Significant Local-Scale Plant-Insect Species Richness Relationship Independent of Abiotic Effects

16 Pages·2017·0.91 MB·English

RESEARCHARTICLE
Significant Local-Scale Plant-Insect Species
Richness Relationship Independent of Abiotic
Effects in the Temperate Cape Floristic Region
Biodiversity Hotspot
JureneE.Kemp*,AllanG.Ellis*
BotanyandZoologyDepartment,StellenboschUniversity,Matieland,SouthAfrica
a1111111111
*[email protected](JEK);[email protected](AGE)
a1111111111
a1111111111
a1111111111
Abstract
a1111111111
Globallyplantspeciesrichnessisasignificantpredictorofinsectrichness.Whetherthisis
theresultofinsectdiversityrespondingdirectlytoplantdiversity,orbothgroupsresponding
insimilarwaystoextrinsicfactors,hasbeenmuchdebated.Hereweassessthisrelation-
OPENACCESS
shipintheCapeFloristicRegion(CFR),abiodiversityhotspot.TheCFRhashigherplant
Citation:KempJE,EllisAG(2017)Significant diversitythanexpectedfromlatitude(i.e.,abioticconditions),butverylittleisknownabout
Local-ScalePlant-InsectSpeciesRichness
thediversityofinsectsresidinginthisregion.Wefirstquantifydiversityrelationshipsatmulti-
RelationshipIndependentofAbioticEffectsinthe
plespatialscalesforoneofthedominantplantfamiliesintheCFR,theRestionaceae,and
TemperateCapeFloristicRegionBiodiversity
Hotspot.PLoSONE12(1):e0168033.doi:10.1371/ itsassociatedinsectherbivorecommunity.Plantandinsectdiversityaresignificantlyposi-
journal.pone.0168033 tivelycorrelatedatthelocalscales(10–50m;0.1–3km),butnotattheregionalscales(15–
Editor:HanY.H.Chen,LakeheadUniversity, 20km;50–70km).Thelocalscalerelationshipremainssignificantlypositivelycorrelated
CANADA evenwhenaccountingfortheinfluenceofextrinsicvariablesandothervegetationattributes.
Received:June17,2016 Thissuggeststhatthediversityoflocalinsectassemblagesmaybemorestronglyinflu-
encedbyplantspeciesrichnessthanbyabioticvariables.Further,vegetationageandplant
Accepted:November23,2016
structuralcomplexityalsoinfluencedinsectrichness.Theratioofinsectspeciesperplant
Published:January11,2017
speciesintheCFRiscomparabletoothertemperateregionsaroundtheworld,suggesting
Copyright:©2017Kemp,Ellis.Thisisanopen
thattheinsectdiversityoftheCFRishighrelativetootherareasoftheglobewithsimilarabi-
accessarticledistributedunderthetermsofthe
oticconditions,primarilyasaresultoftheunusuallyhighplantdiversityintheregion.
CreativeCommonsAttributionLicense,which
permitsunrestricteduse,distribution,and
reproductioninanymedium,providedtheoriginal
authorandsourcearecredited.
DataAvailabilityStatement:Wehavedeposited
thedataintheOpenScienceFramework(URL:
https://osf.io/sx545/).
Introduction
Funding:NationalResearchFoundation(ZA)
providedpersonalfundingtoJEK(nogrant Arthropodsassociatedwithplantsconstituteamajorpartoftheearth’sbiodiversity[1].Since
number;URL:http://www.nrf.ac.za/).TheSouth herbivoresfeedonplants,itisgenerallyacceptedthatherbivorousinsectdiversityshould
AfricanBiosystematicsInitiativeprovidedproject
increasewithanincreaseinplantdiversity[2–4],andthisshouldsubsequentlyalsoleadtoan
fundingtoAGE(grantnumber:74458;URL:http://
increaseinpredaciousinsectdiversity[5].However,boththestrengthoftheassociation
www.sassb.co.za/sabi.htm).Stellenbosch
betweenthesegroupsandthekeydeterminantsoftherelationship,arestilldebated[6].The
UniversityprovidedprojectfundingtoAGE(no
grantnumber;URL:http://www.sun.ac.za/english). strengthandslopeofthisrelationshipcouldvarybetweenregionsduetovariationininsect
PLOSONE|DOI:10.1371/journal.pone.0168033 January11,2017 1/16
Plant-InsectRichnessRelationshipintheCFR
Thefundershadnoroleinstudydesign,data hostspecialisation[7],plantbiomass[8],plantphylogeneticdiversity[4,9],orabioticfactors
collectionandanalysis,decisiontopublish,or [10–12].
preparationofthemanuscript.
Speciationresultingfromevolutionarytransitionsofspecialistherbivoresfromonehostto
CompetingInterests:Theauthorshavedeclared anotherhasbeeninferredasanimportantmechanismdrivingarthropoddiversity[13].If
thatnocompetinginterestsexist. insectsarehighlyspecialisedonplants,ahighdiversityofplantsshouldleadtoahighdiversity
ofinsectherbivoresatboththecommunityandregionalscales,andthusapositiveassociation
betweenplantandinsectrichnessisexpected[2,4,5].Becauseinsectsareoftenspecialisedon
plantsatthegenericorfamilylevel[14–16],acommunityofdistantlyrelatedplantspecieswill
morelikelyfallwithinthehostrangeofalargervarietyofherbivores,resultinginhighherbi-
vorediversityinphylogeneticallydiverseplantcommunities[4].Castagneyroletal.[9]suggest
thestrengthofplant-insectrichnessrelationshipversusthestrengthofinsectrichnessand
plantphylogeneticrelationshipsiscontingentonthepatternsofherbivorespecialisation.The
relationshipbetweeninsectrichnessandplantphylogeneticdiversitywillbestrongerthanthe
alternativeifinsectsarespecialisedathighertaxonomiclevels.
However,thepositivecorrelationbetweenplantandinsectdiversityisnotnecessarilythe
resultofadirectassociation.Forexample,Hawkins&Porter[10]foundthatonceenviron-
mentalvariableswerecontrolledfor,planthostdiversityandCalifornianbutterflydiversity
werenotcorrelated.Craftetal.[17]showedthatwhilegeneticstructureoftropicalinsectsisin
somecasesassociatedwithhostspecialisation,itoftenmirrorsotherlandscapegradients.
Thesepatternssuggestthattheplant-insectdiversityrelationshipmightariseduetosimilar
responsesofbothgroupstoenvironmentalgradients.Inaddition,propertiesofplantcommu-
nitiesbesidesspecies/phylogeneticdiversitymayinfluenceinsectherbivorediversity,suchas
vegetationstructure[18],orplantphenophase[19].
Themajorityofstudiesinvestigatingplant-insectdiversityrelationshipshavefocusedon
diversetropicalsystems[20–23],whileotherhyperdiversesystemshavereceivedmuchless
attention.TheCapeFloristicRegion(CFR)ofSouthAfricaisarecognisedbiodiversityhotspot
[24]thatcontainsmorethan9000plantspeciesin90000km2[25].Whileplantspeciesrich-
nessofthemajorfloristickingdomscanbepredictedbyenvironmentalfactors,theCFRrepre-
sentsaclearexception,containingmorethantwicetheplantspeciesrichnessperunitarea
thanpredictedfromenvironmentalconditions[26].ThelimitedworkonCFRinsectdiversity
suggeststhatCapeplantandinsectdiversitymaybepositivelycorrelated[27–30].Authors,
however,disagreeonwhetherCFRinsectdiversityishigh[27,31],depauperate[32,33]or
comparabletoneighbouringregions[30].
HereweexploreCFRplant-insectdiversityrelationshipsusingthespeciesrichRestionaceae
andtheirassociatedinsectherbivoresasamodelsystem.Thisisanidealsystemforinvestigat-
ingthelinksbetweenplantandinsectdiversitysincetheRestionaceaesupportadiverseassem-
blageofinsectherbivores,areadominantcomponentofCFRvegetationandspeciesrichness
withincommunitiesvariessubstantiallyacrosstheregion[34].Inaddition,theirfairlyuniform
growthformsfacilitatestandardisedsamplingofherbivorecommunities.TheRestionaceae
arewind-pollinated,andthusassociatedinsectsarelikelyinteractingantagonistically.Further,
oursamplingareainthecoreCFRhaslowclimaticvariation,whichallowsustoassessthe
plant-insectdiversityrelationshipwhilstcontrollingforbroadclimaticdifferences.
First,weaskwhetherherbivorediversityiscorrelatedwithplantspeciesandphylogenetic
diversityintheRestionaceaesystemanduseaspatiallynestedsamplingdesigntoexplorethe
strengthofthisrelationshipatvariousspatialscales(i.e.localvs.regional)withintheCFR.We
predictthatifinsectsarespecialisedonplantsatthespecieslevel,weshouldseeastrongrela-
tionshipbetweenplantandinsectrichness.Ifinsectsarespecialisedathighertaxonomiclevels,
plantphylogeneticdiversityshouldbeabetterpredictorofinsectrichness.Itisimportantto
notethat,becausewesampledasingleplantfamily,andinsectherbivoresareoftenspecialised
PLOSONE|DOI:10.1371/journal.pone.0168033 January11,2017 2/16
Plant-InsectRichnessRelationshipintheCFR
attheplantgenusorfamilylevel[15,16,35],ourstudyisinfactbiasedagainstdetectionofsig-
nificantplant-insectdiversityrelationships.Secondly,weaskwhichaspectsofplantdiversity
(species,phylogenetic,structuralorphenophasediversity)bestpredictinsectdiversityand
whethertheserelationshipsareinfluencedbyextrinsicfactors(i.e.post-firevegetationage,
andelevationasaproxyforfinerscaleclimaticvariance).Ifplant-insectdiversityrelationships
emergethroughevolutionofhostspecialisation,weexpectthatplantdiversityshouldremain
thebestpredictorofinsectdiversitywhencontrollingfortheinfluenceofenvironmentalgra-
dientsandothercharacteristicsofplantcommunities.Finally,weassembleavailabledatasets
todeterminehowtheCFRinsect-plantdiversityrelationshipcomparestootherregions,and
toallowpreliminaryassessmentofwhetherinsectdiversityinthisregionfollowsthesame
trendastheplants,wherediversityishigherthanexpectedfromtheCFR’slatitude.
MethodsandMaterials
Studysystem
TheAfricanRestionaceae(hereafterrestios)isoneoftheoldestplantcladesintheCFRand
originatedapproximately91.5millionyearsago[36].Thiswind-pollinatedmonophyletic
cladeofreed-likeplantscontains350species[37];allofwhicharedioeciousandsomeshow
dimorphismbetweenmaleandfemalereproductivestructures.Restioleaveshavebeen
reducedtosheathsrolledaroundtheculmsatintermittentnodes.WhileRestionaceaehavea
typicalgraminoidgrowthform,speciesexhibitdifferencesinplantheight,culmdiameterand
branchingoftheculms.RestiosoccurthroughouttheCFRinhabitatsthatvaryinsoiltype,ele-
vation,groundwateravailability,slope,aspectandclimate.Thehighabundanceofthisgroup
inavarietyofhabitatsmakesitidealforassessingtheplant-insectdiversityrelationshipandits
correlatesintheCFR.
InsectdiversityintheCFRhasbeensuggestedtobelow[32],butmorerecentstudieshave
foundhighdiversityingallinginsects[27]andbees[29].ThesclerophyllousleavesofCFR
plantsmayactasdeterrenttofolivores[33,38]andthelowsoilnutrients(leadingtolowplant
nutrients)couldfavourgeneralisminherbivorousinsects,whereinsectsmayswitchseasonally
betweenplantspeciestooptimisenutrientintake[19].Alternately,insectsmaybespecialised
onaplantspeciesandonlybepresentinthecommunitywhennutrientuptakefromthatplant
speciesisoptimal.LeafhopperspeciesinthetribeCephalelini(Cicadellidae)havebeenshown
tobespecialisedonRestionaceaetaxa[19,39–41],whereasBetiscoidesgrasshoppersarespecial-
isedontheRestionaceaeatthefamilylevel[42].
Samplingdesign
Aspatiallynestedsamplingapproachwasused(Fig1).Thirtyrestio-dominatedsiteswere
selectedforsampling.Theseweresituatedonthreeofthemajormountainblocksinthesouth-
westernCape,namelyHottentots-Holland,KogelbergandtheCapePeninsula,50–70km
apart.Thesemountainsexperiencesimilarclimaticconditions,whichallowsustoassessthe
plant-insectdiversityrelationshipwhilstcontrollingforbroadclimaticdifferences(seeS1
Table).Eachmountainblockcontainedtwoclustersoffivesamplingsites(15–20kmapart).
Thefivesitesineachclusterweresituated100mto3kmfromoneanother.Eachsiteconsisted
oftwo10x10msamplingsquareslocated10to50mapart.Five2.5x2.5msamplingplots
werelocatedwithineachsquare(theplotsweresituatedinthecornersandcentreofeach
square).Eachofthethreemountainsthusconsistedoftwoclusters,tensites,twentysquares
andahundredplots.Thisallowedustoexploretheplant-herbivorerelationshipatvariousspa-
tialscales,i.e.theplot,square,site,cluster,mountainandregionallevel.Siteswithknownres-
tiospeciescompositionwerechosentoallowustosamplesitesthatvariedinplantspecies
PLOSONE|DOI:10.1371/journal.pone.0168033 January11,2017 3/16
Plant-InsectRichnessRelationshipintheCFR
Fig1.Mapdepictingsamplingareaandsamplingdesign.Aspatiallynestedsamplingdesignwasemployed.Thirtysites(representedbyblack
circles)weresampledtwice,onceineachseasonofpeakinsectactivity(autumnandspring).Groupsoffivesiteswerespatiallyaggregatedtoformsix
clusters(smallgreysquares).Twoclusterswerepresentineachofthethreemountainblockssampled(largerectangles).Sitesconsistedoftwo10x
10msquaressituated10–50mapart(insertontheright).Eachsquarecontainedfive2.5x2.5mplots(fourcornersandcentreofthesquare).These
plotsweresampledbothforRestionaceaeplantsandallinsectherbivorespresentonRestionaceaeplants.
doi:10.1371/journal.pone.0168033.g001
richnessindependentlyofelevationandvegetationage.Plotsweresampledtwice,onceduring
thesuggestedinsectpeakseason[43](i.e.Australspring:August-October2013)andoncesix
monthsbeforethis(i.e.Australautumn:March-April2013).Thesesamplingperiodscoincide
withthetwopeaksinCephaleliniabundance[19].Thesewillhereafterbereferredtoasspring
andautumnrespectively.
Insectsamplinganddiversityestimation
InsectswerecollectedfromallRestionaceaeplantsoccurringineachplotusingamodified
leaf-blowerwitha15cmdiameternozzleandplacedin70%ethanol.Allrestioswereexhaus-
tivelyvacuum-sampledforapproximately15–20secondsperplantandthenozzlewasmoved
systematicallyupanddowntheculms.Nwokwu&Sanderson[44]foundthatusingamodified
leaf-blowercapturedmoreinsectsthansweep-nettingorpitfalltrapping,bothintermsofrich-
nessandabundance.Restiosweresearch-sampledforinsectsaftervacuumsamplingtoassess
itsefficiencyandalsotoseewhethergalling/mininginsectswerepresent.Extremelyfew
insectswerefoundbysearch-samplingandnogallingormininginsectswerepresent.Insects
wereidentifiedtosuperfamilyorfamilyandthensortedintomorphospecies.Oliver&Beattie
[45]showedmorphospeciestobesufficientsurrogatesforspecies,especiallyinestimatesof
speciesrichness.Sampleswerematchedacrossseasons.Insectfamiliesknowntobenon-her-
bivorouswereexcludedfromthedataset.Insectfamiliesknowntoonlyfeedonnectarof
plants(absentinrestios)wereviewedastransientvisitorsandalsoexcludedfromthedataset.
Nojuvenileinsectswereincludedinthedatset.
PLOSONE|DOI:10.1371/journal.pone.0168033 January11,2017 4/16
Plant-InsectRichnessRelationshipintheCFR
TheCephalelini(adominantinsectgroupinoursurveys)wereidentifiedtospeciesbydis-
sectingmalegenitaliaandusingthespeciesdescriptions[39,46],andmatchingspecimensto
museumcollections(StellenboschUniversity,ConservationEcologyandEntomologydepart-
ment).Femaleswerematchedtomalesusingexternalmorphologyandmuseumspecimens.
TheinsectmorphospeciescollectionishousedintheBotanyandZoologydepartmentatStel-
lenboschUniversity.
Samplingeffectivenesswasevaluatedbyconstructingindividualbasedspeciesaccumula-
tioncurves(numberofspeciesfoundpernumberofindividualssampled)fortheentireregion.
Accumulationcurves(S1Fig)tendedtowardssaturationandsamplingwasthusdeemedsuffi-
cienttoallowcomparisonsbetweenourstudyandotherregions.Rarefactioncurvesbasedon
thenumberofplotsateachsamplingscaleshowedsomevariationinthelevelofapproachto
saturationbetweensamplingscales(S2–S5Figs).
InsectalphadiversitywascalculatedintermsofHillnumbers(ornumbersequivalents)of
theShannondiversityindex.Thisdiversitymetrichastheadvantageofexhibitingadditivity
andisnotbiasedtowardsrareorcommonspecies[47].Alphadiversitywascalculatedforeach
plot,square,site,cluster,mountainandtheentireregion.
Plantsamplinganddiversitycomponents
Restionaceaespeciesoccurringineachplotwereidentifiedusingtheonlineinteractivekeyof
[48].Abundancesofeachrestiowererecordedforeachplot.Plantheightwasrecordedand
eachplantwasplacedinastructuralheightcategory:0–0.5m,0.5–1m,1–1.5m,1.5–2m,
and>2m.Thebranchingorderofplantswasrecordedasunbranched,branched(eachculm
branched1–3times)orhighlybranched(eachculmbranchedmorethanthreetimes).Each
plantwasplacedinadiscretestructuralgroupcategorybasedonacombinationofitsheight
andbranchingorder(e.g.a0.7mtallunbranchedplantwouldfallinthe“Unbranched_0.5–1
m”bin).Thenumberofplantsinflowerineachplotwasalsorecorded.
Plantalphadiversitywascalculatedinthesamemannerasinsectalphadiversity(basedon
Hillnumbers).Plantphylogeneticdiversity(PD)wascalculatedfromtheRestionaceaephylog-
eny(Linder&Bouchenak-Khelladi,unpublished)usingtheRpackagepicantetocalculate
Faith’sPD.Plantstructuraldiversitywascalculatedusingthealphadiversitymetric(Hillnum-
bers)forplantstructuralgroupsdescribedabove.Structuraldiversityindicatesthenumber
(andabundance)ofdifferentstructuralgroupsorformsinaplot.Amonospecificplotwithtall
highlybranchedrestioswouldthusscorethesameasamonospecificplotofshortunbranched
restios.However,tallorhighlybranchedrestiosmayprovidemorefeedingnichesandsupport
ahigherdiversityofherbivores.Thuswealsousedaplantstructuralcomplexitymetricwhich
aimedtocapturetherelativeamountofculmspaceavailabletoherbivoresineachplot.This
wascalculatedforeachplotbymultiplyingthebranchingscoreofeachplant(where
unbranchedplantsscored1,branchedplantsscored3andhighlybranchedplantsscored5)by
itsheight,andsummingacrossallplantsinaplot.Alloftheabovecalculationswererepeated
forplots,squares,sites,clustersandmountains.
Environmentalpredictors
Theelevationandageofthevegetationafterthemostrecentfirewasdocumentedforeach
square(allplotsineachsquaresharedthesameelevationandvegetationage).TheCFRburns
regularly(every10–15years)andvegetationinsquaresrangedinpost-fireagefrom2to20
years.Elevationofsquaresvariedbetween44and968mabovesealevel.Elevationalgradients
oftenreflectenvironmentalgradients,suchasprecipitation,temperature,andwindspeeddif-
ferences,andhereweuseitasaproxyforfinerscaleclimaticdifferencesbecauseavailable
PLOSONE|DOI:10.1371/journal.pone.0168033 January11,2017 5/16
Plant-InsectRichnessRelationshipintheCFR
climaticdata(i.e.Worldclim)didnothavespatialresolutionappropriatetooursamplingscale
andexhibitednosignificantdifferencesacrosssampledmountainblocks(S1Table).
Dataanalysis
WefirstusedlinearregressionsimplementedinR[49]toassesstheplant-insectdiversityrela-
tionshipintheRestionaceaesystemindependentofotherpotentialpredictors(e.g.vegetation
structure,vegetationage,environmentalvariables,etc.).Thisalloweddirectcomparisonto
previousdiversitystudies[27,28,50].Wetestedforrelationshipsbetweeninsectspeciesrich-
nessandplantspeciesrichnessandphylogeneticdiversity,bothlocally(i.e.plot,square,site
scale)andregionally(i.e.clusterscale).Analysisofvariance(ANOVA)wasusedtoassess
whetherplantandinsectspeciesrichnessinplots,squaresandsitesvariedbetweenmountains.
Insectalphadiversitywasalsoregressedontoplantalphadiversitytotestwhetherplantcom-
munitieswithmoreevenabundanceshostinsectcommunitieswithevenabundances.Totest
whetheranincreaseinplantrichnesslinearlyinfluencedbothinsectrichnessandevenness,we
alsoregressedplantrichnessontoinsectalphadiversity.
Next,generallinearmodels(GLMs)wereusedtoassesstheinfluenceoftheadditional
plantdiversitycomponents(structuraldiversity,structuralcomplexityandphenophase
diversity)andenvironmentalvariablesoninsectspeciesrichness.Weconductedfivesepa-
rateGLMstoexploretheinfluenceofenvironmentalvariablesandadditionalplantdiversity
componentsontheinsect-plantspeciesrichnessrelationship.Wefirstlyanalysedthese
effectsonthe,thefulldataset(i.e.seasonaldatawascombinedforeachsquare),andthen
dividedthedataintosubsetsusing1)theautumndataset,2)thespringdataset,3)Hemiptera
only,4)Coleopteraonly.Thisallowedustodeterminewhetherdiversityrelationshipsdiffer
betweenseasonsandacrossthedominantinsectorders.WeusedtheAkaikeinformationcri-
terion(AIC)fromstepwisebackwardeliminationtodeterminewhichpredictorvariables
shouldbeincludedineachoftherespectiveGLMs.Modelfitwascalculatedwithallpredic-
torsincludedinthemodelandpredictorswerethenremovedonebyonetoassesswhether
modelfitimproved.ThemodelwiththebestfitwasthenusedforeachofthefiveGLMs.The
function“stepwise”inR[49]wasusedforthis.Responsevariables(insectspeciesrichness)
werelog (x+1)transformedwherenecessarytoimprovenormality.Squareswereusedas
10
inputforallGLMsasthisisthesamplingscaleatwhichenvironmentalvariableswere
recorded.
TotestfordifferencesintheratioofinsectspeciestoplantspeciesbetweentheCFRand
otherregions,wecompileddatafromexistingstudieswhichassessedtheinfluenceofplant
diversityoninsectdiversity.Variouscombinationsofthesearchterms“insect”,“plant”,“rich-
ness”,“relationship”,“diversity”,“CapeFloristicRegion”and“tropics”wereusedtolocate
studiesthatrelateplantandinsectrichnessonGoogleScholarandISIWebofScience.Studies
which1)onlyassessedasingletaxonofinsects,2)werebasedonexperimentaltreatmentsor
3)onlysampledavegetationtype(ratherthanreportingplantspeciesrichness)wereexcluded.
Ratiosofthetotalnumberofinsectspeciessampledtototalnumberofplantspeciessampled
wereextractedfromthe13studieswecouldfindthatgenerateddatasetscomparabletoour
own(Table1).Asthesestudiesdifferedinthegrowthformsampled(tropicalstudiesonlysam-
pledtrees,whileCFRstudiessampledshrubs,graminoidsandtrees)whichislikelytostrongly
influenceinsect:plantratios,wefirsttestedforaninfluenceofgrowthformonratiosusing
ANOVA.Thisrevealedastronginfluenceofgrowthform(F =11.14,p=0.001)withtree
2,12
specieshousingsignificantlymoreinsectspeciesthanshrubsorgraminoids.Thisisnotsur-
prisingsince[8]showedleafbiomasstobeastrongpredictorofinsectabundance.Asaresult,
treeswereexcludedfromtheregionalcomparison(usingWelch’st-test)ofinsect:plantratios
PLOSONE|DOI:10.1371/journal.pone.0168033 January11,2017 6/16
Plant-InsectRichnessRelationshipintheCFR
Table1. Numberofinsectspeciesperplantspeciesforvariousregions(i.e.theCapeFloristic
Region,temperatezones,tropicalzones). Thenumberofinsectspeciesperplantspeciesissignificantly
higherinthetropicsthantheCFR,althoughthisisentirelyconfoundedbyplantgrowthform.TheCFRdoes
nothavesignificantlymoreinsectspeciesperplantspeciesthanothertemperatezones,suggestingregional
plantrichnessandgrowthformarebetterpredictorsofregionalinsectrichnessthanenvironmentalvariables.
Insect:Plantratio Growthfrom Region Study
5.82 grass CFR Thisstudy
1.45 shrub CFR Prochessetal.(2009)[28]
2.06 shrub CFR Pryke&Samways(2008)[43]
4.34 grass temperate Zhuetal.(2012)[51]
3.24 grass temperate Collinge(2000)[52]
13.21 tree temperate Novotny´etal.(2006)[20]
21.60 tree temperate Marquesetal.(2006)[53]
1.24 shrub temperate Frenzel&Brandl(2003)[54]
2.18 shrub temperate Schaffersetal.(2008)[55]
23.27 tree tropical Basset&Novotny´(1999)[56]
39.86 tree tropical Novotny´etal.(2006)[20]
22.76 tree tropical Novotny´etal.(2012)[50]
20.20 tree tropical Lepsetal.(2001)[57]
35.00 tree tropical Novotny´&Basset(2000)[58]
doi:10.1371/journal.pone.0168033.t001
fromtheCFRandothertemperateregions(i.e.highplantdiversityandlowplantdiversity
temperateareas).Wealsousedat-testtocompareCFRandtropicalinsect:plantratioseven
thoughregionandgrowthformwereentirelyconfoundedinthiscomparison.
Results
Plantandinsectcomposition
Atotalof321herbivorousinsectmorphospecieswerecollected(7276individuals),221insect
species(3619individuals)duringautumn(March-April2013),and195species(3657individu-
als)duringspring(August-September2013).Therestioherbivorecommunitywasstrongly
dominatedbyHemiptera,bothintermsofspeciesrichness(42.6%ofspecies)andabundance
(58.9%ofindividuals)(Fig2).Thehemipterangroups,CephaleliniandFulgoroidea,respec-
tivelycomprisedapproximately10%(773individuals)and17%(1237individuals)ofallinsect
individualssampled.AlthoughColeopteraarethelargestorderofdescribedspeciesglobally,
theydonotdominatethissystem.
Theinsectsweresampledfrom5248Restionaceaeplants(55species;11genera).Mean(±
SD)ofplantspeciesrichnessforplotswas3.04±1.80(range:1–11),forsquares4.48±2.43
(range:1–11),forsites5.57±2.79(range:2–14),forclusters18±7.95(range:8–29)andfor
mountains28.33±12.50(range:16–41)(Table2).
Plant-insectdiversityrelationship
Insectalphadiversityandspeciesrichnesswerealwayssignificantly(ornearlysignificantly)
positivelyassociatedwithplantdiversity(alphadiversity,PDandspeciesrichness)attheplot
(i.e.2.5x2.5msamplingunit)andsquare(10x10m)scales(Table3).However,plant-insect
diversityrelationshipswerenotsignificantatlargersamplingscales(Fig3;Table3),although
therelationshipwasstillpositiveatthesitescale.Whilesignificant,plantdiversitycomponents
onlyexplainedamaximumof11%ofvarianceininsectdiversityatthesquarescale.Plant
PLOSONE|DOI:10.1371/journal.pone.0168033 January11,2017 7/16
Plant-InsectRichnessRelationshipintheCFR
Fig2.Themorphospeciesrichnessandabundanceperinsectorder.Totalmorphospeciesrichness(blackbars)and
abundance(greybars)foreachinsectordercapturedduringvacuumsamplingsurveysofRestionaceaecommunities
acrosstheCapeFloristicRegion.Hemipteradominatedherbivorecommunitiesonrestios.Onlyherbivorousinsectsare
included.Themean(±SD)insectmorphospeciesrichnessforplotswas9.47±4.76(range:0–27),forsquares29.25±10.09
(range:12–60),forsites44.80±12.80(range:26–70),forclusters126.17±9.06(range:118–144)andformountains
192.67±22.55(range:171–216)(Table2).
doi:10.1371/journal.pone.0168033.g002
speciesrichnesswasastrongerpredictorofinsectspeciesrichnessthanplantphylogenetic
diversity.
PlotsintheKogelbergmountainhadsignificantlyhigherinsectspeciesrichnessthanplots
ontheCapePeninsula,butnootherdifferenceswerepresent(ANOVA:F =3.231,
2,297
Table2. Themeaninsectandplantrichnessforeachsamplingscale. Themean(±SD)insectandplantrichnessofeachmountainblock(HH:Hotten-
tots-Holland;KB:Kogelberg;CP:CapePeninsula)isshownforthevarioussamplingscales.
Samplingscale Group HH KB CP
Plot Insect 9.62±4.29 10.2±5.64 8.56±4.1
n=300 Plant 3.02±1.73 3.84±2.11 2.25±1.04
Square Insect 27.7±7.77 32.6±13.1 27.4±8.13
n=60 Plant 4.35±2.35 6.15±2.43 2.95±1.23
Site Insect 42.5±10.3 42.7±13.8 49.2±14.1
n=30 Plant 5.1±2.28 7.8±2.94 3.8±1.4
Cluster Insect 124±1.41 121±4.24 134±14.8
n=6 Plant 17±2.83 27±2.83 10±2.83
doi:10.1371/journal.pone.0168033.t002
PLOSONE|DOI:10.1371/journal.pone.0168033 January11,2017 8/16
Plant-InsectRichnessRelationshipintheCFR
Table3. Insect-plantdiversityrelationshipatvariousspatialsamplingscales. Whiletherelationshipbetweeninsectandplantdiversitycomponents
wasalwayspositive,itwasonlysignificantatthesmallersamplingscaleswiththemoststatisticalpower,andplantdiversitycomponentsexplainedamaxi-
mumof11%ofvarianceininsectdiversity.
Scale Insectspeciesrichness~ Insectspeciesrichness~ Insectspeciesαdiversity~ Insectspeciesαdiversity~
Plantspeciesrichness PlantspeciesPD Plantrichness Plantspeciesαdiversity
R2 p R2 p R2 p R2 p
Plot(n=300) 0.076 <0.001 0.040 <0.001 0.052 <0.001 0.031 0.001
Square(n=60) 0.107 <0.001 0.052 0.043 0.046 0.055 0.049 <0.001
Site(n=30) 0.006 0.288 -0.020 0.522 0.084 0.065 0.066 0.092
Cluster(n=6) 0.035 0.338 0.103 0.278 0.047 0.429 0.085 0.292
doi:10.1371/journal.pone.0168033.t003
p=0.041).Nodifferenceswerepresentbetweenmountainsforsquares(F =1.751,
2,57
p=0.183)orsites(F =0.880,p=0.427).Plantspeciesrichnesspermountainwassignifi-
2,27
cantlyhigherinKogelbergthaninHottentotsHolland,andbothofthesewerehigherthanthe
CapePeninsulaforplots(F =22.19,p<0.001),squares(F =11.92,p<0.001),and
2,297 2,57
sites(F =9.711,p=0.002).
2,27
Fig3.Relationshipbetweenplantandinsectrichnessatvarioussamplingscales.Theplotscaleis
representedbyasolidlineandcircles,thesquarescaleisrepresentedbyadashedlineandcrosses,andthesite
scaleisshownbyadottedlineandtriangles.Theassociationispositiveattheplotandsquarescales(p<0.001),but
notsignificantatthesitescale.
doi:10.1371/journal.pone.0168033.g003
PLOSONE|DOI:10.1371/journal.pone.0168033 January11,2017 9/16
Plant-InsectRichnessRelationshipintheCFR
Table4. ResultsofGLMmodelswithinsectspeciesrichness(atthesquarelevel)astheresponsevariable. Themodelwasfirstrunwiththeentire
dataset(i.e.allspeciesforbothseasons),andthenonsubsetsofthedata(i.e.seasonsseparately,andmaininsectordersseparately).Datafromall60sam-
pledsquareswereincluded.
Insectrichness Plantcomponents Extrinsiccomponents Modelfit
dataset Plantrichness Structuraldiversity Structuralcomplexity Flowering Vegetationage Elevation R2
Overall All (+)0.012* (+)0.015* 0.16
Subset Autumn (+)0.005* (-)0.157 (+)0.004* 0.18
Spring (-)0.008* 0.10
Hemiptera (+)0.022* (-)0.486 0.07
Coleoptera (+)0.006* (-)0.025* 0.10
P-valuesofallpredictorsincludedineachmodelareshownandsignificantvaluesareindicatedwithanasterisk(*).Thelastcolumndepictsmodelfit(R2).
Thedirectionofeffectsareindicatedinbrackets.
doi:10.1371/journal.pone.0168033.t004
Forthefulldataset,additionalvarianceininsectspeciesrichnessacrosssquareswas
explainedonlybypost-firevegetationage,andnotbyelevationorplantstructuralattributes
(Table4).Oldervegetationhadmoreinsectspecies.Themodelincludingbothplantrichness
andvegetationageexplained16%ofvarianceininsectrichness,whileplantrichnessalone
explained11%(Tables3and4).Seasonalpatternsdifferedinthatduringspringplantstruc-
turalcomplexitywasthestrongestpredictorwhileplantstructuraldiversityandnumberof
plantsfloweringwerestrongerpredictorsinautumn.Similarly,Hemipteranrichnesswasmost
stronglyrelatedtoplantstructuraldiversity,whileColeopteranrichnesswasbestpredictedby
plantrichnessandstructuralcomplexity.
Comparisonofinsect:plantspeciesratiosbetweentheCFRandother
regions
ThemeannumberofinsectspeciesperplantspeciesintheCFRdidnotdifferfromothertem-
perateregionswhentheconfoundingeffectoftreeswasexcludedfromanalysis(t=0.2350,d.f.
=2.9626,p=0.83).
Althoughdifferencesinplantgrowthformentirelyconfoundthecomparisonbetween
tropicalregionsandtheCFR,thenumberofinsectspeciesperplantspeciesissignificantly
lowerintheCFRthaninthetropics(t=-4.3214,d.f.=5.753,p=0.005),despiteplantdiversity
beingsimilarbetweentheseregions[59].
Discussion
ThepositiverelationshipbetweenthespeciesrichnessofRestionaceaeplantsandtheirassoci-
atedherbivorousinsectassemblagesconfirmspreviousreportsofplant-insectdiversitylinkage
intheCFR[27,28,30,43].Othercomponentsofplantdiversity(i.e.structuraldiversityand
complexity)andpost-firevegetationagealsoexplainedsignificantamountsofvariancein
insectspeciesrichnessandthecontributionofthesepredictorsvariedbetweenseasonsand
acrossthedominantinsectorders.However,whencombiningseasonsandinsectgroups,
plantspeciesrichnesswasapersistentsignificantpredictorofinsectrichnessevenwhen
accountingfortheinfluenceoftheseothervariables.
Insectsoftenspecialiseonplantsatthegenusorfamilylevel[15,16,35].Thusthepositive
diversityrelationshipwedemonstratewithinasingleplantfamilyisquiteremarkableand
likelyindicatesspecificityatfinertaxonomiclevels.Thisisfurthersupportedbyplantrichness
beingabetterpredictorofinsectrichnessthanplantphylogeneticdiversity.Similarresults
werefoundbyKempetal.[60]whereplant-herbivoreinteractionsintheCFRshowedhigh
PLOSONE|DOI:10.1371/journal.pone.0168033 January11,2017 10/16