ebook img

Lessons from Red Data Books: Plant Vulnerability Increases with Floral Complexity PDF

18 Pages·2015·1.89 MB·English
by  
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Lessons from Red Data Books: Plant Vulnerability Increases with Floral Complexity

RESEARCHARTICLE Lessons from Red Data Books: Plant Vulnerability Increases with Floral Complexity AnastasiaStefanaki1,2*,AphroditeKantsa1,ThomasTscheulin1,MarthaCharitonidou1, TheodoraPetanidou1 1 LaboratoryofBiogeographyandEcology,DepartmentofGeography,UniversityoftheAegean,Mytilene, Greece,2 NaturalisBiodiversityCenter–NationalHerbariumoftheNetherlands,Leiden,Netherlands *[email protected] Abstract Thearchitecturalcomplexityofflowerstructures(hereafterreferredtoasfloralcomplexity) maybelinkedtopollinationbyspecializedpollinatorsthatcanincreasetheprobabilityof OPENACCESS successfulseedset.Asplant—pollinatorsystemsbecomefragile,alossofsuchspecial- izedpollinatorscouldpresumablyresultinanincreasedlikelihoodofpollinationfailure.This Citation:StefanakiA,KantsaA,TscheulinT, CharitonidouM,PetanidouT(2015)Lessonsfrom isanissuelikelytobeparticularlyevidentinplantsthatarecurrentlyrare.Usinganovel RedDataBooks:PlantVulnerabilityIncreaseswith indexdescribingfloralcomplexityweexploredwhetherthisaspectofthestructureofflowers FloralComplexity.PLoSONE10(9):e0138414. couldbeusedtopredictvulnerabilityofplantspeciestoextinction.Todothiswedefined doi:10.1371/journal.pone.0138414 plantvulnerabilityusingtheRedDataBookofRareandThreatenedPlantsofGreece,a Editor:JeffOllerton,UniversityofNorthampton, Mediterraneanbiodiversityhotspot.Wealsotestedwhetherotherintrinsic(e.g.lifeform, UNITEDKINGDOM asexualreproduction)orextrinsic(e.g.habitat,altitude,range-restrictedness)factorscould Received:March14,2015 affectplantvulnerability.Wefoundthatplantswithhighfloralcomplexityscoresweresignifi- Accepted:August30,2015 cantlymorelikelytobevulnerabletoextinction.Amongallthefloralcomplexitycomponents Published:September21,2015 onlyfloralsymmetrywasfoundtohaveasignificanteffect,withradial-flowerplantsappear- ingtobelessvulnerable.Lifeformwasalsoapredictorofvulnerability,withwoodyperen- Copyright:©2015Stefanakietal.Thisisanopen accessarticledistributedunderthetermsofthe nialplantshavingsignificantlylowerriskofextinction.Amongtheextrinsicfactors,both CreativeCommonsAttributionLicense,whichpermits habitatandmaximumrangeweresignificantlyassociatedwithplantvulnerability(coastal unrestricteduse,distribution,andreproductioninany plantsandnarrow-rangedplantsaremorelikelytofacehigherrisk).Althoughextrinsicand medium,providedtheoriginalauthorandsourceare inparticularanthropogenicfactorsdetermineplantextinctionrisk,intrinsictraitscanindicate credited. aplant’spronenesstovulnerability.Thisraisesthepotentialthreatofdecliningglobalpolli- DataAvailabilityStatement:Allrelevantdataare natordiversityinteractingwithfloralcomplexitytoincreasethevulnerabilityofindividual withinthepaperanditsSupportinginformationfiles. plantspecies.Thereispotentialscopeforusingplant—pollinatorspecializationstoidentify Funding:Thisstudywasco-financedbythe plantspeciesparticularlyatriskandsotargetconservationeffortstowardsthem. EuropeanUnion(EuropeanSocialFund[ESF])and GreeknationalfundsthroughtheOperational Program"EducationandLifelongLearning"ofthe NationalStrategicReferenceFramework(NSRF)– ResearchFundingProgram:THALES–Investingin knowledgesocietythroughtheEuropeanSocialFund (seeref.20inthetext).Additionalsupporthasbeen providedtoASbytheFP7EuropeanprojectSTEP (seeref.21inthetext).Thefundershadnorolein PLOSONE|DOI:10.1371/journal.pone.0138414 September21,2015 1/18 PlantVulnerabilityIncreaseswithFloralComplexity studydesign,datacollectionandanalysis,decisionto Introduction publish,orpreparationofthemanuscript. Thecriteriaforthedesignationofendangeredspecies,asestablishedworldwidebytheInterna- CompetingInterests:Theauthorshavedeclared tionalUnionfortheConservationofNature(IUCN),arebasedontrendsinpopulationsizes thatnocompetinginterestsexist. andgeographicalranges[1].Thesedataareenrichedwithinformationonotherexternal threats,mostlyanthropogenic,includingoverexploitation,habitatdegradation,pollutants,and competitionfromintroducedspecies[1].Variouseffortshavebeenmadetocomplementthe IUCNcriteria(e.g.[2–4]),whichhavefocusedprimarilyongeographicaland/orecologicalfac- tors,withendemismorhabitatoftencitedasreliablepredictorsofplantvulnerability(e.g.[2]). Itis,however,possiblethatotherintrinsicfactors,includingthefloweringduration,reproduc- tionorpollinationmodeofplants,mayaffecttheirvulnerability[5].Incompleteknowledgeof thereproductionbiologyofmanyrareandthreatenedplants[6,7]impedestheconsideration ofsuchintrinsictraitsassociatedwiththereproductivesuccessofplants. Adeterminantofsuccessfulplantreproductionispollination.Acomplexflowerstructure mayincreaseaplant’schancesofsuccessfulreproductionbyenhancingpollinatorfidelityto flowersandreducinglarcenybyillegitimatefloralvisitors([8]andreferencestherein).Onthe otherhand,plantswithcomplexflowersmaybelikelytosufferfrompollinationfailure,iftheir handlingrequirestoospecializedpollinatorsthatarerareorforsomereason(e.g.human impact)becomelost,oriftheydependonlarge-sizedpollinatorsthatarenaturallylessabun- dantoronpollinatorswithfluctuatingpopulationsoverspaceandtime.Particularlyforrare plants,pollinationfailuremaybecritical,becausetheseplantsaremorelikelytobepollenlim- itedduetoco-floweringspeciescompetingforlimitedpollinatornumbers[9,10].Thevulnera- bilityofplantspeciescausedbythelackofpollinatorshasbeenlongdiscussedthroughthe prismoftheplants’interspecificcompetitionforsharedpollinators([11]andreferences therein,[12–14]).Theimportanceofthisproblembecomesevenmorerelevantastheconcern forhuman-inducedbreakdownofplant—pollinatorsystemsincreases[15–21]. Hereweinvestigatewhetherplantvulnerabilitycanbepredictedbyfloralstructure(simple tocomplex)vis-à-visotherpotentialpredictorsofvulnerability,usingasacaseexamplethe rareandthreatenedplantsofGreece.With6600planttaxa(1072genera,185families)Greece isregardedasahotspotforbiodiversityandendemismwithintheMediterranean[22,23].A largenumberofthesetaxa(1462,i.e.22.15%oftheflora)areconsideredendemictoGreece includingnumerousnarrowendemicsrestrictedtoasinglemountainorisland[23].The endangeredGreekfloraisdescribedintwoeditionsoftheRedDataBooksofRareandThreat- enedPlantsofGreece(hereafterGreekRedDataBook).Thefirstwaspublishedin1995with accountsof263taxa[24].Atwo-volumebookbasedonthecurrentIUCNcriteria[1]waspub- lishedin2009ascomplementarytothatof1995andincludedupdatesofsomeprevious accountsplusaccountsofmanyadditionaltaxa[25].Intotal,thenumberofplanttaxa includedintheGreekRedDataBookcorrespondstoc.7%oftheGreekflora,whilehalfofthe country’sendemic,rareandthreatenedplantshavenotbeenevaluated[26]. Thisstudyaddressesthefollowingquestions:(1)Areplantswithcomplexflowerstructures morevulnerablethanthosewithsimplerflowerstructures?(2)Areotherintrinsicfactorsasso- ciatedwithplantvulnerability,suchaspollination-related(floweringtimes,floralcolorand size)andlifestrategyones(lifeform,capabilityofasexualreproduction)?(3)Whatistheeffect ofextrinsicfactors(e.g.geographical,ecological)onplantvulnerability?Floralcomplexitywas quantifiedusingtheFloralComplexityIndexintroducedhere,i.e.anindextakingintoaccount severalfloraltraitsrelatedtopresumedpollinatorspecialization.Wediscusstheeffectsoffloral complexityalongwithotherfactorsshownheretoaffectplantvulnerabilityandaddressthe necessityofjointplant—pollinatorconservationassessments. PLOSONE|DOI:10.1371/journal.pone.0138414 September21,2015 2/18 PlantVulnerabilityIncreaseswithFloralComplexity MaterialsandMethods Dataset Adatasetofintrinsicandextrinsicvariables(describedindetailbelow)wasgeneratedforthe planttaxaincludedintheGreekRedDataBook[24,25].Thelatterwasthemainsourceof informationforthecompilationofthedatasetwithsupplementarysourcescomprisingmajor floristicworks[23,27–34],onlineHerbariumcollections(mainlyfromtheHerbariaB,E,K, LD,MAICandW),originaldescriptionsoftaxaandothertaxonomicandfloristicpapers, booksandonlinedatabases[35–58].NomenclaturewasupdatedaccordingtoAPGIII[59]for plantfamiliesandaccordingtoDimopoulosetal.[23]forspeciesandsubspeciesconsidering synonymiesandexcludingerroneousrecordsasreportedin[23].Wind-andwater-pollinated taxawerealsoexcludedfromthedataset,whichfinallyincludedatotalof427planttaxa(S1 Table),allpollinatedbyinsects,theonlypollinatorsintheMediterraneanarea. Intrinsicvariables FloralcomplexityvariablesandFloralComplexityIndex. Weusedseveralfloralvari- ablesshownintheliteraturetobefunctionallyimportantforpollination,whichwepresumed tobeinvolvedinplants'selectivityforpollinatorsandthusrelatedtoplantvulnerability.These were(1)shape,(2)depth,(3)symmetry,(4)corollasegmentation,and(5)functionalreproduc- tiveunit.Anadditionalapproachinvolvedtheassemblingofthesevariablesintoafloralcom- plexityindexdescribedbelow. Floralshape.Accordingtothefunctionalshapeofflowers(oroffloweraggregationsfunc- tioningassingleattractionunits,seefunctionalreproductiveunitbelow),taxawereassignedto 11levelsbasedontheclassificationbyFægriandvanderPijl[60],Barth[61]andPetanidou [62].Theselevelswere:(1)bell(adownwardfacingbell-shapedflowerthattheinsectenters withmuchofortheentirebody,notonlyproboscis);(2)brush(asinglefloweroraflower aggregationwithnumerousprotrudinganthers);(3)disk(ashallowflowerwithpetalsmoreor lessspreadoutinaflatcircle);(4)tube(atubularflower);(5)disk-tube(aflowerwithaflat- tenedpartabruptlyarisingonatubularstalk);(6)funnel(anupwardfacingfunnel-shaped flowerthattheinsectenterswithmuchofortheentirebody);(7)flag(the“butterfly”-shaped floweroftheFabaceaeandPolygalaceae);(8)gullet(aflowerwithalipservingasalanding platformforinsectstoinserttheirheadorwholebodyintothecorollatube);(9)head(a densely-packedfloweraggregationwithmoreorlesssphericalorflatappearance;(10)lip(an orchidflowerwithanextendedlipusedbyvisitinginsectsasalandingplatform);and(11)trap flowers(abowlormorecomplicatedtubularstructurewithsteepandsmoothsurfacewhere theinsectsget“trapped”forsomeperiodoftime). Floraldepth.Floraldepth,measuredaslengthofthecorollatube,expressestheaccessibility ofnectarrewardaccumulatedinthebottomoftheflowertubetoaflowervisitorstandingat thefreesurfaceoftheflower.Planttaxawereassignedtothreelevels,viz.thosehaving:(1)low- depthflowers,i.e.withcorollatubelength<4mm,includingthosewithoutfloraldepth(disk, brushflowers);(2)medium-depthflowers,withcorollatubelength4–10mm;and(3)high- depthflowers,withcorollatubelength>10mm[62].Particularlyinflagflowersthedepthwas expressedasthelengthofthekeel,intrapflowers(Araceae)asthelengthofthespathetube, andinspurredflowers(e.g.taxaofViolaceae,Fumariaceae,Ranunculaceae,Orchidaceae)as thelengthofthespur. Floralsymmetry.Taxawereassignedtotwolevelsbasedonthenumberoffloralsymmetry axes,viz.thosehaving:(1)severalaxes(radialsymmetry);and(2)oneaxis(bilateral PLOSONE|DOI:10.1371/journal.pone.0138414 September21,2015 3/18 PlantVulnerabilityIncreaseswithFloralComplexity symmetry).FloralsymmetryinAsteraceaewasestimatedforthewholeinflorescence(seefunc- tionalreproductionunitbelow). Corollasegmentation.Threedegreesofcorolla(orperianth)segmentationwereconsidered: (1)sympetaly,whenpetals(orperianthsegments)arefusedalongtheirentirelength;(2)chori- petaly,whenpetals(orperianthsegments)arecompletelyfree;and(3)semichoripetaly,when petals(orperianthsegments)arefusedtosomedegree,formingamoreorlessdistinctcorolla tubewithfreelobes.Calyxsegmentationwasnotconsidered. Functionalreproductiveunit.Thiscorrespondstotheinflorescenceandincludesflower aggregationsfromafunctionalratherthanamorphological(planttaxonomy)viewpoint[62]. Taxawereassignedaccordinglytothreelevels,describedashaving:(1)singleflowers,including anytypeofinflorescencewithuptofiveconcurrentlyfunctional(open)flowers;(2)aggrega- tionsofflowersarisingonaflatorsphericalsurfaceoftheplant(heads,umbels,corymbs);and (3)aggregationsofflowersarisingalongacylindricalsurfaceoftheplant(spikes,racemes, panicles). TheFloralComplexityIndex(FCIhenceforth)considersfloralcomplexityastheplants' “selectivity”forpollinatorsandisbasedonthefivefloralvariablesdescribedabove.Eachvari- ablewasgivenaweight(w)basedonitspotentialcontributiontofloralcomplexity:highfor floralshape(w =0.3);moderatefordepth,symmetryandcorollasegmentation(w = shape depth w =w =0.2);andlowforfunctionalreproductiveunit(w symmetry corollasegmentation functionalrepro- =0.1).Thesumofthefiveweightsequals1.Eachlevelofthevariablesdepth,symme- ductiveunit try,corollasegmentationandfunctionalreproductiveunitwasassignedavalue(V)basedona scalefrom1to3,where1indicatesalessspecializedtraitand3amorespecializedtrait.Forflo- ralshape,duetogreaternumberoflevels(i.e.shapetypesofdifferentlevelsofdifficulty),values (V)wereassignedbasedonascalefrom1to5.Eachvariablelevel’svalue(V)wasthemeanof assignedscoresbyfourpollinationexperts(seeS1Text).Thefinalweightofeachvariable definingitscontributiontotheindexwasthencalculatedbymultiplyingthevariable’sweight (w)bythevariablelevel’svalue(V)(Table1).TheFCIforeachtaxonwasfinallyexpressedas thesumoffinalweightsofthefivevariables:FCI=∑(w (cid:1)V ),wherew =theweightoftheflo- j ij j ralvariablejandV =thevalueofthefloralvariablejfortaxoni.Forexample,theFCIvalueof ij ataxonwithracemesofgullet-shapedflowers(i.e.bilaterallysymmetricalandpartlyfused) andalong(>10mm)floraltubewouldbethesumofthesefinalweights:FCI=0.30+1.13 +0.60+0.40+0.60=3.03(seeTable1).TheFCIvalueassignedtoeachtaxonofthedatasetis giveninS1Table. Otherintrinsicvariables. Floralcolor.FivelevelswereadoptedaccordingtoPetanidou andLamborn[63]:(1)white;(2)yellow;(3)violet,purple,pink,red;(4)blue;and(5)green.In casesoftwoormoreconcurrentcolors,thedominant(>50%offlowersurface)wasselected, andintaxawithindividualsbearingflowersofdifferentcolorsthemostfrequentwastaken. Floralsize.Thiswasexpressedasthelength—widthaverageoftheexposedsurfaceofthe corollaina2D-projection(correspondinge.g.tothefloraldiameterformoreorlesscircular flowers).Taxawereassignedtothreelevels:(1)small,1–10mm;(2)medium,10–20mm;and (3)large,>20mm[62].Incasesofstronglycompactinflorescences,whichgivethe“impression offunctionallysingleflowers”,e.g.headsandcompoundcorymbsofAsteraceaeorcompound umbelsofApiaceae,floralsizewasmeasuredforthewholeinflorescence. Floweringseason.Threefloweringseasonswereconsidered:(1)FebruarytoMay;(2)June toSeptember;and(3)OctobertoJanuary.Whenthefloweringtimeofataxonfellintotwodif- ferentseasonsitwasassignedtothedominantone(biggestoverlap).Ifataxonfellequallyin twoseasonsitwasassignedtotheearliest. Floweringduration.Thetotalnumberofmonthsduringwhichataxonhasbeenreportedto beinflower. PLOSONE|DOI:10.1371/journal.pone.0138414 September21,2015 4/18 PlantVulnerabilityIncreaseswithFloralComplexity Table1. WeightsofthefloralvariablesusedintheFloralComplexityIndex.FordetailsontheestimationofthevariablelevelvaluesseeS1Text. Alltermsareexplainedinthetext. Floralvariable(w:variableweight) Levelortrait Levelvalue(V)a Finalweightb Shape(w=0.3) bell 3.33 1.00 brush 3.50 1.05 disk 1.00 0.30 tube 3.25 0.98 disk-tube 2.75 0.83 funnel 2.65 0.80 flag 4.25 1.28 gullet 3.75 1.13 head 2.25 0.68 lip 4.50 1.35 trap 4.25 1.28 Depth(w=0.2) low-depth 1.00 0.20 medium-depth 2.00 0.40 high-depth 3.00 0.60 Symmetry(w=0.2) bilateral 3.00 0.60 radial 1.13 0.23 Corollasegmentation(w=0.2) sympetalous 2.50 0.50 semichoripetalous 2.00 0.40 choripetalous 1.50 0.30 Functionalreproductiveunit(w=0.1) single 1.50 0.15 spikes/racemes 3.00 0.30 heads 1.50 0.15 aBasedonascaleof1–3(forfloraldepth,symmetry,corollasegmentationandfunctionalreproductiveunit)orascaleof1–5(forfloralshape).SeeS1 Text. bResultingfromthefloralvariableweight(w)multipliedbythevariablelevel’svalue(V). doi:10.1371/journal.pone.0138414.t001 Lifeform.TaxawereassignedtofourgroupsaccordingtoPetanidouetal.[64]:(1)thero- phytes,i.e.annualplantsthatsurviveunfavorableseasonsintheformofseeds;(2)geophytes, i.e.perennialplantsthatsurviveunfavorableseasonsintheformofundergroundstorage organs(bulbs,tubers,corms,rhizomes);(3)herbaceousperennials,i.e.perennialplantsthat areentirelyherbaceousabovethegroundoronlywoodyatthebase;and(4)woodyperennials, i.e.perennialplantsmoreorlessentirelywoody. Asexualreproduction.Taxawereclassifiedasknownornotknowntoreproducebyvegeta- tivemeans,e.g.bulbs,tubers,corms,rhizomes,stolons,suckers. Extrinsicvariables Habitat. Taxawereassignedaccordingtothehabitatstheyoccurin.Thiswasbasedon habitatassociationsdescribedinDimopoulosetal.[23]as:(1)aquatic(freshwater)habitats; (2)cliffsandotherrockyhabitats;(3)lowlandtomontanegrasslands;(4)high-mountainvege- tation;(5)coastalandmarinehabitats;(6)phrygana(lowscrubland);(7)agriculturaland ruderalhabitats;and(8)woodlandsandscrub.Whenmorethanonehabitatwasprovidedfor ataxon,thedominanthabitataccordingtotheRedDataBookdescriptionwasretained. Minimumaltitude. Thisdenotestheminimumaltitude(m),whereataxonhasbeen recorded. PLOSONE|DOI:10.1371/journal.pone.0138414 September21,2015 5/18 PlantVulnerabilityIncreaseswithFloralComplexity Maximaldistance. Aquantitativemeasureofrange,denotingtheEuclideandistance(in km)betweenthemostdistantlocalitiesofataxonasprovidedintheRedDataBookdotmaps. ForitsmeasurementwedigitizedthemapsfromtheRedDataBookinArcGIS9.3(ESRI),con- sideringalsorecentupdatesindistributionsofthetaxabyDimopoulosetal.[23]. Range-restrictedstatus. Accordingtothisqualitativemeasureofrangeintroducedby Dimopoulosetal.[23],taxawereassignedtotwolevels:(1)range-restricted,i.e.taxawithadis- tributionareawhoselongestdimensiondoesnotexceed500km[23];and(2)notrange- restricted.Thisvariablewasselectedinsteadofendemismasnon-affectedbycountryborders. Phytogeographicalregion. Taxawereassignedtothethreephytogeographicalregionsof GreecesensuBrummitt[65]:(1)Kriti—Karpathosgroup;(2)EastAegeanIslands;and(3) remainingGreece.Whenataxonfellintomorethanoneregion,itwasassignedtotheonewith thehighestnumberoflocalitiesitwasrecordedin.Ifataxonoccurredinequalnumbersof localitiesinmorethanoneregion,itsextentofoccurrence(sensuIUCN[1])ineachregionwas additionallyestimatedandthetaxonwasconsequentlyassignedtotheregioninwhichits extentofoccurrenceislarger. Dataanalysis Thepossibleoutcomesoftheresponsevariable“plantvulnerability”wereexpressedbythe IUCNcategoriesasgivenforeachtaxonintheGreekRedDataBook,updatedwhereapplicable accordingtotheIUCNRedListofThreatenedSpecies(www.iucnredlist.org)andaggregated intwogroups:(1)“Lessthreatened”,includingVulnerable(VU),NearThreatened(NT),Rare (R)andLeastConcerned(LC)taxa;and(2)“Morethreatened”,includingCriticallyEndan- gered(CR),Endangered(EN)andpresumedExtinct(EX)taxa.IncontrasttotheIUCN approach[1],i.e.assemblingVUtogetherwithCRandENtaxaina“threatened”group,we adoptedastricterdistinctionamong“lessthreatened”and“morethreatened”taxa,withVU taxaplacedinthefirstgroup.Thiswasalsonecessaryinordertoavoidpotentialinconsistency inourdatasetthatmightarisefromthepresenceoftheoldIUCNcategory“R”oftheplants drawnfromthe1995RedDataBook[24]:whenpullingthedataoutofbothRedDataBooks [24,25]werealizedthatassigningataxonasRin[24]occasionallyappearedtooverlapwith thatforVUascurrentlyformedin[1]andasusedinthe2009RedDataBook[25]. AGeneralizedLinearModel(GLM)wasfittedinR3.1.1[66]usingthe‘glm’functionwitha binomialprobabilitydistribution,inordertoexploretherelationofplantvulnerabilitytothe independentintrinsicandextrinsicvariables(maineffects).Astepwisebackwardmodelselec- tionprocedureusingtheAkaikeInformationCriterion(AIC)wasperformedtoselectthebest model.ThefinalmodelwasconstructedonthebasisoftheAIC.Asisthecasewithbinary traitsinlogisticGLM,goodness-of-fitinthefinalmodelwastestedgraphicallywithvalidation plotsoftheempiricalprobabilitieswiththeirstandarderrors[67].Allindependentvariables werecheckedforcollinearitybeforeregressionanalysis.Inordertotestfortheeffectsofeach levelofthecategoricalindependentvariablesonplantvulnerability,weranthe‘glm’function foreachlevelseparatelyandcalculatedthesignificanceoftheireffectswiththelikelihoodratio chi-squaretestsofthedevianceanalysis. InordertocheckforthepotentialeffectsofthecomponentsoftheFCI(viz.floraldepth,flo- ralshape,floralsymmetry,functionalreproductiveunit,andcorollasegmentation),wecon- ductedthefollowingsteps:Weran(1)bivariatetestswheretheeffectsofeachoneofthesingle floralcomplexitycomponentsandofFCIitselfweretestedindividually(S2Text);(2)afull GLMmodelwithallindividualintrinsicvariablesincludingfloralcomplexitycomponents(viz. floraldepth,floralshape,floralsymmetry,functionalreproductiveunit,corollasegmentation, floralcolor,floralsize,floweringduration,floweringseason,lifeformandasexual PLOSONE|DOI:10.1371/journal.pone.0138414 September21,2015 6/18 PlantVulnerabilityIncreaseswithFloralComplexity reproduction)andtheextrinsicones(viz.habitat,minimumaltitude,maximaldistance,range- restrictedstatusandphytogeographicalregion)(S2Text);(3)aGLMmodelwitheachsingle floralcomplexitycomponentandallotherintrinsicandextrinsicvariables(S2Text);(4)the GLMwiththefloralcomplexitycomponentsreplacedbytheFCIitselfinthefullmodel. Beforechoosingourstandardnon-phylogeneticstatisticalmodelingapproach,wetested ourdatasetforphylogeneticdependenceregardingthevulnerabilityoftheplants.Thephylog- enyofthe427plantsofourdatasetwasconstructedaccordingtoAPGIII[59]usingtheonline softwarePhylomaticv3[68].Thebranchlengthsofthisphylogenetictreewereadjustedsoas tocorrespondtoevolutionarydivergencetimebetweenclades,usingthe‘bladj’algorithmin thesoftwarePhylocomv4.2[69].Forthiscalibration,weusedtheupdatednodeagesprovided byGastauerandMeira-Neto[70],whichaddresstheinconsistenciesobservedinthedefault agesfileofPhylocombasedonWikstrom’sdatingofAngiospermfamilies[71].Aphylogenetic tree(Fig1)wascreatedusingtheweb-basedInteractiveTreeofLifev2[72].Boththeraw responsevariableanditsresidualsinthefinalmodelweretestedforphylogeneticsignal(see [73]andreferencestherein).Toestimatethephylogeneticsignal,i.e.totestwhetherplantvul- nerabilityfollowsarandomphylogeneticpatternoritisphylogeneticallyclumpedcorrespond- ingtoaBrownianmotionmodel,wemeasured(i)fortherawvariable,theDstatisticfor binarytraits,whichisbasedonthesumofthesister-cladedifferencesofthegiventraitina phylogeny[74]usingthe‘phylo.D’functionintheRpackagecaper0.5.2and(ii)fortheresidu- als,bothBlomberg’sK[75]andPagel’sλ[76]usingfunctions‘phylosignal’intheRpackage picante1.6–2and‘pgls’(byfittingthemodel:vulnerability~1)incaper0.5.2.Allmetrics revealednosignificantphylogeneticsignalofplantvulnerability(seeS3Text),allowingusto proceedwiththeabove-mentionedGLMapproach. Results FloralComplexityIndex TheFCIscoresforthe427taxaexaminedvariedbetween1.15and3.25.Thelowestscore (1.15)wasrecordedforseveralplantsineightfamilies(e.g.Papaveraceae,Geraniaceae,Paeo- niaceae),whilethehighest(>2.50)forplantsofAraceae,Fumariaceae,Orobanchaceae,Faba- ceae,Orchidaceae,Lamiaceae,Polygalaceae,LentibulariacaeandViolaceae.Twosubspeciesof theorchidDactylorhizakalopissii,subsp.kalopissiiandsubsp.pythagorae,designated,respec- tively,asEndangeredandCriticallyEndangered,obtainedthehighestcomplexityscoreof3.25 (Fig1;S1Table). Plantvulnerability BivariatetestsfortheeffectsofeachofthecomponentsoftheFCIshowedthatnosinglefloral complexitycomponent(viz.floralshape,floraldepth,floralsymmetry,corollasegmentation, functionalreproductiveunit)isasourceofvariationforplantvulnerability(S2Text).When componentswereindividuallyenteredinthefullGLMincludingallotherintrinsic(viz.floral color,floralsize,floweringduration,floweringseason,lifeform,asexualreproduction)and extrinsicfactors(viz.habitat,minimumaltitude,maximaldistance,range-restrictedstatus, phytogeographicalregion),wefoundseparatestatisticallysignificanteffectsonlyfortwocom- ponents,i.e.floralsymmetryandfunctionalreproductiveunit.Whenallsinglefloralcomplex- itycomponentswereenteredtogetherinthefullmodel,onlyfloralsymmetryshoweda significanteffect(S2Text).Morespecifically,taxawithradiallysymmetrical(actinomorphic) flowersshowedanegativecorrelationtovulnerability(vulnerability~radialfloralsymmetry estimate:-0.86,z-statistic=-2.95,p-value=0.003).Consequently,eachdimensionoffloral complexityalonedoesnotseemtobeapredictorofplantvulnerability,andfloralsymmetryis PLOSONE|DOI:10.1371/journal.pone.0138414 September21,2015 7/18 PlantVulnerabilityIncreaseswithFloralComplexity Fig1.TheFloralComplexityIndex(FCI)valuesdistributedacrosstheGreekrareandthreatenedplants’phylogeny.Greybarsindicatetherelative magnitudeoftheFCI(highestvalue:3.25,lowest:1.15).Redrectanglesmarkthe“morethreatened”(CR,ENorEX)taxa. doi:10.1371/journal.pone.0138414.g001 thestrongestpredictorofallthecomponentswhenaddedtothemultivariatemodel.Lastly,we comparedtwofullmodelsincludingeithersymmetryorFCIplusallotherintrinsicandextrin- sicfactorsmentionedabove;wefoundthatthemodelincludingFCIperformedbetterthanthe modelincludingsymmetrybasedonAICandtheBayesianInformationCriterion(S2Text). OurfinallytestedGLMincludedFCIplusallotherintrinsicandextrinsicfactorsasinde- pendentvariables.BasedontheAICstepwisebackwardmodelselectionthebestfittingmodel includedFCI,lifeformandfloralcoloramongtheintrinsicfactors;andmaximaldistance,hab- itatandminimumaltitudeamongtheextrinsicones(Table2).Thevariablesasexual PLOSONE|DOI:10.1371/journal.pone.0138414 September21,2015 8/18 PlantVulnerabilityIncreaseswithFloralComplexity Table2. Resultsofthebestfitting(basedonAIC)GLMshowingtheeffectsoftheintrinsicandextrinsicvariablesontheGreekrareandthreatened plants’vulnerability. Independentvariables LRχ2 Df P-value Intrinsic Lifeform 10.183 3 0.0171 Floralcolor 8.602 4 0.0719 FloralComplexityIndex 10.230 1 0.0014 Extrinsic Habitat 18.129 7 0.0114 Minimumaltitude 3.012 1 0.0826 Maximaldistance 12.683 1 0.0004 doi:10.1371/journal.pone.0138414.t002 reproduction,floweringseason,floweringduration,floralsize,phytogeographicalregionand range-restrictedstatuswerenon-significantandthusdiscardedfromthefinalmodelbasedon theAIC. ThemeanFCIwasfoundtobesignificantlyhigherinthe“morethreatened”thanthe“less threatened”taxa(Fig2a).Boththeseplantgroupsarecharacterizedbythehighproportionof herbaceousperennialsandgeophytesbutdiffersignificantlyintheproportionofwoodyperen- nials(higherinthe“lessthreatened”);thepresenceoftherophytesisslightlyhigherinthe “morethreatened”thaninthe“lessthreatened”group,albeitwithoutsignificantdifference (Fig2b).Amongallvariablesexamined,maximaldistancebetweenataxon’smostdistantpop- ulationswasfoundtobethemostsignificant(Table2)withthismeasureofmaximumrange beingsmallestinthe“morethreatened”comparedtothe“lessthreatened”plants(Fig3a).Both plantgroupsoccurinallhabitatcategories,withthe“morethreatened”beingsignificantly morefrequentincoastalhabitats.“Morethreatened”taxatendtobemorefrequentinruderal habitats,whereas“lessthreatened”aremorelikelytobeencounteredincliffs,woodland/scrub andhigh-mountainhabitats—althoughthesedifferenceswerenotstatisticallysignificant(Fig 3c).Floralcolorandminimumaltitude,althoughpresentinthebestmodel,havep-valuesmar- ginallyoutsidethesignificancelevel(Table2)andsodonotdifferbetweenthetwoplant groups(Figs2cand3b). Discussion Effectsofintrinsicfactorsonplantvulnerability Amongallfloralvariablesexamined(viz.FCI,floralsize,floralcolor,floweringdurationand season),theFCIwasfoundtobethemostpredictiveforataxon’svulnerability:taxawithhigher floralcomplexityaremorelikelytobemorethreatened.Apossibleexplanationforthismaybe pollinationlimitation;suchlimitationmaybeduetospecializedpollinatorscarcityortothelim- itedattractionbythesmallpopulationsizesthreatenedplantsnormallyoccurin[16,77]. Akeyrolemaybeattributedtofloralsymmetry,whichwastheonlyoneamongFCIcompo- nentshavingperseasignificanteffectonplantvulnerability.The“morethreatened”plants examinedherewerefoundtohavemorefrequentlybilaterallysymmetricalflowers(44.3%of the“morethreatened”plants)comparedtothe“lessthreatened”(28%).Bilateralfloralsymme- tryissuggestedtoprovidemultipleadvantagesforaplant,suchashighervisitation,moreeffi- cientrecognitionandfasterhandlingbypollinatorsandhigheroutcrossingratesthanradial symmetry[78–80].Moreover,inacommunitycontextfloralsymmetryhasbeenshowntopre- dictdifferencesinpollinatorvisitationwithinaplant—pollinatornetwork[81].However,bilat- eralsymmetrymightalsoturnintoadisadvantageintimesthatpopulationsofassociated specializedpollinatorsdecline[78].Furtherthanfloralsymmetryhavinganimportanteffect PLOSONE|DOI:10.1371/journal.pone.0138414 September21,2015 9/18 PlantVulnerabilityIncreaseswithFloralComplexity Fig2.IntrinsicvulnerabilityfactorsoftheGreekrareandthreatenedplants.Effectsoftheintrinsicvariablesonplantvulnerabilityasincludedinthebest fittedlogisticGLMbasedonthebackwardAICselectionprocess.(a)MeanvalueoftheFCI(±SE);(b)lifeform;and(c)floralcolor.Categoricalindependent variables(b,c)arepresentedinspinograms.Thewidthofthecolumnscorrespondstotherelativefrequencyofthe“morethreatened”and“lessthreatened” plantsinthedataset;theheightsofthecellsrepresenttherelativefrequencyoftheresponsevariableineveryleveloftheexplanatoryvariables.Coloredcells denotestatisticalsignificanceoftherespectivelevel(**:(cid:3)0.01,***:(cid:3)0.001). doi:10.1371/journal.pone.0138414.g002 onplantvulnerability,thehighersignificanceoffloralcomplexityvs.thatofitsindividualcom- ponentsimpliesthatflowercomplexityisamoremeaningfulpredictor,andwhenconsidered togethernotonlythesymmetrybutalsotheshapeanddepthoftheflowers,theirsegmentation andtheiraggregationininflorescencesareinfluentialforplantvulnerability.Evidencefrom empiricalstudiespointstowardsthesamedirection.Forexample,ithasbeenfoundthatsym- petaloustubular(deep)flowers,whosehandlingrequirespollinatorswithlongmouthparts, receivefewerinsectvisitsthanopenchoripetalousflowers[82],thelatterbeingmorecom- monlyvisitedbyinsectsthatarelessspecialized[83].Likewise,plantspecieswithcomplexflo- ralshape(e.g.flagorgullet)wereshowntodependonspecializedpollinatorbehavior comparedtospecieswithlesscomplexflowers(e.g.diskorhead),whichhaveamoregeneral- izedpollinationsystem[82].Finally,theorganizationoffloralreproductiveunitsatdifferent PLOSONE|DOI:10.1371/journal.pone.0138414 September21,2015 10/18

Description:
Official Full-Text Paper (PDF): Lessons from Red Data Books: Plant Among all the floral complexity components only floral symmetry was found to
See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.