Correlatedevolutionoffruitandleafsizeinbird-dispersedplants:
species-levelvarianceinfruittraitsexplainedabitfurther?
CarlosM.Herrera, Estacio´nBiolo´gicadeDon˜ana,ConsejoSuperiordeInvestigacionesCient´ıficas,Apdo 1056,
E-41080Sevilla,Spain().
Theastoundingmorphologicaldiversityexhibitedbythefruitsof vertebrate-dispersedplantshasbeentraditionallyinterpretedas theadaptiveoutcomeofdivergentselectivepressuresexertedon plantsbythebroadarrayoffrugivorousanimalsinvolvedinseed dispersal.Althoughtheselectivecapacityoffrugivoresprovides supporttothisinterpretation,recentstudieshavechallengedit bydocumentingastrongphylogeneticcomponentassociatedto interspecificvariationinmostfruitcharacteristics.Size-related fruittraits providea conspicuousexceptionto this pattern, because theyexhibitconsiderablevariationatthebetween-spe- cies level whichis largely independentof phylogenyandis correlatedwithconsumptionby differently-sized dispersal agents.Substantialspecies-levelvarianceinsize-relatedtraits mayreflectgenuinedisperser-driven diversification,butmayalso bepartlyinfluencedbycorrelatedevolutionoffruitsizewiththe sizeofotherplantstructures.Thislatterpossibilityistestedhere forbird-dispersedplantsoftheIberianPeninsulausingphyloge- neticallyindependent contrasts.Resultsdemonstratetheexis- tenceofcorrelatedevolutionoffruitandleafsizeatthespecies level.Asalltheplanttaxaconsideredhavetheirfruitseaten,and seedsdispersed,bythesamerelativelyreducedsetoffrugivorous birdspecies,resultssuggestthata significantfractionof the variationinfruitsizerepresentedinthespeciessamplemaybe explainedasanindirectconsequenceofvariationinleafsize, ratherthanbeingassociatedwithadaptivedivergencerelatedto seeddispersalagents.
individuals with fruits differing incharacteristics such ascolor,sizeandnutritionalreward (HoweandVande Kerckhove 1980,McPherson1988,Willsonetal.1990, Avery et al. 1993,Sallabanks 1993,Rey et al. 1997). Thisundisputableselectivecapacityoffrugivores,along with the potential influence of their activity on the fitnessof fruiting plants (Herrera et al. 1994,Schupp
1995),isconsistent with the evolutionaryrole granted tothembyadaptive interpretationsofinterspecificvari- anceinfruitfeatures. Nevertheless, anincreasing num- ber of studies have recently shown that phylogenetic effectsmaybeequallyormoreimportant thanseed dispersers inexplaining patterns ofinterspecific varia- tion in fruit features (Debussche et al. 1987,Herrera
1987, 1992, Ehrle´n and Eriksson 1993, Fischer and
Chapman 1993, Jordano 1995, Eriksson and Ehrle´n
1998;seeHerrera 2002forreview),thuschallengingthe prevailing role of frugivores in the diversification of fruit characteristics.
Size-related fruit traits represent conspicuous excep-
tionstothestrongphylogenetic componentofinterspe- cific variation in morphological and nutritional fruit traits revealed bythe‘‘phylogeny-aware’’studies men-
The fruits ofvertebrate-dispersedplants are notorious forthebewilderingdiversityofsizes,colors,structures, and amount and nutritionalcharacteristics ofthepulp reward offered to vertebrate frugivores (van der Pijl
1969).Thisextraordinaryinterspecificvarianceinfleshy fruitfeatureshasbeentraditionally interpreted asthe ultimate result ofdivergent selection byabroad array of frugivorousmutualiststhatdifferin fruitselection patterns, mobility, trophic apparatus, sensory capac- ities,andfitness consequencestoplantsoftheirseed- processing behavior (van der Pijl1969,McKey 1975, Snow1981).Frugivorousanimalseffectivelypossessthe ability to selectdifferentially among plant speciesand
tioned above. Some of these investigations (Herrera
1992,Fischer and Chapman1993,Jordano1995;see alsoMack1993)haveshownthat asubstantial fraction ofinterspecificvariation insize-relatedfruittraitsisnot accounted forbyphylogeny, and issignificantlycorre- latedwithconsumptionbydifferent kindsofdispersers withcontrastingbodysize distributions(e.g.birdsvs mammals; Jordano1995).Does this residual interspe- cificvarianceinsize-relatedfruittraits(afteraccounting for phylogenetic correlations at the genus level and above)mainlyreflectgenuinedisperser-driven diversifi- cation generated by fruit selection patterns by dis- persersdifferinginbodysizeand/or trophic apparatus?
Thisquestionmaybe addressedby seekingcorrelative evidence relating interspecific variation in size-related fruit traits with differences inthe sizedistributionsof theirconsumerfrugivores.Suchrelationshipshavebeen reported byanumber ofstudies,irrespectiveofwhether theyaccounted (Mack1993,Jordano 1995)ornot (Herrera 1985, 1989, 1992, PrattandStiles 1985, De- busscheandIsenmann 1989)fortheeffectsofphyloge- neticcorrelations. Unfortunately, however,such correlativeevidencefails criticallyatdiscriminating be- tween the adaptive hypothesis that ‘‘differently-sized fruits have come into existence because they have evolvedin responsetoselectionby thedifferently-sized animals that feedonthem’’ from theequally plausible non-adaptive onestatingthat‘‘becauseofinherent morphological orbehavioral restrictions, differently- sizedanimals tend tofeedondifferent segmentsofthe fruit sizespectrum available tothem’’.
Onealternative wayofaddressing thequestion above wouldbe toevaluatethemeritsofplausiblealternative hypotheses that donottakeintoconsiderationtherole ofdispersers,yet mayaccountforsomefractionof interspecificvarianceinfruitsize. Forexample,an increasing number of studies show that the sizes of differentplantstructures,including branches,leaves, flowers,inflorescencesandseeds,aresignificantlycorre- lated across species(White 1983,Primack 1987,Bond andMidgley1988,MidgleyandBond1989,Andersson
1993,Ackerlyand Donoghue1998,Cornelissen 1999). Iffruitsize ofvertebrate-dispersedplantsweresignifi- cantly correlated with the size of some other plant structure, theninterspecificvarianceinsize-relatedfruit traitsmightbeatleastpartlyexplainedbythatrelation- ship. The possibility that fruit and leaf size may be linked has already been suggested byPrimack (1987), whofoundthepredictionsupported inthreeofthesix intragenericcorrelationstested(includingone fleshy- fruited, vertebrate-dispersedgenus).Tomyknowledge, thepossiblerelationship betweenfruitandleafsizehas not been explored to date for vertebrate-dispersed plantsin abroadtaxonomiccontext,despitethecon- siderable effort devoted in the last two decades to exploretheadaptivebasis andecological correlatesof interspecific variation in fruit characteristics (Herrera
1987,Jordano 1995,andreferencestherein).Usingan independent comparisons approach(Felsenstein 1975, Burt 1989,Harvey and Pagel1991),Iexamine inthis paper the relationship between fruit and leaf size among speciesofIberian bird-dispersed plants. Results clearly indicate that fruit size and leaf size exhibit correlated evolutionary change at the terminal, be- tween-specieslevel ofdiversification,andsuggestthat climate-driven diversificationinleafsizemayultimately account foranimportantfraction ofinterspecificvari- anceinfruitsize inthesetofbird-dispersed species considered.
Taxasampleandmethods
Information on mean length, width (cross diameter) andfreshmassofbird-dispersed fruitswasdrawnfrom a large database containing informationon structural and nutritionalcharacteristics ofIberian fleshy-fruited plants. Except for minor additions, the database used was essentially the same analyzed by Herrera (1987) and summarized inTables A1and A2ofthat publica- tion.Thedatabase contains informationonthecharac- teristics of the fruits of 120 species of vertebrate- dispersed plants native to the Iberian Peninsula. This represents about 95%of the families, 87%of genera and anestimated 62%ofspecieswithfleshyfruitsthat occur on the Iberian Peninsula (seeHerrera 1987for sampling methods and geographical locations). I screenedthedatabase forsetsofcongenericspeciesthat would qualify to test for a relationship between fruit and leaf sizeusing phylogenetically independent con- trasts for the two traits. Only speciesknown or pre- sumedtobeexclusivelyorpredominantlydispersed by a relatively small group of small (mainly Sylvia spp.) and medium-sized (mainly Turdus spp.) frugivorous passerine birds wereconsidered, based on information inHerrera (1995),and references therein; and unpub- lishedobservations).Asthefocusofthisstudywason patterns of variation below the level of genus, and infrageneric phylogenies arenot available fortheplant genera involved, species sets had necessarily to be definedonthebasisoftheiraffiliation togeneraor,in a few instances, to well-supported infrageneric (sub- genus or section) categories. Using taxonomic affilia- tion instead of resolved phylogenies as the basis for selectinggroups oftaxa forindependent contrasts may lead to biased results ifthe taxonomic entities chosen are not strictly monophyletic (Burt 1989,Harvey and Pagel1991).Inthepresent instance, however,itseems reasonable toassume that thegenericand infrageneric categories considered represent monophyletic lineages. Toqualify foranalysis, theleavesofallspeciesina givensetshould havethe samebasicshape and differ only by a scale factor. In this way, one could safely assume that differencesinlinear measurements (length andwidth,seebelow)accurately reflecteddifferencesin leafsize(i.e.areaormass).Tothesameend,taxawith compound leaves(e.g.Pistacia,Rosa,Sambucus, some Sorbusspecies)wereexcludedfromanalysisbecausethe sizeofsingleleaflets,theinformationusually reported infloras(seebelow),maynot adequately reflectactual differencesinoverallleafsize.Eventually atotal of21 setsofcongeneric bird-dispersed specieswereretained for the independentcontrasts analysis (Table 1).Two setsweremade upofspecieswithin distinct subgenera ofthesamegenus(Prunus),andfourcomprised species within distinct sections of two genera (Lonicera and Rhamnus).Mean fruit sizeofeachspecieswasinitially assessedusingboth linearand massmeasurements. As
analyses based onthetwometrics offruit sizeyielded identicalresults,onlythe resultsbasedonlineardimen- sions willbe presented here for simplicity. For each species,thegeometricmeanofaveragefruitlengthand widthwill beusedasthedescriptor of‘‘fruitsize’’. Informationonmature leafsizeofallthespeciescon- sideredwasgathered fromregional(Valde´setal.1987), national (Castroviejo 1986–1999)orcontinental(Tutin et al. 1964–1980) standard floras, mentioned in de- creasingorder ofsearchpriority. Withonlyoneexcep- tion(Atropaspp.),allleafdimension data forspeciesin
the same set were taken from the same literature source,to avoidpossible inconsistenciesamongmonog- raphers inreporting leafsize. Forthepurposesofthe analyses,‘‘leafsize’’wasdefinedasthegeometricmean ofthemidpoints forthereported ranges ofleaflength andwidth.Three ofthe21 setsofspecieschosenfor analysiswerecomposedofthreespecies, andonewas composed of four species (Table 1). In these cases, linearcontrasts werecomputed usingPagel’s (1992) method.Species inthesetwerefirstdividedintotwo groupsaccordingtotheirvaluesofleafsize.Thosewith
Table 1. Speciessets(‘‘independent contrasts’’, IC)ofIberian bird-dispersed plants chosen fortheanalysisofthecorrelation betweenfruit and leafsize.NomenclaturefollowsTutin etal.(1964–1980).
IC ca / Genus (Family) / Species / Fruit size(mm)b / Leafsize(mm)c1 / Asparagus / acutifolius / 6.8 / 1.4
(Liliaceae) / aphyllus / 6.9 / 3.7
2* / Atropa / baetica(S) / 11.0 / 68.7
(Solanaceae) / belladonna(N) / 12.4 / 86.6
3* / Berberis / hispanica(S) / 7.2 / 10.0
(Berberidaceae) / vulgaris(N) / 7.6 / 22.8
4 / Cotoneaster / granatensis / 7.2 / 22.9
(Rosaceae) / integerrimus / 6.3 / 19.8
5 / Crataegus / laciniata / 11.8 / 25.0
(Rosaceae) / monogyna / 10.6 / 20.0
6 / Daphne / gnidium / 5.6 / 11.2
(Thymelaeaceae) / laureola / 7.8 / 41.1
oleoides / 5.4 / 11.8
7 / Euonymus / europaeus / 5.9 / 41.7
(Celastraceae) / latifolius / 6.0 / 73.9
8 / Juniperus / communis / 6.2 / 3.6
(Cupressaceae) / oxycedrus / 9.5 / 3.8
9 / Lonicera(sect.Caprifolium) / etrusca / 7.4 / 34.1
(Caprifoliaceae) / implexa / 6.7 / 28.3
splendida / 9.7 / 30.7
10* / Lonicera(sect.Lonicera) / arborea(S) / 7.8 / 24.5
(Caprifoliaceae) / xylosteum (N) / 8.5 / 41.8
11 / Osyris / alba / 8.8 / 7.1
(Santalaceae) / quadripartita / 7.4 / 12.0
12 / Phillyrea / angustifolia / 5.2 / 24.5
(Oleaceae) / latifolia / 7.3 / 33.2
13* / Prunus(subgen. Cerasus) / mahaleb / 8.1 / 31.0
(Rosaceae) / padus(N) / 7.9 / 56.5
avium(N) / 13.1 / 76.0
prostrata(S) / 7.1 / 11.6
14* / Prunus(subgen. Prunus) / ramburii(S) / 8.9 / 9.5
(Rosaceae) / spinosa(N) / 13.3 / 15.8
15 / Rhamnus(sect.Alaternus) / alaternus / 5.9 / 34.5
(Rhamnaceae) / myrtifolius / 6.5 / 10.3
16 / Rhamnus(sect.Rhamnus) / lycioides / 5.9 / 17.7
(Rhamnaceae) / saxatilis / 5.5 / 13.2
17 / Ruscus / aculeatus / 11.5 / 25.5
(Liliaceae) / hypophyllum / 12.8 / 49.0
18 / Solanum / nigrum / 6.9 / 39.3
(Solanaceae) / dulcamara / 9.2 / 51.2
19 / Sorbus / latifolia / 13.2 / 87.5
(Rosaceae) / aria / 11.0 / 66.3
mougeotii / 10.2 / 72.5
20* / Viburnum / lantana(N) / 7.5 / 75.0
(Caprifoliaceae) / tinus(S) / 6.6 / 52.6
21* / Viscum / album(N) / 7.9 / 40.9
(Viscaceae) / cruciatum(S) / 6.0 / 23.6
a:IC’smarked withasterisks areconsidered intheDiscussion, andconsistofspecieswithvicariant distributionsintheIberian Peninsula, occupying either moist temperate forests in the north (species coded with N) or summer-dry Mediterranean habitats inthesouth (S).
b:Geometric mean offruit length and width (crossdiameter).
c:Geometric mean ofleaflength and width.
Fig.1. Relationshipbetweenfruitandleafsizeinasampleof bird-dispersed,fleshy-fruitedIberianplants.Thetwo values corresponding toeachphylogeneticallyindependent contrast (Table 1)are connected byaline.For setscomprising more than two species,contrasts werecomputed asdetailed inthe text.
leafsizeabove themean forthesetwereput into one group, whiletherestwereputintoother group. Linear contrast coefficientswerethen obtained astherecipro- calofthenumber ofspeciesineachgroup, withthose inthesecondgroupbeinggiven anegativesign.Raw contrasts forleafsizeandfruitsizewerethenobtained asthedifferenceofthecorrespondinggroup meansfor thesetwovariables(HarveyandPagel1991).Standard- ized contrasts were obtained using Pagel’s (1992) method asimplemented intheCAIC program (Purvis and Rambaut1995).
Results
Thesign andmagnitudeoftheraw(nonstandardized) independent contrasts (IC’shereafter) forfruitandleaf sizeare depicted inFig. 1.Out ofatotal of21IC’s, therewere18positiverelationships betweenthetwo variables, i.e.simultaneous increases in both leaf and fruit size.Thishighproportionofpositiveassociations in the direction of change (85.7%) is significantly greater than that expected bychance alone, i.e.ifthe twovariablescovaried independently (P=0.0015,two- sidedbinomial test). This qualitative findingthus sug- geststhat fruitandleafsizearecorrelated inthesetof speciesconsidered.
This iscorroboratedbythe quantitativeanalysis of therelationship betweenthestandardizedcontrasts for leafandfruitsize showninFig.2.Inthisgraph,the contrasts in leaf size (horizontal axis) are all made positivebyconvention, andthoseforfruitsize(vertical axis)arepositiveornegativedepending onwhetherthe changes in the two traits covary positively or nega- tively,respectively.Correlationorregressionanalysisof
thesedatamusttakeintoconsideration thesymmetry arising from the arbitrarydirection of contrasts,and thus regressions orcorrelations must befittedthrough theorigin(Garlandetal.1992).Thesignificanceofthe relationship between the standardized contrasts for fruit and leafsizewastested bymeans ofthecorrela- tioncoefficient through theorigin(computedusingthe formula inMartins and Garland1991),asthere were noapriorireasonstoassigntherolesofdependent and independentvariablestothesevariablesin alinear regression. Thecorrelationcoefficientwaspositiveand significantly different from zero (r±SE=0.613±
0.120, N=21 contrasts, 95% confidence interval=
0.328–0.802;SE and CI determined using accelerated bias-corrected bootstrapresamplingofthestandardized contrastsdata).Resultsremainedunchangedafter ex- cluding from thecomputationstheextremedata point inthe top right corner ofFig. 2,correspondingto IC c13(r±SE=0.523±0.125,CI=0.256–0.747).
Discussion
Resultsof thisstudyprovidestrongsupportforthe existenceofcorrelated evolution offruitandleafsizeat the specieslevelamong Iberian bird-dispersed plants. Thisconclusion isfurther strengthenedbythefactthat thetwoanalyticalmethodsusedhere,asign testand correlationbased oncontrasts computed exclusivelyat thebetween-specieslevel,possesslowstatistical power, thusproviding highlyconservative testsoftraitassocia- tions (Ackerly2000).Current availability ofseedplant phylogenies would havemade possible the application ofindependent contrasts atdeepernodes.Nevertheless, the decision ofrestricting the analysis to the terminal
Fig. 2. Scatterplotof standardizedindependent contrasts of fruit sizevsleafsizecomputed forthecongeneric speciessets listedinTable1.Eachcontrast represents thedifferenceinthe correspondingtrait values between two sister taxa. The line shown isthereduced major axisregression fittedthrough the origin (y=0.0981x).
divergences and studying correlated evolution of leaf andfruittraitsatthebetween-specieslevelwasdictated bytheultimate goalofthisinvestigation. Asnoted in theIntroduction,mypurpose herewastoevaluate the possibility that thelargeresidual (i.e.after accounting for pervasive phylogenetic correlations above the spe- cies level) interspecific variances in size-related fruit traitsdocumented bysomerecentinvestigations maybe unrelated tovariation indispersers.
Giventherelativelysmallsampleoftaxaexaminedin this study, results of independent contrasts analysis may beinfluenced bytaxon sampling effects(Ackerly
2000). Therearereasons,however,tosuggestthatthe correlationbetweenfruitandleafsizefound hereprob- ablyreflectsageneralrelationship occurring inawider phylogenetic context, and is not the spurious conse- quenceof aphylogeneticallybiasedorotherwiseid- iosyncratic species set. First, the dataset analyzed includesspeciesfrom 12different plant familieswidely distributed overtheseedplant phylogeny (Soltisetal.
2000). Bothgymnospermsandangiospermsarerepre- sented inthe sample and, among the latter, there are both dicot and monocot genera. Results thus suggest that correlated variation between leaf and fruit size occursrepeatedlyandconsistentlyin differentphyloge- neticlines.Secondly,thesampleincludesspeciesfroma varietyof habitat types,fromMediterranean-climate woodlands and shrublandsthrough temperate conifer and broadleaved forests (Herrera 1987: Fig. 1). The wholespectrum ofgrowth forms isalsorepresented in the sample, which includes perennial herbs, shrubs, trees and woody climbers. This reduces the likelihood that the character correlationfound stems from a bi- ased selection of species from plant communities or growthformswhere theassociationbetweenthechar- acters isstronger than average.
Considerable attentionhasbeenpaid totheecologi- cal significance of variation in leaf attributes,both acrossandwithinhabitats, andpreviouscomparative studies have found a variety of relationships of leaf traits among themselvesand withother plant traits. In somecases,suchrelationships presumablyreflectthe coordinatedresponsetoselectionofthetraits involved. Thismostlikely applies,forexample,tocorrelations predicted byleafbiophysical modelsbasedonoptimal- ityprinciples(ParkhurstandLoucks1972,Givnishand Vermeij 1976,Givnish 1979).The recurrent combina- tion of functional leaf traits in distantly related taxa acrossbiomes(Reichetal.1999) andthecorrelated evolutionof someof them(AckerlyandReich1999) strongly suggest that, in these instances, the correla- tionsthemselvespossess adaptivevalue.Inothercases, however,theadaptivevalueof thecorrelatedvariation betweenleaftraitsandotherplanttraitsis muchless straightforward.Thisappliestoassociations frequently reported between leaf sizeand plant height, stem di- ameter, inflorescence size, or seed size (White 1983,
BondandMidgley1988, MidgleyandBond1989, Andersson 1993,AckerlyandDonoghue 1998,Cor- nelissen1999).Correlationsbetween leafsizeand any ofthesecharactersmayreflect parallel,independent adaptive responses tosomeabiotic orbiotic factor, or be theconsequenceofsomeallometricrelationship mediated by a general ‘‘size factor’’ (e.g. pleiotropic effectsofgenescontrolling size)affectingallplantparts (Primack 1987,Thompson andRabinowitz1989,Ship- leyand Dion 1992).Therelationship documented here between fruit and leaf sizefor Iberian bird-dispersed plantsprobablybelongstooneof thesetwolatter categories. Irrespective of its ultimate adaptive value, an hypothesis of functional correlation may be ad- vanced which could account for the observed correla- tion between leafsizeand fruit size.Both leavesand fruitsmustbesupported biomechanically bythesame branches. If there are a limited number of branches there willbeparallel size/number tradeoffs forvegeta- tiveandreproductivestructures atthewholeplantlevel and, ifresources forfruit development aresupplied by subtending leaves,then larger fruits willbeassociated withlargerleaves(AckerlyandDonoghue 1998,D. Ackerly,pers.comm.).As thishypothesisappliespar- ticularlytowholeinfructescencesrather thantoindivid- ualfruits(MidgleyandBond1989, Cornelissen1999), assessingitsexplanatory valuewouldrequiresimulta- neousconsiderationoftherelationships betweenleaf, fruit, and infructescence size.
Partoftheinterspecificvarianceinleafsizeoccurring inmyspeciessampleisattributabletothebroad range ofecologicalconditionssampled.As notedearlier,the sample ofbird-dispersed taxa considered includes spe- cies inhabiting diversehabitat typesfromcontrasting climaticregionsin theIberianPeninsula,rangingfrom moisttemperateforestsin thenorthtosummer-dry, Mediterranean scrublands inthesouth.Arelationship between climate and leaf sizehas been known for a longtime(Parkhurstand Loucks 1972,and references therein), and Mediterranean-climate regions of the Worldarenoexceptiontothepattern ofdecreasingleaf sizewithincreasingaridity andinsolation (Parsons and Moldenke1975, Parsons1976, Ackerlyetal.2002). A similartrendexistsintheIberian Peninsulafordecreas- ingleafsizeasonemovesfromnorthernmoisttemper- ate forests to southern dry Mediterraneanshrublands. This pattern isillustrated inthe speciessample exam- inedherebyseven setsofvicariantcongenershaving disjunct geographical distributions,withspeciesoccur- ing in either northern(moist temperate) or southern (dryMediterranean)habitats (Table1).Inalltheseven instances, southern taxa had smaller leavesthan their sisternorthern taxa,whichdeparts significantlyfrom random expectation (P=0.016, two-sided binomial test). This finding,along with theobservationthat six oftheseven southern, smaller-leavedvicariantsare endemic to Mediterranean-climateregions ofsouthern
SpainandnorthwesternAfrica,providecompellingevi- dencethat part ofthevariation inleafsizerepresented in my species sample most likely originated as an adaptive responsetovariation inabiotic conditions. In this scenario, and keeping in mind that all the plant taxa considered herelargelyshare the same speciesof seed dispersers, the correlated evolution of fruit and leaf size found in this study would reflect either (1) changesinfruitsize weresubsequent to,andamere allometric consequence of,adaptive changesinleafsize occurringin responsetothebioticenvironment;or(2) changes infruit sizereflectparallel adaptive modifica- tioninresponsetotheabioticenvironment.Developing fleshy fruits are photosynthesizing structures them- selves,and thus perhaps also susceptible to the same abioticfactorsthataregenerallyacceptedtoselect for evolutionary modificationsofleaftraits.Decreased moisture levelsinMediterranean-climateregions could thusnotonlyselectforareduction inleafsize,butalso foraparallelreduction infruitsize. Theinformation available does not allow for discriminating between thesetwoexplanations.
Regardlessofitsultimatecauseandproximate mech- anismsinvolved, thecorrelationbetweenleafand fruit sizedocumented hereforasetofbird-dispersed plants serves tosuggestplausiblealternative scenariosand hypothesesfortheevolutionof interspecificvariancein size-related fruit traits that are independent of the selectiveaction ofdispersal agents. Results also high- lighttheimportanceofconsidering allometric relation- shipsinstudiesofplant reproduction,becauseofboth itsinherent explanatorypotential and itsvalueassuit- able‘‘nullmodels’’ incomparative studies(Bondand Midgley 1988,Thompson and Rabinowitz1989,Ship- leyandDion 1992).Among other consequences, recog- nition ofthese possibilities callsfor areassessment of thepresumedadaptivebasisin relationtodispersersof some geographic or elevational trends in fruit size reported previously (e.g.Herrera 1985).Inthelightof the results reported here, these patterns in fruit size might merely ensue from geographical variations in abiotic factors, either directly or indirectly through theireffectsonleafsize. Concomitant changesindis- persercomposition andsize distributioncouldbea secondary ecologicalconsequence brought about bythe operationof ‘‘ecological fitting’’ processes (sensu Janzen 1985), rather than its ultimate evolutionary cause.Further studiesareneededtoevaluatethese possibilities.
Acknowledgements –Foruseful commentsandsuggestionson earlierdrafts, Iamverygrateful toD.Ackerly,J.Herrera, M. SalvandeandM.Verdu´.D.Ackerlyalsokindlyallowedaccess tounpublished work. During thepreparationofthispaper, I wassupportedbygrants PB96-0856from Ministerio deEdu- cacio´nyCultura, and BOS2000-1122-C03-01from Ministerio deCienciayTecnolog´ıa.
References
Ackerly, D. D. 2000. Taxon sampling, correlated evolution and independent contrasts.–Evolution 54:1480–1492. Ackerly, D.D.and Donoghue,M.J.1998.Leafsize,sapling
allometry, and Corner’s rules: phylogeny and correlated evolution inmaples (Acer). –Am.Nat. 152:767–791.
Ackerly,D.D.andReich,P.B.1999. Convergenceand correlations among leafsizeand function inseedplants: a comparative test using independent contrasts. – Am. J. Bot. 86:1272–1281.
Ackerly, D. D., Knight, C.A.,Weiss,S.B.etal.2002.Leaf size, specific leaf area and microhabitat distribution of woody plants in a California chaparral:contrastingpat- ternsinspecieslevel andcommunity level analyses. – Oecologia 130:449–457.
Andersson, S.1993.PopulationdifferentiationinCrepistecto-
rum(Asteraceae):patterns ofcorrelationamongcharacters.
–Biol.J.Linn. Soc.49:185–194.
Avery,M.L.,Goocher, K.J.andCone,M.A.1993.Handling efficiencyand berry sizepreferences ofcedar waxwings. – Wilson Bull.105:604–611.
Bond,W.J.andMidgley,J.1988. Allometryandsexual differencesinleafsize.–Am.Nat. 131:901–910.
Burt,A.1989.Comparative methodsusingphylogenetically independent contrasts.– In:Harvey, P.H and Partridge, L.(eds),Oxford SurveysinEvolutionaryBiology.Vol.6. Oxford Univ. Press,pp. 33–53.
Castroviejo, S.(ed.)1986–1999.Flora Iberica. Plantas vascu- lares de la Pen´ınsula Ibe´rica e Islas Baleares. Volumes I–VIII. –Real Jard´ın Bota´nicodeMadrid.
Cornelissen, J.H. C.1999.Atriangularrelationship between leafsize andseedsize amongwoodyspecies: allometry, ontogeny, ecologyand taxonomy. – Oecologia 118:248–
255.
Debussche, M.andIsenmann, P.1989.Fleshyfruitcharacters andthechoicesofbirdandmammalseed dispersersina Mediterraneanregion. –Oikos 56:327–338.
Debussche, M.,Cortez, J.andRimbault,I.1987.Variation in fleshy fruitcomposition intheMediterraneanregion:the importance ofripeningseason,life-form,fruittypeand geographical distribution.–Oikos 49:244–252.