalResearchCouncil(CNR),Bari,Italy
a1111111111
a1111111111 *[email protected]
a1111111111
a1111111111
a1111111111 Abstract
Understandingthedistributionofgeneticvariationsandaccessionstructuresisanimportant
factorformanaginggeneticresources,butalsoforusingpropergermplasminassociation
mapanalysesandbreedingprograms.Theglobeartichokeisthefourthmostimportanthor-
OPENACCESS
ticulturalcropinEurope.Here,wereporttheresultsofamolecularanalysisofacollection
Citation:PagnottaMA,Ferna´ndezJA,SonnanteG,
includingglobeartichokeandleafycardoongermplasmpresentintheItalian,Frenchand
Egea-GilabertC(2017)Geneticdiversityand
accessionstructureinEuropeanCynara Spanishgenebanks.Theaimsofthisstudywereto:(i)assessthediversitypresentinEuro-
cardunculuscollections.PLoSONE12(6): peancollections,(ii)determinethepopulationstructure,(iii)measurethegeneticdistance
e0178770.https://doi.org/10.1371/journal.
betweenaccessions;(iv)clustertheaccessions;(v)properlydistinguishaccessionspresent
pone.0178770
inthedifferentnationalcollectionscarryingthesamename;and(vi)understandthediversity
Editor:RobertoPapa,UniversitàPolitecnicadelle
distributioninrelationtothegenebankandthegeographicoriginofthegermplasm.Atotal
Marche,ITALY
of556individualsgroupedinto174accessionsofdistincttypologieswereanalyzedbydiffer-
Received:January10,2017
enttypesofmolecularmarkers,i.e.dominant(ISSRandAFLP)andco-dominant(SSR).
Accepted:May18,2017 Thedataofthetwocrops(globeartichokeandleafycardoon)wereanalyzedjointlyandsep-
Published:June1,2017 aratelytocompute,amongotheraims,thegenediversity,heterozygosity(He,Ho),fixation
indexes,AMOVA,geneticdistanceandstructure.Thefindingsunderlinethehugediversity
Copyright:©2017Pagnottaetal.Thisisanopen
accessarticledistributedunderthetermsofthe presentintheanalyzedmaterial,andtheexistenceofallelesthatareabletodiscriminate
CreativeCommonsAttributionLicense,which amongaccessions.Theaccessionswereclusterednotonlyonthebasisoftheirtypology,
permitsunrestricteduse,distribution,and
butalsoonthebasisofthegenebanktheycomefrom.Probably,theenvironmentalcondi-
reproductioninanymedium,providedtheoriginal
authorandsourcearecredited. tionsofthedifferentfieldgenebanksaffectedgermplasmconservation.Theseoutcomes
willbeusefulinplantbreedingtoselectaccessionsandtofingerprintvarieties.Moreover,
DataAvailabilityStatement:Allrelevantdataare
withinthepaperanditsSupportingInformation theresultshighlighttheparticularattentionthatshouldbepaidtothemethodusedtocon-
files. servetheCynaracardunculusgermplasmandsuggesttothepreferenceofusingacces-
Funding:AGRIGENRESCommunityProgramme sionsfromdifferentgenebankstorunanassociationmap.
(EuropeanCommission,Directorate-Generalfor
AgricultureandRuralDevelopment,underCouncil
Regulation(EC)No870/2004)financially
supportedtheproject.Financialsupportwasalso
providedbyTheSenecaFoundationforitsprogram
"EstanciasdeInvestigadoresVisitantes",for
mobility,cooperationandinternationalization.
PLOSONE|https://doi.org/10.1371/journal.pone.0178770 June1,2017 1/23
Cynarastructure
Competinginterests:Theauthorshavedeclared Introduction
thatnocompetinginterestsexist.
Cynaracardunculusisadiploidspecieswith2n=2x=34belongingtotheAsteraceaefamily.
Arecentbotanicalclassificationgroupedtwovegetablecropsandtheirwildprogenitor,which
werepreviouslyconsidereddistinctspecies,underC.cardunculus.Thethreetaxawithinthe
speciesare:wildcardoon(C.cardunculusvar.sylvestris(Lamk)Fiori),globeartichoke(C.car-
dunculusvar.scolymus(L.)Fiori),andcultivatedorleafycardoon(C.cardunculusvar.altilis
DC).Inlinewiththisclassification,theupdatedCPVO/UPOVprotocol(enteredintoforceon
27.02.2013)totestsondistinctness,uniformityandstability,hasbecomethesameforboththe
globeartichokeandtheleafycardoon[1].
Globeartichokegermplasmcanbeclassifiedaccordingtodifferentcriteria,themost
importantofwhichare(i)theharvestingtimeand(ii)theheadshape.(i)Thefirstdividesthe
globeartichokeinto(a)autumnandspring(earlyor“re-bloom”varieties)or(b)onlyspring
(lateor“spring”varieties).(ii)Thesecondcriterionclassifiestheglobeartichokeonthebasis
ofthemorphologyofitscommercialpart,theimmatureflowers(buds)called“capitula”or
“heads”.Inparticular,thecharacteristicstakenintoconsiderationare:shape,color,andthe
presenceofspines.Itisthereforepossibletoidentifyfourgroups:"Spiny"withlongsharp
spinesonbothbractsandleaves,"Violet"withviolet-coloredcapitula,"Romanesco"with
sphericalorsub-sphericalnon-spinycapitula,and"Catanese"withrelativelysmall,elongated
non-spinycapitula[2].SpinyandCatanesetypesarenormallyre-bloomingtypologies,while
VioletandRomanescovarietiesareusuallyharvestedinspring.
Theglobeartichokeiscultivatedallaroundtheworld,butitisparticularlywelladaptedto
thedifferentpedo-climaticconditionsoftheMediterranean[3].Itiswidelygrownthroughout
theMediterraneanBasin,inSouthAmerica(Chile,ArgentinaandPeru)andinCalifornia
(USA).ItisthefourthmostimportanthorticulturalcropinEuropewithaproductionof815
kt,[4]afterpotato,tomato,andleafyvegetables.Theleafycardoonisaminorcropandis
mainlygrowninNorthernItaly,SouthernFranceandinSpain,forlocalconsumption.The
speciesisalsoimportantsinceithasnutraceutical,biochemicalandmedicinalproperties(see
[5]andreferencestherein).
ThecenteroforiginoftheglobeartichokeistheMediterraneanBasin,probablysouthern
Italy.Evidencebasedonitsdiversityandonmorphologicalandmoleculardataindicatesthat
theglobeartichokewaspossiblydomesticatedinSicilyatthebeginningofthefirstmillennium
[6–8],whilecardoonprobablyoriginatedintheIberianPeninsulaandtheSouthofFrance[8,
9].Infact,Italyhastherichestbiodiversityofglobeartichokeandcardoon,whichhasresulted
inthelocalcultivationofmanytypesofvarietiesandlandraces,veryoftenwelladaptedtospe-
cificlocalclimaticconditions[5].However,despitethiswidebiodiversity,thegreatestpartof
Italianglobeartichokecultivationisbasedonveryfewclones[10].
ThemainC.cardunculusgermplasmcollectionsareheldbyItaly,FranceandSpain,where
traditionalglobeartichokecultivarsarepredominant.However,cultivationsofsingleuniform
varietieshaveexpandedwithaconsequentreductionindiversitycomparedwiththeoriginal.
Indeed,inspiteofthehugegermplasmdiversityavailableinItaly,onlyafewvarietiesareculti-
vatedoverlargeareas:‘ViolettodiSicilia’,‘Brindisino’,‘VioletdeProvence’(allthreebelong-
ingtothesameCatanesegroup),‘Romanesco’(mainlythecloneC3,nowsubstitutedbysome
clonesderivedfrommicro-propagation),and‘SpinosoSardo’.InFrance,onlyaboutfiveorsix
varietiesarecommonlycultivated.AlmostallFrenchproductionisbasedontheglobearti-
chokes‘CamusdeBretagne’,‘GrosVertdeLaon’,‘BlancHyerois’,‘VioletduGapeau’,‘Castel’
andthesmallcylinderglobeartichoke‘PetitVioletdeProvence’.InSpain,the‘Blancade
Tudela’ecotyperepresents90%ofthewholeproduction.
PLOSONE|https://doi.org/10.1371/journal.pone.0178770 June1,2017 2/23
Cynarastructure
InordertounderstandandpreserveC.cardunculusgeneticdiversityitisimportanttoknow
thepropagationsystem,whichdiffersbetweenglobeartichokeandcardoon.Theyareboth
allogamousplants,butthefirstismainlypropagatedvegetativelybymeansofbasalshootsor
semi-dormantshootswithalimitedrootsystem[11],whilethesecondisseedpropagated.Asa
result,thelevelofheterozygosityinglobeartichokeishigherthanincardoon,bothwildand
cultivated[5,9,12,13];moreover,theglobeartichokeoftenhasamulti-clonalstructure.One
probleminC.cardunculusconservationandcharacterizationisthatthelocalvarietiesareoften
namedonthebasisoftheareawheretheyarecultivated[14],regardlessoftheirrealgenetic
diversityorsimilarity.Asaresult,thenamesofaccessionsorlocalvarietiesarenotalwaysuniv-
ocalandinsomecasescangeneratesynonymiesorhomonymies.Inaddition,previousstudies
havedemonstratedthatthehugediversitywithineachbotanicalvarietyisquiteoftennotrelated
withthecorrespondinggeographicorigin(e.g.[15–19]).Consideringallthis,itisclearthatthe
characterizationofC.cardunculusisessentialforitscorrectconservationandutilization.
Moreover,itissometimedifficulttodefineglobeartichokevarietiessince,aftercultivation
overseveraldecadesinvariousgeographicalareas,theymightbesubjectedtodivergentselec-
tion.Thus,theaccessionsstoredingenebankfieldcollectionsneedtoberationalizedby
improvingcorecollectionsandavoidingduplications[20–22].
Molecularmarkersarepowerfultoolsthatcanbeusedtoidentify,cluster,andfingerprint
individualsoraccessions[23].Severaltypesofmolecularmarkershavebeenusedforcardoon
andglobeartichokecharacterizationinthelastfifteenyears.Theseincludebothdominant
markers,whichdonotrequireaprioriDNAsequenceinformation,andco-dominantmarkers,
forwhichsequenceknowledgeisnecessary.ThedominantmarkersincludeRAPD(Random
AmplifiedPolymorphicDNA,[24–27]),AFLP(AmplifiedFragmentLengthPolymorphism),
andISSR(InterSimpleSequencesRepeats),whichwereusedwithglobeartichokevarieties
andcardoon[28],orspecificglobeartichokevarietalgroups,suchasSpiny[29],Romanesco
[15,30,31],VioletandCatanese[32–35],globeartichokehybrids[36]andcardoon[37].
Amongco-dominantmarkers,SSRs(SimpleSequenceRepeats)[8,34,38–42]andSRAP
(Sequence-RelatedAmplifiedPolymorphism)havebeenusedforglobeartichokes,andinboth
cultivatedandwildcardoon[43].
Theassociationbetweengenotypeandphenotypecanbeachievedeitherbycontrolledbipa-
rentalcrosses(linkagemapping)orviaassociationmapping,controllingthelinkagedisequilib-
rium,i.e.thenon-randomassociationofallelesbetweenloci,regardlessoftheirpositionacross
thechromosomes[44].Theassessmentofgeneticvariationandpopulationstructureisapre-
requisitebeforeperformingassociationmapping.Linkagemappinghasaseriesoflimitations,
includingitshighcost,lowresolution,theneedforpolymorphismbetweentheparentsused,
largesegregantpopulationandthedistributionofchiasmaacrossthegenome.Conversely,an
associationmap,usingaccessionsthatarenotrelatedbycommonparents,candetectseveral
allelesateachlocusandahigherlevelofpolymorphism.Associationmappingisatoolthatcan
beusedtoinvestigateelitegenesbystructuringthenaturalvariationpresentinagermplasm.
Possibleerrorsinassociationmapsmayariseduetounequalallelefrequencydistribution
betweensubgroups,whichmayleadtospuriousassociationsbetweenmolecularmarkersand
thetraitsofinterest[45].Inordertoreducesucherrors,beforeperforminganassociationanaly-
sisinapopulation,itisessentialtodeterminethepopulationstructure,aswedointhisstudy.
PreviousworksonC.cardunculuscharacterizationwereaddressedatspecificaccession
typologiesandusedlimitedcollectionsbelongingtorestrictedgeographicareas.Theonly
exceptionisourpreviouspaper,whichreportedpreliminarydatausingonlysomeofthedomi-
nantmarkersintheEuropeancollection[18].Theaimsofthepresentpaperare(i)toassess,for
thefirsttime,thediversitypresentinC.cardunculusEuropeancollections,usingbothdominant
andco-dominantmarkers,(ii)todeterminethepopulationstructure,(iii)tomeasurethe
PLOSONE|https://doi.org/10.1371/journal.pone.0178770 June1,2017 3/23
Cynarastructure
geneticdistancebetweenaccessions;and(iv)toclustertheaccessionsaccordingtothemolecu-
lardata.Moreover,twootherquestionsareaddressed:(v)aretheaccessionspresentinthedif-
ferentnationalcollectionsandcarryingthesamenamereallythesamematerial?and(vi)howis
diversitydistributedinrelationtothegeographicoriginofthegermplasm?
Materialandmethods
Plantmaterial
Atotalof556individualsbelongingtoItalian(264)(CNR-IBBR,Bari;CNRISAFOMCatania;
ARSIALRome),French(162)(GEVES,Cavaillon)andSpanish(130)(ITGA,Navarra)collec-
tions,andrepresenting174accessionswerejointlyanalyzed.Theaccessionlistisreportedin
S1Tabletogetherwiththeirtypologyandcountryofconservation.Theaccessionsaredivided
accordingtothefourtypologiesdescribedintheintroductionandidentifiedbyPorceddu
etal.[2]:Romanesco(225individuals),Violet(34individuals),Catanese(116individuals)and
Spiny(11individuals),plustheaccessionsbelongingtoleafycardoon(72individuals).More-
over,twoadditionalglobeartichokecategorieswereadded:theBlancadeTudelatypology(39
individuals)duetoitsimportanceintheIberianPeninsula,andOFF(59individuals),which
includestheaccessionsnotunivocallyclassifiableasbelongingtotheprevioustypologies.
Molecularmarkers
TotalgenomicDNAwasextractedfrom100mgoffrozentissue,usingtheQiagenDNeasy
PlantMiniKitandsharedamonglaboratoriestoperformmarkeramplifications.FortheISSR
markers,11primerswereused:810,818,827,834,840,841,855,857,857c,857gand872,all
developedbytheUniversityofBritishColumbia,Canada.Thereactionswereperformedin
10μlcontaining10ngDNA,0.3μMprimer,100μMdNTP,10mMTris-HCl(pH9.0)and1U
Taqpolymerase.Amplificationconditionswere94˚C/5min,followedby35cyclesof94˚C/1
min,43–59˚C(specificforeachprimer;seeS2Table)/1min60˚C/1minand72˚C/2min,and
endingwithanextensionstepof72˚C/10min.
ForAFLPanalysis,MseIandPstIorEcoRIandMseIwereusedtodigesttheDNAtemplate.
FollowingtheproceduredescribedbyVosetal.[46],pre-amplificationwasperformedwith
non-selectiveprimers,while7primercombinationswereusedfortheselectiveamplification
step:EcoACC/MseCTA,EcoACG/MseCTT,EcoAGC/MseCTT,MseAC/PstCA,MseAC/
PstCG,MseGC/PstCAandMseGC/PstCG.
For19SSRmarkersdevelopedbyAcquadroetal.[47](CMAL06,CMAL-108,CMAL11,
CMAL117,CMAL21,CMAL24,CMAL-25),Acquadroetal.[48],(CDAT-01,CLIB-02,CLIB-
12),Acquadroetal.[40](CMAFLP-01,CMAFLP-04,CMAFLP-05,CMAFLP-18),Sonnante
etal.[39](CsCiCaCa05,CsPal02,CsPal03,CsEST03)andFA2-GAT(Primers,F:GCCGAAGA
AGACGGAAGAATCTGA,R:CATCACGCTTGGTTAAAGATCGGG)wereused.
ForallamplificationstheforwardprimerswerefluorescentlylabeledtoresolvePCRampli-
consonanABI3130xl(AppliedBiosystems)oraCEQ8800(BeckmanCoulter)sequencer.
Thedetectedbandswerecheckedforreproducibilityevenifthevisualizationbysequencer
showedhighsensitivityandprecision.Fortheco-dominant(SSR)markers,eachallelewas
scoredinaccordancewithitsmolecularweightinbp,whileforthedominant(AFLPand
ISSR)markers,0–1matriceswereobtained,withoutknowingtheallelicrelationships.Inthis
case,eachpossiblebandwasconsideredasalocuswith2possiblealleles,0(absence)or1
(bandpresence).Insomecases,the0–1matrixwasconsideredashaplotype,whileinothersit
wasconvertedintoaco-dominantmatrixwith1asdominanthomozygote(A/A)and0asthe
otherhomozygote(a/a).
PLOSONE|https://doi.org/10.1371/journal.pone.0178770 June1,2017 4/23
Cynarastructure
Statisticalanalyses
ThegenediversityindexwascalculatedforeachlocusandpopulationaccordingtoNei[49],
usingtheHardy-Weinbergformula
P
He ¼1(cid:0) n p2:
i¼1 i
Thepolymorphisminformationcontent(PIC)wascomputedas[50].
Xn Xn(cid:0) 1Xn
PIC ¼1(cid:0) p2(cid:0) 2p2p2
i i j
i¼1 i¼1j¼iþ1
TocomparedifferencesamongandwithinaccessionsandgroupsWright’sfixationindices
wereused[51].TheF-statisticsarebasedontheexpectedlevelofheterozygosity.Themeasure-
mentswerecomputedforthedifferentlevelsoftheaccessionstructures,suchasthevariance
ofallelefrequencieswithinaccessions(F ),varianceofallelefrequenciesamongaccessions
IS
(F ),inbreedingcoefficientwithinindividualtotaldiversity(F ),varianceamongacces-
ST IT
sionswithintypes(F )andvariancepermutingaccessionsamonggroups(F ),whichare
SC CT
relatedtothedegreeofheterozygosityatvariouslevelsoftheaccessionstructure.Theterms
mentionedabovearerelatedthroughtheformula:1-F =1-F +1-F ,whereIindicatesthe
IT IS ST
individual,Sthesub-accessionandTthetotalaccession;F referstotheindividualcompared
IT
withthetotal;F istheindividualcomparedwiththesubaccession;andF isthesubacces-
IS ST
sioncomparedwiththetotal.ThetotalF,indicatedbyF ,canbepartitionedintoF (orf)
IT IS
andF (orθ).F canbecomputedusingtheformula:F =(H -H )/H whereH isthe
ST ST ST T S T, T
proportionoftheheterozygotesinfullaccessionsandH theaverageproportionofheterozy-
S
gotesinsub-accessions.TheFstatisticwasalsousedintheAMOVA(AnalysisofMOlecular
Variance)tomeasurethepartitionofvariationamongtypologies,amongaccessionswithin
typologies,amongindividualswithinaccessions,andwithinindividuals.
Thegeneticdiversity(He)andgeneticidentity(JorHo)werealsousedtoestimatethe
geneticdistance.If
Xn
J ¼ p2
x xi
i¼0
istheprobabilityofidentityinthexaccessionand
Xn
J ¼ p2
y yi
i¼1
istheprobabilityofidentityintheyaccession,theprobabilityofidentityinbothaccessionsis
Xn
J ¼ p p
xy xi yi
i¼1
asdescribedbyNei[52].Theprobabilityofidentityinthexaccessionforallnormalizedlociis
qffiffiffiffiffiffiffi
I ¼J (cid:4) J J
xy x y
and,inturn,thegeneticdistanceis:
qffiffiffiffiffiffiffi
D¼(cid:0) LnI(cid:0) LnJ (cid:4) J J
xy x y
PLOSONE|https://doi.org/10.1371/journal.pone.0178770 June1,2017 5/23
Cynarastructure
ThedistancesbetweenaccessionswerealsocomputedusingtheEuclideandistance
qffiffiffiffiffiffiffiffiffiffiffiffiffiffi
d¼ ðp p Þ2
1 2
Theobtaineddistancewasthenusedtoclustertheaccessionsaccordingtodifferentcluster-
ingmethodologiessuchastheUPGMAalgorithm.TheclusteringwasalsoperformedbyK-
mean,whichisanon-hierarchicalmethodofclassificationthatpartitionsasetofsamplesinto
themostappropriatenumberofclustersdecidedinadvance[53].Runlength,inSTRUCTURE
software,wasgivenasa150kburningperiodlengthfollowedby150kMarkovChainMonte
Carlo(MCMC)replications.Assuggested,severalanalyseswerefirstrunusingdifferentKval-
uesfrom2to9.Finally,inaccordancewithlikelihood,ΔK[54]thatidentifiedtwopicksfor
K=3andK=6(S1Fig),F distributionamonggroupsandwiththefactthatthegermplasm
ST
couldbedividedinto6groupsonthebasisoftypology(Romanesco,Violet,Catanese,Spiny,
Cardoon,andOFF),bothK=3andK=6wereadoptedandarepresentedbelow.Individuals
wereassignedtosubgroupsbythe“Noadmixturemodel”.Theoutputreportsthesubsequent
probabilitythatindividualiisfromaccessionk.Thepriorprobabilityforeachaccessionis1/
K.Thismodelisappropriateforstudyingfullydiscreteaccessions.The“admixturemodel”
wasalsorun,butnobetterresolutionwasobservedwiththismodel(datanotshown).Linkage
disequilibriumbetweenloci,whichmeasuresthedeviationfromrandomassociationbetween
allelesatdifferentloci[55],anditssignificance(Pvaluesofχ2with1000permutations),was
alsocomputed.
Insomecases,theco-dominantanddominantdatatypologieswereanalyzedseparately,
sinceitwasnotpossibletocomputesomeparametersforthedominantmarkers.Inaddition,
analyseswererunbothjointlyandseparatelyforglobeartichokeandcardoonaccessions.
Tocomputetheabovementionedparameters,GenAlEx[56,57],PowerMarker[58],Arle-
quin[59],andSTRUCTUREv2.3[45]softwarewereused.
Results
Geneticdiversity
Theidentifiedco-dominantalleleswere147intotal,fromthe556individualsanalyzed.For
eachco-dominantmarker,theallelesrangedfrom1to15withanaverageof7.4allelesper
marker(Table1).Nevertheless,mostofthealleles(about62%)wererare,havingafrequency
lowerthan5%;asaresult,themajoralleleshadanoverallaveragefrequencyof70%.Forthe
co-dominantmarkers,themajorallelesforeachmarkerhadafrequencyrangingfromabout
23%toaround87%,excludingtheCMAFLP-05marker,whichwasmonomorphic.Someof
theSSRalleleswerespecificforasingleaccession(Table2).Intotal,33privatealleleswere
foundin24accessions.Asexpected,thefrequencywashigherandmoreconstantinthedomi-
nantmarkerswithonlytwoalleles.Infact,averagefrequencyofprivatealleleswas0.816and
0.805forAFLPandISSRrespectivelyversus0.700forSSR,whileitrangedfrom0.743to0.878
andfrom0.232to0.869fordominantandco-dominantmarkersrespectively.
Thegenediversitycomputedasexpectedheterozygosity(He)andthepolymorphisminfor-
mationcontent(PIC)provideinformationofamarker’sabilitytodeterminepolymorphism.
Inthepresentstudy,Hevaluesforthedominantmarkerswerequiteuniform,withanaverage
ofaround0.26and0.28forAFLPandISSR,respectively(Table1).ForSSR,thegenediversity
rangedfrom0.24(FA2-GATandCLIB-02II)to0.83(CLIB-02I)(Table1).Notethatthe
markerCLIB-02identified2loci,herelabeledasCLIB-02IandCLIB-02II,sothatCLIB-02
altogetherhadhighervalues.ThePICvalueshadasimilarbutnotequalrankingamongmark-
erscomparedwiththegenediversityparameter.PICrangedfrom0.21ofFA2-GATto0.80of
PLOSONE|https://doi.org/10.1371/journal.pone.0178770 June1,2017 6/23
Cynarastructure
Table1. Markersusedtogenotype556Cynaracardunculusindividuals,withtheirgeneticdiversityparameters.
Marker Alleles PAF He PIC Ho Fis Fit Fst F
AFLP
EaccMcta 0.799 0.255 0.203
EacgMctt 0.801 0.291 0.242
EagcMctt 0.835 0.219 0.176
MacPca 0.796 0.283 0.231
MacPcg 0.820 0.257 0.213
MgcPca 0.828 0.249 0.207
MgcPcg 0.832 0.244 0.204
Mean 0.816 0.257 0.211
ISSR
810 0.848 0.245 0.210
818 0.781 0.310 0.250
827 0.766 0.331 0.271
834 0.884 0.203 0.181
840 0.756 0.333 0.267
841 0.852 0.231 0.197
855 0.817 0.279 0.234
857 0.878 0.201 0.175
857c 0.757 0.341 0.277
857g 0.780 0.321 0.263
872 0.743 0.332 0.264
Mean 0.805 0.284 0.235
SSRloci
CsCiCaCa05 10 0.829 0.306 0.298 0.243 -0.287 0.228 0.400 -0.252
CDAT-01 7 0.311 0.768 0.734 0.806 -0.930 -0.048 0.457 -0.923
CLIB-02I 7 0.232 0.825 0.800 0.391 -0.655 0.446 0.665 -0.635
CLIB-02II 4 0.869 0.236 0.223 0.049 0.079 0.817 0.801 0.110
CLIB-12 3 0.365 0.663 0.589 0.790 -0.977 -0.232 0.377 -0.956
CMAFLP-01 6 0.741 0.425 0.397 0.375 -0.850 0.154 0.543 -0.804
CMAFLP-04 13 0.632 0.554 0.516 0.123 -0.554 0.815 0.881 -0.507
CMAFLP-05 1 1.000 0.000 0.000 0.000 1.000 1.000
CMAFLP-18 12 0.500 0.605 0.532 0.750 -0.894 -0.301 0.313 -0.895
CMAL06 13 0.474 0.633 0.568 0.532 -0.749 0.050 0.457 -0.723
CMAL-108 4 0.561 0.582 0.513 0.628 -0.958 -0.090 0.444 -0.936
CMAL11 3 0.619 0.473 0.363 0.478 -0.612 -0.024 0.365 -0.555
CMAL117 14 0.737 0.415 0.371 0.240 -0.407 0.307 0.507 -0.336
CMAL21 10 0.632 0.553 0.515 0.527 -0.732 -0.024 0.409 -0.705
CMAL24 6 0.366 0.683 0.617 0.164 -0.511 0.649 0.767 -0.423
CMAL-25 2 0.504 0.500 0.375 0.992 -0.995 -0.925 0.035 -0.994
CsPal02 15 0.604 0.562 0.508 0.317 -0.476 0.372 0.575 -0.432
CsPal03 5 0.512 0.637 0.580 0.723 -0.851 -0.189 0.358 -0.836
CsEST03 8 0.666 0.484 0.418 0.403 -0.314 0.124 0.333 -0.289
FA2-GAT 4 0.864 0.236 0.209 0.244 -0.764 -0.024 0.419 -0.696
Mean 7.4 0.601 0.507 0.456 0.439 -0.655 0.155 0.505 -0.620
Overallmean 0.700 0.397 0.347
NumberofallelesinthecaseofSSR(Alleles),frequencyoftheprincipalallele(PAF),GenediversityandExpectedHeterozygosity(He),polymorphism
informationcontent(PIC),observedheterozygosity(Ho),varianceofallelefrequencieswithinaccessions(Fis),inbreedingcoefficientwithinanindividual
totaldiversity(Fit),varianceofallelefrequenciesamongaccessions(Fst),Wright’sfixationindices(F).
https://doi.org/10.1371/journal.pone.0178770.t001
PLOSONE|https://doi.org/10.1371/journal.pone.0178770 June1,2017 7/23
Cynarastructure
Table2. Summaryofprivateallelesbyaccession.
Accession Locus Allele Frequency
Ascolano CMAFLP-04 262 0.313
BlancHye´rois CLIB-02I 221 0.333
BlancoPeralta CMAFLP-04 253 1.000
BlancoValencia EST03 195 0.100
BlancoValencia EST03 198 0.100
Brindisi CLIB-02II 223 0.500
Camard FA2-GAT 187 0.100
CamusBretagneI FA2-GAT 223 0.250
Caribou CMAL117 168 0.100
Caribou CMAFLP-04 272 0.125
CaribouSp EST03 183 0.250
Chrysanthème CMAL117 192 0.100
Chrysanthème CsPal03 331 0.100
DelCortijo EST03 201 0.100
DelCortijo EST03 204 0.100
Mola CMAL06 137 0.038
MonteluponeA CsPal02 381 0.250
Paestum CsPal02 343 0.063
Pertosa CsPal02 355 0.125
Pisa CMAL117 174 0.063
PuvisAme´liore´ CMAL-108 101 0.167
PuvisAme´liore´ CMAFLP-04 286 0.375
SErasmo CMAL117 170 0.250
VerdeCalahorra CMAL06 157 0.125
VerdePeralta CsPal02 361 0.400
VertVaulxVelin CMAL06 147 0.750
VertVaulxVelin CsPal02 349 1.000
VertVaulxVelin CMAFLP-01 224 0.167
VertVaulxVelin CMAFLP-01 323 0.333
VertVaulxVelin CMAFLP-04 273 0.250
ViolettoSicilia13 CMAL06 140 0.250
ViolettoSicilia98 CMAL11 275 0.250
ViolProvence41S CMAL24 236 0.500
https://doi.org/10.1371/journal.pone.0178770.t002
CLIB-02I,followedbyCDAT-01with0.73(Table1).Thevalueswere,onaverage,lowerin
leafycardoonthaninglobeartichoke(S3andS4Tables).
Fortheco-dominantmarkers,itisalsopossibletocomputetheobservedheterozygosity
(Ho)andtheWrightfixationindicesbyconsideringbothHeandHo.Theobservedheterozy-
gosityrangedfrom5%forCLIB-02IIto99%forCMAL-25,withanaverageof44%(Table1).
Asaconsequence,thepartitionofvariation,intoitscomponentsi.e.withinaccessions(F ),
IS
withinindividuals(F ),andamongaccessions(F )wasquitedifferentfromonemarkerto
IT ST
another.TheaveragevaluesfortheF ,F andF were-0.66,0.16and0.51,respectively
IS IT ST
(Table1).Generally,thevalueswerelowerinleafycardoonthaninglobeartichoke(S3andS4
Tables).Ingeneral,thelociwereinLDwitheachother(S5Table)exceptforCMAL-25,which
wasnotinLDwithCMAL11,CMAL117,CMAL24,CsCaCa05,CsEST03andCsPal02.
Theaccessionsanalyzedhadquitedifferentlevelsofdiversity,asdetectedusingbothdomi-
nantandco-dominantmarkers,asshowninS6andS7Tables.Thepolymorphismranged
PLOSONE|https://doi.org/10.1371/journal.pone.0178770 June1,2017 8/23
Cynarastructure
from0toabout55%,withanaverageof20%,forthedominantmarkersandfrom10to85%,
withanaverageof54%,fortheco-dominantmarkers.Theexpectedheterozygosityranged
from0to17%,withanaverageof7%,fordominantandfrom5to42%,withanaverageof
26%,forco-dominantmarkers.Theobservedheterozygosityrangedfrom10to67%,withan
averageof45%.Thefixationindexwasmainlynegative,rangingfrom-1to0.24,withanaver-
ageof-0.74.Theaccessionshadonaveragemorethanoneallelebut,inspiteofthehighnum-
berofallelesidentifiedbythemarkers,theallelesweregenerallyspecificforeachaccession.
Thealleleswithafrequencyhigherthan5%inasingleaccessionrangedfrom0.6to2.3,and
thenumberoflocallycommonallelesfoundinlessthan25%oftheaccessionsbutpresentin
morethan5%inthespecificaccessionrangedfrom0to1.2,withanaverageof0.2.Some
accessionshadmorethanasingleprivateallele,reachingamaximumoffiveprivateallelesspe-
cificforVertVaulxVelin(Table2).TheShannonInformationIndexindicatedrichness,and
theevennessrangedfrom0to0.26forthedominantandfrom0to0.66fortheco-dominant
markers,withanaverageof0.11and0.36,respectively(S6Table).Onaverage,themarker
parametersforeachaccessionrangedfrom0.403to3.286forthenumberofalleles;from1to
2.55forthenumberofeffectivealleles,from0to0.934fortheShannon’sInformationindex,
from0to0.517fortheexpectedheterozygosity;from0to0.583fortheunbiasedexpectedhet-
erozygosity;andfrom6to192forthenumberofampliconsofthedominantmarkers(S7
Table).
Accessionstructureandgeneticrelationships
Thestructureofthe174accessionswasanalyzedbymeansofaBayesianbasedapproachinthe
STRUCTUREprogram,consideringonlytheco-dominantloci.AccordingtotheEvanno[54]
calculation,themostprobableKwasthree(S1Fig).TheresultsobtainedusingSTRUCTURE
withK=3identifiedafirstgroupwiththeleafycardoonaccessions,asecondgroupwith
mainlytheCataneseandTudelaaccessionsandathirdgroupwithalltheotherglobeartichoke
varieties(Fig1).Tobetterseparatethenon-Cataneseglobeartichokes,thestructureanalysis
wasrepeatedwithKsetequalto6,whichwasthesecondmostprobableKintheEvannoanaly-
sis(S1Fig).Fig2showstheQvalueforeachofthe556individualsanalyzed;thecoloredseg-
mentshavelengthsproportionaltoeachoftheKinferredclusters.
Thefirstsubgroup(SG1),labeledingreeninFig2,containedindividualsfrom1to72,
whichcorrespondedtothefirst14accessions,allofthembelongingunequivocallytotheleafy
cardoongroup.Thesecondsub-group(SG2),labeledinblue,includedmostoftheaccessions
ofCatanesetypology,butalsosomeoftheTudela,someoftheViolet,andsomeoftheunas-
signedindividualssuchasAquara,Francesco,MC12,MC14,MC6,MO10,MO5,andMT1.
Thethirdsub-group(SG3),labeledinturquoise,includedalltheSpiny,someRomanescoand
someVioletindividuals,andfourunassigned(BiancoOstuni,Calimera,NeroCastrignano,and
NeroOstuni).Thefourthsub-group(SG4),labeledinpurple,groupedmostoftheRomanesco
typesplusHydesandVelours(Violet),Cacique(Catanese),andCarlit(unassigned).Thefifth
sub-group(SG5),labeledinyellow,contained12accessionsofRomanescotypes,twounas-
signed(ItalianaandVertProvence),andtwoViolettypes(ViolettoToscanaandSErasmo).
Finally,thesixthsub-group(SG6),labeledinred,includedtheTudelaaccessionsplussome
Romanesco(Macau,CamusBretagneBH8)andCatanese(VioletProvenceF,VioletPro-
vence41S).InthecaseofKequal3(Fig1)thefirstgroupcorrespondedtoSG1includingall
theleafycardoonaccessions.ThesecondgroupwassimilartoSG2+SG6,whilethethird
groupincludedSG3,SG4andSG5.
Geographicallocalizationofsomeaccessionswasprovidedbythegenebanks,whichcol-
lectedthem.Fortheaccessionsgeographicallylocalizedwithcertainty,theaverageproportions
PLOSONE|https://doi.org/10.1371/journal.pone.0178770 June1,2017 9/23
Cynarastructure
Fig1.STRUCTUREanalysesof556individualsbasedon150_000permutations,Noadmixturemodel
andK=3.Theaccessionorderwas:PleinBlancInerme,PuvisAme´liore´,RougeAlger,VertVaulxVelin,
PLOSONE|https://doi.org/10.1371/journal.pone.0178770 June1,2017 10/23