The Community-Level Consequences of Seed Dispersal Patterns Author(s): Jonathan M. Levine and David J. Murrell Source: Annual Review of Ecology, Evolution, and Systematics, Vol. 34 (2003), pp. 549-574 Published by: Annual Reviews Stable URL: http://www.jstor.org/stable/30033786 Accessed: 07/10/2010 14:27 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=annrevs. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. Annual Reviews is collaborating with JSTOR to digitize, preserve and extend access to Annual Review of Ecology, Evolution, and Systematics. http://www.jstor.org Annu. Rev. Ecol. Evol. Syst. 2003. 34:549-74 doi: 10.1146/132400 Copyright@ 2003 by AnnualReviews. All rightsreserved First publishedonline as a Review in Advance on July 30, 2003 OF THE COMMUNITY-LEVEL CONSEQUENCES PATTERNS SEEDDISPERSAL M.Levine'andDavidJ.Murrell2 Jonathan andMarine Evolution, Biology, University of California, ofEcology, 1Department SantaBarbara, 93106;email:[email protected] California atSilwood 2Center Park,Ascot, forPopulation Biology, Imperial College Berkshire email:[email protected] SL57PYUnited Kingdom; distribution, KeyWords abundance, aggregation, diversity U Abstract Becauseit laysthetemplate fromwhichcommunities develop,thepatternof dispersed seedis commonly believedtoinfluence structure. Totest community thevalidityofthisnotion,weevaluated theoretical andempirical worklinkingdispersal kernelsto therelativeabundance, andcoexistence of species. distribution, dispersion, Wefoundconsiderable theoretical evidencethatseeddispersal affectsspeciescoexistenceby slowingdownexclusionthrough localdispersal anda competition-dispersal wasscant.Instead,mostempirical extrade-off, yet empirical support investigations aminedhowdispersalaffectsspeciesdistribution anddispersion, with little subjects theory.Thisworkalsoreliedheavilyondispersal proxiesandcorrelational analysesof methodsunableto excludealternative community patterns, hypotheses. Owingto the overalldichotomy betweentheoryandempirical results,we arguethattheimportance of dispersal cannotbetakenforgranted. Weconcludeby advocating that experiments the seed and models that documanipulate dispersal pattern, incorporate empirically menteddispersal kernels. INTRODUCTION the importanceof dispersalfor the Althoughecologistshave long appreciated and of spread persistence populations(Harper1977,Howe& Smallwood1982, Skellam1951),thelastdecadehaswitnesseda surgeof interestin howthisphase of the life cycle influencescommunitystructure (Bullocket al. 2002, Cainet al. 2000,Clobertet al. 2001,Nathan2003,Nathan& Muller-Landau 2000,Wang& Smith2002).Theoreticalmodelshavebeenan important motivatingforce,with numerousstudiesemphasizingthe importance of spatialstructure in influencing speciesinteractions. Particularly tantalizingareresultssuggestingthatlocalseed colonizationtrade-offsfavorthecoexistenceof competidispersalorcompetition tors (Murrellet al. 2001, Pacala 1997, Rees et al. 1996, Tilman 1994). Results such as these are mirroredin the tropical forest literature,where theories emphasizing the importanceof limited dispersalhave gained prominence(Hubbell 1543-592X/03/1215-0549$14.00 549 550 LEVINEE MURRELL 2001). Meanwhile, empiricalecology is increasinglyimplicatingdispersalas an andobcontroloverspeciesdiversity.Numerousrecentexperimental important servationalresults suggest that local communities are seed limited, with diversity limited largely by the regional species pool (Cornell 1993, Srivastava1999, et al.2000).Interesthasalsobeenheightened Turnbull proby dispersal-mediated cesses in conservationbiology, includinghabitatfragmentation(Hanski& Gilpin 1997), species abilityto migratewith climatechange (Clark1998), andthe spread et al. 1989). invasions of biological (Drake Althoughthe influenceof dispersalon communitystructureis only beginningto causalrelationship. seemtoexpectanimportant berigorously examined, ecologists This is likely because dispersalis believed to set the templatefrom which com- to influencepopulationspread munitypatternsdevelop,andis well documented andpersistence(Hanski& Gilpin1997,Harper1977,Skellam1951).However, spreadandpersistenceareverydifferentresponsevariablesfrommostmeasures of communitystructure,includingpatternsof abundance,distribution,and coex- the templatelaidby dispersal,oftenreferredto as the istence.Moreimportantly, seedrain,is influencedby otherpotentiallymoreimportant factors,includingthe distribution,density,andfecundityof parentplants(Clarket al. 1998, Platt 1975), as well as landscapefeaturesthattrapseeds (Schneider& Sharitz1988) (Figure1). Moreover,as notedby severalauthors(Nathan& Muller-Landau2000, Schupp& Fuentes 1995, Wang& Smith 2002), a numberof importantlife historyprocesses to adultplants.Speciesinteractions occurbetweendispersalandtheprogression andenvironmentalfactorsstronglyinfluencethese transitionsandcan significantly changethe templatelaid by dispersal(Figure1). Thus,unless populationsare substantiallyseed limited, the importanceof the dispersalkernel for the abundance, distribution,and diversityof species should not be assumeda priori. In thispaper,we criticallyreviewtheoreticalandempiricalworkrelatingpatternsof dispersalto spatialpatternsin communitiesin an attemptto providea critical frameworkfor researchin this area. More specifically, we ask how importantthe specific seed dispersalkernelor seed shadowis for explainingrelative abundance,distributions,andcoexistencein naturalcommunities.By dispersalkernel, we mean the probabilitydensity function describingthe probabilityof seed transportto various distances from the parentplant. Although both theoretical andempiricalapproachesaddressthe importanceof seed dispersalfor community structure,a large gap exists in the coverage and motivationbehind the different methods. As we demonstrate,the modeling work is focused largely on coexistence, motivatedby interest in how incorporatingspatial processes changes the predictionsof earliernonspatialmodels. Meanwhile, the empiricalwork focuses largely on dispersal'seffect on species distributions,motivatedby naturalhistory observationsrelated to dispersalkernels, the movement of animal vectors, seed trappingpatterns,or the clumping of parentplants. Thus, of primaryinterestin our review is how empiricalresultsmatchthe predictionsof theory. To encourage a greaterexchange of ideas and questions between empirical and theoreticalapproaches,we organize our review aroundthe influence of dispersal kernels on four key features of community structure:patternsof relative DISPERSAL EFFECTS ONCOMMUNITIES 551 Distribution of SeedSources Densityof SeedSources Dispersal Kernel SeedProduction features that Landscape trapseeds Pattern of Seed Rain Environmental Factors Species Interactions Abundance, Distribution, andDiversity of Species Figure 1 Pathwayby whichthe shapeof speciesdispersalkernelscan influencethe abundance,distribution,and diversityof species. Shownin bold is the link between dispersalkernalsandcommunitypatternsthatwe examinein thisreview.Weemphasize the multitudeof factorsotherthandispersalkernelsinfluencingthe seed rain,andthe multitudeof factorsotherthanseed raininfluencingthe abundance,distribution,and diversityof species. abundance,distribution,dispersion,and coexistence. For each of these features, we firstcharttheoreticalpredictionson the importanceof dispersalandthenreview evidence for these predictionsin the empiricalliterature.We located our papers by electronicallysearchingthe Science CitationIndex and examiningreferences therein.Because of the large body of theoreticalwork on coexistence, we divide this literatureinto two sections, one describingthe effects of local dispersal,the otherfocused on competition-dispersaltrade-offs.For similarreasons, we review the work examiningdispersaleffects on dispersionseparatelyfrom work on other types of distributionpatterns.We do not reviewdispersaleffects on persistenceand spread(Hanski& Gilpin 1997, Howe & Smallwood 1982, Shigesada& Kawasaki 2002, Skellam 1951), the evolution of dispersal (Clobertet al. 2001), or the effects of colonization on communitypatternsachieved throughdormancyor high fecundity.Still, it shouldbe notedthatboth dormancyandfecunditycombinewith dispersalto influencecolonization(Harper1977). RELATIVE ABUNDANCE Theoretical Background Over the past decade, it has become increasinglycommon to regardlocal populations as embeddedwithin a largermetapopulationstructure.In classic (Levins &Gilpin1997)metapopulation 1969)andcontemporary (Hanski models,species abundance at the metapopulationscale is a function of the colonization and 552 LEVINEm MURRELL or patches.Thus,speciesabilityto disextinctionof individualsubpopulations of relativeabundance-whichspeciesare determinant persemaybe animportant commonandwhicharerare. alsoemerges thatdispersalis positivelyrelatedto abundance Theexpectation in severalothertypes of models(Turnbullet al. 2000). In spatialmass effect habimodels(Shmida& Ellner1984), speciesoccupymarginallyunfavorable tatsbecauseof seed inputfromother,morefavorable,locations.Greaterinput in thesesystems.Similarly, causedby increaseddispersalcanenhanceabundance if communitydynamicscan be conceivedas a competitivelottery(Chesson& Warner1981),speciescan dominatesystemsby havinghighcolonizationrates, achievablethrougheffectivedispersalor high fecundity.Greaterabundanceof hypothesis better-dispersing speciesis also predictedunderthe Janzen-Connell (Connell1971,Howe1989,Janzen1970,Schupp&Fuentes1995),wheredispersalawayfromtheparentplantconfersgreatlyreduceddensity-dependent mortality (Ellner2001,Lawet al. 2003,Pacala& Silander1985).Althoughsimilarpredictionsemergefromthesemodelscollectively,it is themetapopulation predictions how abundance at tests of in motivating thathavebeenmostinfluential empirical scalerelatesto dispersalability(Eriksson1997,Eriksson& the metapopulation Jakobsson1998). Thesetests, however,may not have consideredthatlong-rangedispersalis in models.Forexample,if the landscapeis variable sometimesdisadvantageous in (abiotic)quality,dependingon the spatialscaleof favorableandunfavorable thanlong-rangedisdispersalmayleadto higherabundance patches,short-range & Travis et al. 1998, Dytham1999).In thesecases persal(Bolker2003,Dockery thanoutweighsthe increased more in a of the importance remaining goodpatch from often dispersal. short-range resulting competition intraspecific EmpiricalEvidencefor a RelationshipBetween Dispersaland Abundance levelwould Inadditionto theoretical empiricalworkatthepopulation predictions, seemto suggestthatspeciesabilityto disperseis an important predictorof relaatleastin somesystems.A numberof seedadditionexperiments tiveabundance, (see Turnbullet al. 2000 for review) suggest that for early successional systems seedlimitationis common,andthusgreaterdispersalshouldconfer in particular, Nonetheless,in ourreviewof empiricalevidence,we found greaterabundance. was controlledby dislittle supportfor the expectationthatrelativeabundance persalpatterns.Forexample,Rabinowitz& Rapp(1981)foundno relationship betweenabundanceand dispersalunderfield conditionsin a Missouriprairie. Similarly,Eriksson& Jakobsson(1998)foundthatdispersalmodewasunrelated to abundanceandgeographicrangesize for over 80 grasslandplantspeciesin Sweden. Also in Sweden, the ability of 17 species to disperse on the fur of animals was unrelatedto theiroccupancyratesin seminaturalgrasslands(Kiviniemi & Eriksson 1999). In the herbaceousflora of centralEngland, seed dispersalas DISPERSAL EFFECTS ONCOMMUNITIES 553 estimated through terminal velocity was very weakly related to species abundance as assessed by their UK range (Thompsonet al. 1999). Thompson et al. (1999) andEriksson& Jakobsson(1998) reviewedthe literaturerelatingdispersal mode to geographic range size more generally and found similarly ambiguous results. Support for the notion that dispersal dictates commonness and rarity might seem to come from annual plant communities. It is often found that common species are small seeded, whereaslarge-seededspecies are consistentlyrare(Guo et al. 2000, Levine & Rees 2002, Maranon& Grubb1993, Rees 1995, Rees et al. 1996). However,althoughsmall-seeded species may be bettercolonizers, this is more likely the resultof a relationshipbetween seed size andfecundityratherthan seed size and dispersal. Small-seeded species consistently produce more seeds (Jakobsson& Eriksson2000, Leishman2001, Rees 1995, Shipley & Dion 1992, Turnbullet al. 1999), yet seed size is very poorly related to dispersal (Carey & Watkinson1993, Westobyet al. 1996). How can we explain the absence of a relationshipbetween abundanceand seed dispersalwhen experimentsoften show seed-limitedpopulations?One likely possibility is that the evidence for seed limitationis more a productof low seed productionor a low density of adult plants than of poor dispersal. In addition, numerousotherprocesses, such as competition,dormancy,and habitatvariation, may simplybe moreimportantthandispersalin controllingrelativeabundancepatternsin communities(Figure1). Because dispersalcan clearlyinfluencethe rateof species arrival,it may be more importantin controllingtemporalpatternsof abundance afterdisturbance.In a simulationstudy,Hovestadtet al. (2000) showedthat long-rangedisperserswere more abundantin the colonizationphase immediately after disturbance,but dependingon model details, short-rangedisperserssubsequentlyobtaineda higherrelativeabundance.Indeed,on lake islands in Sweden, species lacking traitsfor water,wind, or animaldispersalare rareor absentearly in succession (Rydin & Borgegard1991), but become abundantyears later.Despite these temporalpatterns,we conclude that little empiricalevidence supports the expectationthatinterspecificdifferencesin dispersalcontrolcommonnessand rarityat metapopulationscales. SPATIALDISTRIBUTION TheoreticalBackground The distance and direction a plant disperses its seeds should play an important role in the distributionof that species. However, as illustratedin Figure 1 (left), the influence of the dispersal kernel will depend strongly on the importanceof other factors influencing the seed rain. For example, Levine (2003) modeled a streamsideplant assemblage where species occur in discrete habitatpatches linearly arrayedalong the channel. Species dispersed 60% of their seeds to other patches and 95% of those seeds to patches downstream.In such a system, one 554 LEVINEm MURRELL thanupstream. However,Levine naturally expectsgreaterdepositiondownstream of parentplantsthroughout the thatwith a uniformdistribution demonstrated comes from the one a reasonable most seed and kernel, input dispersal system of the firstfew or two patchesupstreamof the targetpatch.Thus,downstream in the seed increases Still, only marginally. thisrelatively system, input patches increases variationin seeddepositioncoulddrivedownstream smalldownstream are in population size anddiversity,butonlyif thecomponent populations highly Thisresult seedlimited,asresultingfromverylowfecundityorrecentdisturbance. between 1. the two pointsintroduced demonstrates by Figure First, relationship the seed-dispersal kernelandthe seedraincanbe complicated, deservingsignifiandempiricists. fromboththeoreticians Second,forpatterns cantlymoreattention the componentpopulationsmust of seed rainto influencespeciesdistributions, or environmental andnot speciesinteractions be constrained by seedavailability factors. The Importanceof Seed-TrappingAgents In contrastto theory,numerousempiricalstudieshaverelatedseeddepositionto yet in mostof the work,spatialvariationin depositionis speciesdistributions, not easily relatedto the dispersalkernels.Instead,it is drivenlargelyby landstill Differencesin seedmorphology scapeelementsthattrapseedsorpropagules. withthe influencedepositionpatternsbutdo so largelythroughtheirinteraction trappingagents.Forexample,Schneider& Sharitz(1988)trackedthe dispersal UnitedStates.Seedswereiniof treeseedsin a swampforestin the southeastern levels as water but then rose,theywereredistributed by tiallygravitydispersed, structures to locationsagainstemergent includingtwigs,trees,logs, hydrochory branches,knees,and stumps.Some of these substrates,such as treesand tree knees,providedelevatedstablemicrosites,andtreeseedlingsweredisproportioninfluenced seedmorphology deposition atelyfoundintheselocations.Importantly, andseedlingpatternsacrossspecies.Thesmall,angularcypressseedsweremore frequentlytrappedby kneesthanthe largerellipsoidtupelofruits.Consequently, 45%of the cypressseedlingsoccurredon kneesin comparisonto only 27%of tupeloseedlings. reSeveralstudiesconductedin northernSwedishrivershaveexperimentally theirdepositionwithspecies mimicsandcorrelated orpropagule leasedpropagules and diversity.Althoughnot a seed-dispersal distributions study,Johansson& Nilsson(1993)taggeduniformlysizedrametsof the vegetativelydispersingRatheirdepositionwiththe presenceof established nunculuslinguaandcorrelated found that ramet high in curves depositionwas disproportionately plants.They standstended established where locations also the the andatobstaclesalong river, a similar to be found(see Anderssonet al. 2000 for focusingon diverapproach sity). Patternssuch as these may relateto propagulecharacteristics.Nilsson et al. (1991) found that long-floatingspecies tended to be more abundantin areasthat trappedfloating seed mimics, whereas the distributionof short-floatingspecies EFFECTS ONCOMMUNITIES 555 DISPERSAL was unrelatedto trappingpatterns.Danvind& Nilsson(1997) foundno correlationbetweenfloatingtime anddistribution patternsalongan alpineSwedish river. hasalsobeencorrelated withspeciesdistriSeeddepositiondrivenby trapping studies.Rabinowitz(1978)hypothesized thatthe different butionsin nonriverine sized seedsof the threemangrovedominantsin Panamaweretrappedat differentpointsalongthetidalelevationgradient,matchingtheelevationaldistribution of mangrove of adulttrees.Supportwas providedby the experimental transport outside their tidal which showed across theelevation zone, survivorship propagules recent work Sousa & B. inreview) Nonetheless, Mitchell,manuscript (W. gradient. in the kernels the same forests has found no eviestimating dispersal mangrove In a similarsystem,Rand(2000)foundthat dencefortidalsortingof propagules. of severalsaltmarshspeciesto particular therestricted distribution tidalelevations andphysiological stressesthanseeddeposition wasmorea functionof competition forms patterns.In contrast,in theNegevdesertof Israel,Anasticahierochuntica reticulatedistribution thatrelateto seeddepositionin thecracksthatform patterns withthewettinganddryingof thesoil (Friedman & Stein1980).Peart& Clifford thesegregation of grassspeciesacrossdifferentsoil typesto the (1987)attributed interactionof the soil textureandcrackingwith the differentawn typesof the variousspecies. In thesestudies,patternsof seeddepositionresultfromstructures or topograthat or sorts Seed then influences not traps phy propagules. morphology onlyhow seedsmove,butmaybemoreimportantly, whereseeds stop,a less appreciated andmoredifficultto predictcomponentof thedispersalkernel.Therealizeddistributionof seedsmaythereforebe verydifferentthanthe potentialdistribution surroundbythedispersalkernelalone.Inaddition,thelocalcommunity predicted inga dispersingindividual maystronglyinfluencethedispersalkernelby trapping seedsandinfluencing thesizeof thedisperserthroughcompetition. Consequently, furtherworkclarifyingtherelationship betweendispersalkernelsandpatternsof for clarifyingits influenceon deposition(Figure1, left) is particularly important distributions et al. 2000). (Alcantara Distributions Driven by Dispersal Kernels kernelsto distribution Relativelyfew studiesrelatedseed-dispersal patterns.In one of the only studiesrelatingdispersalto the seed rain,Dallinget al. (2002) demonstrated thatspatialpatternsof seedrainformanyneotropical pioneertrees were drivenlargelyby limiteddispersal.The resultingseed rainpatternswere an important in forestgaps. Similarly,Platt& predictorof seedlingabundance Weiss(1977)examinedfield-derived dispersalkernelsandcompetitivetraitsof on mounds in anIowaprairie.Thespeciessegregate fugitiveplantsliving badger acrossa habitatgradientin mounddensitysuchthatthe morepoorlydispersing species dominate the portion of habitat with closely spaced mounds, whereas better-dispersingspecies dominatethe areawith distantlyspacedmounds.Parallel 556 LEVINEm MURRELL differences inphysiological tolerance in soilmoisture andinterspecific gradients maintain thespeciessegregation. thatasymmetric Several riverine studies downstream beginwiththeassumption in seed in these drives variation systems(butsee deposition dispersal longitudinal increases in diversity todispersal. Levine2003).Theythenattribute downstream oftheVindelriverinSwedencontained Nilssonetal.(1994)foundthattributaries tothe onlya subsetofthosespeciesfoundinthemainstem,aresulttheyattributed seeds. et al. downstream and accumulation of Honnay (2001)analyzed transport networks ofsmall inplantcommunities ofdiversity alongdendritic spatial patterns see Friedman & came similar conclusions Stein foreststreams and to (also 1980). of community Thesestudies, aresimplycorrelational however, analyses patterns. orlimitation aremade,andthusas of seeddispersal, Nomeasurements deposition, environotherprocesses, Nilssonetal.(1994)acknowledge, changing particularly thediversity couldalsoexplain conditions mental gradient, alongthedownstream wastakenbyLevine(2001),whodocuA moremanipulative approach pattern. in diversity andindividual increases menteddownstream plantabundance along mimics released theEelriverinCalifomrnia. successfully Experimentally propagule were andthestreambed tosuitable microsites downstream, assemblages dispersed whenseedsupplywasexperimentally seedlimited.Mostimportantly, equalized colonization of themost therewasno greater acrossthedownstream gradient, even withthose5 kmupstream. habitats as compared downstream Nonetheless, to drive downis sufficient whether downstream withallthisevidence, dispersal inthisandotherwork. untested remains increases in seeddeposition stream ourreviewdocumented in of the relative rarity hydrochorynature, Considering influenced where water number of studies a disproportionate species dispersal movement to the obvious directional attributed Inpart,thiscanbe distributions. thatlineriversand theplantcommunities Inaddition, of waterin manysystems. suchascompetition arereadilydisturbed, lakeshores processes likelypreventing theirfullimpacts. fromexerting Last,thefloodingthatoccursin thesehabitats seedlimitation canexportmostof theseed,favoring (Levine2001).Despitethe few insystems withhydrochory, fordistributions ofdispersal importance apparent orexamined testedforseedlimitation studiesintheseorotherhabitats empirical to adultplants. andtheprogression betweenseedarrival theprocesses occurring to resultfrom are of whether thedistribution Thus,regardless argued patterns as a whole is highly evidence the or kernels, empirical agents dispersal trapping for the alternative unable to exclude often correlational, putative explanations distribution patterns. dispersal-driven DISPERSION The aggregationof species in communitiescan influence the rate of competitive (Stoll& Prati2001),patternsof resourceavailability(Pastoret al. displacement and 1999), patternsincludingspecies-areacurvesandspeciesmacroecological distributions abundance (Chaveet al. 2002).Becauselocaldispersalis anobvious EFFECTS ONCOMMUNITIES 557 DISPERSAL sourceof clumping,numerousstudiesattributeadultaggregationto restricted seed transport. However,as is truewith distribution patternsmoregenerally,a numberof conditionshaveto be met beforedispersalcontrolsdispersionpatterns(Schupp& Fuentes1995).Not only mustthe componentpopulationsbe so sparselyarrayedthatlocal dispersaldrivesclumpingof dispersedseeds,but cannotbe so severeasto eliminatetheaggregation of mortality density-dependent seedlings. Models Where LocalDispersalGeneratesAggregation Bleheret al. (2002)usedsimulationmodelsto showthatlocaldispersaland,to a lesserdegree,low adultdensitystronglyinfluencedthe clumpingof foresttrees. Bleheret al. accomplishedthis by constraining fecunditysuchthatall species one but the relevance of suchgrowthratesto natueffectivelyproduce offspring, ralsystemsis unclear.Presumably, withdifferentfecundityacrossspecies,some wouldgrowandtheirpatcheswouldcoalesce,whereasotherswould populations declineandtheirclumpscontract.LikeBleheret al., Chaveet al. (2002)examined clumpingpatternsin neutralmodels(wherebirthratesjust balancedeath rates),andshowedgreaterclumpingwithincreasedlocaldispersal.Chaveet al. alsoexploredtheinfluenceof localdispersalin modelswherestablecoexistence is achievedthrougha competition-fecundity trade-offanddensitydependence. Theyshowedthatwhentheseotherprocessesgeneratecoexistence,localdispersal can stronglyinfluenceclumping,whichcausesthe species-areacurveto rise moregraduallythanwhenspeciesaremorediffuselyspreadthroughthehabitat. Dispersal-generated clumpingcanalsoinfluenceecosystemprocesses.In a simulationmodelof borealforestwherespeciesdifferin decomposability, Pastoret al. thataltersthe (1999) showedhow local dispersalgeneratesspeciesaggregation spatialpatternof nitrogenavailability. Furthersupportfortheinfluenceof localdispersalon clumpingis explicitlyor its impacton coexistence(reviewed implicitlyprovidedby studiesdemonstrating it is to remember that below).Still, important dispersalis justone of severalspatialprocessesthatshapethepatternof individuals acrossthelandscape(Figure1) & Fuentes For because interactions betweenindividuals 1995). example, (Schupp alsotendtobelocalizedin space,offspringexperience intensekincompetition with localdispersal.Theoryshowsthatlocalinteractions cancausetherealizedpattern of adultsto be random,overdispersed (spatiallysegregatedor evenlyspaced)or clumped(Ellner2001, Lawet al. 2003, Molofskyet al 2002, Pacala& Silander to achieveaggregation in homogeneous 1985).Generally, spacewithlocaldispermustoccuron a spatialscalesimilarto thatof thedispersalkernel sal,interactions (Ellner2001,Lawet al. 2003, Pacala& Silander1985).If dispersaloccursover muchlargerscalesthanneighborhood interactions thenanevenspacingof conspecificindividuals is expected(Lawet al. 2003).Alternatively, processesotherthan dispersal(Figure 1), such as spatialheterogeneityin the externalenvironment,or reducedpollinationof isolatedindividuals,may also generateaggregation.In sum, theory suggests that dependingon the spatial scale of density dependence,local 558 LEVINEN MURRELL dispersalcandrivea widerangeof dispersionpatterns,patternsthatcanalsobe explainedby variationin environment. EmpiricalEvidence localseeddispersalandanaggregated A largebodyof empiricalworkdocuments of adults,and suggestsa causallink betweenthe two (e.g., Bleher distribution & Bohning-Gaese 2001, Fragaso 1997, Westelaken & Maun 1985). However, as arguedby Schupp& Fuentes(1995)andputforthby theoreticalworkmore betweenseedarrivaland generally,becauseof theprimacyof processesoccurring to demonstrate is insufficient evidence correlative to adult this plants, maturity local documented Overton Indeed, (1996) dispersalof patterns. dispersal-driven in the numberof mistletoe,butthis hadno effecton the spatialautocorrelation studieshavefoundthattheaggregated numerous mistletoepertree.Moregenerally, as of seedsandseedlingsfollowinglocaldispersaltendsto disappear distribution theseedlingsmature(Barotet al. 1999,Houle1995,Rey&Alcantara 2000,Schupp and ourreviewon severalmorequantitative &Fuentes1995).Wethusconcentrate for of the to assessing importance dispersal aggregation approaches comparative patterns. to dispersalfor explaininga clumpedspecies Becausethe majoralternative distributioninvolves patchyenvironmentalvariables,one approachfor testing the toaskwhetherspeciesare of localdispersalis tousemultipleregression importance moreclumpedthancan be explainedby spatialvariationin the environmentalone. of fourpalmspeciesin an AndeanrainSvenning(2001)foundthataggregation forestcouldnotbe explainedby variationin foreststructure, topographic-edaphic conditions,altitude,or aspect.This,coupledwithgreaterdensitiesof seedlings near adult plants, caused the investigatorto implicate dispersalas a control over the clumped distributionof these species. With a similar approach,Svenning & Skov (2002) found that in a managed forest in Denmark,20 of 60 understorey plant species exhibited aggregationunexplainableby environmentalparameters to dispersal.Moreover,animal-andwind-dispersed andthusattributable species showedless clumpingthanspecieswithless-efficientdispersalmodes.Incontrast variablesin influencingtree of environmental to theseresults,thepredominance clumping patternswas suggested for the palm Borassa aethiopumin an African savanna(Barotet al. 1999)andforspeciesoccupyingSouthPacificisland(Webb & Fa'aumu1999)andMalaysian(Plotkinet al. 2002)forest. havebeenusedto attribute Similarapproaches throughspace speciesturnover to dispersal.Tuomistoet al. (2003) found thatdecreasingfloristicsimilaritywith distance in Amazonian rainforestscould not be entirely explained by changing environmentswith distance,pointingto the potentialimportanceof local dispersal. Moreover,the importanceof geographicdistancealone (a proxyfor local dispersal) versus environmentalfactors was comparativelyless importantin pteridophytes than in the more poorly dispersedMelastome shrubs,furthersupportingthe role of dispersal.Similarly,Condit et al. (2002) used predictionsof Hubbell's (2001) DISPERSAL EFFECTS ONCOMMUNITIES 559 of speciesturnover neutralmodeltoarguethatlocaldispersalcouldexplainpatterns overthe scaleof 0.2 to 50 km. in tropicalrainforests influencespatterns An alternative approachfor assessinghow seed transport of aggregationassumesthatdispersalmodeis a predictorof dispersalkernels (Willson1993),andcomparesclumpingacrossspeciesdifferingindispersalmode. Workingin a CostaRicandryforest,Hubbell(1979)foundthattherateat which densitydeclinedwithdistancefromadultswas steepestfor speciesdispersedby mammals,followedby wind,andthenby birdsandbats.Conditet al. (2000) foundthatwind-orexplosivelydispersedspecieswereless clumpedthananimalforest,butthisdifferencewas nonsignificant. dispersedspeciesin a Panamanian withtheirpoorlydispersedseeds Nonetheless,in a Malaysianforest,dipterocarps, weremoreaggregatedthannondipterocarps (Conditet al. 2000).Also studying howseed-dispersal mode rainforesttrees,Plotkinetal.(2002)described Malaysian of largeandsmallindividualswithina cluster. influencesthe spatialdistribution Inearlierwork,Plotkinet al. (2000)showedclusteringof individuals independent of driven of topography, dispersal suggestive aggregation. Resultsfor speciesotherthantropicaltreesaremoreequivocal.Niederet al. (2000) foundthatwhetherepiphyteswere dispersedvia wind or animalswas unrelated to theirdegreeof clumping.Incontrastto theHubbell(1979)results,but at a muchsmallerscale(2 m x 0.5 m plots),Myster& Picket(1992)foundthat treesinvadinganoldfieldweremoreclumpedthanthosedispersed bird-dispersed Theperchingbehaviorof birdspresumably wind or mammals. by generatesclumps of seedatthissmallerspatialscale.A numberof otherstudieshavedemonstrated of seeds(Howe1989). thatanimaldispersalcausesverylocalaggregation Limitationsof CurrentApproaches One very importantlimitationof the work relatingaggregationto dispersalis that nearly all of the evidence is correlationalor indirect.These results are thus best aspatterns thatgeneratehypotheses,ratherthandefinitivesupportforthe regarded of areimplicitwhendispersal importance dispersal.A numberof keyassumptions is assessedastheresidualclumpingunexplained environmental variables(asin by & Skov Tuomisto et al. Most 2002, 2003). Svenning2001,Svenning significantly, we mustassumethatall potentiallyimportantenvironmental variables,past or variablesmust present,areaccountedforin thestatisticalmodel.Anyunmeasured at leastbe correlated withthosethatarequantified. Thisis probablyanunreasonableassumption becausean historicalprocesssuchas a forestgapfourdecades agocouldeasilyinfluenceclumpingpatternstoday,yet unlessit left a signaturein variables,suchan effectwouldbe incorrectly currentlymeasuredenvironmental todispersal. attributed Inaddition,thesestudiesassumethatpositivedensitydepenorsharedsoilmutualists, is notimportant. dence,asmightarisethoughpollination Otherconcernsexistforstudiesusingdispersalmodeasa proxyforthedispersal kernel or mean dispersal distance. Dispersal mode is often shared across members of a family, such as wind dispersalin the Asteraceae,and is thus likely to be 560 N MURRELL LEVINE correlated withotherfactorsthatcouldinfluenceclumping,suchas seedproductionorcompetitive ability.Indeed,Horvitz&LeCorff(1993)foundthatwithinthe herbsof the Marantaceae, ant-versusbird-dispersed tropicalunderstorey plants didnotdifferin theirdegreeof clumping.Thepotentialpowerof alternative experimentalapproaches forassessingtheinfluenceof dispersalpatterns on aggregation is emphasized et al. of limestone Turnbull When annual (1999). by species grasslandweredispersedrandomly thecommunity thatdeveloped by theinvestigators, stillshowedstrongaggregation, variablesexertimindicatingthatenvironmental controls over in the portant dispersionpatterns system. is a scale-dependent A moregeneralproblemin thisworkis thataggregation at measure,andthusspeciesclumpedat one scalemaybe randomlydistributed to the degreeof another.Thus,if the questionis whetherdispersalis important thenthe measurement of aggregation needsto be madeoverscales aggregation, at scaleslargerthandispersimilarto the spatialscaleof dispersal.Aggregations of larger-scale sal mayindicatethe importance processessuchas environmental Forexample,Peres& Baider's(1997)hypothesisthatclumpingin heterogeneity. thatthespatial Brazilnut treesis dispersaldrivenmayseemreasonable considering scaleof the seed-dispersal kernelgeneratedby agoutismatchesthe scaleof tree thatdispersalmodedoesnotinfluence clumping.By contrast,it is notsurprising x of 10 km at the scale 10 gridcellsdividingtheUnitedKingdom plantclumping et al. 1994). (Quinn COEXISTENCETHROUGHLOCALDISPERSAL, AND SEGREGATION AGGREGATION, ClassicLotka-Volterra competitionmodelsassumethatdispersalandspeciesinMuchof therecentinterestin theinfluenceof teractionsarespatiallyunrestricted. work hasbeenmotivated on structure by theoretical dispersal community species slows rates of local often Results that this dispersal suggest relaxing assumption. Murrell & Law Law et al. & Pacala 2003, 1999, (Bolker competitive displacement 2003).This,alongwithcoexistenceachievedthrougha competition-colonization trade-offhasgenerated a surgeof interestin spatialprocessesinplantcommunities Murrell & Law 2003). However,as we explainbelow,the et al. 2001, (Murrell local literature dispersalto coexistenceis plaguedby ambiguityoverthe relating temporalscalesoverwhichlocaldispersalinfluencescoexistence. Models ShowingLocalDispersalEffectson Coexistence Anincreasingly amongecologistshasbeenthatlocaldispersal popularperception indipromotescoexistenceby causingthe spatialsegregationof heterospecific viduals across a landscape (e.g., Green 1989, Murrellet al. 2001, Pacala 1997, Pacala & Levin 1997,Weiner & Conte 1981). In homogeneous environments (as assumed in most community theory), mathematicalmodels for two-species DISPERSAL EFFECTS ONCOMMUNITIES 561 moreto theaggregation of concompetitionshowthatlocaldispersalcontributes proximitythanit does to aggregationof hetspecificsthroughparent-offspring &Law2003).Thissegregation of speciesreducestherelative erospecifics(Murrell of inter-versus interactions (Pacala 1997);moregenerally, frequency intraspecific weakerinter-versus favors coexistence (Chesson2000). intraspecific competition Localinteractionsare also key in thesemodels(Bolker& Pacala1997, 1999; Dieckmann et al.2000;Lawet al. 2001,2003;Murrell&Law2003).If speciesinteractwithotherindividuals in a systemovera sufficiently largespatialscale,local will have little effect on the of versusconspecific dispersal frequency heterospecific interactions. of suchspatialsegregation forlong-termcoexistence However,theimportance hasrecentlybeenchallengedon severalcounts.First,it is apparent thatinequality in dispersalkernelsbetweenotherwisesimilarspecieswill leadtotheexclusionof themoreaggregated allelsebeingequal)(Bolker& species(theshortestdisperser, Pacala1999,Durrett& Levin1998,Murrell& Law2003,Pacala1986).Second, even if dispersalis symmetric,segregationon its own is not enoughto prevent exclusionwhereit is predicted in thenonspatial case(Chesson&Neuhauser 2002, Durrett& Levin1998,Ghandiet al. 1998,Neuhauser& Pacala1999,Takenaka et al. 1997).In these models,the dynamicsmay be thoughtof as havingtwo phases:the initialphasewheremonospecificclustersbuildup owing to local Itis thesecond dispersal,andthesecondphasewheretheseclustersstarttointeract. that is most for the outcome phase important determining long-term(equilibrium) of competition,andin particular it is the interactions at the clusterboundaries thatareof greatestimportance 2002,Ghandiet al. 1998). (Chesson& Neuhauser interactions are and Here,heterospecific relativelyfrequent thestronger competitor wins.Thus,clustersof the strongerspecieswill slowlybutsurelyoverwhelmthe clustersof theweakerspecies.It is notablethatthe speedwithwhichthisoccurs is dependenton the geometryof the clusterboundary;the largerthe curvature of the clusterinterfacethe fasterthe rateof exclusion(Ghandiet al. 1998).The curvature of theclustersis greatlydependenton thedispersalkernel,providinga mechanisticlinkbetweenthedispersaldistanceandtherateof exclusion. Thus,ratherthana stabilizingforceforcoexistence,localizeddispersalmaybe thoughtof as an equalizingprocessbecauseit slowsdownthe dynamicswithout (Chesson2000,Ghandiet al. 1998).Nonethechangingtheexpectedequilibrium less, by slowingexclusionto suchlongtimescales,otherprocessessuchas immigration(Hubbell2001)andselection(Aarssen1984,Park&Lloyd1955,Pimentel Inneutralmodelsexploredby Hubbell(2001) 1968)mayactto maintaindiversity. andChaveet al. (2002),local dispersalslows the rateat whichspeciesdriftto extinction.Becausethesemodelsalso incorporate externalimmigration or speincreasesspeciesrichnessby changingthe ciation,this slowingof displacement balancebetweenextinctionandspeciation(or colonization).In doingso, local distribution dispersalalso changesthe species-abundance (Chaveet al. 2002). Similarresults are achievedwhen species coexist throughdensity dependenceor fecundity-competitiontrade-offs. 562 LEVINE0 MURRELL In a comparativelyrareexampleof theoryexaminingthe effects of local dispersal in spatiallyheterogeneousenvironments,Snyder& Chesson (2003) found that local dispersalincreasedthe degreeof coexistence as comparedwith thatachieved under global dispersal. This result requiresniche differentiationamong species and dispersaloccurringover smaller spatial scales than the environmentalautocorrelation.However,in the model, environmentalheterogeneityis fixed, and as the authorscaution,longer-rangedispersalmay be more importantin temporally heterogeneousenvironments. A possibly moresurprisingresultof segregationis that,at least in discrete-space models where interactionsoccur in very small neighborhoods(nearestneighbor on a grid), local dispersalcan hindercoexistence (Bolker et al. 2003, Neuhauser & Pacala 1999). This occurs when species are competitivelyunequalbut intrais greaterthan interspecificcompetition, allowing coexistence in the nonspatial analog. Nonetheless, this effect only serves to reduce the amountof coexistence where it is marginalin the firstplace. Much of the above theoryis based on the well knownLotka-Volterracompetition equations,which areextendedto includelocal interactionsandlocal dispersal. Clark& Ji (1995) used a more mechanistic,patch-basedmodel to show thatlocal dispersalbetweenneighboringpatchesanda disturbanceregimethatresetpatches periodically could aid coexistence as long as there was spatial variationin seed deposition (caused by both seed limitation and local dispersal),and a nonlinear relationshipbetween fecundity and patch density. As in a spatial-storageeffect (Chesson 2000), high seed productionin relativelyuncrowdedpatchesmore than outweighs the poor seed productionin crowdedpatches, allowing stable coexistence. Incorporatinga more mechanisticapproach,such as that used by Clark& Ji, while still including local interactionsand local dispersal,would be of much value in extendingthe theoryon dispersaland coexistence. Field Evidence that Local Dispersal Influences Coexistence Motivatedby the results of recent theory, a large number of empirical studies now speculatethat local interactionsfavor coexistence (e.g., Hubbellet al. 1999, Tuomistoet al. 2003). However,largely owing to the only recent developmentof the models, the rigorousempiricaltests requiredto justify such claims have yet to be completed.Rees et al. (1996) used models fit to monitoringdata of sand dune annualsin Britainto show that species sufferedintense intraspecificcompetition butlittle interspecificcompetitionin the field. They attributedthis resultto the high degree of intraspecificaggregationfound in the system. The degree to which this clumping resultedfrom local dispersalversus species specializationon different habitattypes was unclear.Stoll & Prati(2001) examinedcompetitiveinteractions among species experimentallyplanted in clumped versus randomdistributions. Consistent with aggregationenhancing the impacts of intra-versusinterspecific interactions,competitively inferior species performedbetter, and competitively superiorspecies performedworse in the aggregatedtreatment.These results are similarto thosein a longer-termstudyby Schmidt(Rejmanek2002, Schmidt1981), DISPERSAL EFFECTS ONCOMMUNITIES 563 whichshowedthatovera periodof threeyears,intraspecific allowed aggregation the exotic Solidago canadensis to coexist with the native Urtica dioicia, whereas withouttheaggregation, S. canadensiswasquicklyexcluded.Bothstudiesfollow froma historyof appliedresearchexamininghow spatialpatterninginfluences theimpactsof weedson crops(Garrett & Dixon1998).Weemphasize,however, thatlocal dispersalneed not alwaysbe important.Webb& Peart(2001) used modelsparameterized withfielddatato suggestthatlocaldispersalwasrelatively for tree coexistence in a Borneantropicalforest. unimportant Weconcludefrompublishedworkthatlocaldispersaldoesnotfavorthelongtermcoexistenceof speciesbut insteadsimplyslows the rateof displacement. thecoexistenceof speciesin a naturalcommunityto Thus,if we wantto attribute thelocalnatureof dispersal,we needto firstclarifythetimescalesoverwhichwe areattempting to explaincoexistenceandthendemonstrate thatlocaldispersalis sufficientto explaincoexistenceoverthosetimescales.Thisis verydifferentthan the almosttrivialresultthatlocal dispersalslowsdowndisplacedemonstrating ment.Incorporating timescaleswill almostundoubtedly requiretheuseof models withfielddata,as advocatedin ourconclusions. parameterized COEXISTENCE THROUGHCOMPETITION-DISPERSAL TRADE-OFFS andsuperiorcolonizercan Ecologistshavelongknownthata superiorcompetitor coexistin homogeneous modelsystems(Hastings1980,Holmes& Wilson1998, Horn&MacArthur 1972,Hutchinson 1951,Levins& Culver1971,Tilman1994). thatthis mechanismof coexistenceextends However,the recentdemonstration to anynumberof species(Tilman1994)has generateda surgeof interestin this area.Coexistenceoccursbecausethe superiorcompetitorlacksthe colonization abilityto fill all availablehabitats,leavingspacefor morepoorlycompetingbut better-colonizing species.Althoughcolonizationis a functionof bothfecundity anddispersalability,mosttheoryhasassumeda competition-fecundity trade-off (Hastings1980,Tilman1994);only morerecentlyhas dispersaldistancebeen consideredexplicitly(Bolker& Pacala1999,Dytham1994,Holmes& Wilson interestis coexistencethrough 1998,Murrell& Law2003).Althoughourprimary resultsincortrade-offs,we alsoreviewrelevanttheoretical competition-dispersal trade-offs. poratingcompetition-fecundity Assumptions UnderlyingTheoreticalResults Theresultthatnumerousspeciescancoexistthrougha competition-colonization trade-off(Tilman1994)dependson severalquestionable First,there assumptions. mustbe a stricthierarchyof competitioninverselyrelatedto colonization.Second,the colonizationof a strongercompetitormustalwayseliminateanyestablished weaker competitorfrom its location, and do so prior to its reproduction (Tilman 1994, Yu & Wilson 2001). This rapid displacementassumptionseems particularlyunrealisticfor manyplants(Levine & Rees 2002, Yu & Wilson 2001) 564 LEVINE0 MURRELL because seedlings are unlikely to outcompeteestablished individuals.Relaxing this assumptionrecovers so-called lottery models, where competitionfor empty sites occurs at the juvenile stage (Yu & Wilson 2001). In such lotterymodels, the competition-colonizationtrade-offis insufficientto producecoexistence,although coexistencecanbe achievedwith environmentalheterogeneity,fecundity-dispersal trade-offs,or stochasticvariationin seed arrival(Chesson& Warner1981, Comins & Noble 1985, Kisdi & Geritz2003, Kohyama1993, Yu & Wilson 2001). Althoughthe above assumptionspertainto competition-colonizationtrade-offs generally,thenatureof competitionis also key to coexistenceproducedspecifically by a competition-dispersaltrade-off.Species with longer-rangedispersalcan coexist with superiorcompetitors(Bolker & Pacala 1999, Law & Dieckmann2000, Murrell& Law 2003), but this requiresa relativelylarge asymmetryin competition (Murrell& Law 2003). Withoutstronglyasymmetriccompetition,the species possessing an optimal combinationof competition and colonization dominates the system. A similarcompetitivehierarchyis requiredin competition-fecundity trade-offmodels (Adler & Mosquera2000, Geritzet al. 1999). If this assumption is met, Holmes & Wilson (1998) show thatspecies with greaterdispersalcan also coexist with those that are simultaneouslybetter competitorsand more fecund. Nonetheless,this resultwas only found for a small set of parametercombinations. The ability of the competition-colonizationtrade-offto generatecoexistence is also sensitive to the density and fecundity of the superiorcompetitor.If the superiorcompetitorhas fecundity high enough that it leaves only a small fraction of unoccupiedpatches, the inferiorcompetitoris unlikely to persist (Bolker et al. 2003). High density can also be achievedthroughhigh dispersal,which reduces parent-offspringcompetition,favoringa more even distributionof individuals across space (Bolker & Pacala 1997, Ellner2001, Law et al. 2003). However, this morediffuse distributionof individualsmay still leave temporarygaps thatthe weakercompetitorcan exploit as long as it more rapidlycompletes its life cycle. weakercompetitorscan coexist with a long-rangeIn fact, short-range-dispersing dispersing dominantby having a higher turnoverrate (Bolker & Pacala 1999). In sum, the main conclusion from the theory is that while competitiondispersal trade-offsfavorcoexistence, this requiresseveralassumptionsthatmay not apply to real systems. EmpiricalSupport Likethe theoreticalwork,mostof the empiricalstudiesexaminingcompetitionforcoexistenceexaminefecunditycolonizationtrade-offsandtheirimplications competitiontrade-offs(Levine& Rees 2002). Evidencefor a dispersal-competition trade-offis sparseand comes largelyfrom systemswith regulardisturbance. Breweret al. (1998) foundonly weak evidencefor a competition-dispersaltrade-off amongclonal grassescoexistingin a regularlydisturbedsaltmarshhabitat.Instead they attributedcoexistence to interspecificdifferencesin physiological tolerance of gap conditions.Yeaton& Bond (1991) showed that ant dispersalgives a competitively inferior South African fynbos shrub an advantagein colonizing the DISPERSAL EFFECTS ONCOMMUNITIES 565 open areas after disturbance.However,Markovchain models suggested that this dispersal advantagewas not sufficientto explain long-termcoexistence with the competitivedominantshrubin the system. In some of the best supportfor coexistence achievedthrougha competitiondispersaltrade-off,Platt(1975) showed that the species occupying matureIowa prairie(the competitivedominants)tended to dispersemorepoorlythan"fugitive"species living primarilyon badgermounddisturbances.However,dispersalcapacity and propaguleproductionwere positively correlated;thus, the source of the colonizationadvantageis unclear. The best support for coexistence achieved through a competition-dispersal trade-offcomes from models parameterizedwith field data, althougheven here evidence is not definitive.Workingin temperateforest with two codominanttree species, Nanami et al. (1999) documentedclumpingin the gravity-dispersedand dioecious-competitivedominant,which they hypothesizefavoredthe persistence of the bird-dispersedcompetitiveinferior.Althoughthis is simply a patternanalysis, they followed the work with mathematicalmodels testing the suitabilityof a competition-dispersaltrade-offas a coexistencemechanismin the system(Nanami et al. 2000). Resultsindicatedthatbecause male trees of the competitivedominant do not dropseeds below the parentcanopy,this createsgaps thatare differentially colonized by the better-dispersingcompetitiveinferior.Still, this result depends stronglyon the dioecy, restrictingits generalityto other systems. In Pacala et al.'s (1993) analysis of their SORTIEforest simulation model, which they parameterizedwith field-derivedmeasuresof vital rates, coexistence in EasternU.S. forest trees is achieved in part througha documentedtrade-off between dispersaland the ability to cast and surviveshade.Althoughothertradeoffs are also importantto coexistence in their simulation,Ribbens et al. (1994) show a tremendousimpact of changing mean dispersal distance on coexistence and dominance.Still, althougha numberof studiesimplicitly or explicitly invoke competition-colonizationtrade-offs,relativelylittle definitiveempiricalevidence exists to supportcoexistence achievedthroughthis mechanism. CONCLUSIONS Ourreview of the literatureuncovereda dichotomyof supportfor the importance of seed dispersalfor communitypatterns.Although we found an ever expanding body of theory suggesting that seed dispersalaffects species coexistence through local dispersaland a competition-dispersaltrade-off,empiricalsupportwas scant. Instead,most empiricalinvestigationsexaminedhow dispersaland seed-trapping agents affect species distributionand dispersion,subjectswith little theory.In addition, this work relied heavily on dispersalproxies and correlationalanalyses of communitypatterns,methods unable to exclude alternativehypotheses. Interestingly, this relativelytepid supportfor dispersal'sinfluenceon communitypatterns is in contrastto work at the populationlevel, where dispersalis well appreciated to stronglyinfluence fitness, colonization, spread,and persistence(Harper1977, Howe & Smallwood 1982, Skellam 1951). 566 LEVINEm MURRELL Thus, the main conclusion of our review is that it may be prematureto expect influence structure. Ofcourse, thatpatterns of seeddispersal community strongly species withoutany dispersalcould not spreadbeyond a foundingindividual,but it remainsunclearwhetherthespecificpatternof dispersedseedis a strongdeterminantof relativeabundance, distribution, dispersion,andcoexistencein natural systems.Althoughthis lack of claritystemsfromthe weak empiricalsupportin the thisweaksupportcouldbe interpreted in severalways.Itmaybethatthe literature, butcurrent is important, approaches methodological shapeof thedispersal kemrnel in thefield. to definitively demonstrate itsimpactsonpattemrns aretoocorrelational believe. Resoluthan we be less commonly dispersalmay important Altemrnatively, to the more tionof thesealternatives rigorousapproaches understanding requires of dispersal. importance Evidence theEmpirical Improving in the field is difBecause estimating and manipulatingseed-dispersalkemrnels for community structurecan ficult, testing the importanceof dispersal kemrnels be challenging. In principle,the ideal empiricalstudy would (a) documentseed shadows and dispersionpatterns,(b) correlatethese with patternsof community structure,and then (c) demonstratethat experimentallymanipulatingthe seed shadow changes populationor community structure(Schupp & Fuentes 1995). Althoughexamplesof each of these steps can be found in the literature,we found no single study that performedall of them. Instead,most of the empiricalwork relating dispersal to community propertiestended to be correlational,often involving proxies for dispersalkernels, such as dispersalmode. These studies thus reliedheavily on severalpotentiallyprecariousassumptions,includingthe fact that dispersalmode is a reasonablepredictorof dispersalkernelsand that aggregation unrelatedto currentenvironmentalvariablesis attributableto dispersal.These assumptionsmay be valid in much of the work (Willson 1993), but withoutfurther supportsome conclusions may be incorrect.This is illustratedby the often-cited example where dispersalinfluences patternsof mangrovetree zonation through the tidal sorting of propagules (Rabinowitz 1978). Although Rabinowitz's hypothesis has intuitiveappeal,recent work actuallyquantifyingthe seed-dispersal kernelsof mangrovetrees in the same forests suggests the tidal sortingof propagules is unlikely to generatethe zonation (W. Sousa & B. Mitchell, unpublished data). How can we more definitively demonstratethat dispersal influences relative abundance,distribution,dispersion, and coexistence? To this end, we strongly encourageexperimentsthat directly manipulatethe seed-dispersalkernel, as advocatedin more specific cases by Bolker & Pacala(1999), Pacala& Rees (1998), and Schupp & Fuentes (1995). Such experimentsare not substitutesfor quantifying dispersalkernels and communitypatterns,but they are complementaryand uniquelypoised to definitivelytest the mechanismssuggestedby the patterns.The basic methodology involves manipulatingthe seed-dispersalkernel in replicate DISPERSAL EFFECTS ONCOMMUNITIES 567 distriplotsandcomparingtheresultingcommunitypattern(relativeabundance, control.Thekernelis manipulated bution,coexistence)to thatin anunmanipulated by firstcollectingall seedproducedin a plotandthendependingon theresearch question,dispersingit randomlyor locally.Community patternsin an additional treatment,whereseed is dispersedfollowingthe estimatednaturalkernels,can be comparedwith the unmanipulated controlto test for any artifactsassociated with seed handling.Thisbasicapproachhas beensuccessfullyusedto examine thepopulation-level consequencesof thedispersalkernelin a tropicalforesttree & (Augspurger Kitajima1992). If dispersalcontrolscommonnessandrarity,thenforcingall species'dispersal to follow the across-speciesmeanor mediankernelshouldcausethe more commonspeciesto declineandthe rarerspeciesto increaserelativeto controls. If dispersalkernelscontrolaggregationor otherdistribution patterns,thendispersingall seed globallyshouldeliminate,or at leastbeginto homogenize,the distribution or aggregation patterns(e.g., Levine2001, Turnbullet al. 1999).If speciescoexistthroughlocaldispersal,thendispersingseedsgloballyshouldenhancecompetitivedisplacement (Bolker& Pacala1999).Lastly,if speciescoexist througha competition-dispersal trade-off,thenequalizingthe dispersalkernel acrossspeciesshouldresultin the displacement of species(Pacala& Rees 1998). Theexperiments will be mosttractablewithcommunities of annualor shortlivedperennialplantswheredispersaloccursoverrelativelysmallspatialscales. trendsin thepredicteddirectionwithearly Still,evenwithlonger-lived organisms, life stages(Augspurger & Kitajima1992) can be used to bolstercorrelational evidence.In addition,seed additionexperimentslocatedat differentlocations relativetoexistingdistributions couldfollowtherationale of theexperiments above butbe conductedoverlargerspatialscales.For example,in studiesattributing adultclumpingto localdispersal,experiments shouldshowseedlimitationaway fromtheclump.In otherwords,if seedsdidtravelfurther,adultscouldestablish. conductedin realecologicalhabitatswhereother Last,we encourageexperiments factors are freeto exerttheirimpacts.Ourreviewdoesnot potentiallyimportant anddiversity, distribution, questionwhetherdispersalhasanyeffectonabundance, butratherquestionstheimportance of suchaneffectin comparison to otherfactors (Figure1). FutureDirections for Theory doesnotlead trade-offs,aggregation Theoryhasshownthatwithoutinterspecific to stablecoexistence(Bolker& Pacala1999,Durrett& Levin 1998,Murrell& Law2003,Neuhauser &Pacala1999,Takenaka et al. 1997).Yetthemodelsusedto this result are based on the Lotka-Volterra produce largely competition equations, andmostincorporate lineardensitydependence. morecomplexproIncorporating cesses such as nonlinearcompetitionmay change some of these predictionsor at the very least lead to new strategies for coexistence (Bolker & Pacala 1999). 568 LEVINEN MURRELL However,few modelshavetakenthis approach(see Pacala1986for a rareexception).Incorporating processessuchas nonlineareffectsof densitymaybe at theexpenseof analyticaltractability, butthetechniquesavailableto spatialtheory aresophisticated to make it possible(Dieckmannet al. 2000, Tilman& enough Kareiva1997). Theoryhasalsofocusedlargelyon thequestionof coexistence.Yetothercomfromtheoretical munitydynamicsrequirecloserattention ecologists.Forexample, theworkof Chaveet al.(2002)showedthatthescaleof dispersalis themostimportantinfluenceon manymeasuresof communitystructure. Still,it is notknownif theseresultsarerobustto inequalityin dispersalabilityacrossspecies.Althoughit is oftenassumedthatlonger-range arethemoreabundant, thisprediction dispersers favoringshort-range mightchangewhenthelandscapeis spatiallyheterogeneous, dispersal(Bolker2003,Travis& Dytham1999). Last,empiricalworkshouldmotivateinteresting modelingquestions.Mostnotably,we founda surprisingnumberof empiricalstudiesthatdocumentedthe of trapping patternsof seedarrival.Interestingly, importance agentsin controlling whereasclumpingthatresultsfromlocaldispersalshouldenhanceintraspecific agclumpingthatresultsfromgeneraltrapping gregationandthusslowdisplacement, contact.Alternatively, trappingmay segreagentscouldincreaseheterospecific influences if seed trappingpatterns,as in Schneider& morphology gatespecies Sharitz(1988). How trappingagents,theirspeciesspecificity,andtheirspatial andcoexistenceareissuesripefor influenceabundance, distribution, arrangement theoretical exploration. Closing the Gap BetweenTheory and EmpiricalWork literature a cleardichotomybetweentheempiricalandtheoretical Wedocumented this is to kernels gap essential communitypatterns.Closing relatingdispersal of dispersal.To this end,we stronglyadvocatethe to clarifyingthe importance documented withempirically modelsparameterized of mathematical development kernels and other realistic (as in Wu & parameters demographic seed-dispersal can uniquelyaddresswhetherthe predictionsof Levin 1994).Suchapproaches modelsreasonablydescribethe dynamicsof real ecological the mathematical to askthe key longwill also forceinvestigators communities. Suchapproaches orwhatcreates thedispersal-driven termquestionssuchaswhatmaintains patterns, for exploring informative seed limitation.Modelsshouldalso proveparticularly on theinfluenceof seed-dispersal kernels,parentplantdensity,andseedproduction betweendispersalkernels seeddepositionacrossthelandscape.Therelationship but critical,linkagebetweendispersaland and seed rainis an underexplored, communitypatterns. of seeddispersalforcommunitydynamicshaslong Thepotentialimportance been acknowledged,and recenttheoreticalresults supportthis expectation.However, the results of our review suggest that the importanceof dispersal cannot be taken for granted;empirical supportfor the theoreticalpredictionsis largely DISPERSAL EFFECTS ONCOMMUNITIES 569 lacking. A firmerunderstandingof the role of dispersalin communitystructureis achievableif we begin with more rigorousfield approachesdirectlymanipulating seed-dispersalkernels.These resultsmust thenby coupled with the predictionsof models incorporatingempiricallyderiveddispersalkernels. The Annual Reviewof Ecology,Evolution,and Systematicsis online at http://ecolsys.annualreviews.org LITERATURE CITED tounderstand AarssenLW.1984.Ecologicalcombiningabildriven equations stochastically and in formation in spatialpattern ecologicalsysity competitivecombiningability tems. Theor Popul. Biol. 52:179-97 plants:towarda generalevolutionarytheoryof coexistencein systemsof competition. BolkerBM,PacalaSW.1999.Spatialmoment Am. Nat. 122:707-31 understandequationsforplantcompetition: AdlerFR, MosqueraJ. 2000. Is spacenecesof ing spatialstrategiesandthe advantages Interference and limits to short Am. Nat. 153:575-602 sary? competition dispersal. BolkerBM, PacalaSW,NeuhauserC. 2003. Ecology81:3226-323 biodiversity. AlcantaraJM, Rey PJ, ValeraF, SanchezSpatialdynamicsin model plantcommunities:Whatdo we reallyknow?Am.Nat. LafuenteAM. 2000. Factorsshapingthe seedfallpatternof a bird-dispersed 162:135-48 plant. Brewer 1937-50 SJ,RandT, LevineJM,BertnessMD. Ecology ME.2000. Andersson 1998.Biomassallocation,clonaldispersal, E, NilssonC, Johansson Plantdispersal inborealriversanditsrelation andcompetitivesuccessin threesaltmarsh to thediversityof riparian flora.J. Biogeogr plants.Oikos82:347-53 27:1095-106 BullockJM, KenwardRE, Hails RS. 2002. K. 1992. CK, Augspurger Kitajima Experimen- Dispersal Ecology. Malden,MA: Blackwell. talstudiesof seedlingrecruitment fromcon458 pp. AE.2000.Longtrastingseeddistributions. Ecology73:1270- CainML,MilliganBG,Strand 84 distanceseeddispersalin plantpopulations. BarotS, GignouxJ,MenautJ. 1999.Demogra- Am. J. Bot. 87:1217-27 from CareyPD,Watkinson AR. 1993.Thedispersal phyof a savanna palmtree:predictions andfatesof seedsof thewinterannualVulpia comprehensive spatialpattern analysis.Ecolciliata. J. Ecol. 81:759-67 ogy 80:1987-2005 Bleher B, Bohning-GaeseK. 2001. Conse- ChaveJ, Muller-Landau HC,LevinSA. 2002. of for seed disclassical quences frugivorediversity Comparing communitymodels: andthe spatheoreticalconsequences for patternsof dipersal,seedlingestablishment tialpatternof seedlingsandtrees.Oecologia versity.Am. Nat. 159:1-23 129:385-94 ChessonP.2000.Mechanisms of maintenance BleherB, Oberrath K.2002. of species diversity.Annu. Rev. Ecol. Syst. R,Bohning-Gaese Seeddispersal, 31:343-66 breedingsystem,treedensity andthespatialpatternof trees-a simulation ChessonP, NeuhauserC. 2002. Intraspecific andspeciescoexistence.Trends approach.Basic Appl. Ecol. 3:115-23 aggregation BolkerBM.2003.Combining and Ecol.Evol.17:210-11 endogenous exogenousvariabilityin analyticalpopula- ChessonP, WarnerRR. 1981.Environmental tion models. Theor Popul. Biol. In press variabilitypromotescoexistencein lottery BolkerBM, PacalaSW. 1997.Usingmoment competitivesystems.Am.Nat. 117:923-43 570 LEVINE 0 MURRELL Clark JS. 1998. Why trees migrate so fast: confronting theory with dispersal biology and the paleorecord.Am. Nat. 152:204-24 ClarkJS, Ji Y 1995. Fecundityanddispersalin plantpopulations:implicationsfor structures and diversity.Am. Nat. 146:72-111 ClarkJS, MacklinE, WoodL. 1998. Stages and scales of recruitmentlimitation in southern Appalachianforests. Ecol. Monogr.68:21335 ClobertJ, DanchinE, DohndtAA, Nichols JD. 2001. Dispersal.Oxford:OxfordUniv.Press. 452 pp. Comins HN, Noble IR. 1985. Dispersal, variability, and transientniches: species coexistence in a uniformly variable environment. Am. Nat. 126:706-23 ConditR, AshtonPS, BakerP, Bunyavejchewin S, GunatillekeS, et al. 2000. Spatialpatterns in the distributionof tropicaltreespecies.Science 288:1414-18 Condit R, PitmanN, Leigh EG, Chave J, TerborghJ, et al. 2002. Beta-diversityin tropical forest trees. Science 295:666-69 Connell JH. 1971. On the role of naturalenemies in preventing competitive exclusion in some marine animals and in rain forest trees.InDynamicsofPopulations,ed. PJdenBoer, GR Gradwell,pp. 298-312. Wageningen, Neth.: Cent. Agric. Publ. Doc. Cornell HV. 1993. Unsaturated patterns in species assemblages:therole of regionalprocesses in setting local species richness. In Species Diversity in Ecological Communities, ed. RE Ricklefs, D Schluter,pp. 243-52. Chicago/London:Univ. Chicago Press Dalling J, Muller-Landau HC, Wright J, Hubbell SP. 2002. Role of dispersal in the recruitmentlimitationof neotropicalpioneer species. J. Ecol. 90:714-27 DanvindM, Nilsson C. 1997. Seed floatingability and distributionof alpine plants along a northernSwedish river.J. Veg.Sci. 8:271-76 Dieckmann U, Law R, Metz JAJ. 2000. The Geometryof EcologicalInteractions:Simplifying Spatial Complexity.Cambridge:CambridgeUniv. Press. 564 pp. Dockery J, Hutson V, Mischaikow K, PernarowskiM. 1998. The evolutionof slow dispersal rates: a reaction diffusion model. J. Math.Biol. 37:61-83 Drake JA, Mooney HA, DiCastri F, Groves RH, KrugerFJ, et al. 1989. Biological Invasions: A Global Perspective. Chichester, Engl.: JohnWiley. 525 pp. DurrettR, Levin S. 1998. Spatialaspects of interspecific competition. Theor.Popul. Biol. 53:30-43 DythamC. 1994. Habitatdestructionandcompetitive coexistence: a cellular model. J. Anim.Ecol. 63:490-91 Ellner SP. 2001. Pair approximationfor lattice models with multiple interactionscales. J. Theor.Biol. 210:435-47 Eriksson O. 1997. Colonization dynamics and relative abundance of three plant species (Antennaria dioca, Hieracium pilosella and Hypochoerismaculata)in seminaturalgrasslands.Ecography20:559-68 Eriksson O, Jakobsson A. 1998. Abundance, distributionand life histories of grassland plants:a comparativestudy of 81 species. J. Ecol. 86:922-33 FragasoJM. 1997. Tapir-generatedseed shadows: scale dependentpatchinessin the Amazon rainforest. J. Ecol. 85:519-29 FriedmanJ, Stein Z. 1980. The influence of seed-dispersalmechanismson the dispersion of Anastatica hierochuntica(Cruciferae)in the Negev desert,Israel.J. Ecol. 68:43-50 GarrettA, Dixon PM. 1998. Whendoes the spatialpatternof weeds matter?Predictionsfrom neighborhoodmodels. Ecol. Appl. 8:125059 Ghandi A, Levin S, Orszag S. 1998. Critical slowing down in time to extinction: an example of critical phenomenain ecology. J. Theor.Biol. 192:363-76 GeritzSA, van der MeijdenE, Metz JAJ.1999. Evolutionary dynamics of seed size and seedling competitive ability. Theor. Popul. Biol. 55:324-43 Green DF. 1989. Simulated effects of fire, dispersal and spatial pattern on competition withinforestmosaics. Vegetatio82:13953 DISPERSALEFFECTSON COMMUNITIES Guo Q, Brown JH, Valone TJ, KachmanSD. 2000. Constraintsof seed size on plant distributionand abundance.Ecology 81:214955 Hanski IA, Gilpin ME. 1997. Metapopulation Biology. San Diego: Academic. 512 pp. HarperJL. 1977. PopulationBiology of Plants. London:Academic. 892 pp. Hastings A. 1980. Disturbance, coexistence, history and competition for space. Theor Popul.Biol. 163:491-504 Holmes EE, Wilson HB. 1998. Runningfrom trouble:long-distancedispersalandthe competitive coexistence of inferior species. Am. Nat 151:578-86 Honnay O, Verhaeghe W, Hermy M. 2001. Plant community assembly along dendritic networks of small forest streams. Ecology 82:1691-702 Horn HS, MacArthurRH. 1972. Competition among fugitive species in a harlequinenvironment.Ecology 53:749-52 HorvitzCC, LeCorffJ. 1993. Spatialscale and dispersionpatternof ant- andbird-dispersed herbs in two tropical lowland rain forests. Vegetatio107/108:351-62 Houle G. 1995. Seed dispersal and seedling recruitment:the missing link(s). Ecoscience 2:238-44 Hovestadt T, Poethke HJ, Messner S. 2000. Variabilityin dispersal distances generates typical successionalpatterns:a simple simulation model. Oikos 90:612-19 Howe HF. 1989. Scatter-and clump-dispersal and seedling demography:hypotheses and implications.Oecologia 79:417-26 Howe HF, SmallwoodJ. 1982. Ecology of seed dispersal.Annu.Rev.Ecol. Syst. 13:201-28 Hubbell SP. 1979. Tree dispersion,abundance, and diversityin a tropicaldry forest. Science 203:1299-309 Hubbell SP. 2001. The UnifiedNeutral Theory of Biodiversity and Biogeography. Princeton/Oxford:PrincetonUniv. Press. 375 pp. HubbellSP,FosterRB, O'BrienST,HarmsKE, ConditR, et al. 1999. Light-gapdisturbances, recruitmentlimitation,andtree diversityin a neotropicalforest. Science 283:554-57 571 Hutchinson GE. 1951. Copepodology for the ornithologist.Ecology 32:571-74 JakobssonA, ErikssonO. 2000. A comparative studyof seed number,seed size, seedling size and recruitmentin grassland plants. Oikos 88:494-502 JanzenDH. 1970. Herbivoresand the number of tree species in tropical forests. Am. Nat. 104:501-28 JohanssonME, Nilsson C. 1993. Hydrochory, populationdynamics and distributionof the clonal aquatic plant Ranunculus lingua. J. Ecol. 81:81-91 Kisdi E, Geritz SAH. 2003. On the coexistence of perrenialplantsby the competitioncolonizationtrade-off.Am. Nat. 161:350-54 Kiviniemi K, Eriksson O. 1999. Dispersal, recruitment and site occupancy of grassland plantsin fragmentedhabitats.Oikos86:24153 Kohyama T. 1993. Size structuredtree populations in gap forest-the forest architecture hypothesis for the stable coexistence of species. J. Ecol. 81:131-43 Law R, DieckmannU. 2000. A dynamicalsystem for neighborhoodsin plantcommunities. Ecology 81:2137-48 Law R, MurrellDJ, Dieckmann U. 2003. On populationgrowthin space and time: spatial logistic equations.Ecology 84:252-62 Law R, PurvesDW,MurrellDJ, DieckmannU. 2001. Dynamics of small-scale spatialstructurein plantpopulations.InIntegratingEcology and Evolution in a Spatial Context,ed. J Silvertown,J Antonovics, pp. 21-44. Oxford:Blackwell Sci. Leishman MR. 2001. Does the seed size/ number trade-off model determine plant community structure? An assessment of the model mechanismsand their generality. Oikos 93:294-302 Levine JM. 2001. Local interactions,dispersal, and native and exotic plant diversityalong a Californiastream.Oikos 95:397-408 Levine JM. 2003. A patch modeling approachto the community-levelconsequences of directional dispersal. Ecology 84:121524 572 LEVINE MURRELL Levine JM,Rees M. 2002. Coexistenceandrelativeabundancein annualplantassemblages: the roles of competition and colonization. Am. Nat. 160:452-67 LevinsR. 1969. Some demographicandgenetic consequences of environmentalheterogeneity for biological control.Bull. Entomol.Soc. Am. 15:237-40 Levins R, Culver D. 1971. Regional coexistence of species and competition between rare species. Proc. Natl. Acad. Sci. USA 68:1246-48 MaranonT, GrubbPJ. 1993. Physiological basis and ecological significance of the seed size and relative growthrate relationshipin Mediterraneanannuals.Funct. Ecol. 7:59199 Molofsky J, Bever JD, Antonovics J, Newman JT. 2002. Negative frequency dependence and the importanceof spatialscale. Ecology 83:21-27 MurrellDJ, Law R. 2003. Heteromyopiaand the spatial coexistence of similar competitors. Ecol. Lett. 6:48-59 MurrellDJ, PurvesDW, Law R. 2001. Uniting patternand process in plant ecology. Trends Ecol. Evol. 16:529-30 Myster RW, Pickett STA. 1992. Effects of palatability and dispersal mode on spatial patternsof trees in old fields. Bull. Torrey Bot. Club 119:145-51 NanamiS, KawaguchiH, KuboT. 2000. Communitydynamicmodelsof two dioecioustree species. Ecol. Res. 15:159-64 Nanami S, Kawguchi H, YamakuraT. 1999. Dioecy-induced spatial patterns of two codominant tree species, Podocarpus nagi andNeolitsea aciculata. J. Ecol. 87:678-87 NathanR. 2003. Seeking the secrets of dispersal. TrendsEcol. Evol. 18:275-76 Nathan R, Muller-LandauHC. 2000. Spatial patternsof seed dispersal,theirdeterminants and consequences for recruitment.Trends Ecol. Evol. 15:278-85 NeuhauserC, Pacala SW. 1999. An explicitly spatial version of the Lotka-Volterramodel with interspecific competition. Ann. Appl. Probab.9:1226-59 Nieder J, Engwald S, Klawun M, Barthlott W. 2000. Spatial distribution of vascular epiphytes (including hemiepiphytes) in a lowland Amazonian rain forest (Surumoni CranePlot) of southernVenezuela.Biotropica 32:385-96 Nilsson C, Ekblad A, Dynesius M, Backe S, GardfjellM, et al. 1994. A comparison of species richness and traitsof riparianplants between a main river and its tributaries.J. Ecol. 82:281-95 Nilsson C, Gardfjell M, Grelsson G. 1991. Importance of hydrochory in structuring plant communitiesalong rivers. Can. J. Bot. 69:2631-33 Overton JMC. 1996. Spatial autocorrelation and dispersalin mistletoes: field and simulation results. Vegetatio125:83-98 Pacala SW. 1986. Neighborhood models of plant populationdynamics.2. Multi-species models of annuals. Theor. Popul. Biol. 29:262-92 Pacala SW. 1997. Dynamics of plant communities. In Plant Ecology, ed. MJ Crawley,pp. 532-55. Oxford:Blackwell Sci. Pacala SW, Levin SA. 1997. Biologically generatedspatialpatternand the coexistence of competing species. In Spatial Ecology: The Role of Space in PopulationDynamics and InterspecificInteractions, ed. D Tilman, P Kareiva,pp. 204-32. New York:Princeton Univ. Press. 368 pp. Pacala SW, Rees M. 1998. Models suggesting field experimentsto test two key hypotheses explaining successional diversity.Am. Nat. 152:729-37 Pacala SW, Silander JA Jr. 1985. Neighorhood models of plant populationdynamics. I. Single-speciesmodels of annuals.Am.Nat. 125:385-411 Pacala SW, Canham CD, Silander JA. 1993. Forest models defined by field measurements:I. The design of a northeasternforest simulator.Can. J. For Res. 23:1980-88 Park T, Lloyd M. 1955. Natural selection and the outcome of competition.Am. Nat. 89:235-40 PastorJ, Cohen Y, Moen R. 1999. Generation DISPERSAL EFFECTS ONCOMMUNITIES 573 theimpactof competition andspatialheterogeneityon the structureanddynamicsof a PeartMH,CliffordHT.1987.Theinfluenceof fourspeciesguildof winterannuals. Am.Nat. and soil surface 147:1-32 propdiasporemorphology ertieson thedistribution of grasses.J. Ecol. RejmanekM. 2002. Intraspecific aggregation and species coexistence. TrendsEcol. Evol. 75:569-76 Peres CA, BaiderC. 1997. Seed dispersal, 17:209-10 JM.2000. Recruitment spatial distributionand populationstruc- Rey PJ,Alcantara dyture of Brazilnuttrees (Bertholletiaexcelsa) namics of fleshy-fruitedplant (Olea euAmazonia.J. Trop.Ecol. in southeastern ropaea):connectingpatternsof seed dis13:595-616 J. Ecol. persalto seedlingestablishment. PimentelD. 1968. Populationregulationand 88:622-33 RibbensE, SilanderJA, PacalaSW. 1994. genetic feedback.Science 159:1432-37 PlattWJ. 1975.The colonizationandformain forests:calibrating Seedlingrecruitment tionof equilibrium associations models to plantspecies predictpatternsof tree seedling onbadgerdisturbances in a tall-grassprairie. dispersion. Ecology75:1794-806 Ecol. Monogr 45:285-305 S. 1991.PlantcharacterisRydinH,Borgegard PlattWJ,WeisIM. 1977.Resourcepartition- tics overa centuryof primarysuccessionon withina guildof fugitive islands:LakeHjalmaren. ingandcompetition Ecology72:1089Am. Nat. 111:479-513 101 prairieplants. PlotkinJB,ChaveJ, AshtonPS. 2002.Cluster SchmidtW.1981.UberdasKonkurrenzverhaltree ten von Solidago canadensisund Urticadioianalysisof spatialpatternsin Malaysian cia. Verh.Ges. Okol.9:173-88 species.Am.Nat. 160:629-44 PlotkinJB,PottsMD,LeslieN, Manokaran N, SchneiderRL, SharitzRR. 1988.Hydrochory LaFrankie in a Bald Cypress-Water J, AshtonPS. 2000.Species-area and regeneration andhabitatspeforest. curves,spatialaggregation, Tupelo swamp Ecology 69:1055cializationin tropicalforests.J. Theor.Biol. 63 207:81-99 SchuppEW, FuentesM. 1995. Spatialpatternsof seeddispersalandtheunification of QuinnRM,LawtonJH,EvershamBC, Wood SN. 1994.Thebiogeography of scarcevasplantpopulation ecology.Ecoscience2:267cularplantsin Britainwithrespectto habitat 75 preference, dispersalabilityandreproductive ShigesadaN, KawasakiK. 2002.Invasionand the rangeexpansionof species:effects of biology. Biol. Conserv.70:149-57 RabinowitzD. 1978. Earlygrowthof manlong-distancedispersal.In DispersalEcolRSHails, groveseedlingsin Panama,andan hypothogy,ed.JMBullock,REKenward, esis concerning therelationship of dispersal pp. 350-73.Malden:BlackwellSci. 458 pp. andzonation.J. Biogeogr.5:113-33 ShipleyB, DionJ. 1992.Theallometryof seed RabinowitzD, RappJK. 1981.Dispersalabilin herbaceous Am. production angiosperms. ities of seven sparseand commongrasses Nat. 139:467-83 froma Missouriprairie.Am.J. Bot.68:616- ShmidaA, EllnerS. 1984.Coexistenceof plant 24 58:29specieswithsimilarniches.Vegetatio RandT. 2000. Seeddispersal,habitatsuitabil55 of halophytes acrossa SkellamJG. 1951.Randomdispersalin theoityandthedistribution saltmarshtidalgradient. J. Ecol.88:608-21 reticalpopulations. Biometrika 38:196-218 Rees M. 1995. Communitystructurein sand SnyderRE, ChessonP. 2003. Localdispersal duneannuals:Is seedweighta keyquantity? can facilitatecoexistencein thepresenceof J. Ecol. 83:857-63 Ecol.Lett. permanent spatialheterogeneity. ReesM, GrubbPJ,KellyD. 1996.Quantifying 6:301-9 of spatialpatternsin borealforestlandscapes. Ecosystems2:439-50 574 LEVINE MURRELL SrivastavaDS. 1999. Using local-regionalrichness plots to test for species saturation:pitfalls and potentials.J. Anim.Ecol. 68:1-16 Stoll P, PratiD. 2001. Intraspecificaggregation alterscompetitiveinteractionsin experimental plant communities.Ecology 82:319-27 SvenningJC.2001. Environmentalheterogeneity, recruitmentlimitationand the mesoscale distributionof palms in a tropical montane rainforest (Maquipucuna,Ecuador).J. Trop. Ecol. 17:97-113 SvenningJC, Skov F. 2002. Mesoscale distribution of understoreyplantsin temperateforest (Kalo,Denmark):the importanceof environment and dispersal.Plant Ecol. 160:169-85 TakenakaY, MatsudaH, Iwasa Y. 1997. Competition and evolutionarystability of plants in a spatially structuredhabitat.Res. Popul. Ecol. 39:67-75 Thompson K, Gaston KJ, Band SR. 1999. Range size, dispersal and niche breadthin the herbaceous flora of central England. J. Ecol. 87:150-55 Tilman D. 1994. Competition and biodiversity in spatially structuredhabitats.Ecology 75:685-700 Tilman D, KareivaP. 1997. Spatial Ecology: The Role of Space in Population Dynamics and InterspecificInteractions.New York: PrincetonUniv. Press. 368 pp. Travis JMJ, Dytham C. 1999. Habitatpersistence, habitatavailabilityand the evolution of dispersal. Proc. R. Soc. London Ser. B 266:723-28 Tuomisto H, Ruokolainen K, Yli-Halla M. 2003. Dispersal, environment,and floristic variationof westernAmazonianforests.Science 299:241-44 TurnbullLA, CrawleyMJ, Rees M. 2000. Are plant populationsseed-limited?A review of seed sowing experiments.Oikos 88:225-38 TurnbullLA, Rees M, CrawleyMJ. 1999. Seed mass andthe competition/colonizationtradeoff: a sowing experiment.J. Ecol. 87:899912 Wang BC, Smith TB. 2002. Closing the seed dispersal loop. TrendsEcol. Evol. 17:37985 Webb EL, Fa'aumu S. 1999. Diversity and structureof tropical rain forest of Tutuila, AmericanSamoa:effects of site age andsubstrate.Plant Ecol. 144:257-74 Webb CO, Peart DR. 2001. High seed dispersal rates in faunally intact tropical rain forest:theoreticaland conservationimplications. Ecol. Lett. 4:491-99 WestelakenIL,MaunMA. 1985. Spatialpattern and seed dispersalof Lithospermumcaroliniense on LakeHuronsanddunes. Can.J. Bot. 63:125-32 WestobyM, LeishmanM, LordJ. 1996. Comparative ecology of seed size and dispersal. Philos. Trans. R. Soc. London. Ser B 351:1309-18 WeinerJ, ContePT. 1981. Dispersalandneighborhoodeffects in an annualplant competition model. Ecol. Model. 13:131-47 Willson MF. 1993. Dispersalmode, seed shadows, and colonization patterns. Vegetatio 107/108:261-80 Wu J, Levin SA. 1994. A spatialpatchdynamic modelingapproachto patternsandprocessin an annualgrassland.Ecol. Monogr 64:44764 Yeaton RI, Bond WJ. 1991. Competitionbetween two shrub species: dispersal differences andfirepromotecoexistence.Am.Nat. 138:328-41 Yu DW, Wilson HB. 2001. The competitioncolonization trade-offis dead; long live the competition-colonizationtrade-off.Am.Nat. 158:49-63
© Copyright 2026 Paperzz