Download K26-05228 ( .pdf )

PAGE 1

RESEARCHARTICLEOpenAccess AncientoriginofaWesternMediterranean radiationofsubterraneanbeetles IgnacioRibera 1,2* ,JavierFresneda 3 ,RuxandraBucur 4,5 ,AnaIzquierdo 1 ,AlfriedPVogler 4,5 ,JoseMSalgado 6 , AlexandraCieslak 1 Abstract Background: Caveorganismshavebeenusedasmodelsforevolutionandbiogeography,astheirreducedabovegrounddispersalproducesphylogeneticpatternsofareadistributionthatlargelymatchthegeologicalhistoryof mountainrangesandcavehabitats.Mostcurrenthypothesesassumethatsubterraneanlineagesaroserecently fromsurfacedwelling,dispersivecloserelatives,butforterrestrialorganismsthereisscantphylogeneticevidence tosupportthisview.Westudyherewithmolecularmethodstheevolutionaryhistoryofahighlydiverse assemblageofsubterraneanbeetlesinthetribeLeptodirini(Coleoptera,Leiodidae,Cholevinae)inthemountain systemsoftheWesternMediterranean. Results: Ca.3.5KBofsequenceinformationfromfivemitochondrialandtwonucleargenefragmentswas obtainedfor57speciesofLeptodiriniandeightoutgroups.Phylogeneticanalysiswasrobusttochangesin alignmentandreconstructionmethodandrevealedstronglysupportedclades,eachofthemrestrictedtoamajor mountainsystemintheIberianpeninsula.Amolecularclockcalibrationofthetreeusingtheseparationofthe Sardinianmicroplate(at33MY)establishedarateof2.0%divergenceperMYforfivemitochondrialgenes(4%for cox1 alone)anddatedthenodesseparatingthemainsubterraneanlineagesbeforetheEarlyOligocene.The colonisationofthePyreneanchain,byalineagenotcloselyrelatedtothosefoundelsewhereintheIberian peninsula,begansoonafterthesubterraneanhabitatbecameavailableintheEarlyOligocene,andprogressed fromtheperipherytothecentre. Conclusions: OurresultssuggestthatbytheEarly-MidOligocenethemainlineagesofWesternMediterranean Leptodirinihaddevelopedallmodificationstothesubterraneanlifeandwerealreadypresentinthemain geographicalareasinwhichtheyarefoundtoday.Theoriginofthecurrentlyrecognisedgeneracanbedatedto theLateOligocene-Miocene,andtheirdiversificationcanthusbetracedtoMioceneancestorsfullyadaptedto subterraneanlife,withnoevidenceofextinctepigean,lessmodifiedlineages.Theclosecorrespondenceof organismalevolutionandgeologicalrecordconfirmsthemasanimportantstudysystemforhistorical biogeographyandmolecularevolution. Background Isolatedorextremeenvironments,suchasislandsor highmountains,havebeenpreferredsystemsforthe studyofspeciationandprocessesofadaptation[1,2]. Oneofthese “ naturallaboratories ” forevolutionisthe deepsubterraneanenvironment,whichcombines extremebuthomogeneousandconstantconditionswith adiscontinuousdistributionpromotingisolation[3,4]. Despitetheearlyrecognitionofthepotentialvalueof thesubterraneanfaunainevolutionarybiology(e.g.DarwindevotesthreepagesoftheOriginstodiscussthe effectofdisuseandtheconvergenceamongcavespecies inNorthAmericaandEurope,[5]pp.137-139),studies ofcaveorganismshavebeenhamperedbytheirgeneral scarcityandthedifficultyofaccessingtheirhabitat. Mostoftheevolutionarystudiesonsubterranean faunahavefocusedonitsoriginandadaptations[4-6]. Caveorendogeanspeciesarecharacterisedbyanumber ofsharedcharacters,assumedtobeeithercausedby lossoffunctiondueto “ lackofuse ” (apterism,depigmentation,reductionorcompletelossoreyes[5,7]),or *Correspondence:ignacio.ribera@ibe.upf-csic.es 1 MuseoNacionaldeCienciasNaturales,JoséGutiérrezAbascal2,28006 Madrid,Spain Ribera etal . BMCEvolutionaryBiology 2010, 10 :29 http://www.biomedcentral.com/1471-2148/10/29 ©2010Riberaetal;licenseeBioMedCentralLtd.ThisisanOpenAccessarticledistributedunderthetermsoftheCreativeCommons AttributionLicense(http://creativecommons.org/licenses/by/2.0),whichpermitsunrestricteduse,distribution,andreproductionin anymedium,providedtheoriginalworkisproperlycited.

PAGE 2

adaptationstotheharshenv ironmentalconditionsin cavesordeepsoil,suchaselongatedbodyandappendages,coldadaptation,modifiedlifecycles,modifiedfat metabolism,ordevelopment ofsensoryorgans[3,6,8] (see[9-11]forspecificexamplesofPyreneanLeptodirini).Subterraneanspeciesareusuallyconsidered “ super specialists ” ,whichcannotsurviveoutsidethenarrow rangeofhighlystableconditionsfoundintheirhabitats, andhaveverylimitedpotentialtodisperse[4].Asa result,thegeographicalrangeofsubterraneanspeciesis usuallyveryrestricted,inmanycasestoasinglecaveor akarsticsystem[8,12].Speciestraditionallyconsidered tohavewidedistributionshadoftenbeenshowntobe complexescomposedofmultiplecrypticlineageswhen molecularmethodswereapplied(e.g.[13,14]). Awidelyacceptedviewonlineageevolutionincave organismsisthatsubterraneanspeciesareevolutionary dead-endsthatdonotdisperseoncefullyadaptedtothe environmentofdeepcavesorsoil.Theyareproneto becomeextinctbeforebeingreplacedbylineagesnewly derivedfromepigeanspeciesthatre-colonisethesubterraneanmedium[4,15].Therearetwogeneralhypothesesregardingtheoriginofthesubterraneanfauna(see [4,16,17]forreviews):theclimaticrelictandthehabitat shifthypotheses.Briefly,undertheclimaticrelict hypothesis[18]thesubterraneanmediumactsasa refugeforepigeanfaunaintimesofunfavourableclimaticconditions.Thepopulationsthatareforcedintoa subterraneanhabitatbecomeisolatedfromtheirepigean relativesandeventuallyde velopthemorphologicaland physiologicaladaptationstothisnewenvironment.In lineageswithexclusivelysubterraneanspeciesthelack ofcloseepigeanrelativesisduetoextinctionofthelatter,leavingthesubterran eanspeciesastheonlysurvivorsthatescapedextinction[18].Theadaptiveshift hypothesis[19,20]surmisesthatthecolonisationofthe subterraneanmediumisdrivenbytheopportunityto exploitnewresources.Epigeanpopulationsarenot forcedbelowgroundbychangingconditions,andthere maybelimitedgeneticfluxbetweenthetwoenvironmentsforsometime.Bothscenariosassumemultiple originsofthesubterraneanlineagesfromcloselyrelated epigeanrelatives,withrecurrentcolonisationofthesubterraneanmedium. Thereis,however,littlephylogeneticevidencetosupportthesehypotheses,esp eciallyforthemostdiverse lineagesofexclusivelysubterraneanspecies,suchas somegroupsofbeetles.AmongColeopterathereare multipleexamplesofphylogeneticallyindependentcave orendogeangroups,inparticularinthefamiliesCarabidae,StaphylinidaeandLeiodidae[21,22].Manyofthese areeitherspeciespoorsubterraneanlineagesoramix ofhypogeanandepigeanspecies,withdifferentdegrees ofmorphologicalmodificationsandecological specialisation.Onlyafewgroupsappeartorepresent extensivemonophyleticradiationsofexclusivelysubterraneanspecies.Amongthem,thetribeLeptodiriniin thefamilyLeiodidae[21]i ncludesatpresentca.240 recognisedgeneraandca.1,800species[23],whichare almostexclusivelysubterranean.Thehighestdiversityis foundintheMediterraneanbasin,inparticularinthe northandeastoftheIberianpeninsula,someMediterraneanislands(CorsicaandSardinia),thesouthern Alps,Balkanpeninsula,RomaniaandsouthernRussia, theCaucasus,MiddleEastandIran[23].ThefewLeptodirinifoundeastofIran,andtheonlytwoNearcticspecies(inthegenus Platycholeus ),areofuncertain phylogeneticaffinities([24,25],seeDiscussion). ThemonophylyofthewesternPalaearcticLeptodirini isstronglysupportedbymorphologicalcharacters [24-26],buttheirinternalphylogenywasonlyrecently addressedwithnumericalmethods,andmostlywith morphologicalanalysisofth einternalstructureofthe malegenitalia,asotherexte rnalmorphologicalcharactersshowextremehomoplasy[25](seeFig.1foran overviewofthemorphologicaldiversityofthegroup). Theresolutionandsupporta ttainedwiththesecharactersetsis,however,verylimited,andwasunabletoprovidearobustphylogeneticframework. Weprovidehereacomprehensivemoleculardataset forthemainlineagesofLeptodirinipresentintheWesternMediterranean,includingtheIberianpeninsula (plusadjacentmountainmassifsinSouthernFrance) andSardinia.Theknownfaunaofthisregionincludes ca.40generawithsome230mostlyobligatorytroglobionticspecies,butalsosomeendogeanormuscicolous speciesinthegenera Bathysciola and Notidocharis [23] (Additionalfile1).Weaimtoestablisharobustphylogenytostudytheevolutiono fthisextensivesubterraneanspeciesradiation,andtoprovideatemporal frameworkforthediversificationofvariouslineagesand thecolonisationofthegeographicalareasinwhichthey occur.MethodsTaxonsamplingWefollowtheclassificationofLawrence&Newton[27] andPerreau[23]downtotribelevel.Allingrouptaxain thestudyarecurrentlyincludedinLeiodidae:Cholevinae:Leptodirini,althoughth eformalclassificationof Leptodiriniisinneedofaphylogeneticrevision[25]. Weusehere “ groupofgenera ” (or “ group ” forsimplicity)forlineagesfoundtobemonophyleticinouranalyses,toavoidconfusionwith thetraditionallydefined “ series ” whichintheFrenchschoolofmid-20thcentury entomologydidnotnecessarilyimplymonophylyinits modernsense[28].Wesampledrepresentativesofall majorlineagespresentintheWesternMediterraneanRibera etal . BMCEvolutionaryBiology 2010, 10 :29 http://www.biomedcentral.com/1471-2148/10/29 Page2of14

PAGE 3

Figure1 PhylogramobtainedwithMrBayeswiththeMAFFTalignmentincludingtheSardinianspecies,withthegeographicalareain whichthemainlineagesaredistributed(seeFig.2) .Blackcircles,wellsupportednodes(Bayesianpp>0.95,MLbootstrap>70%)forthefour measures(BayesianinMrBayes,MLinGarli,eachfortheMAFFTandPRANKalignments).Greycircles,goodsupportinatleasttwomeasures,the othersatleastbootstrap>50%orpp>0.5.SeeAdditionalfile3forthedetailedvaluesofsupport,andFig.2forthedistributionofthemain clades.Habitusofspecies,fromtoptodown:(1) Lagariellacolominasi (Zariquiey),(2) Salgadoiabrieti (Jeannel),(3) Troglocharinuskiesenwetteri (Dieck),(4) Perriniellafaurai Jeannel,(5) Phacomorphusfratyi (Dupré),(6) Pseudospeonomusraholai (Zariquiey),(7) Aranzadiellaleizaolai Español,(8) Bathysciolamystica Fresneda&Fery,(9) Quaestusarcanus Schauffus,(10) Spelaeochlamysehlersi Dieck,(11) Bathysciolazariquieyi Bolívar,(12) Notidocharisuhagoni (Sharp),(13) Catopsnigricans (Spence). Ribera etal . BMCEvolutionaryBiology 2010, 10 :29 http://www.biomedcentral.com/1471-2148/10/29 Page3of14

PAGE 4

(31outofthe42generaoccurringintheIberianpeninsula),plusthetwoendemicSardiniangenerausingdata from[29]andtwogenerafromtheCarpathiansas examplesofEasternlineages, Drimeotus and Pholeuon [23](Additionalfile1).NoLeptodiriniareknownfrom theBalearicislands.Sardiniahasthreeadditionalspecies currentlyincludedin Speonomus subgenus Batinoscelis andfourspeciesof Bathysciola ,andCorsicafivespecies of Parabathyscia ,agenusofthe Bathysciola series sensu Perreau[23]withAlpineandEasternEuropeanaffinities.MostWesternMediterraneanmissinggenera couldreadilybeassociatedtocladesofthesampledtaxa accordingtothestructureoftheaedeagus,whichwe showheretobeacharacterwithreliablephylogenetic information(see[25,30]foradetailedtaxonomic discussion). Thefinaldatamatrixincluded57speciesofLeptodiriniand8outgroupsfromdifferenttribesofCholevinae (Anemadini,Ptomaphagini,Cholevini;Additionalfile1) and Agathidium asamemberofthephylogenetically separatedsubfamilyLeiodinae[26,31].DNAextraction,amplificationandsequencingThespecimensusedinthestudywerepreservedin absoluteethanolinthefield.Afulllistofcollectorsand localitiesisgiveninAdditionalfile1.Extractionsofsinglespecimenswerenon-destructive,usingastandard phenol-chloroformmethodortheDNeasyTissueKit (QiagenGmbH,Hilden,Germany).VouchersandDNA samplesarekeptinthecollectionsoftheNaturalHistoryMuseum,London(NHM)andMuseoNacionalde CienciasNaturales,Madrid(MNCN)(Additionalfile1). Weamplifiedfragmentsofsevengenes,fivemitochondrialandtwonuclear:3 ’ endofcytochromecoxidasesubunit( cox1 );5 ’ endofthelargeribosomalunit plustheLeucinetransferplusthe3 ’ endofNADH dehydrogenasesubunit1( rrnl + trnL + nad1 );aninternal fragmentofcytochromeb( cob );5 ’ endofthesmall ribosomalunit,18SrRNA( SSU );andaninternalfragmentofthelargeribosomalunit,28SrRNA( LSU ).PrimersusedaregiveninTable1,andPCRprotocolsare givenin[32,33].Sequenceswereassembledandedited usingSequencherTM4.1.4(GeneCodes,Inc.,Ann Arbor,MI).Newsequenceshavebeendepositedin GenBank(NCBI)withAcc.NosGU356744-GU356993 (Additionalfile1).Intwocases( Quaestusnoltei (Coiffait)and Bathysciolaovata (Kiesenwetter))thefinal sequencewasacompositeoftwodifferentspecimensof thesamespecies(Additionalfile1). NospecimensfromSardiniawereavailablefor sequencing,butsequencesofthesamefragmentofthe cox1 geneusedhere(776overla ppingpositions)could beobtainedforseveralSardinianspeciesfromGenBank [29].Weexcludedsequencesreportedtobeobtained fromdrymaterialin[29],astheywereplacedinunlikelypositionsinthephylogenyorhadlargenumbersof autapomorphies,PhylogeneticanalysesWeusedtwodifferentapproachesformultipleprogressivepair-wisealignment,eitherwithsecondaryrefinementusingtheMAFFTonlinev.6andtheQ-INS-i algorithm[34]("MF ” inthefollowing)orwithmodelling theevolutionofindelswithPRANK[35]("PR ” ). Table1Primersusedinthestudy.F,forward;R,reverse.GeneNameSenseSequenceReference cox1 Jerry(M202)FCAACATTTATTTTGATTTTTTGG[72] Pat(M70)RTCCA(A)TGCACTAATCTGCCATATTA[72] ChyFT(A/T)GTAGCCCA(T/C)TTTCATTA(T/C)GTA.Cieslak,thiswork TomRAC(A/G)TAATGAAA(A/G)TGGGCTAC(T/A)AA.Cieslak,thiswork Tom-2RA(A/G)GGGAATCATTGAATAAA(A/T)CCA.Cieslak,thiswork cob CB3FGAGGAGCAACTGTAATTACTAA[73] CB4RAAAAGAAA(AG)TATCATTCAGGTTGAAT[73] rrnL nad1 16saR(M14)FCGCCTGTTTA(A/T)CAAAAACAT[72] 16SaRATGTTTTTGTTAAACAGGCG[72] 16SbRCCGGTCTGAACTCAGATCATGT[72] 16SAlf1RGCATCACAAAAAGGCTGAGG[74] ND1A(M223)RGGTCCCTTACGAATTTGAATATATCCT[72] 16SbiFACATGATCTGAGTTCAAACCGG[72] FawND1RTAGAATTAGAAGATCAACCAGC[72] SSU 5 ’ FGACAACCTGGTTGATCCTGCCAGT[32] b5.0RTAACCGCAACAACTTTAAT[32] LSU KaFACACGGACCAAGGAGTCTAGCATGNHM(2000) KbRCGTCCTGCTGTCTTAAGTTACNHM(2000) Ribera etal . BMCEvolutionaryBiology 2010, 10 :29 http://www.biomedcentral.com/1471-2148/10/29 Page4of14

PAGE 5

Bayesiananalyseswereconductedonacombineddata matrixwithMrBayes3.1.2[36],usingsixpartitionscorrespondingtothesixgenes,butcombiningtheshort trnL togetherwiththe rrnL gene.Evolutionarymodels wereestimatedpriortotheanalysiswithModelTest3.7 [37].MrBayesranfor5×10^6generationsusingdefault values,savingtreesateach500thgeneration. “ Burn-in ” valueswereestablishedaftervisualexaminationofa plotofthestandarddeviationofthesplitfrequencies betweentwosimultaneousruns. Wealsoconductedmaximumlikelihoodsearchesin Garliv.0.951http://www.bio.utexas.edu/faculty/antisense/garli/Garli.html,whichusesastochasticgenetic algorithm-likeapproachtosimultaneouslyfindthe topology,branchlengthsandsubstitutionmodelparameters[38].WeusedanestimatedGTR+I+ modelfor thecombinedsequence(asestimatedwithModelTest), andthedefaultsettings.Supportwasmeasuredwith 1,000bootstrapreplicates,reducingthenumberofgenerationswithoutimprovingthetopologynecessaryto completeeachreplicateto5,000.EstimationofmolecularratesanddatesofdiversificationFortheagecalibrationofthephylogenetictreeweused thepointofseparationbetweentheSardinianandthe continentallineages.TheSardinianclademostlikelyoriginatedbyallopatricseparationwhentheSardinian microplatessplitfromcont inentalEuropethroughtectonicmovements[29].Recen ttectonicreconstructions [39,40]allowbothadetailedgeographicalanalysisofthe relativepositionsofthemicroplatestothecontinent andfixingaminimumdatefortheoriginoftheSardinianclade.Thetectonicmovementsleadingtothe separationoftheWesternMe diterraneanmicroplates startednotbefore33MYago,andby25MYagothe separationfromIberiaandtheGulfofLyonwascomplete,althoughsomeconnectionswiththeItalianpeninsulamayhavepersisted[40].However,giventhelimited dispersalamongsubterraneanpopulationsthegeological uncertaintyastotheexistenceofremnantlandbridges arelikelytobeunimportantforthedatingofthe separationofLeptodirini.Theuseofahardbound minimumagecalibrationof25MYasthelatestpossible populationsplitwouldbiastheestimationstowards youngerages[41],soweusetheearlierdateof33MY asthemostplausibletimeofseparationbetweenthe Sardiniancladeanditssister,whichmaystillbeconservative(toorecent)ifthesep arationofthelineagespredatedthesplitofthemicroplates. BecauseavailablesequencesforthespeciesoftheSardiniancladewerelimitedtothe cox1 marker(seeabove) weconductedtheBayesiantreesearchesincludingor excludingthesequencesoftheSardinianspecies,totest fortheeffectofmissingdata.Thetopologiesofthe resultingtreeswereidentical(seeResults).Wetherefore usedthetopologyobtainedwiththeBayesiananalyses forthewholedatasetwiththePRalignment(whichwas fullycompatiblewiththatobtainedwithMF,butmore resolved;seebelow),toapplytwomethodstoobtainan ultrametrictree:Bayesianes timations,asimplemented inBeast1.4.7[42],andpenalizedlikelihood[43],as implementedinr8s1.71http://ginger.ucdavis.edu/r8s/. Weusedthebranchlengthscorrespondingtothesingle cox1 sequence(pruningthespecieswithmissing cox1 fromthetree),withaGTR+I+ modelwithparameters estimatedinPAUP[44]. FortheBayesiananalysesinBeast,wellsupported nodesintheanalyseswithth ecombinedmitochondrial andnuclearsequencewereconstrainedtobemonophyletic(seeResults),andaGTR+I+ modelwasenforced withanuncorrelatedlognormalrelaxedclockanda Yulediversificationmodel[42].Allparameterswereset todefaultvalues,withtheexceptionofthepriorofthe ageofthecladeformedbytheSardinianspeciesand theirsister,whichwassettoanormaldistributionwith meanof33MYandastandarddeviationof2.0MY (correspondingtoa95%confidenceintervalof28.8and 36.3MY)toallowforvariancefromstochasticsampling errorofnucleotidechanges[41].Theresultsoftwo independentrunsweremergedwithTracerv1.4and TreeAnnotatorv1.4.7[42]. Inr8sweusedtheTruncatedNewtonalgorithmand performedacross-validationprocedurewithsmoothing factorsbetween1and500(aftereliminationoftheoutgroups)[43].Oncetheoptimalsmoothingfactorwas found,theprogramwasrunagainusing33MYasa minimumconstraintforthen odeincludingtheSardiniancladeanditssister.Aconfidenceintervalofthe rateswasestimatedusingtheprofilingoptioninr8s withthetopologiesofthelast1,000treesinoneofthe MrBayesruns,withbranchlengthsfor cox1 estimatedin PAUPasabove.Notethatbecausethebranchlengths arenotthoseoriginallyestimatedinMrBayesforthe wholecombinedsequence,butthoseestimatedinPAUP for cox1 alone,thestochasticvariationreflectedinthe confidenceintervalistheresultofthedifferenttopologiesofthe1,000treesused,plusthevariationintroducedsubsequentlybyratesmoothinginr8s. Totestforpossibleartefactsduetotheinclusionof onlythe cox1 sequence,andtoobtainaratecalibration forothermitochondrialgenes,weperformedadditional analysesinBeastincludingallthemitochondrial sequencesbutexcludingtheSardinianspecies.The nodebelowthatoftheinsertionoftheSardinianspecies wasconstrainedtohavetheageestimatedwith cox1 alone,withanormaldistributionwithstandarddeviationof2.0MY.Beastwasrunfor5×10^6generations, savingtreesatevery500thgeneration.Ribera etal . BMCEvolutionaryBiology 2010, 10 :29 http://www.biomedcentral.com/1471-2148/10/29 Page5of14

PAGE 6

ResultsPhylogenyofWesternMediterraneanLeptodiriniTheprotein-codinggenes( cox1,cob,nad1 )hadno indels,withthesingleexceptionofathreetofour nucleotideinsertionbeyondtheendofthe cox1 gene(in theregionofthe trnL )inthe Quaestus groupofgenera andin Aranzadiella .Mostofthelengthvariationwas concentratedinthehypervariableregionsofthe LSU fragment,whileforthe rrnL and SSU genesthelength variationbetweenthetwoalignmentmethodsused(MF andPR)wasrelativelysmall(Table2).Forthe LSU gene thePRalignmentwasmorethan1,000positionslonger thantheMFalignment,withalossof79informative characters(Table2).Despitethelargedifferencesinthe lengthofthecombinedsequencebetweenthetwoalignmentmethods(mostlydueto LSU ),thechangesinthe topologyorlevelofsupportoftheresultingtreeswere negligible(seebelow). Theoptimalevolutio narymodelwasGTR+I+ ,both forthecombineddataandtheindividualgenefragments.ThetwoindependentBayesianrunsreached valuesofthestandarddeviationofthesplitfrequencies betweenthemofca.0.005forboththePRandMF alignments.Theestimatedparametersforthetwoalignmentswerewithinthe95%confidenceintervalsofeach other,withthesingleexceptionofalpha( a )ofthe SSU and LSU partitions,andtheprobabilityofinvariantsites (I)ofthe LSU partition(Additionalfile2).ThetopologiesofthetreesobtainedwithBayesianprobabilitiesfor thetwoalignmentsandthoseobtainedwithMLin Garliwereeitheridenticalorcompatiblewitheach other(i.e.consensusconsistentwithresolvednodes). Thedifferencesintreetopologyonlyaffectedtherelationshipsofoneoftheoutgroups( Speonemadusangusticollis Kraatz)andtheinternalphylogenetic relationshipsoftheCarpathi anclade(Fig.1,Additional file3).Nodalsupportwasgenerallyhigh,withmostposteriorprobabilityvalues>0. 95andmaximumlikelihood boostratpvalues>70%(Additionalfiles3,4),withgenerallyhighersupportvalues atsomedeepernodeswith Bayesianmethods(Additionalfile3),ashadbeen observedinotherstudies[45]. TheexclusionoftheSardinianspecies(duetomissing data,seeMethods)inaBayesiananalysisusingtheMF alignmentproducedatopologyidenticaltothat obtainedwithallspecies(Additionalfile4). AllspeciesofLeptodiriniformedawellsupported clade,totheexclusionofothertribesofCholevinae (Fig.1).WithintheLeptodirini,wellsupportedmajor lineageswerecomposedofspeciesrestrictedtoparticularbiogeographicalregions(Figs1,2).Inmostcasesthe mainlineageswerefullyallopatric,asseparateclades wereconfinedtothePyrenees( Speonomus groupof genera),theCantabrianMountains( Quaestus and Speonomidius groups),thecentralMediterraneancoastof Iberia( Spelaeochlamys group),Sardinia( Ovobathysciola group),andtheCarpathians.Inaddition,twospeciesof Bathysciola , B.zariquieyi Bolívardistributedthroughout thecoastalmountainsystemsinCatalonia[46]and B. ovata ,withawidedistributioninPyreneesandSE France,constitutephylogen eticallyisolatedlineages separatedfromothermembersofthegenus(Figs1,2). ThemonophyleticPyreneanlineage,correspondingto the Speonomus seriesofFresneda etal .[25],wassister tothegenerafromtheCarpathians(Fig.1),andwas splitintwomainlineagesdefinedbynodes2and5in Fig.1.Theformerincludesgenerafromthelowerwesternareas(node3,BasqueCountryandNavarra)plus someendogeanandmuscicolous Bathysciola witha widerdistributiontowardstheeast.Thecladedefined bynode5includesgenerafromtheeasternlowland areas(node4, Parvospeonomusand Pseudospeonomus ) plusalargecladeofgenerafromthecentralPyrenees (node7).Thelattercladewasalsogeographicallystructured,withgeneraroughlydistributedwest(node6)and east(node8)oftheriverGállegointhecentralPyrenees (Figs1,2).Thegenus Troglocharinus (includedinnode 8)hasadisjunctdistribution,withsomespeciesdistributedinthesoutherncoastalrangesofCatalonia(Fig. 2b;[30]),althoughthemonophylyofthegenuscould notbetestedasthesouthernspeciesweremissingfrom ourstudy. TheremainingLeptodiriniconsistedofseveralhighly supportedcladeswithverywelldefineddistributions: (1)the Quaestus groupintheCantabrianmountains; (2)the Spelaeochlamys groupintheMediterranean coastofSpain;(3)the Ovobathysciola groupinSardinia; (4)thedivergentspecies Bathysciolazariquieyi fromthe Catalonianmountains;(5)the Speonomidius groupof Cantabria;and(6) Bathysciolaovata fromthePyrenees (Figs1,2a).Therelationshipsamongtheselineages wereingeneralpoorlyresolved,althoughthesisterrelationshipbetweenthe Quaestus and Spelaeochlamys groups,andof Bathysciolazariquieyi andtheSardinian Table2Lengthofthesequencedfragments,with maximumandminimumlengthandnumberof nucleotidesinthematrix(No.)andnumberof informativecharacters(Inf.)inthetwoalignments.RawlengthMAFFTPRANK geneminmaxNo.Inf.No.Inf. cox1 826830830380830380 rrnL + trnL 668705727248756257 nad1 378378378176378176 cob 358358358176358176 SSU 5916066476961256 LSU 5797749232522141173 Ribera etal . BMCEvolutionaryBiology 2010, 10 :29 http://www.biomedcentral.com/1471-2148/10/29 Page6of14

PAGE 7

clade,werewellsupportedinsomeBayesiananalyses (Additionalfile3). TemporaldiversificationoftheWesternMediterranean Leptodirini ThesplitbetweentheSardiniancladeanditsmainland sister Bathysciolazariquieyi wasusedforestimatingthe evolutionaryrateof cox1 ,usingacalibrationpointof33 MYagoasthetimeofvicariantseparationofboth lineages(Fig.1).Whenusing Beast,therateestimate was0.020+/-0.005substitutionspersiteperMY,i.e. anoverallpair-wisedivergenceof4.0%perMY(Table 3,Additionalfile5).Thenodethatgrouped B.zariquieyi plustheSardiniancladewiththeirsisterswas estimatedtodatebacktoca.38+/-7MYago.This constraintwassubsequentlyusedfortheestimationof theevolutionaryrateforthecombinedmitochondrial genes(excludingthespeciesoftheSardiniancladefor whichthesedataweremissing).Therateofthefive combinedmitochondrialmarkers( cox1 , rrnL , trnL , nad1 and cob ,comprising2,293bpintheMFalignment; Table2)was0.010+/-0.002,equivalenttoanoverall pairwisedivergenceof2.0%perMY(Table3).Theestimationofnodeagesusingthecombinedmitochondrial Figure2 DistributionofthemainlineagesofWMediterraneanLeptodiriniasestablishedinthisstudy .(a),Generaldistributionofthe mainlineages(seeFig.1);(b)detaileddistributionofthe Speonomus groupofgenerainthePyrenees(seeFig.1). Ribera etal . BMCEvolutionaryBiology 2010, 10 :29 http://www.biomedcentral.com/1471-2148/10/29 Page7of14

PAGE 8

markerswasverysimilartothatobtainedusing cox1 only,andfullywithinthe95%confidenceintervalof eachother(Table3,Fig.3,Additionalfiles3,6). Forcomparisonweestimatedtheageofthenodeswith thePenalizedLikelihood(P L)methodinr8s[43],using cox1 onlyandacalibrationdateof33MYfortheSardinian-continentalsplit.Estimatednodeagesthroughout wereverysimilartothoseob tainedwithBayesianmethods,andwellwithintheir95%confidenceinterval,with thesoleexceptionofthenodedefiningthesisterrelationshipbetween Antrocharis and Gesciella (olderwiththePL estimation;Table3andAdditionalfile6).However,the estimatedoverallratewas0 .0115+/-0.0005(2.3%per MY,Table3),i.e.lowerthantheestimateobtainedwith Beastforthe cox1 gene,suggestingsubstantialdifferences inthebranchlengthestimatesfrombothmethods. ThemainlineagesofWesternMediterraneanLeptodiriniwereestimatedtohavediversifiedbetweenca.4533MY,duringtheMidtoLateEocene(Fig.3).The relativeageofthenodessuggeststhatbythetimethe Sardinianlineageformed(presumablybyLateEoceneEarlyOligocene)allothermainlineageswerealready presentanddistributedovertheirpresentgeographical areas.Alllineageswithawelldefinedgeographicaldistributionwereestimatedto haveoriginatedbytheLate Oligocene(Figs2,3).Theageofdiversificationwithin thegeneraforwhichmorethanonespecieswas sequencedrangedfromca.20to5MY(mostlyMiocene),disregardingsomecaseswherethecurrently recognisedgenerawerefoundtobeparaphyletic( Euryspeonomus , Quaestus )orpolyphyletic( Speonomus , Bathysciola )(Fig.1). Table3Estimatedratesofmolecularevolutionandagesoftheconstrictednodes,withconfidenceintervals.Beast,mitochondrialBeast, cox1 r8s, cox1 meanloweruppermeanloweruppermeanlowerupper Rate0.00990.00810.01160.02040.01510.02600.01150.01130.0117 1root44.1637.4650.7645.4135.8756.0339.62.. 2 Sardinianclade[ cox1 ] n.a.n.a.n.a. 32.7528.9636.7633.00 .. 3 Sardinianclade[mt]37.7933.8641.84 37.9231.4345.4336.92.. 4infraflagellates41.1835.6046.8241.4233.0450.2836.9220.6939.09 5 Speonomus group33.6127.6739.8236.3227.5346.2330.4522.8132.50 6 Speonomidius group18.5113.7023.6922.6714.3731.3120.70.. 7 Spelaeochlamys group15.0411.3118.9813.639.0818.3215.97.. 8 Quaestus group29.2024.0634.3730.5423.0337.9431.11.. 9node117.5913.3922.1217.5611.0024.2318.02.. 10node224.1619.4729.1223.5917.2930.5821.88.. 11node322.5918.0327.2322.4116.4629.0421.88.. 12node419.4014.6424.4722.3414.6930.3220.80.. 13node527.9422.9533.3431.4623.3939.9026.94.. 14node621.1817.0625.6323.2917.0029.9421.24.. 15node725.0820.7129.8926.9520.2834.1825.08.. 16node821.2217.2825.3721.4416.0627.2919.63.. 17 Aranzadiella + Speocharidius 10.097.3912.898.995.8112.529.85.. 18 Euryspeonomus + Josettekia 18.2714.2022.8918.0712.6224.2918.92.. 19 Phacomorphus Bellesia 16.1812.7619.7516.7011.8321.7315.84.. 20 Phacomorphus 5.704.117.375.713.637.884.83.. 21 Perriniella + Ceretophyes 14.7210.8618.6517.9312.0024.1217.37.. 22 Gesciella Antrocharis 9.456.5612.4212.117.3217.0813.72.. 23 Speonomites 9.967.3612.659.616.2513.169.49.. 24 Gesciella Stygiophyes 18.3814.9222.1719.7414.6925.3618.95.. 25 Speonomites Stygiophyes 13.0610.5515.6312.489.2815.9612.26.. 26 Stygiophyes Speonomus 11.719.5214.0411.218.3214.3211.67.. 27 Speonomus 9.857.6812.119.696.9612.6710.37.. 28 Stygiophyes Troglocharinus 10.708.6712.839.767.1412.4910.59.. 29 Stygiophyes Salgadoia 7.856.189.597.675.4610.118.61.. 30 Stygiophyes + Lagariella 5.634.327.105.553.707.486.48..SeeFig.3&Additionalfile5forthedefinitionofthenodes.Inbold,nodesusedforthecalibration(seeMethods).Ribera etal . BMCEvolutionaryBiology 2010, 10 :29 http://www.biomedcentral.com/1471-2148/10/29 Page8of14

PAGE 9

DiscussionThemostprominentfeatureofthephylogenyofthe WesternMediterraneanLeptodiriniisthestronggeographicaldivisionamongthemainlineages.Supportfor thesegeographicallyrestrictedcladesisgenerallyvery strongandstabletochangesinalignmentorreconstructionmethods,whiletherelationshipsamongthemis lesswellsupported.Allwellsupporteddeepcladeswere restrictedtoasinglemountainmassif,andeachofthese isoccupiedbyasinglecladewiththesoleexceptionsof theCantabrianmountains,withtwoseparatelineages (the Quaestus and Speonomidius groupsofgenera)and somespeciesof Bathysciola inthePyrenees.Inboth casesoneofthesesympatriclineagesiscomposedof cavespeciesonly( Quaestus and Speonomus groups) whiletheotherismuscicolousorendogean( Speonomidius and Bathysciola fortheCantabrianmountainsand thePyreneesrespectively)(seeAdditionalfile1forthe Figure3 UltrametrictreeobtainedwithBeastusingthecomb inedsequence,excludingtheSardinianspecies .Blackcircles,well supportednodes(seeFig.1,Additionalfile3)constrainedtobemonophyletic.Numbersinsidenodes,ageestimate(MY),usingtheseparation of Bathysciolazariquieyi fromitssister(rednode),withapriorageof37.9MY(seetextandAdditionalfile5). Ribera etal . BMCEvolutionaryBiology 2010, 10 :29 http://www.biomedcentral.com/1471-2148/10/29 Page9of14

PAGE 10

habitatofthestudiedspecies).Itwouldbeinterestingto comparetheratesofdiversificationofthedifferent lineagesinthesamegeographicalarea,andtheoccurrenceofspeciesinsympatrytotestforthepossible effectofcompetitiveexclusionwithineachhabitattype.PhylogenyofWesternMediterraneanLeptodiriniOurresultsbasicallyagreewiththetraditionaltaxonomicdivisionofLeptodiriniin “ infraflagellates ” [25] and “ supraflagellates ” [8,28],basedonthecomplex structurethatischaracteristicoftheaedeagusinthe former(the “ Ypiece ” [25]).InourstudytheinfraflagellatescorrespondtothePyreneanandtheCarpathian species,albeitsupportedon lyintheBayesianinference andwiththeexclusionofsomespeciesof Bathysciola . Thepolyphylyofthegenus Bathysciola hasalreadybeen established(see[25]),buttheinclusionofsomespecies amongthe “ supraflagellates ” impliesmultipleindependentoriginsofacomplexaedeagus.Alternatively,the presenceofthe “ Ypiece ” couldbetheancestralconditionofLeptodirini,whichunderwentrepeatedsimplificationsinatleasttheCantab rianandSardinianspecies ofthe Ovobathysciola group(Fig.1),contrarytoexpectations(e.g.[8,26,28]). Ourresultsalsosupportmostofthetraditional “ phyleticseries ” ofJeannel[8,28],whileinFresneda etal . ’ s [25]studyusingcharactersoftheexternalmorphology andtheinternalgenitalia, supportwasfoundonlyfor themonophylyofthe Speonomus (including Bathysciola ) and Spelaeochlamys plus Ovobathysciola series.The molecularanalysesalsosupportedthetraditionally defined Speonomidius , Quaestus and Spelaeochlamys series[30,47-49].The Speonomidius seriesincludesthe genus Notidocharis [47],withsevenmuscicolousrather thansubterraneanspecies,allwithreducedeyes(Additionalfile1).Othermuscic olousoculatedLeptodirini includesome Bathysciola -liketaxainAnatolia,theCaucasusandnorthIran,plus Adelopsellabosnica (Reitter) fromtheBalkansandtwoNearcticspeciesinthegenus Platycholeus [50].Thegenera Adelopsella , Platycholeus and Sciaphyes (tribeSciaphyinifromeastSiberia[23]) sharesomeuniquecharactersnotpresentintheremainingLeptodirini[8],whichsuggeststheymaynotbecloselyrelated.AccordingtoFresneda etal. [25]the presenceofeyesplaced Notidocharis assistertotherest ofstudiedLeptodirini.Inthecurrentanalysisthispositionisnotstronglycontradicted,asthesupportofthe basalnodesofLeptodiriniinthetreesobtainedwith MLwasveryweak(Fig.1,Additionalfile3). ThePyreneancladecorrespondstotheSpeonomus seriesasdefinedbyFresneda etal .[25]withtheexclusionofsomespeciesof Bathysciola ,butconfirmingthe inclusionofthegenus Pseudospeonomus (= Pseudochlamys Comas),distributedintheextremeeastofthe Pyreneesandpreviouslygroupedwithspeciesinthe mountainsoftheMediterraneancoastofSpain[25]. TheSardinianspeciesofthe Ovobathysciola group, althoughgroupedwiththe Mediterraneanspecies( Spelaeochlamys group)instudiesofmorphology[25,29], seemmorecloselyrelatedto Bathysciolazariquieyi accordingtotheDNAdata.ThesupportforthisrelationshipwaslowunderML,possiblyduetomissing data(onlythe cox1 sequencewasavailablefortheSardinianclade[29]),butreasonablyhighwithBayesian methods(pp>0.9forboththeMFandPRalignments). Bathysciolazariquieyi hasarelativelywidedistribution intheCataloniancoast,fromsouthofBarcelonatothe provinceofGirona[46].TheCatalonia-Sardinialinkis moreinagreementwithgeologicalreconstructions[40], accordingtowhichtheSardinianmicroplateswerelast connectedtothemainlandinthenorth(Frenchcoast), ratherthanthesouthIberia nMediterraneancoast,as previouslyassumed[29]. AccordingtoourresultstheLeptodirinifaunaofthe Iberianpeninsulaismostlikelynotmonophyletic.The PyreneancladeseemsmostcloselyrelatedtosomeEasternEuropeanlineages,andthelineageoftheMediterraneancoast( Spelaeochlamys groupofgenera)isrelated totheSardinian Ovobathysciola groupandtosomespeciesnorthoftheEbrobasinandinsouthernFrance.CalibrationTheuseofmoleculardatatoestimatetheagesofextant taxahasrevolutionisedmanyaspectsofevolutionary biology[51,52],despitetheirknownlimitationsand shortcomings[51,53].Thereis,however,agreatscarcity ofgoodreferencecalibrationsformanygroups,andthe “ standard ” arthropodmitochondrialrateof2.3%perMY [54]isoftenuseduncriticallyforavarietyofgenes,ages andorganisms.Subterraneanspecieshavebeenusedto calibratemolecularphylogen ies,astheirdistributions presumablyaremorelikelyt oreflectancientvicariant eventsduetotheirlowdispe rsal(e.g.[29]).However, theseancientvicariantscenarioscouldbeconfounded byoccasionaldispersalofminuteendogeanspecies,or inlineageswhereabove-grounddispersalofepigean formsprecededtheshifttothesubterraneanenvironment.Ourcalibrationsdependonthevicariance betweentheSardinianandtheEuropeanplates,i.e.the presenceofacommonancestorinthisregionpriorto thesplitoftheselandmasses.Strictly,wecannotexclude anindependentoriginofsubterraneanhabitsintheSardinian Ovobathysciola group,ordriftingofa Bathysciola -likeendogeanancestortotheislandsubsequentto itsseparationfromthecontinent,buttheseseemtobe unlikelyscenarios.Usingthiscalibrationpoint,theaveragerateforthesequencedmitochondrialgenes( cox1 , cyb , rrnL and nad1 )was2%perMY,surprisinglycloseRibera etal . BMCEvolutionaryBiology 2010, 10 :29 http://www.biomedcentral.com/1471-2148/10/29 Page10of14

PAGE 11

tothestandard2.3%[54],andinagreementwithresults fromothercalibrationsforbeetles[12,55-57].The cox1 genealoneproducesatwofoldfasterrate(4%)comparedtotheaveragedmitochondrialrateincludingthe slower rrnL and trnL genes,alsoinagreementwithpreviousresults(e.g.[58,59]).Theageestimationsusing Beastandr8s,whetherusing cox1 orthecombinationof severalproteincodingandribosomalmitochondrial genes,wereremarkablysimilar(Table3,Additionalfile 6),despitedifferencesintheestimationoftheratefor the cox1 genebetweenr8sandBeast.Asnotedabove, thismayberelatedtothedifferencesinthemethod usedforthecomputationofthebranchlengths(BayesianmethodsinBeast,penalizedlikelihoodinr8s [42,43]).EvolutionoftheWesternMediterraneanLeptodiriniTheestimationsofnodeagesdatetheoriginofthemain lineagesofLeptodirinitotheEarlyEocene,withthe initialdiversificationtaki ngplaceduringtheOligocene (ca.35-20MY,Fig.3).Ours amplingdoesnotinclude somelineagesofLeptodirini includinggenerafromthe Alps,BalkansormainlandItaly,butthesearemostly “ infraflagellates ” ,[26,28],i.e.likelytobeembeddedinthe existingclades.Duetothef indingoftheCarpathianas sistertothePyreneanlineage,andtheinclusioninour datasetoftheoculated,muscicolous Notidocharis ,itis unlikelythatthesemissinglineagescouldmovetheoriginofthelineagesfurtherbackintime. Thediversificationofth emainlineagesduringthe PaleogenecoincideswiththeclosureoftheTethysSea andthecollisionoftheIberianandEurasianplates[39]. Underthisscenario,theoriginofthePyreneanlineage ofLeptodiriniat34+/-6MYagofollowedshortlyupon theformationofthePyreneesduringtheAlpineorogenywhichwaslargelycompletedbytheEarlyOligocene[60,61].ThecolonisationofthePyrenean subterraneanmediumthereforetookplacewithout delayaftersuitablehabitatwasavailable,i.e.afterthe karstificationbywatererosionoftherecentlyraisedcalcareousmassifs.Thisviewisstrengthenedbythefact thatthebasalcladogeneticsplitsinvolvethespeciesat bothedgesofthechain(nodes2and4inFig.3,see Fig.2b),withsubsequentsplitsincludingspeciesmore centrallydistributed(node6,plus Ceretophyes and Perriniella ),andthespeciesinthecentralpart,withthe highestelevations,amongthemostrecentlyderived lineages(node8).Thiscentri petalearlycolonisation, completedca.10+/-3MYago(Fig.3),wasfollowedin somecasesbyanintra-genusdiversificationineachof thevalleysormountainma ssifstowhichthesegroups arecurrentlyconfined.Theoriginofthemainlineages ofLeptodiriniintheCantabrianmountainsandthe Mediterraneancoastappear sslightlyolderthaninthe Pyrenees.Thisisinagreementwithalikelyolderageof theavailablehabitat,asthesemountainareaswerepresentintheIberianplatebeforetheformationofthePyrenees[39].TheOligocenetomidMioceneoriginofthe mainlineages(abovewhatiscurrentlyrecognizedatthe levelofgenus)isolderthanforotherlineagesofsubterraneanEuropeanfaunawhoseoriginwasplacedinthe LateMiocenetoPleistoce ne([62,63]andreferences therein),includingthePyreneanradiationofsubterraneanTrechinigroundbeetles[22]. Thestronggeographicals tructureofsubterranean taxaatmultiplehierarchicallevelsisacommonpattern alsoevidentinotherrecentstudies(seee.g.[64,65]for Crustacea;or[12,22]forColeoptera).Eachwelldefined biogeographicalunitisoccupiedbyasinglemonophyleticlineage,whileanyphylogeneticsubstructurewithin suchgroupsfrequentlyreflectsthegeographicalsubdivisionofthesewiderareas,inmanycasescontraryto expectationsfrommorphologicalsimilarities.Insome casesthesegeographically confinedlineagesinclude bothsubterraneanandepigeanspecies,e.g.stygobiontic Crustacea[66,67],spidersandbeetlesinthetribeTrechiniintheCanaryIslands[56,68],orstygobionticdivingbeetlesinAustralia[12 ].Theseexamplesprovide strongevidenceformultiplecolonisationofthesubterraneanmedium,astheyincludelineagesthatexhibitdifferentdegreesofmorphologicalmodifications.Their adaptationstothesubterraneanenvironmentisin accordancewiththeadaptiveshifthypothesis[19],that invokesmultipleindepende ntcolonisationsofthesubterraneanmediumandcontinuedsurvivaloftheepigean ancestors(seeIntroduction).IntheWesternMediterraneanLeptodirini,theonlylineagewithnon-subterranean,oculatedmuscicolousspecies( Notidocharis )is placedinanambiguouspositionclosetotheoriginof thewesternMediterraneanli neages,i.e.hasbeenseparatedfromthetroglobiticlineagesformorethan40MY ago.Allothermajorlineages,withOligoceneorEarly Mioceneorigin,includeonlyanophthalmous,endogean, interstitialorcavespecies(A dditionalfile1).Thispatternstronglysuggeststha teachofthegeographical lineagesdiversifiedafterthec olonizationofthesubterraneanmediumtookplace,assuggestedforotherterrestrial[15,22,69,70]andstygobiontic[71]groups.This isindisagreementwithboththeclimaticrelictandthe adaptiveshifthypotheses,whichassumemultiplecolonisationsofthesubterraneanmediumbycloselyrelated epigeanancestorsandperceivethetroglobitictaxaas evolutionarydeadends[18,20].ConclusionsWehaveshownherethatthemainlineagesofWestern MediterraneanLeptodirinihaveanoriginintheEarlyMidOligocene,andmostlikelydevelopedallRibera etal . BMCEvolutionaryBiology 2010, 10 :29 http://www.biomedcentral.com/1471-2148/10/29 Page11of14

PAGE 12

modificationstothesubterraneanlifeinthemaingeographicalareasinwhichtheyarefoundtodaypriorto theLateOligocene-Miocene.Thediversificationwithin eachofthesemaingeographicalareasseemsthusto havetakenplacesinceearlyMiocenefromancestors fullyadaptedtothesubterraneanmedium,contraryto mostcurrentassumptionsabouttheevolutionofthe subterraneanfauna.Additionalfile1: Specimensusedinthestudy,withlocality,collector, voucherreferencenumbersandaccessionnumbersforthesequences.In grey,specimenscombinedinacompositesequence. Clickhereforfile [http://www.biomedcentral.com/content/supplementary/1471-2148-1029-S1.XLS] Additionalfile2: EstimationoftheparametersfortheBayesiananalyses. Clickhereforfile [http://www.biomedcentral.com/content/supplementary/1471-2148-1029-S2.XLS] Additionalfile3: CladogramobtainedwithMrBayeswiththeMAFFT alignmentincludingtheSardinianspecies,withdetailednodesupport values.Upperrow,MrBayesposteriorprobabilities,MF/PR;lowerrow, bootstrapsupportvalues(1,000replicas)inGarli,MF/PR. Clickhereforfile [http://www.biomedcentral.com/content/supplementary/1471-2148-1029-S3.PDF] Additionalfile4: PhylogramobtainedwithMrBayeswiththeMAFFT alignment,withtheexclusionoftheSardinianspecies. Clickhereforfile [http://www.biomedcentral.com/content/supplementary/1471-2148-1029-S4.PDF] Additionalfile5: UltrametrictreeobtainedwithBeastusingthe cox1 sequenceonly,includingtheSardinianspecies(redclade).Blackcircles, wellsupportednodes(seeFig.1,Additionalfile3)constrainedtobe monophyletic.Numbersinsidenodes,ageestimate(MY)usingthe separationoftheSardinianspecieswithapriorageof33MY(seetext). Clickhereforfile [http://www.biomedcentral.com/content/supplementary/1471-2148-1029-S5.PDF] Additionalfile6: AgeestimationsofthenodesinFig.3andAdditional file5.Verticalaxis,estimatedage(MY).Horizontalaxis,nodes.White circles,estimationusingthecombinedmitochondrialgenes( cox1 , rrnL + trnL , nad1 , cob )inBeast,with95%confidenceintervals(dashedlines) (Fig.3).Blackcircles,estimationusingonly cox1 inBeast,with95% confidenceinterval(solidline)(Additionalfile5).Redcircles,estimation usingthegen cox1 inr8s.Estimationsusing cox1 alonewerecalibrated withthenode “ Sardinianclade[cox1] ” ,andthoseusingthecombined mtDNAwiththenode “ Sardinianclade[mt] ” .Notethanintheestimation usingthecombinedmtDNAgenestheSardinianspecieswerenot included. Clickhereforfile [http://www.biomedcentral.com/content/supplementary/1471-2148-1029-S6.PDF] Acknowledgements WeparticularlythankV.Assing,M.Baena,C.Bourdeau,A.Castro,Ph.Déliot, F.Fadrique,A.Faille,C.HernandoandI.Zabaleguiforsendingmaterialfor study.WealsothankX.Bellés,C.Bourdeau,A.Casale,A.FailleandC. Hernandoformultiplediscussionsandcommentsonthesystematicsand evolutionofthesubterraneanfauna,andFaunaIbéricaforpermissionto reproducesomeofthehabitusdrawingsinFig.1.Thecommentsofthree Refereeshelpedtoimproveanearlierversionofthemanuscript.Thiswork hasbeenfundedbyNERCgrantstoAPV(labworkattheNHM)andprojects CGL2006-11403andCGL2007-61943toAC. Authordetails1MuseoNacionaldeCienciasNaturales,JoséGutiérrezAbascal2,28006 Madrid,Spain.2InstituteofEvolutionaryBiology(CSIC-UPF),PasseigMaritim delaBarceloneta37-49,08003Barcelona,Spain.3CadeMassa,E-25526 Llesp,Lleida,Spain.4NaturalHistoryMuseum,CromwellRoad,LondonSW7 5BD,UK.5ImperialCollegeLondon,SilwoodParkCampus,AscotSL57PY, UK.6DepartamentodeBiologíaAnimal,FacultaddeBiología,Universidadde León,León,Spain. Authors ’ contributions IR,ACandJFdesignedthestudy.JFandJMSobtainedandidentifiedthe material.IR,RB,AIandACdidthemolecularworkandobtainedthe sequencedata.IRdidthephylogeneticanalyses.IRandACwroteafirst draft,whichfinalversionwascompletedbyIR,AC,APV,JFandJMS.All authorsreadandapprovedthefinalmanuscript. Received:14September2009 Accepted:28January2010Published:28January2010 References1.SchluterD: Theecologyofadaptiveradiations. Oxford,UK:Oxford UniversityPress2000. 2.CoyneJ,OrrHA: Speciation. Sunderland,MA:SinauerAssociates2004. 3.VandelA: Labiologiedesanimauxcavernicoles. Paris:Gauthier-Villars 1964. 4.CulverDC,PipanT: TheBiologyofCavesandOtherSubterranean Habitats. Oxford,UK:OxfordUniversityPress2009. 5.DarwinC: TheOriginofSpecies. London:JohnMurray1859. 6.CulverDC: CaveLife:EvolutionandEcology. Cambridge,MA:Harvard UniversityPress1982. 7.LamarckJB: Philosophiezoologique,ouexpositiondesconsidérations relativesàl ’ histoirenaturelledesanimaux. Paris:Museumd ’ Histoire Naturelle,Paris1809, 1. 8.JeannelR: MonographiedesBathysciinae. ArchZoolexpgen,Paris 1924, 63 :1-436. 9.DeleuranceS: RecherchessurlesColéoptèrestroglobiesdelasousfamilledesBathysciinae. AnnScinat,Paris(Zool)ser12 1963, 1(5) :1-172. 10.DelayB: Milieusouterrainetecophysiologiedelareproductionetdu developpementdescoléoptèresBathysciinaehypoges. MémBiospél 1978,1-349. 11.Corbiere-TichaneG,LoftusR: Antennalthermalreceptorsofthecave beetle, Speophyeslucidulus Delar.2.Coldreceptorresponsetoslowly changingtemperature. JCompPhysiol 1983, 153(3) :343-351. 12.LeysR,WattsCHS,CooperSJB,HumphreysWF: Evolutionofsubterranean divingbeetles(Coleoptera:Dytiscidae:Hydroporini,Bidessini)inthearid zoneofAustralia. Evolution 2003, 57 :2819-2834. 13.TronteljP,DouadyCJ,FiserC,GibertJ,GorickiS,LefebureT,SketB, ZaksekV: Amoleculartestforcrypticdiversityingroundwater:how largearetherangesofmacro-stygobionts?. FreshwBiol 2009, 54 :727-744. 14.ZaksekV,SketB,GottsteinS,FranjevicD,TronteljP: Thelimitsofcryptic diversityingroundwater:phylogeographyofthecaveshrimp Troglocaris anophthalmus (Crustacea:Decapoda:Atyidae). MolEcol 2009, 18 :931-946. 15.BarrTC,HolsingerJR: Speciationincavefaunas. AnnuRevEcolSyst 1985, 16 :313-337. 16.Desutter-GrandcolasL,GrandcolasP: Theevolutiontowardstroglobitic life:aphylogeneticreappraisalofclimaticrelictandlocalhabitatshift hypothesis. MémBiospéol 1996, 23 :57-63. 17.BellesX: Survival,opportunismandconvenienceintheprocessesof cavecolonizationbyterrestrialfaunas.OecolAquat 1992, 10(1991) :325-335. 18.PeckSB,FinstonTL: Galapagosislandstroglobits:thequestionoftropical troglobits,parapatricdistributionsandeyedsister-species,andtheir originbyparapatricspeciation. MémBiospél 1993, 20 :19-37. 19.HowarthF: TheZoogeographyofspecializedcaveanimals:abioclimatic model. Evolution 1980, 34 :394-406. 20.HowarthF: Theevolutionofnon-relictualtropicaltroglobites. IntJSpeleol 1987, 16 :1-16. 21.DecuV,JuberthieC: Coléoptères(généralitésetsynthèse). Encyclopædia BiospeologicaII SociétédeBiospéologieJuberthieC,DecuV1998, II :1025-1030.Ribera etal . BMCEvolutionaryBiology 2010, 10 :29 http://www.biomedcentral.com/1471-2148/10/29 Page12of14

PAGE 13

22.FailleA,RiberaI,DeharvengL,BourdeauC,GarneryL,QuéinnecE,DeuveT: AmolecularphylogenyshowsthesingleoriginofthePyrenean subterraneanTrechinigroundbeetles(Coleoptera:Carabidae). Mol PhylogenEvol 2010, 54 :97-106. 23.PerreauM: CataloguedesColéoptèresLeiodidaeetPlatypsyllinae. Mém SocentFr 2000, 4 :1-460. 24.NewtonAF: Phylogeneticproblems,currentclassificationandgeneric catalogofworldLeiodidae(includingCholevidae). Phylogenyand evolutionofsubterraneanandendogeanCholevidae(=Leiodidae,Cholevinae) ProceedingsofXXInternationalCongressofEntomology,Firenze,1996 Torino: AttiMuseoRegionalediScienzeNaturaliGiachinoPM,PeckSB1998,41-178. 25.FresnedaJ,SalgadoJM,RiberaI: PhylogenyofwesternMediterranean Leptodirini,withanemphasisongenitalcharacters(Coleoptera: Leiodidae:Cholevinae). SystEnt 2007, 32 :332-358. 26.GiachinoPM,VailatiD,CasaleA: Majorquestionsinthephylogenyand biogeographyofCholevidae(Coleoptera),withemphasisonthe subfamilyLeptodirinae. Phylogenyandevolutionofsubterraneanand endogeanCholevidae(=Leiodidae,Cholevinae)ProceedingsofXX InternationalCongressofEntomology,Firenze,1996 Torino:AttiMuseo RegionalediScienzeNaturaliGiachinoPM,PeckSB1998,179-209. 27.LawrenceJF,NewtonAF: FamiliesandsubfamiliesofColeoptera(with selectedgenera,notes,referencesanddataonfamily-groupnames). Biology,phylogeny,andclassificationofcoleoptera:paperscelebratingthe 80thbirthayofRoyACrowson Varsovia:MuseumiInstytutZoologii PANPakalukJ,SlipinskiSA1995,779-1006. 28.JeannelR: Lagenèsedesfaunesterrestres.Elémentsdebiogéographie. Paris:PressesUniversitairesdeFrance1942. 29.CacconeA,SbordoniV: Molecularbiogeographyofcavelife:Astudy usingmitochondrialDNAfromBathysciinebeetles. Evolution 2001, 55 :122-130. 30.SalgadoJM,BlasM,FresnedaJ: FaunaIberica.Vol.31:Coleoptera: Cholevidae. Madrid:CSIC2008. 31.NewtonAF: LeiodidaeFleming,1821. HandbookofZoologyVolIV, Arthropoda:Insecta;Coleoptera,VolI,MorphologyandSystematics (Archostemata,Adephaga,Myxophaga,Polyphagapartim) Berlin,NewYork: DeGruyterBeutelRG,LeschenRAB2005,261-280. 32.ShullVL,VoglerAP,BakerMD,MaddisonDR,HammondPM: Sequence alignmentof18SribosomalRNAandthebasalrelationshipsof Adephaganbeetles:evidenceformonophylyofaquaticfamiliesandthe placementofTrachypachidae. SystBiol 2001, 50 :945-969. 33.RiberaI,BeutelRG,BalkeM,VoglerAP: DiscoveryofAspidytidae,anew familyofaquaticColeoptera. ProcRSocLondB 2002, 269(1507) :2351-2356. 34.KatohK,TohH: RecentdevelopmentsintheMAFFTmultiplesequence alignmentprogram. BriefBioinform 2008, 9 :286-298. 35.LoytynojaA,GoldmanN: Analgorithmforprogressivemultiple alignmentofsequenceswithinsertions. ProcNatlAcadSciUSA 2005, 102 :10557-10562. 36.HuelsenbeckJP,RonquistF: MRBAYES:Bayesianinferenceofphylogenetic trees. Bioinformatics 2001, 17 :754-755. 37.PosadaD,CrandallKA: MODELTEST:testingthemodelofDNA substitution. Bioinformatics1998, 14 :817-818. 38.ZwicklD: Geneticalgorithmapproachesforthephylogeneticanalysisof largebiologicalsequencedatasetsunderthemaximumlikelihood criterion. Austin:Texas 2006. 39.RosenbaumG,ListerGS,DubozC: RelativemotionsofAfrica,Iberiaand EuropeduringAlpineorogeny. Tectonophysics 2002, 359 :117-129. 40.SchettinoA,TurcoE: PlatekinematicsoftheWesternMediterranean regionduringtheOligoceneandEarlyMiocene. GeophysJInt 2006, 166 :1398-1423. 41.HoS,PhillipsM: Accountingforcalibrationuncertaintyinphylogenetic estimationofevolutionarydivergencetimes. SystBiol 2009, 58 :367-380. 42.DrummondA,RambautA: BEAST:Bayesianevolutionaryanalysisby samplingtrees. BMCEvolBiol 2007, 7 :214. 43.SandersonMJ: Estimatingabsoluteratesofmolecularevolutionand divergencetimes:apenalizedlikelihoodapproach. MolBiolEvol 2002, 19 :101-109. 44.SwoffordDL: PAUP*:PhylogeneticAnalysisUsingParsimony(*and OtherMethods)Version4. Sunderland,MA.:SinauerAssociates2002. 45.ErixonP,SvennbladB,BrittonT,OxelmanB: ReliabilityofBayesian posteriorprobabilitiesandbootstrapfrequenciesinphylogenetics. Syst Biol 2003, 52 :665-673. 46.FresnedaJ,SalgadoJM: Thegenus Bathysciola Jeannel,1910inthe IberianPeninsulaandPyrenees.TaxonomicrevisionofthesectionsIV, VIandVII(Jeannel,1924)(Coleoptera,Cholevidae,Leptodirinae). Graellsia 2006, 62 :45-54. 47.SalgadoJM: AnewrevisionandtaxonomicpositionfortheCantabrian Leptodirini:sections Quaestus and Speonomidius (Coleoptera: Cholevidae). AnnSocentFr(NS) 2000, 36 :45-59. 48.SalgadoJM,FresnedaJ: Revisionofthesection Anillochlamys Jeannel, 1909(Coleoptera:Leiodidae:Cholevinae:Leptodirini). AnnSocentFr(NS) 2003, 39 :361-384. 49.BellésX,ComasJ,EscolàO,EspañolF: LosBathysciinaeibéricos, propuestadeordenacióntaxonómica(Col.Catopidae). Speleon 1978, 24 :59-68. 50.ZoiaS,RampiniM: LeptodirinaedelCaucasoedell ’ Iransettentrionale (ColeopteraCholevidae). RevsuisseZool 1994, 101:771-827. 51.BromhamL,PennyD: Themodernmolecularclock. NatRevGenet 2003, 4 :216-224. 52.DonoghuePCJ,BentonMJ: Rocksandclocks:calibratingtheTreeofLife usingfossilsandmolecules. TrendsEcolEvol 2007, 22 :424-431. 53.PulquerioMJF,NicholsRA: Datesfromthemolecularclock:howwrong canwebe?. TrendsEcolEvol 2007, 22 :180-184. 54.BrowerAVZ: Rapidmorphologicalradiationandconvergenceamong racesofthebutterfly Heliconiuserato inferredfrompatternsof mitochondrial-DNAevolution. ProcNatlAcadSciUSA 1994, 91(14) :6491-6495. 55.PonsJ,BarracloughTG,Gomez-ZuritaJ,CardosoA,DuranDP,HazellS, KamounS,SumlinWD,VoglerAP: Sequence-basedspeciesdelimitation fortheDNAtaxonomyofundescribedinsects. SystBiol 2006, 55 :595-609. 56.Contreras-DíazHG,MoyaO,OromíP,JuanC: Evolutionanddiversification oftheforestandhypogeanground-beetlegenus Trechus intheCanary Islands. MolPhylogenetEvol 2007, 42 :687-699. 57.BalkeM,RiberaI,HendrichL,MillerMA,SagataK,PosmanA,VoglerAP, MeierR: NewGuineahighlandoriginofawidespreadarthropod supertramp. ProcRSocLondB 2009, 276 :2359-2367. 58.RiberaI,HernandoC,AguileraP: Agabusalexandrae sp.n.fromMorocco, withamolecularphylogenyoftheWesternMediterraneanspeciesof the A.guttatus group(Coleoptera:Dytiscidae). InsSystEvol 2001, 32 :253-262. 59.TrewickSA,WallisGP: Bridgingthe ‘ beech-gap ’ :invertebrate phylogeographyimplicatesrecentratherthanancientprocessesinNew Zealandbiogeographicpatterns. Evolution 2001, 55 :2170-2180. 60.CasasSainzAM,FaccennaC: Tertiarycompressionaldeformationofthe Iberianplate. TerraNova 2001, 13 :281-288. 61.BarnolasA,PujalteV: LacordilleraPirenaica. GeologíadeEspaña Madrid: SGE-IGMEVeraJ2004,233-241. 62.TronteljP: Theageofsubterraneancrayfishspecies.Acommenton Buhay&Crandall(2005):subterraneanphylogeographyoffreshwater crayfishesshowsextensivegeneflowandsurprisinglylargepopulation sizes. MolEcol 2007, 16 :2841-2843. 63.TronteljP,GorickiS,PolakS,VerovnikR,ZaksekV,SketB:Ageestimatesfor somesubterraneantaxaandlineagesintheDinaricKarst. TimeinKarst Meeting:March2007;Postojna,Slovenia 2007, 18 :3-189. 64.FiserC,SketB,TronteljP: Aphylogeneticperspectiveon160yearsof troubledtaxonomyof Niphargus (Crustacea:Amphipoda). ZoolScr 2008, 37 :665-680. 65.FoulquierA,MalardF,LefebureT,GibertJ,DouadyCJ: Theimprintof Quaternaryglaciersonthepresent-daydistributionoftheobligate groundwateramphipod Niphargusvirei (Niphargidae). JBiogeogr 2008, 35 :552-564. 66.VillacortaC,JaumeD,OromíP,JuanC: Underthevolcano: phylogeographyandevolutionofthecave-dwelling Palmorchestia hypogaea (Amphipoda,Crustacea)atLaPalma(CanaryIslands). BMCBiol 2008, 6. 67.CarliniDB,ManningJ,SullivanPG,FongDW: Moleculargeneticvariation andpopulationstructureinmorphologicallydifferentiatedcaveand surfacepopulationsofthefreshwateramphipod Gammarusminus . Mol Ecol 2009, 18 :1932-1945.Ribera etal . BMCEvolutionaryBiology 2010, 10 :29 http://www.biomedcentral.com/1471-2148/10/29 Page13of14

PAGE 14

68.ArnedoMA,FerrándezMA: Mitochondrialmarkersrevealdeeppopulation subdivisionintheEuropeanprotectedspider Macrothelecalpeiana (Walckenaer,1805)(Araneae,Hexathelidae). ConservGenet 2007, 8 :1147-1162. 69.BarrTC: ASynopsisofthecavebeetlesofthegenus Pseudanophthalmus oftheMitchellPlaininSouthernIndiana(Coleoptera,Carabidae). Am MidNat 1960, 63 :307-320. 70.ChristmanM,CulverD,MaddenM,WhiteD: Patternsofendemismofthe easternNorthAmericancavefauna. JBiogeogr 2005, 32 :1441-1452. 71.BuhayJE,CrandallKA: Subterraneanphylogeographyoffreshwater crayfishesshowsextensivegeneflowandsurprisinglylargepopulation sizes. MolEcol 2005, 14 :4259-4273. 72.SimonC,FratiF,BeckenbachA,CrespiB,LiuH,FlookP: Evolution, weighting,andphylogeneticutilityofmitochondrialgene-sequences andacompilationofconservedpolymerasechain-reactionprimers. Ann entSocAm 1994, 87 :651-701. 73.BarracloughTG,HoganJE,VoglerAP: Testingwhetherecologicalfactors promotecladogenesisinagroupoftigerbeetles(Coleoptera: Cicindelidae). ProcRSocLondB 1999, 266(1423) :1061-1067. 74.VoglerAP,DesalleR,AssmannT,KnisleyCB,SchultzTD: Molecular populationgeneticsoftheendangeredtigerbeetle Cicindeladorsalis (Coleoptera,Cicindelidae). AnnentSocAm 1993, 86 :142-152. doi:10.1186/1471-2148-10-29 Citethisarticleas: Ribera etal .: AncientoriginofaWestern Mediterraneanradiationofsubterraneanbeetles. BMCEvolutionary Biology 2010 10 :29. Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color Þgure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit Ribera etal . BMCEvolutionaryBiology 2010, 10 :29 http://www.biomedcentral.com/1471-2148/10/29 Page14of14