The role of habitat area and edge in fragmented landscapes - PDF document

conductedseparately,withrelativelyfewattemptstoexplic-itlycomparetheinfluenceofbothlandscapeelements.In-vestigationsonhabitatareahavecenteredoneithertheETIBormetapopulationtheoryanddeterminingtheirap-plicabilityinfragmentedlandscapes(e.g.,Hanski1998;Ricketts2001;Brotonsetal.2003).Researchershavepro-posedseveralmechanismsfortheinfluenceofhabitatarea(MacArthurandWilson1967;Root1973;GilpinandDia-mond1976;AmbuelandTemple1983;Matter1997;Fletcher2006),yetthereismuchoverlapinthedescriptionandinterpretationofthesemechanisms(seeBrotonsetal.2003),withnocurrentframeworkthatdescribesandlinksfundamentalmechanismsforareaeffects.Similarly,alargeliteraturehasprovidedasuiteofpotentialmechanismsunderlyingedgeeffects(Wiensetal.1985;Murcia1995;McCollin1998;Faganetal.1999;Cadenassoetal.2003).Riesetal.(2004)recentlysynthesizedtheproposedmecha-nismsforedgeeffectsintoaunifiedconceptualframework,focussingonhowhabitatedgesinfluencethedistributionandabundanceoforganisms.Below,webrieflyreviewtheirframeworkandapplyasimilarapproachtounderstandhowareainfluencesecologicalpatternsandprocesses,highlight-ingempiricalexamplesthatfindevidenceforsuchmecha-MechanismsofedgeeffectsHabitatedgesfundamentallyinfluenceecologicalproc-essesintwodirectways:(1)byalteringtheflowsofenergy,materials,and(or)organismsand(2)byprovidingaccesstospatiallyseparatedresources(Riesetal.2004).Thesetwodirecteffectscanleadtoindirecteffectsthroughresourcemappingandspeciesinteractions(Fig.2),allofwhichmayleadtochangesinpopulationsandcommunitystructurenearedges(Faganetal.1999;Riesetal.2004).Resourcemappingoccurswhenanorganism’sdistributionreflects(‘‘maps’’onto)thatofitsresources(seeRiesetal.2004).Changesinspeciesinteractionsincludevariationinpredator–preyinteractions(RiesandFagan2003),competition(Suarezetal.1998),orfacilitation(JulesandRathcke1999)nearedges.Thesedirectandindirecteffectscapturemostdynamicsthatoccurnearedges(McCollin1998;Faganetal.1999;Riesetal.2004).Muchoftheobservedvariationindynamicsnearedgescanbeexplainedbytheedgetypeinvestigated,wherethetypeofhabitatsurround-ingfragmentshasthepotentialtoinfluenceeachoftheaforementionedmechanisms(RiesandSisk2004;Riesetal.Theflowofmaterialsandorganismscanbeinfluencedbyedges,wherehabitatboundariescanamplify,attenuate,orreflectflowsintoandoutofadjacentpatches(Strayeretal.2003).Inturn,variationinecologicalflowscanleadtoabundanceand(or)diversitygradientsnearedges,whichcanbeinfluencedbyedgecontrast(Stampsetal.1987),thesurroundingmatrix(Ricketts2001;Chalfounetal.2002andspecies’lifehistories(Henleetal.2004).Ecologicalflowsarepredictedtocausedeclinesinabundancenearedgeswithinpreferredhabitat,butincreasesinabundancewithintheadjacenthabitatcausedby‘‘spillover’’(RiesandSisk2004).Therearenumerousexamplesofedgesinfluenc-ingecologicalflows,includingenergyflowsthatimpactmicroclimate(Chenetal.1995),seeddispersal(Cadenasso InfluenceofdecliningpatchsizeNewcommunityfordifferentpatchsizesRMSIPopulationsizesImmigrationImmigrationamountordiversityofresourcesIncreasedEmigrationIncreasedEmigrationRMSIInfluenceofadjacenthabitatSessileorganismMobileanimalNewcommunityattheedge Fig.2.Conceptualframeworksthatidentifydifferentprocess-basedpathwaysbywhichthedistributionoforganismsareinfluencedbyhabitatedgeandpatchsize,wherethickarrowsdenotedirectef-fectsofedgeandpatchsizeandarrowsdenoteindirecteffects.)Habitatedgesdirectlyinfluencetheecologicalflowormove-mentofmaterials,energy,andorganisms,aswellasdirectlypro-vidingaccesstospatiallyseparatedresourcesformobileorganisms.Theseissuesindirectlyinfluenceresourcemapping(RM)andspe-ciesinteractions(SI).(ModifiedfromRiesetal.2004,reproducedwithpermissonofAnnu.Rev.Ecol.Evol.Syst.,vol.35,p.496,2004AnnualReviews.)()Declinesinpatchsizefromhabitatlossandfragmentationdirectlyresultinareductionoftheamountand(or)diversityofavailableresources.Decliningpatchsizealsoaltersthelikelihoodofmovementintoandoutofpatchesbydis-persingindividuals,withmosttheoryassumingthatdecliningpatchsizeincreasesemigrationanddecreasesimmigrationrates.Emigra-tionandimmigrationcanbedirectlyinfluencedbychangesinpatchgeometrywithinlandscapes(e.g.,the‘‘target’’effect)orin-directlythroughchangesinresourceconcentration.Theseissuesarepredictedtogenerallydecreasepopulationsizeandincreaselocalextinctionriskoforganisms,whichhaveindirecteffectsonre-sourcemappingandspeciesinteractions.Indirecteffectsofre-sourcemappingandspeciesinteractionscanbevariable,however,leadingtopotentiallycomplexeffectsonresultingpopulationandcommunitydynamics.Seetextfordetails.Fletcher,Jr.etal.2007NRCCanada manyimmigrationbehaviorsshouldactuallycontributetoadeclineindensitywithincreasingpatchsize.Thisrelation-shipmayoccurbecauseimmigrationratesperunitareaarelikelytodeclineowingtothenegativeexponentialnatureofpatchcircumferenceperunitareawithincreasingpatchsize(seealsoHambackandEnglund2005).Conversely,individ-ualsmovingwithinsmallerpatchesaremorelikelytoen-counterboundariesthanthosemovinginlargerpatches(Stampsetal.1987),thusconsistentlyincreasingemigrationratesoutofsmallerpatches(Bowmanetal.2002).Forex-ample,Cronin(2003)foundthataplanthopper((VanDuzee,1879))exhibitedhigheremigrationratesinsmallpatchesthaninlargerpatches.Suchvariationinemigrationratesispredictedtocauselowerdensitiesandcommunitydiversityinsmallpatches(Fig.2Bothemigrationandimmigrationcanalsobeindirectlyinfluencedbypatchsizethroughchangesinresourcecon-centration(Fig.2;Root1973),wherehigherconcentrationsofresourcesinlargepatchesarethoughttodecreaseemigra-tionratesandincreaseimmigrationrates.Forinstance,somesocialbehaviors,suchasconspecificattraction(i.e.,theuseofconspecificsaspositiveproximatestimuliforselectinglocationsofhabitat),maycauseanincreaseinimmigrationandadecreaseinemigrationinlargepatches,leadingtopatchsizeeffects,and,albeitweak,edgeeffectsondensity(Fletcher2006).However,suchbehaviorsarebetterex-plainedbypatchsizethanedge,becauseincreasingpatchsizeincreasestheopportunityforthepresenceofsocialcues,whichmaybeusedasindirectmeasuresofresourceSpeciesinteractionscanbeindirectlyinfluencedbyde-cliningpatchsizewhendisproportionatechangesinonespe-cies’densityorbehavioralteritsinteractionswithotherspecies(e.g.,OrrockandFletcher2005).Boththeenemieshypothesis(Root1973;Risch1981)andthedensitycom-pensationhypothesis(MacArthuretal.1972)havebeende-velopedtoexplainhowspeciesinteractionsarealteredwithdecliningpatchsize(Fig.4);however,eachfocussesondif-ferenttypesofinteractionsandmakesdifferentpredictionsforchangesinpopulationdensity.Theenemieshypothesisstatesthatpredatorsaremoreabundantormoreeffectiveinsmallerareasofhabitat(Root1973;Risch1981;Askinsetal.1987),therebyalteringspeciesinteractions.Forexample,predatorsmayforageforrelativelylongerperiodsperunitareainsmallerpatchesbecausethedistancetootherresour-cesisoftengreaterthanthatinlargerpatches(StephensandKrebs1986).Wenote,however,thattheoriginallogicfortheenemieshypothesis(Root1973;Risch1981)isspecifictoinsectpredator–herbivoreinteractionsinmonoculturesandpolyculturesandmaynotapplybroadlyacrosstaxainfragmentedlandscapes.Furthermore,someempiricalevi-dencesuggeststhatpredatordensitycanactuallyincreasewithpatchsize(Chapinetal.1998;Pardini2004).Con-versely,thedensitycompensationhypothesispredictsthatdecliningspeciesrichnesswithreducedarea(MacArthurandWilson1967)willcausedensitiestoincreaseowingtoadecreaseininterspecificcompetition(MacArthuretal.1972).Othertypesofinteractionsmayalsobeaffected,suchthatdensitymayincreaseordeclinebasedonthespe-cificsofeachinteraction.Althoughsomestudiesinfragmentedlandscapeshaveat-temptedtotestforchangesinmovement,resourcemapping,andspeciesinteractionswithpatchsize,surprisinglyfewhaveshownstrongmechanisticsupportforobservedareaef-fects.Manyoftheseinvestigationshaveattemptedtounder-standwhysomebirdspeciesareareasensitive,oravoidsmallpatches.ArecentexamplefromJapan(KurosawaandAskins2003)showedthatmanyforestbirdsoccurredatlowerdensitiesinsmallerpatches,andsuchpatchesalsocontainedhigherdensitiesofpredators,providingsomesup-portfortheenemieshypothesis.Similarly,Herkert(1994)foundthatmanygrasslandbirdsweresensitivetoarea,andthatforsomespecies,suchasHenslow’sSparrow(dramushenslowii(Audubon,1829))andSavannahSparrowPasserculussandwichensis(J.F.Gmelin,1789)),smallfrag-mentscontainedfewerpotentialresourcesperunitarea(basedonhabitatstructure).However,inbothoftheseex-amples,andindeedinmanyinvestigationsonareaeffects(Parkeretal.2005),investigatorsdidnotisolatewhetherob-servedareaeffectswerepatchsizeeffects.Finally,therearetwomainfactorsthatcangenerateareaeffects,suchthatobservedeffectsarenotactualpatchsizeeffects.First,theproportionofedgetendstoincreasewithdecliningpatchsize(Fig.1).Ifaspeciesrespondstoedges,theseresponsescanresultinchangesindensity(orothermetrics)inpatchesofdifferentsizes(Sisketal.1997;Fletcher2005),andcouldbefalselyattributedtoapatchsizeeffect.Therefore,tounderstandtheuniquerolesofareaandedgeinfragmentedlandscapes,weneedtonotonlycontrolforedgeswheninvestigatingeffectsofareabutalsotofocusonthedirectmechanismsresponsibleforedgeandareaeffects.Second,assamplingincreases,sodoesthelike-lihoodofdetectinganindividualoraspecies,resultinginincreasesinthelikelihoodofoccurrenceorspeciesrichnesswithareasampled(ConnorandMcCoy1979;Hornetal.2000).Thisrelationshiphasbeentermedtherandomsam-plinghypothesisandcanexplainsomeobservedareaeffects(Hailaetal.1993;Andren1994;Hornetal.2000).Varia-tioninsamplingeffort,however,shouldnotcausechangesindensityorothermeasuresthatdonotincreasewithsam-pling(e.g.,fecundity).Interplayofareaeffects,edgeeffects,andlandscapeTheseconceptualframeworkshighlightthedistinctivenessofdirecteffectsofpatchsizeandedgeonecologicalpat-ternsandprocesses,yetbothedgeandpatchsizecanhaveindirecteffectsonspeciesinteractionsandresourcemap-ping,suchthatobservedeffectscanappearsimilarinnature.Forinstance,ifspeciesinteractionschangeasafunctionofpatchsize(Fig.2),theseinteractionswillalsoinevitablyvarywithdistancetoedgeacrosspatches(Fig.1).Conse-quently,isolatingwhetherspeciesinteractionsvaryfromedgeorpatchsizewilloftenbedifficult.Ontheotherhand,thedirecteffectsofedgeandpatchsizearedistinct(Figs.2),suchthatbytestingforthesepotentialeffects,wecandrawstrongerinferenceregardingtheuniquerolesofareaandedge.Edgeandareaeffectsalsohavethepotentialtointeract.Forexample,KiviniemiandEriksson(2002)foundthatplantspeciesrichnessingrasslandsofSwedenincreasednearedgesinsmallfragmentsbutdeclinedinlargefrag-Fletcher,Jr.etal.#2007NRCCanada beinferredasapatchsizeeffect.Parkeretal.(2005)re-centlynotedthatmost,ifnotall,studiesonforestbirddis-tributioninrelationtoareaconfoundedpotentialareaeffectswithedgeeffectsbecauselargepatcheshadsamplesawayfromedges,whereassmallpatchesinevitablydidnot.Incontrast,similarstudiesfocussingonedgeeffectstendedtoavoidsuchissuesbyselectingrelativelylargepatchesforcomparisons(Parkeretal.2005).Here,wedistinguishstud-iesthatcontrolfortheseissuesstatistically,viacovariates(e.g.,Dunfordetal.2002)orresidualanalysis(e.g.,Winteretal.2000),thosethatcontrolviastudydesign(e.g.,WilderandMeikle2005),andthosethatmakenoattemptstocon-trolfortheconfoundingnatureofareaandedge(e.g.,Hick-ersonetal.2005).Studydesignsthatweacceptedascontrollingforedgeeffectsintestsofareaincludeddesignswheresamplesweretakenatthesamedistancesfromthenearestedgeinpatchesofdifferentsizes.Whilethatap-proachdoesnotcontrolformultipleedges(Fig.1;Fletcher2005),inpracticetherearefewwaystodosowhentestingforareaeffects.Studydesignsthatweacceptedascontrol-lingforareawhentestingforedgeeffectsincludedtestsforedgewithinagivenpatchsizeclass.Basedontheseclassifi-cations,wethencomparedconclusionsfromnonconfoundedanalysestothatofconfoundedanalyses.Wereporttwomainresultsforeachresponsevariablemeasuredinthe38studies.Thefirstiswhethersignificanteffectswereobservedforedge,area,both,orneither.Whenobservedforboth,wereportwhicheffectwasstrongerbasedoninferentialstatisticsprovidedinthearticle(e.g.,statistics,confidenceintervals),whichisreason-ablebecauseweonlyconsideredstudieswithidenticalsam-plesizesforanalysesofedgeandareaeffectsforagivenresponsevariable.Thesecondmainresultiswhetheredgeandarearesponseswereconcordantordiscordant.Wede-finedconcordantresponsesasthosewhereedgeandareaef-fectswereobservedinsimilardirections.Forexample,ifthedensityofanorganismdeclinesnearanedge(anegativeedgeeffect;Riesetal.2004)anddeclinesindensitywithdecreasingpatchsize(apositiveareaeffect;Benderetal.1998),thensuchresponseswouldbeconcordantbecausehabitatfragmentationgenerallyincreasestheproportionofedgeanddecreasespatchsize.Aconcordantresponsesug-geststhathabitatfragmentationhasconsistenteffectsonedgeandarearesponsesofagivenvariableandsuggestsfourpossiblescenarios:(1)areaandedgeeffectsaredrivenbysimilarindirectprocesses(e.g.,speciesinteractions;Figs.2),(2)areaeffectsmaybeexplainedsimplybyedgeeffects(particularlyinconfoundedstudies),(3)edgeeffectsmaybeexplainedsimplybypatchsizeeffects(par-ticularlyinconfoundedstudies),or(4)ourconceptualframeworksareincompleteandedgeandarearesponsesare,infact,drivenbysomesharedbutunknownecologicalmechanism.Weonlyconsideredresponsesdiscordantifop-posingpatternswereobservedforareaandedgeeffects.Discordantresultssuggestthat()edgeeffectsdonotdriveobservedareaeffectsandthat()observededgeandareaeffectsaregeneratedbydistinctecologicalprocesses.Wealsonotethattheresponsescouldbeconcordantordiscord-antasaresultofchancealone.Mixedresponses,whereoneeffectwasobservedbuttheotherwasnot,aredifficulttointerpretbecauselowsamplesizesmaylimitdocumentingareaand(or)edgeeffectsinsomestudies(e.g.,MartMorales2005)andareaeffectsarelikelytobehardertode-tectasaresultoflargersite-to-sitevariation(seebelow).Therefore,wefocusonconcordantanddiscordantresponses.Westratifiedtheresultsbasedontaxonomicgroup,thetypeofresponsevariableinvestigated,focalhabitats,andmechanismstested.Wesummarizedresponsevariablesintothreegeneralcategoriesofmetrics:community(speciesrichness,diversity,orspeciescomposition),distribution(oc-currence,abundance,ordensity),andperformance(e.g.,clutchsize,survival).Weexaminedcoarseeffectsofhabitattypesthathadadequatesamplesizes,includinggrasslandandforest.Fewstudiesprovidedexplicitinformationonlandscapecontext,sowedidnotattempttoaddressthisissue.Finally,wenotedwheninvestigatorsexplicitlyad-dressedpotentialmechanismsforedgeandpatchsizeef-fectsandwhethersuchmechanismsweresupported.Thereviewdatasetisavailableuponrequestfromtheseniorau-ReviewresultsanddiscussionManystudiestestingforedgeandareaeffectswerecon-foundedintheirdesignand(or)analysis,withsimilarcon-foundingfortestsofarea(42%)andedge(39%;Fig.3).Only16ofthe38articles(42%)controlledforbothareaandedgeinatleastsometestsforagivenresponsevariable(nonconfoundedanalyseshereinafter).Manystudiesthatad-dressededgeandareaeffectshad‘‘interior’’plotsthatin-creaseindistancefromedgeinlargerpatches(Hickersonetal.2005;WeaklandandWood2005),whichconfoundspatchsizeandedgeintestsofboth.Investigationsthatdidattempttocontrolforsuchissuespredominantlydidsowithstatisticaltestsratherthanstudydesign(Fig.3).Thestudieswereviewedcontrolledforpotentialconfoundingmore StudydesignStatisticalNotcontrolledUnknownAreatestscontrollingforedge?Edgetestscontrollingforarea?Percentageofinvestigations Fig.3.Summaryofinvestigationsreviewed(=38)thatdidanddidnotattempttocontrolfortheconfoundingnatureofedgeandarea.Wedistinguishedstudiesthatcontrolfortheseissuesstatisti-cally,viacovariatesorresidualanalysis,thosethatcontrolviastudydesign,andthosethatmakenoattemptstocontrolfortheconfoundingnatureofhabitatareaandedge.Onlystudieswherethesameresponsevariablewastestedforbothfactorswerein-cluded.Seetextfordetails.Fletcher,Jr.etal.2007NRCCanada oppositepatternswithrespecttoareaandedge.Thediscord-antresponseswererestrictedtoexperimentsoninvertebratesinforestedhabitat(DaviesandMargules1998;Didhametal.1998).However,theseresultscamefromonlytwoex-periments,thuslimitingourabilitytoassessthegeneralityofthepattern.Concordancewasobservedfor57.8%ofre-sponsevariablesinnonconfoundedanalyses(59.6%forcon-foundedstudies;Fig.6).Giventhatdiscordancewasrare(Fig.6),thatedgeeffectstendedtobestrongerwhenbotheffectswereobserved(Figs.4,5),andthatnonconfoundedanalysesreportedareaeffectslessoftenandedgeeffectsmoreoftenthanconfoundedanalyses(Fig.4),weconcludethatareaeffectsreportedbyconfoundedstudiesmay,infact,oftenbetheresultofedgeeffects.Relativelyfewstudiesexplicitlytestedformechanismsgeneratingedgeorareaeffects(21.1%),withevenfewerex-plicitlytestingformechanismsforbotheffectsforatleastoneresponsevariable(15.8%ofstudiesreviewed).Thiswaslikelydue,inpart,tothedifficultyoftestingformech-anismsofbotheffectsinagiveninvestigation.Articlesthatdidaddressmechanismsfocussedonmeasuresdirectlyorindirectlyrelatedtoresourcemapping(WilderandMeikle2005)and(or)changesinspeciesinteractions(Storch1991;Suarezetal.1998;LienertandFischer2003),bothofwhichcanoccurfromeitheredgeorpatchsizeeffects(orboth),therebylimitinginferenceaboutwhetheredgeorpatchsizegovernedobservedresponses.Noneofthesearticlesad-dressedthedirectprocessesinfluencedbyareaandedgetointerprettheiruniquecontributions.Wenotethatinvestiga-tionsfocussingentirelyonedgeorareahaveaddressedmechanismsresponsibleforpotentialeffectsmoreoften(AmbuelandTemple1983;Askinsetal.1987;McGeochandGaston2000;Cronin2003;FletcherandKoford2003;RiesandFagan2003).Becauseoftheconfoundingnatureoftheseeffects,anyattempttounderstandthemechanismsofedgeorpatchsizeshoulddosobyexplicitlysearchingforthedirecteffectsofeachlandscapeelement(Fig.2).Futureresearch:avoidingthepitfallsofconfoundingfactorsTheresultsofourreview,aswellasotherstudies,suggestthatedgeeffectscansometimesexplainareaeffects(Sisketal.1997;Lauranceetal.1998;SchtickzelleandBaguette2003;Fletcher2005),yettheconverseislesslikelybasedonthefundamentalmechanismsinfluencedbypatchsize.Forexample,inarigorousexperimentonfragmentation,Bowersetal.(1996)suggestedthatspaceuseandreproduc-tivebehaviorsoffemalemeadowvoles(Microtuspennsyl-(Ord,1815))nearedgeswereresponsiblefortheobservednegativeareaeffects(seealsoWilderandMeikle2005).AmbuelandTemple(1983)similarlyinvokedinter- (a)NonconfoundedanalysesCommunityDistributionPerformanceGrasslandForest(b)Confoundedanalyses(10)(118)(20)(33)(94)(9)(27)(29)(11)(30)(4)(23)(4)(9)(22)(1)(8)(6)(3)(9)PercentageofPercentageofCommunityDistributionPerformanceGrasslandForestCommunityDistributionPerformanceGrasslandForestCommunityDistributionPerformanceGrasslandForest Fig.5.Resultsfromstudiesthatexplicitlytestedforbothedgeandareaeffectsonecologicalpatterns,summarizedbymetricused(distri-bution,community,performance)andhabitattype(forestandgrassland)for()nonconfoundedanalysesand()confoundedanalyses.Notethatresultsforthemetricusedandhabitattypeseparatelysummarizethesameresponsevariables.FormatsameasFig.4.Fletcher,Jr.etal.2007NRCCanada patchsizecanhelptoisolatetheeffectsoftheselandscapeelements.Forinstance,edgeeffectsfromecologicalflowspredictthatorganismsmaychangemovementbehaviorsnearhabitatboundaries,whereaspatchsizeshouldonlyin-fluencethelikelihoodofencounteringanedge.Likewise,patchsizeeffectsarisingfrom‘‘target’’effectspredictpercapitaimmigrationandemigrationratestovarywithpatchsizebutnotwithdistancefromedge.Whiletherehavebeenrecentattemptstoproviderigorousapproachestoestimatingmovementincomplexlandscapes(SchtickzelleandBagu-ette2003;HambackandEnglund2005;Martinetal.2006),noneoftheinvestigationswereviewedfocussedonmove-mentbehaviors(e.g.,encounterandturningrates)todisen-tangletheinfluenceofbothpatchsizeandedge.Habitatareaandedgearekeycomponentsoflandscapestructure,yetthesecomponentstendtocovarywithinland-scapes.Becausehabitatareaandedgedescribedifferentas-pectsoflandscapestructure,theimportanceofeachsuggestsdifferentfociforconservationstrategies,makingitimpera-tivetounderstandtheiruniquerolesinhabitatfragmenta-tion.Unfortunately,todatemuchoftheliteraturetestingfortheinfluenceofedgeandareahasbeenconfounded.In-formationfromnonconfoundedstudiessuggeststhatedgeeffectsmayoccurmorefrequentlyinnaturethanareaef-fects.However,edgeresponsesarelikelyeasiertodetectandoursummariesmadenoattempttomeasureoverallef-fectsizes,suchthatareaeffectsmay,insomecases,bemoreimportantthantheseresultssuggest.Inaddition,weemphasizethatvariationinhabitatareaatlandscapescalesviahabitatlossiswidelyknowntobeofconsiderableim-portancetopopulationandcommunitydynamics(e.g.,Fah-rig1997;FlatherandBevers2002).Futureattemptstoisolateedgeandpatchsizeeffectsneedtopaycarefulatten-tiontostudydesignandfocusonthedirect,fundamentalprocessesinfluencedbyareaandedge.R.J.F.wassupportedbygrantsfromtheUSDepartmentofAgriculture–NationalResearchInitiativesandPPLMontanawhilethismanuscriptwaswritten.L.R.wassup-portedbyaNationalScienceFoundationfellowship.J.B.wassupportedbyaNationalResearchCouncilpostdoctoralfellowship.Wethanktwoanonymousreviewersforprovid-ingthoughtful,constructivecommentsonapreviousdraftofthemanuscript.Alencar,A.A.C.,Solorzano,L.A.,andNepstad,D.C.2004.Model-ingforestunderstoryfiresinaneasternAmazonianlandscape.Ecol.Appl.:S139–S149.doi:10.1890/01-6029.Ambuel,B.,andTemple,S.A.1983.AreadependentchangesinthebirdcommunitiesandvegetationofsouthernWisconsinUSAforests.Ecology,:1057–1068.doi:10.2307/1937814.Andren,H.1994.Effectsofhabitatfragmentationonbirdsandmammalsinlandscapeswithdifferentproportionsofsuitablehabitat:areview.Oikos,:355–366.doi:10.2307/3545823.Askins,R.A.,Philbrick,M.J.,andSugeno,D.S.1987.Relationshipbetweentheregionalabundanceofforestandthecompositionofforestbirdcommunities.Biol.Conserv.:129–152.doi:10.1016/0006-3207(87)90030-9.Beier,P.,VanDrielen,M.,andKankam,B.O.2002.Avifaunalcol-lapseinWestAfricanforestfragments.Conserv.Biol.1097–1111.doi:10.1046/j.1523-1739.2002.01003.x.Belisle,M.,Desrochers,A.,andFortin,M.J.2001.Influenceoffor-estcoveronthemovementsofforestbirds:ahomingexperi-ment.Ecology,:1893–1904.Bender,D.J.,andFahrig,L.2005.Matrixstructureobscuresthere-lationshipbetweeninterpatchmovementandpatchsizeandiso-lation.Ecology,:1023–1033.doi:10.1890/03-0769.Bender,D.J.,Contreras,T.A.,andFahrig,L.1998.Habitatlossandpopulationdecline:ameta-analysisofthepatchsizeeffect.:517–533.Bowers,M.A.,andMatter,S.F.1997.Landscapeecologyofmam-mals:relationshipsbetweendensityandpatchsize.J.Mammal.:999–1013.doi:10.2307/1383044.Bowers,M.A.,Gregario,K.,Brame,C.J.,Matter,S.F.,andDooley,J.L.1996.Useofspaceandhabitatsbymeadowvolesatthehomerange,patchandlandscapescales.Oecologia(Berl.),107–115.doi:10.1007/BF00328798.Bowman,J.,Cappuccino,N.,andFahrig,L.2002.Patchsizeandpopulationdensity:theeffectofimmigrationbehavior.Conserv.Ecol.[online]:9.Availablefromwww.ecologyandsociety.org/vol6/iss1/art9/[accessed30March2006].Brand,L.A.,Noon,B.R.,andSisk,T.D.2006.Predictingabun-danceofdesertriparianbirds:validationandcalibrationoftheeffectiveareamodel.Ecol.Appl.:1090–1102.doi:10.1890/1051-0761(2006)016[1090:PAODRB]2.0.CO;2.Brotons,L.,Monkkonen,M.,andMartin,J.L.2003.Arefragmentsislands?Landscapecontextanddensity–arearelationshipsinborealforestbirds.Am.Nat.:343–357.doi:10.1086/376887.PMID:12970842.Cadenasso,M.L.,andPickett,S.T.A.2001.Effectofedgestructureonthefluxofspeciesintoforestinteriors.Conserv.Biol.91–97.doi:10.1046/j.1523-1739.2001.99309.x.Cadenasso,M.L.,Pickett,S.T.A.,Weathers,K.C.,andJones,C.G.2003.Aframeworkforatheoryofecologicalboundaries.:750–758.doi:10.1641/0006-3568(2003)053[0750:AFFATO]2.0.CO;2.Chalfoun,A.D.,Thompson,F.R.,III,andRatnaswamy,M.J..Nestpredatorsandfragmentation:areviewandmeta-analysis.Conserv.Biol.:306–318.doi:10.1046/j.1523-1739.Chalfoun,A.D.,Ratnaswamy,M.J.,andThompson,F.R.,III..Songbirdnestpredatorsinforest–pastureedgeandfor-estinteriorinafragmentedlandscape.Ecol.Appl.:858–Chapin,T.G.,Harrison,D.J.,andKatnik,D.D.1998.InfluenceoflandscapepatternonhabitatusebyAmericanmarteninanin-dustrialforest.Conserv.Biol.:1327–1337.doi:10.1046/j.Chen,J.Q.,Franklin,J.F.,andSpies,T.A.1995.Growingseasonmicroclimaticgradientsfromclear-cutedgesintoold-growthdouglas-firforests.Ecol.Appl.:74–86.doi:10.2307/1942053.Collinge,S.K.,andPalmer,T.M.2002.TheinfluencesofpatchshapeandboundarycontrastoninsectresponsetofragmentationinCaliforniagrasslands.Landsc.Ecol.:647–656.doi:10.1023/A:1021536302195.Connor,E.F.,andMcCoy,E.D.1979.Thestatisticsandbiologyofthespecies–arearelationship.Am.Nat.:791–834.doi:10.Connor,E.F.,Courtney,A.C.,andYoder,J.M.2000.Individuals–Fletcher,Jr.etal.#2007NRCCanada Conserv.Biol.:784–793.doi:10.1046/j.1523-1739.1999.Kareiva,P.1990.Populationdynamicsinspatiallycomplexenvir-onments:theoryanddata.Philos.Trans.R.Soc.Lond.BBiol.:175–190.doi:10.1098/rstb.1990.0191.Kiviniemi,K.,andEriksson,O.2002.Size-relateddeteriorationofsemi-naturalgrasslandfragmentsinSweden.Divers.Distrib.21–29.doi:10.1046/j.1366-9516.2001.00125.x.Kurosawa,R.,andAskins,R.A.2003.Effectsofhabitatfragmenta-tiononbirdsindeciduousforestsinJapan.Conserv.Biol.695–707.doi:10.1046/j.1523-1739.2003.01118.x.Lahti,D.C.2001.The‘‘edgeeffectonnestpredation’’hypothesisaftertwentyyears.Biol.Conserv.:365–374.doi:10.1016/S0006-3207(00)00222-6.Laurance,W.F.,Ferreira,L.V.,Rankin-DeMerona,J.M.,andLaur-ance,S.G.1998.RainforestfragmentationandthedynamicsofAmazoniantreecommunities.Ecology,:2032–2040.Laurance,W.F.,Lovejoy,T.E.,Vasconcelos,H.L.,Bruna,E.M.,Didham,R.K.,Stouffer,P.C.,Gascon,C.,Bierregaard,R.O.,Laurance,S.G.,andSampaio,E.2002.EcosystemdecayofAmazonianforestfragments:a22-yearinvestigation.Con-serv.Biol.:605–618.doi:10.1046/j.1523-1739.2002.01025.Levey,D.J.,Bolker,B.M.,Tewksbury,J.J.,Sargent,S.,andHad-dad,N.M.2005.Effectsoflandscapecorridorsonseeddispersalbybirds.Science(Washington,D.C.),:146–148.doi:10.1126/science.1111479.PMID:15994561.Levins,R.1969.Somedemographicandgeneticconsequencesofenvironmentalheterogeneityforbiologicalcontrol.Bull.Ento-mol.Soc.Am.:237–240.Lienert,J.,andFischer,M.2003.HabitatfragmentationaffectsthecommonwetlandspecialistPrimulafarinosainnorth-eastSwit-zerland.J.Ecol.:587–599.doi:10.1046/j.1365-2745.2003.Lomolino,M.V.1990.Thetargetareahypothesis:theinfluenceofislandareaonimmigrationratesofnon-volantmammals.Oikos,:297–300.doi:10.2307/3565957.pez-Barrera,F.,Manson,R.H.,Gonzalez-Espinosa,M.,andNewton,A.C.2006.Effectsofthetypeofmontaneforestedgeonoakseedlingestablishmentalongforest–edge–exteriorgradi-ents.For.Ecol.Manage.:234–244.doi:10.1016/j.foreco.MacArthur,R.H.,andWilson,E.O.1967.Thetheoryofislandbio-geography.PrincetonUniversityPress,Princeton,N.J.MacArthur,R.H.,Diamond,J.M.,andKarr,J.R.1972.Densitycompensationinislandfaunas.Ecology,:330–342.doi:10.Malcolm,J.R.1994.EdgeeffectsofcentralAmazonianforestfrag-ments.Ecology,:2438–2445.doi:10.2307/1940897.Mancke,R.G.,andGavin,T.A.2000.Breedingbirddensityinwoo-dlots:effectsofdepthandbuildingsattheedges.Ecol.Appl.598–611.doi:10.1890/1051-0761(2000)010[0598:BBDIWE]2.0.Martin,J.,Nichols,J.D.,Kitchens,W.M.,andHines,J.E.2006.MultiscalepatternsofmovementinfragmentedlandscapesandconsequencesondemographyofthesnailkiteinFlorida.J.Anim.Ecol.:527–539.doi:10.1111/j.1365-2656.2006.01073.x.PMID:16638005.Martin,J.L.1956.Thebionomicsoftheaspenblotchminer,colletissalicifoliaCham.(Lepidoptera:Gracillariidae).Can.En-:155–170.nez-Morales,M.A.2005.Landscapepatternsinfluencingbirdassemblagesinafragmentedneotropicalcloudforest.Biol.Con-:117–126.doi:10.1016/j.biocon.2004.04.015.Matter,S.F.1997.Populationdensityandarea:theroleofbetween-andwithin-patchprocesses.Oecologia(Berl.),533–538.doi:10.1007/s004420050191.Matthews,A.,Dickman,C.R.,andMajor,R.E.1999.Theinfluenceoffragmentsizeandedgeonnestpredationinurbanbushland.:349–356.doi:10.1111/j.1600-0587.1999.tb00572.Matthysen,E.2005.Density-dependentdispersalinbirdsandmammals.Ecography,:403–416.doi:10.1111/j.0906-7590.McCollin,D.1998.Forestedgesandhabitatselectioninbirds:afunctionalapproach.Ecography,:247–260.doi:10.1111/j.1600-0587.1998.tb00562.x.McGarigal,K.J.,andMarks,B.J.1995.FRAGSTATS:spatialpat-ternanalysisprogramforquantifyinglandscapestructure.U.S.For.Serv.Gen.Tech.Rep.PNW-GTR-351.McGeoch,M.A.,andGaston,K.J.2000.Edgeeffectsonthepreva-lenceandmortalityfactorsofPhytomyzailicis(Diptera,Agro-myzidae)inasuburbanwoodland.Ecol.Lett.:23–29.doi:10.1046/j.1461-0248.2000.00114.x.Murcia,C.1995.Edgeeffectsinfragmentedforests:implicationsforconservation.TrendsEcol.Evol.:58–62.doi:10.1016/S0169-5347(00)88977-6.Norton,M.R.,Hannon,S.J.,andSchmiegelow,F.K.A.2000.Frag-mentsarenotislands:patchvs.landscapeperspectivesonsong-birdpresenceandabundanceinaharvestedborealforest.:209–223.doi:10.1034/j.1600-0587.2000.230207.Nour,N.,Matthysen,E.,andDhondt,A.A.1993.Artificialnestpredationandhabitatfragmentation:differenttrendsinbirdandmammalpredators.Ecography,:111–116.doi:10.1111/j.1600-0587.1993.tb00063.x.Orrock,J.L.,andFletcher,R.J.,Jr.2005.Changesincommunitysizeaffecttheoutcomeofcompetition.Am.Nat.:107–111.doi:10.1086/430641.PMID:15937793.Pardini,R.2004.EffectsofforestfragmentationonsmallmammalsinanAtlanticForestlandscape.Biodivers.Conserv.:2567–2586.doi:10.1023/B:BIOC.0000048452.18878.2d.Parker,T.H.,Stansberry,B.M.,Becker,C.D.,andGipson,P.S.2005.Edgeandareaeffectsontheoccurrenceofmigrantforestsongbirds.Conserv.Biol.:1157–1167.doi:10.1111/j.1523-Paton,P.W.C.1994.Theeffectofedgeonaviannestsuccess:howstrongistheevidence?Conserv.Biol.:17–26.doi:10.1046/j.1523-1739.1994.08010017.x.Pulliam,H.R.1988.Sources,sinksandpopulationregulation.Am.:652–661.doi:10.1086/284880.Rand,T.A.,Tylianakis,J.M.,andTscharntke,T.2006.Spilloveredgeeffects:thedispersalofagriculturallysubsidizedinsectnat-uralenemiesintoadjacentnaturalhabitats.Ecol.Lett.:603–614.doi:10.1111/j.1461-0248.2006.00911.x.PMID:16643305.Ricketts,T.H.2001.Thematrixmatters:effectiveisolationinfrag-mentedlandscapes.Am.Nat.:87–99.doi:10.1086/320863.Ries,L.,andDebinski,D.M.2001.ButterflyresponsestohabitatedgesinthehighlyfragmentedprairiesofcentralIowa.J.Anim.Ecol.:840–852.doi:10.1046/j.0021-8790.2001.00546.Ries,L.,andFagan,W.F.2003.Habitatedgesasapotentialecolo-gicaltrapforaninsectpredator.Ecol.Entomol.:567–572.Ries,L.,andSisk,T.D.2004.Apredictivemodelofedgeeffects.:2917–2926.doi:10.1890/03-8021.Ries,L.,Fletcher,R.J.,Jr.,Battin,J.,andSisk,T.D.2004.Ecologi-calresponsestohabitatedges:mechanisms,models,andvaria-Fletcher,Jr.etal.#2007NRCCanada REVIEW/SYNTHETheroleofhabitatareaandedgeinfragmentedlandscapes:definitivelydistinctorinevitablyRobertJ.Fletcher,Jr.,LeslieRies,JamesBattin,andAnnaD.ChalfounOverthepastfewdecades,muchresearchhasfocussedontheeffectsofhabitatarea(i.e.,patchsize)andedgesinfragmentedlandscapes.Wereviewandsynthesizetheliteratureonareaandedgeeffectstoidentifywhethertheeco-logicalprocessesinfluencedbypatchsizeandedgearedistinct,tosummarizeevidencefortherelativeeffectofeach,andtodiscusshowestimatingtheirindependenteffectsmaybeaccomplishedinfieldstudies.Areaandedgedirectlyinfluenceecologicalprocessesindistinctways,yetindirecteffectscanbesimilar,makingitdifficulttoisolatetheeffectsofareaandedgeinnature.Manystudiesinvestigatingbothareaandedgehavebeenconfoundedintheirdesignand(or)analysis(i.e.,studiesdidnotcontrolforonepotentialeffectwhiletestingfortheother).Nonconfoundedstudieshavemorefre-quentlyshownsupportforedgeeffects,andcomparisonsbetweennonconfoundedandconfoundedstudiessuggestthatsomeobservedareaeffectscouldbeexplainedbyedgeeffects.Wearguethatbyfocussingonthefundamentalprocessesdirectlyinfluencedbyareaandedge,andbydevelopingmorerigorousstudydesignsandanalysesthatisolatetheirrelativeinfluence,greaterinsightcanbegainedinfutureinvestigationsonhabitatlossandfragmentation.Aucoursdesquelquesdernieresdecennies,beaucoupdetravauxsesontintesauxeffetsdelasurfacedel’habitat(c’est-a-direlatailledestaches)etdeslisieresdanslespaysagesfragmentes.Nouspassonsenrevueetrelalitteraturesurleseffetsdelasurfaceetdeslisieresafindedeterminersilesprocessusecologiquesinfluencesparlasur-faceetleslisieressontdifferents,deresumerl’informationsurleseffetsrelatifsdechacunedecesvariablesetd’examinercommentestimerleurseffetsrespectifsdansdesetudesdeterrain.Lasurfaceetleslisieresaffectentdirectementlesproc-essusecologiquesdefac¸ondistincte,maisleurseffetsindirectspeuventetresemblables,cequirendladistinctionentreleseffetsdelasurfaceetceuxdeslisieresdifficileennature.Plusieursetudesquiconsiderentlasurfaceetleslisierescon-fondentlesdeuxvariablesacausedeleursplansd’experienceet(ou)d’analyse(c’est-a-direellesnetiennentpascompted’uneffetpotentielentestantl’autre).Plusfrequemment,lesetudesquineconfondentpaslesdeuxvariablesontdevertdesindicationsdel’existenced’uneffetdeslisieres;deplus,unecomparaisond’etudesavecousansvariablesconfon-duesindiquequ’unepartiedeseffetsattribuelasurfacepeuts’expliquerparleseffetsdeslisieres.Nouscroyonsqu’enseconcentrantsurlesprocessusfondamentauxinfluencesdirectementparlasurfaceetleslisieresetenmettantaupointdesplansd’experienceetd’analyseplusrigoureuxpourisolerleseffetsrelatifsdechacunedesdeuxvariables,ilserapos-sibled’obtenirdemeilleuresperspectivesdanslesrecherchesfuturessurlaperteetlafragmentationdeshabitats.[TraduitparlaReLandscapesarecomposedofapatchworkofhabitatsthatvaryinsize,shape,andlocation.Ongoinghabitatlosscon-tinuestoinfluencelandscapestructure,withhabitatlossgenerallyleadingtoincreasedfragmentationofhabitats(orthebreakingapartofhabitat,independentofloss;Wiens1995;Fahrig2003),wherebypatchesdeclineinsize,in-creaseinisolation,andincreaseintheproportionofedge Received8September2006.Accepted31August2007.PublishedontheNRCResearchPressWebsiteatcjz.nrc.caon7NovemberR.J.Fletcher,Jr.AvianScienceCenter,DivisionofBiologicalSciences,UniversityofMontana,Missoula,MT59812,USA.L.Ries.DepartmentofBiology,UniversityofMaryland,CollegePark,MD20742,USA.J.Battin.NationalOceanicandAtmosphericAdministration,NorthwestFisheriesScienceCenter,2725MontlakeBoulevardEast,Seattle,WA98112,USA.A.D.Chalfoun.MontanaCooperativeWildlifeResearchUnit,UniversityofMontana,Missoula,MT59812,USA.Thisreviewisoneofaseriesdealingwithsomeaspectsoftheimpactofhabitatfragmentationonanimalsandplants.ThisseriesisoneofseveralvirtualsymposiafocussingonecologicaltopicsthatwillbepublishedintheJournalfromtimetotime.Correspondingauthor(e-mail:robert.fletcher@ufl.edu).Presentaddress:DepartmentofWildlifeEcologyandConservation,UniversityofFlorida,P.O.Box110430,Gainesville,FL32611-0430,USA.Can.J.Zool.:1017–1030(2007)doi:10.1139/Z07-100#2007NRCCanada REVIEW/SYNTHETheroleofhabitatareaandedgeinfragmentedlandscapes:definitivelydistinctorinevitablyRobertJ.Fletcher,Jr.,LeslieRies,JamesBattin,andAnnaD.ChalfounOverthepastfewdecades,muchresearchhasfocussedontheeffectsofhabitatarea(i.e.,patchsize)andedgesinfragmentedlandscapes.Wereviewandsynthesizetheliteratureonareaandedgeeffectstoidentifywhethertheeco-logicalprocessesinfluencedbypatchsizeandedgearedistinct,tosummarizeevidencefortherelativeeffectofeach,andtodiscusshowestimatingtheirindependenteffectsmaybeaccomplishedinfieldstudies.Areaandedgedirectlyinfluenceecologicalprocessesindistinctways,yetindirecteffectscanbesimilar,makingitdifficulttoisolatetheeffectsofareaandedgeinnature.Manystudiesinvestigatingbothareaandedgehavebeenconfoundedintheirdesignand(or)analysis(i.e.,studiesdidnotcontrolforonepotentialeffectwhiletestingfortheother).Nonconfoundedstudieshavemorefre-quentlyshownsupportforedgeeffects,andcomparisonsbetweennonconfoundedandconfoundedstudiessuggestthatsomeobservedareaeffectscouldbeexplainedbyedgeeffects.Wearguethatbyfocussingonthefundamentalprocessesdirectlyinfluencedbyareaandedge,andbydevelopingmorerigorousstudydesignsandanalysesthatisolatetheirrelativeinfluence,greaterinsightcanbegainedinfutureinvestigationsonhabitatlossandfragmentation.Aucoursdesquelquesdernieresdecennies,beaucoupdetravauxsesontintesauxeffetsdelasurfacedel’habitat(c’est-a-direlatailledestaches)etdeslisieresdanslespaysagesfragmentes.Nouspassonsenrevueetrelalitteraturesurleseffetsdelasurfaceetdeslisieresafindedeterminersilesprocessusecologiquesinfluencesparlasur-faceetleslisieressontdifferents,deresumerl’informationsurleseffetsrelatifsdechacunedecesvariablesetd’examinercommentestimerleurseffetsrespectifsdansdesetudesdeterrain.Lasurfaceetleslisieresaffectentdirectementlesproc-essusecologiquesdefac¸ondistincte,maisleurseffetsindirectspeuventetresemblables,cequirendladistinctionentreleseffetsdelasurfaceetceuxdeslisieresdifficileennature.Plusieursetudesquiconsiderentlasurfaceetleslisierescon-fondentlesdeuxvariablesacausedeleursplansd’experienceet(ou)d’analyse(c’est-a-direellesnetiennentpascompted’uneffetpotentielentestantl’autre).Plusfrequemment,lesetudesquineconfondentpaslesdeuxvariablesontdevertdesindicationsdel’existenced’uneffetdeslisieres;deplus,unecomparaisond’etudesavecousansvariablesconfon-duesindiquequ’unepartiedeseffetsattribuelasurfacepeuts’expliquerparleseffetsdeslisieres.Nouscroyonsqu’enseconcentrantsurlesprocessusfondamentauxinfluencesdirectementparlasurfaceetleslisieresetenmettantaupointdesplansd’experienceetd’analyseplusrigoureuxpourisolerleseffetsrelatifsdechacunedesdeuxvariables,ilserapos-sibled’obtenirdemeilleuresperspectivesdanslesrecherchesfuturessurlaperteetlafragmentationdeshabitats.[TraduitparlaReLandscapesarecomposedofapatchworkofhabitatsthatvaryinsize,shape,andlocation.Ongoinghabitatlosscon-tinuestoinfluencelandscapestructure,withhabitatlossgenerallyleadingtoincreasedfragmentationofhabitats(orthebreakingapartofhabitat,independentofloss;Wiens1995;Fahrig2003),wherebypatchesdeclineinsize,in-creaseinisolation,andincreaseintheproportionofedge Received8September2006.Accepted31August2007.PublishedontheNRCResearchPressWebsiteatcjz.nrc.caon7NovemberR.J.Fletcher,Jr.AvianScienceCenter,DivisionofBiologicalSciences,UniversityofMontana,Missoula,MT59812,USA.L.Ries.DepartmentofBiology,UniversityofMaryland,CollegePark,MD20742,USA.J.Battin.NationalOceanicandAtmosphericAdministration,NorthwestFisheriesScienceCenter,2725MontlakeBoulevardEast,Seattle,WA98112,USA.A.D.Chalfoun.MontanaCooperativeWildlifeResearchUnit,UniversityofMontana,Missoula,MT59812,USA.Thisreviewisoneofaseriesdealingwithsomeaspectsoftheimpactofhabitatfragmentationonanimalsandplants.ThisseriesisoneofseveralvirtualsymposiafocussingonecologicaltopicsthatwillbepublishedintheJournalfromtimetotime.Correspondingauthor(e-mail:robert.fletcher@ufl.edu).Presentaddress:DepartmentofWildlifeEcologyandConservation,UniversityofFlorida,P.O.Box110430,Gainesville,FL32611-0430,USA.Can.J.Zool.:1017–1030(2007)doi:10.1139/Z07-100#2007NRCCanada habitat.Becausehabitatlossisoftenimplicatedasthemostimportantthreattobiodiversity(e.g.,Wilcoveetal.1998),understandinghowhabitatlossandfragmentationinfluenceecologicalpatternsandprocesseshasbeen(GatesandGysel1978;AmbuelandTemple1983),andcontinuestobe(TscharntkeandBrandl2004;EwersandDidham2006),acentralfocusoflandscapeecology,populationandcommun-ityecology,andconservationbiology.Bothhabitatareaandisolationhavelongbeenconsideredimportantinaffectingecologicalpatternsandprocesses.Forinstance,theequilibriumtheoryofislandbiogeography(ETIBhereinafter)andmetapopulationtheoryincorporateareaandisolationtoexplainvariationincommunitystruc-tureandpopulationdynamics,respectively(MacArthurandWilson1967;Levins1969;Hanski1998).Thesetheorieshavealsobeenappliedtounderstandhabitatlossandfrag-mentation,butwithlimitedsuccess(DoakandMills1994;Wiens1995;GasconandLovejoy1998;Lauranceetal.2002).Oneexplanationforthislimitedsuccessisthattheboundariesofhabitatfragments—habitatedges—canalsoprofoundlyinfluenceecologicalpatternsandprocesses(GasconandLovejoy1998).Infact,asurveyofrecentex-perimentsonhabitatfragmentationsuggestedthathabitatedgesaretheprimarydriversfortheinfluenceoffragmenta-tion(HarrisonandBruna1999).Overthepastthreedecades,agreatdealofresearchhasfo-cussedontheinfluenceofhabitatedgesandareaonawidediversityofpatternsandprocessesinfragmentedlandscapes.Muchofthisresearchhasbeendrivenbyconcernforrecentpopulationdeclinesofmanyspecies(e.g.,Herkert1994),im-provedtechnologythatallowsrapidassessmentatbroadspa-tialscales(e.g.,Alencaretal.2004),andanincreasingawarenessthatspacecanhavenoveleffectsonavarietyofecologicalandevolutionaryprocesses(e.g.,Kareiva1990).Yet,attemptstoprovideconceptualandpredictiveframe-workstoguideourunderstandingoftheuniquerolesofhabi-tatareaandedgehavebeenrare(butseeRiesandSisk2004).Whilebothhabitatareaandedgemayinfluenceindivi-duals,populations,andcommunities,understandingtherela-tiveroleoftheselandscapecharacteristicsiscrucialforimplementingsoundconservationstrategies.Becausehabitatareaandedgereflectdifferentaspectsoflandscapestruc-ture—landscapecompositionandconfiguration,respec-tively(McGarigalandMarks1995)—theimportanceofeachsuggestsdifferentfociforconservationefforts.Further-more,whenhabitatedgesareimplicatedintheeffectsoffragmentation,thetypeofedgeofteninfluencesobservedoutcomes(Sisketal.1997;Chalfounetal.2002;FletcherandKoford2003;Lopez-Barreraetal.2006),suggestingthatmanagersandconservationistsmayneedtofocusonspecificattributesthatsurroundremainingfragments.Habi-tatareaandedgeinlandscapesnonethelesstendtocovaryintheiramount(Fig.1),andmanystudieshavenotbeenabletoisolatetheuniquecontributionsofeachelement(Parkeretal.2005).Anothercomplicationisthateffectsfromhabitatedgesmayexplainobservedareaeffects(Bowersetal.1996;Lauranceetal.1998;Fletcher2005),makingitdifficulttointerpretifpurportedareaeffectsareactuallyamanifestationofedgeeffects.Theseproblemsareexacerbatedwhenconsideringthecumulativeinfluenceofmultipleedgeswithinfragments(Fig.1),asopposedtoonlytheinfluenceofthenearestedge(Malcolm1994;Fletcher2005).Ourobjectiveistoexaminetheeffectsofbothareaandedgetobetterunderstandtheiruniquecontributionstohabi-tatfragmentation.Wefirstsynthesizeseparateconceptualframeworksforthemechanismsthatunderlieareaandedgeeffects.Wethenexaminetheliteraturewherebothedgeandareaeffectsweremeasuredwithinthesameinvestigationtodeterminetheirrelativecontributiontofragmentationdy-namics.Weconcludebydiscussinghowfuturefieldstudiesshouldbedesignedtolimitthepotentialconfoundingissuesthatarisewheninvestigatingtheroleofareaandedgeinfragmentedlandscapes.Forthepurposesofthisarticle,‘‘patchsizeeffects’’arechangesinecologicalresponses(perunitarea)asafunctionofpatchareaperse,andthusindependentofotherpotentialcauses,suchasthosearisingfromchangesintherelativeamountofedge.Alternatively,weuse‘‘areaeffects’’asabroadertermthatdescribeschangesinecologicalresponseswithhabitatarea,whichmayormaynotbeindependentofotherpotentialcauses(adefinitionimplicitlyusedintheliterature),suchthatpatchsizeeffectsareasubsetofareaeffects.Wedonotincludestudiesthatexaminetotalhabitatareawithinalandscapebe-causeissuesofscaleandlandscapecontextmakeinferencesontheeffectiveareaavailabletoorganismsdifficulttointer-pret.‘‘Edgeeffects’’arechangesinecologicalresponsesthatvarywithdistancefromedge(Riesetal.2004).Interpretingtheuniquerolesofpatchsizeandedgewillimproveourunderstandingofhabitatfragmentationandhelptorefineconservationstrategiesaimedatmitigatingtheeffectsofhabitatlossandfragmentation.ConceptualframeworksforedgeandareaInvestigationsonhabitatareaandedgehavelargelybeen PatchsizeAmountofedgerelativetoarea Fig.1.Theinterplayofpatchsizeandedge.Aspatchsizede-creases,therelativeamountofedgeincreases.Inaddition,thein-fluenceofmultiple(allsurrounding)edgesisexacerbatedinsmallerpatches.Thispatternisshownbythepointineachpatchofincreasingsize,whichallhavethesamenearestdistancetoedge(markedbythethicklinestotheleftofthesolidcircles).However,aspatchsizeincreases,thecumulativedistancetoalledgesalsonecessarilyincreases(illustratedinthefourcardinaldirections),andtypicallydoessoinanexponentialmanner.Can.J.Zool.Vol.85,20072007NRCCanada andPickett2001),andanimalmovement,withbutterflymovementparticularlywellstudied(Haddad1999;RiesandDebinski2001;SchultzandCrone2001;SchtickzelleandBaguette2003).Arigorousexamplehighlightingtheconse-quencesofecologicalflowscomesfromarecentexperimentbyLeveyetal.(2005),whodemonstratedthattheeffectsofcorridorsonseeddispersalbybirdscouldbestbeexplainedbyedgeeffects:individualsencounteringedgestendedtore-flectbackintothefocalpatch,whichinturnledtoextensiveuseofcorridors.Mobileorganisms,asopposedtosessileones(Fig.2nearedgesmayalsogainaccesstoresourcesfromadjacenthabitats(Randetal.2006).Iforganismsgainaccesstosuchspatiallyseparatedresources,anincreaseinthedensityofindividualsispredictednearedges.Inaclassicexample,as-penexperiencedtheheaviestoutbreaksofaleaf-miningin-sect(Phyllonoryctersalicifoliella(Chambers,1875))nearstandsofconiferoustrees,whichP.salicifoliellausedasoverwinteringsites(Martin1956).Changesinecologicalflowsoftenresultindistinctre-sourcegradientsnearedges,whichinfluencethedistributionoftheplantsandanimalsthatmapontochangesinthosere-sources.Suchresourcemappingcanoccurviaplantsandan-imalsmappingontogradientsinabioticorplantresources,orviaanimalsmappingontootheranimalresources(Riesetal.2004).Forexample,foragingbehaviorsofwhite-footedmice(Peromyscusleucopus(Rafinesque,1818))inforestfragmentsofOhiosuggestedthatincreasedfoodavailabilitymayexplainelevatedmousedensitiesnearedges(WilderandMeikle2005).Whileresourcemappingcanresultinei-therincreasesordecreasesinabundanceanddiversity,knowledgeofrelativehabitatusecanallowinvestigatorstopredictresponsestoedges(RiesandSisk2004;Riesetal.Finally,muchtheoreticalandempiricalresearchhascen-teredontheindirecteffectofhabitatedgesonspeciesinter-actions(Faganetal.1999;McGeochandGaston2000;Chalfounetal.2002).Forinstance,manyinvestigationshaveestimatedwhetherpredationonbirdnestsincreasesnearedges,withhighlyvariableresults,yetmuchofthisvariabilitycanbeexplainedbyedgetype,knowledgeofspe-cificpredatorbehavior(whichisoftennotmeasured),andlandscapecontext(Paton1994;HartleyandHunter1998;Lahti2001;Chalfounetal.2002,2002;Stephensetal.2004).Fewerempiricalexamplesexistforchangesinotherspeciesinteractionsnearedges.Inarareexample,Suarezetal.(1998)reportedthatArgentineants(Linepithemahumile(Mayr,1868))foundpredominantlynearedgesappeartooutcompetethelocal,nativeantspecies.MechanismsofareaeffectsToaddressmechanismsofareaeffects,wedistinguishbe-tweenmechanismsthatgeneratepatchsizeeffects,asde-finedabove,andothermechanismsthatcanleadtochangeswithareathatdonotoccurfrompatchsizeeffects.Further-more,wedifferentiatebetweentwoalternativeswhenfocus-singonmeasuresofpopulationsizeanddensity:(1)populationsizechangeswithdecliningpatchsize,butpopu-lationdensityremainsconstant,and(2)populationdensitychangeswithpatchsize.TheETIBfocussedonchangesinspeciesrichnesswithislandareabasedonvariationincolo-nizationandextinctionrates(MacArthurandWilson1967);animplicitassumptionofthistheoryisthatpopulationden-sityremainsconstantwitharea(Andren1994;Connoretal.2000;Brotonsetal.2003).Yetifchangesindensityoccur(BowersandMatter1997;Benderetal.1998;Connoretal.2000),populationsandcommunitiesmaybestronglyinflu-encedbyeffectsofhabitatlossandfragmentation.Wefirstoutlinemechanismsthatgeneratepatchsizeeffects(Fig.2),focussingonhowpatchsizeinfluencespopulationsize,density,andcommunitystructure,andsubsequentlynotetwoprimaryfactorsthatcanleadtoareaeffectswhicharenotpatchsizeeffects.Declinesinpatchsizewithhabitatlossandfragmentationfundamentallyinfluencetheabundanceoforganismsintwodirectways.First,aspatchesdecreaseinsize,declinesintheamountordiversityofresources,orboth,influencepopula-tionsizesofplantsandanimalsbyalteringcarryingcapaci-ties.Second,variationinpatchsizealsoinfluencestheextentofemigrationand(or)immigrationbydispersers(Root1973;GilpinandDiamond1976;Risch1981;Lomo-lino1990).Together,thesedirecteffectscanfurtherleadtoindirectchangesinspeciesinteractionsandresourcemap-ping,leadingtochangesincommunitystructure(Fig.2Aspatchsizedeclines,themostobviouseffectisade-clineintheamountand(or)diversityofresourcesavailabletoorganisms,yetthisphenomenoncanhavepotentiallynovel,cascadingeffectsthroughindirectpathways(Fig.2).Indeed,whenpatchsizedeclinestoverysmallsizes,lackofresourcesorinsufficientspacemaycausepatchestofallbelowtheminimumarearequirementsofin-dividuals(e.g.,territorysize;StratfordandStouffer1999).Theresourceconcentrationhypothesis(Root1973)statesthatsmallerareascontainalowerconcentrationordiversityofresources(Fig.2),resultinginlowerdensitiesofindi-viduals(seealsoAmbuelandTemple1983).Lowerconcen-trationsofresourcescanreducedensitythroughreducedlocalrecruitmentbasedondeclininghabitatquality(Risch1981;Matter1997).MacArthurandWilson(1967)alsohighlightedthatchangesinhabitatheterogeneitywithde-cliningareacouldleadtodeclinesinspeciesrichness.Thus,theETIBandtheresourceconcentrationhypothesispredictthatdecliningresourcesshouldhaveanegativeimpactonpopulationsizeand(or)density,whichsubsequentlyin-creasesextinctionprobability(Didhametal.1998,1998altersspeciesdistributions,andchangescommunitystruc-ture.Indirecteffectsofvariationinresourcemappingcanfurtherinfluencepopulationsandcommunitieswhendeclin-ingpatchsizedifferentiallyaffectssomeresourcesmorethanothers(Fig.2Immigrationandemigrationratesareinfluencedbysev-eralfactors,suchasedgetype(RiesandDebinski2001),populationdensity(Matthysen2005),andhabitatquality(Pulliam1988).Nonetheless,decliningpatchsizeisgener-allythoughttodirectlydecreaseimmigrationratesbecauseindividualsmovingthroughalandscapearelesslikelytoen-countersmallpatches(the‘‘target’’effect;GilpinandDia-mond1976;Lomolino1990),therebyreducingpopulationsizeanddensity(Fig.2).Bowmanetal.(2002)recentlyre-viewedgeneralheuristicargumentsregardingimmigrationandconcluded,however,thatwhileimmigrationmayoftenincreasewithpatcharea,leadingtolargerpopulationsize,Can.J.Zool.Vol.85,2007#2007NRCCanada ments.Ourconceptualframeworks(Figs.2)suggestthatedgeeffectsfromvariationinecologicalflowscouldpotentiallyhavesynergisticeffectsonpatchareainfluencingtheamountofresources,butthereisnoexplicitempiricalevidencethatisolatesthispotentialsynergism.Indeed,littleclearevidencecurrentlyexistsforareaandedgeeffectsin-teractinginfragmentedlandscapes(Nouretal.1993;Mat-thewsetal.1999;Galettietal.2003;LienertandFischer2003;Fletcher2005;butforarecentexampleseeEwersetal.2007).Thereisongoinginterestindeterminingwhen,where,andhowthesurroundinglandscapeinfluencespopulationsandcommunitieswithinfocalareas.Landscapecontextinfluen-cesedgeeffectsprimarilythroughvariationinmatrixqual-ity.Matrixqualitycanalterchangesinthemovement,orflow,ofmaterials,energy,andorganisms—effectsthatarefundamentallylinkedwithvariationinthepermeabilityofhabitatboundaries(e.g.,CollingeandPalmer2002).Whiletherehasbeeninterestinunderstandinghowlandscapecon-textcaninfluenceedgeeffects(Donovanetal.1997;Tewksburyetal.1998;Chalfounetal.2002),anoutstand-ingquestioniswhethermatrixqualityeffectsareinrealitysimplyeffectsofdifferentedgetypes(FletcherandKoford2003;RiesandSisk2004),orviceversa.Thatis,arepur-portedmatrixeffectsdrivenprimarilybyvariationinlocaledgeeffectscausedbydifferentedgetypes,orareedgetypeeffectsactuallygeneratedbylarge-scaledifferencesinmatrixquality?Answerstothisquestionwouldbeextremelyvaluableininterpretingwhetherlarge-scalefragmentationconstrainslocalprocessesorwhetherlocalprocesses‘‘scaleup’’togeneratelarge-scalepatterns.Landscapecontextmayalsoinfluenceareaeffectsinthatdistancesbetweenhabitatscanaltereffectivepopulationsizeswhenindividualsusemorethanonepatch(Diffen-dorferetal.1995).However,therearenocurrentpredic-tionsformatrixqualityalteringpatchsizeeffects,basedonthefundamentalprocessespatchsizeinfluences(Fig.2Whilematrixqualitymayinfluencethemovementoforgan-isms,therebyalteringimmigrationprobabilities,therelative‘‘target’’effectofareashouldnotvary,allelsebeingequal.However,matrixqualitycansometimesbecomemoreim-portantthanareaindrivingimmigrationrates,suchthatpatchsizemayexplainlessvariationinmovementthancharacteristicsofthematrix(BenderandFahrig2005).Whenthesurroundingmatrixprovidesnovelresourcesusedbysomeindividuals,densitiescanincreasewithdecliningpatchsize(Dunningetal.1992;Nortonetal.2000;Daviesetal.2001;Estades2001),yetsuchdensityelevationisnotdrivenbypatchsizeeffects.Instead,sucheffectsaremoredirectlyrelatedtoedgeeffectsallowingaccesstospatiallyseparatedresources(Didhametal.1998Therelativeroleofareaandedge:empiricalToexploretherelativerolesofhabitatareaandedge,wereviewedstudiesthatsimultaneouslytestedforbotheffectsinfragmentedlandscapeswithinaninvestigation.Wefo-cussedonthreekeyquestions.First,doinvestigatorsdesignstudiesthatlimitthepotentialconfoundingofareaandedgewhenaddressingtheseeffects?Second,dostudiesmorefre-quentlyfindevidenceforedgeorareaeffects?Third,doin-vestigatorstestforfundamentalmechanismsresponsibleforpatchsizeandedgeeffects?Wealsoaddresswhetherresultsvarywhencomparingconfoundedandnonconfoundedstud-ies,andwhetherresultsvarydependingontaxonomicgroup,themetricused,orhabitattype.ReviewmethodsWefirstscreened231articlesgleanedfromtheWebof,using‘‘patchsize*ANDedge*ANDfragment*’’,‘‘habitatarea*ANDedge*ANDfragment*’’,‘‘areaeffect*ANDedge*ANDfragment*’’,and‘‘fragmentsize*ANDedge*’’askeywords(accessed14March2006).Wealsosupplementedthislistwithotherrelevantarticles.Fromthesearticles,60testedforboththeeffectsofareaandedge.Forthisreview,weconsideredhabitatareatoincludeonlyissuesofpatchsizeanddidnotconsiderlandscape-levelanalyses(e.g.,Belisleetal.2001;FletcherandKoford2002).Foredgeeffects,weconsideredstudiesthatestimatedresponsesatthewithin-patchscale,usingdistancetoedge,andatthepatchscale,usingmeasuresofpatchshape(e.g.,perimetertoarearatios).Ouroverarchinggoalwastodetermineifstudiesmorecommonlyshowedevidenceforedgeorareaeffects.Assuch,wefocussedonwithin-study,pairwisecomparisonsthatusedsimilarmethodologiestointerprettheevidenceforedgeandareaeffects.Someinvestigationstestedforbothareaandedge,butuseddifferentpatches,differentsamplingunits,ordifferentspeciesforeachtest(DijakandThompson2000;Beieretal.2002;Cronin2003);suchstudieswereomitted.Tolimitbiasesfromvariationinsamplingeffort,wealsoonlyincludedstudieswithequalsamplesizesfortestingtheeffectsofareaandedgeforagivenresponsevar-iable.Weinferredsamplesizebasedoneitherthereporteddenominatordegreesoffreedomoronreportedsamplingunitsusedforanalyses.Thesecriteriareducedthedatasetto38articleswith213responsevariablesmeasured.Ourap-proachcontrolledformanyissuesthatarisewhensummariz-ingeffectsacrossstudies,asisoftenattemptedinmeta-analyticapproaches(e.g.,Benderetal.1998;Connoretal.2000).Yetwechosenottoperformaformalmeta-analysisforseveralreasons.First,weemphasizethatourgoalwasnottoestimateoveralleffectsizes,whichwouldbehighlyvariableacrossstudiesdependingonthetaxaandmetricin-vestigated,butinsteadtoidentifytherelativesupportforareaoredgeeffectswithininvestigations.Second,formalmeta-analysesprimarilyattempttocontrolforvariationinsamplesizeacrossstudies;however,bylimitingoursumma-riestoinvestigationswiththesamesamplesize,wecon-trolledforthisissuedirectly.Finally,aformalmeta-analysiswouldhavefurtherreducedourdatasetbyapprox-imately30%.Wenotethatusingtheentiredataset(all60articleslistedabove)providedsimilarqualitativeresultsasthosepresentedbelow.Foreacharticle,wefirstsummarizedavarietyofinfor-mationrelatedtosamplingdesign.Wewereinterestedinsamplingconcernsthatoccurwheninvestigatorstestforareaandedgeeffects(ConnorandMcCoy1979;Hornetal.2000;ManckeandGavin2000;Parkeretal.2005),particu-larlysituationswhereareatestswereconfoundedbypoten-tialedgeeffects,suchthatanobservedareaeffectcouldnotCan.J.Zool.Vol.85,2007#2007NRCCanada oftenthanthosereportedbyParkeretal.(2005).Thispat-ternisnotsurprising,becauseParkeretal.(2005)includedmanystudiesthatonlytestedforeitheredgeorarea;suchstudiesaremorelikelytobeconfoundedthanthosetestingforbotheffects.Forfurtheranalyses,wecomparethesenon-confoundedanalysestoresultsfromconfoundedanalyses.Nonconfoundedanalysestendedtofindevidenceforedgeandareaeffectsslightlymorefrequently(64.6%)thancon-foundedanalyses(58.1%;Fig.4),withfewernonsignificantresponsesreported.Samplesizevariationbetweencon-foundedandnonconfoundedanalyseswouldactuallypredicttheopposite,becauseoverallsamplesizesweregreaterforconfoundedanalyses(nonconfounded:61.3±10.0,con-founded:95.7±5.4;=3.27,=0.001).Confoundedanal-ysesreportededgeandareaeffectsinapproximatelyequalproportions(39.9%vs.39.2%,respectively;Fig.4),whilenonconfoundedanalysesfoundevidenceforedgeeffects(52%)moreoftenthanareaeffects(35%).Thispatternwasgenerallysimilarregardlessofthemetricinvestigated(Fig.5).Similarpatternsalsooccurredfordifferenthabitats(Fig.5),withedgeeffectsbeingparticularlycommoningrasslands.Whenbotheffectswereobserved,edgeeffectswerestrongerthanareaeffectsfornonconfoundedanalyses(78.6%vs.21.4%ofresponsevariables,respectively)andconfoundedanalyses(56.2%vs.9.4%ofresponsevariables,respectively).Theonlyexceptionwasforinvestigationsonmammals,withareaeffectsbeingreportedasstrongerthanedgeeffectsfornonconfoundedanalyses(Fig.4).Together,theseresultssuggestthatedgeeffectsareeithermorecom-moninnatureorareeasiertodetectthanareaeffects(orboth).Edgeeffectsmayindeedbeeasiertodetectbecausethemostcommonstudydesign,usingwithin-patchmeasuresofdistancesfromedge,controlsforsite-levelvariation.Wecautionthatinferencesabouttherelativeoccurrenceandstrengthofedgeversusareaeffectsbasedoncon-foundedstudiesdifferedfromthosebasedonnoncon-foundedstudies.Forexample,nonconfoundedinvestigationsonbirdsfoundmoreevidenceforedgeeffects,whereascon-foundedstudiesfoundsimilarevidenceforareaandedgeef-fects(Fig.4).Yetrelativelyfewstudieswerenotconfounded,whichlimitsgeneralizationsformosttaxaandhighlightstheneedforrigorousstudydesignandanalysisinstudiesofhabitatfragmentation.Therewasaremarkablelackofdiscordancebetweenob-servedareaandedgeeffects,withonly1.6%ofreportedre-sponses(1of64)fornonconfoundedanalysesand2.4%(5of210;for3responsevariables,concordance–discordancecouldnotbedetermined)forallstudiescombinedshowing (a)Nonconfoundedanalyses(b)ConfoundedanalysesTotalTotalTotalPlantsTotal(7)(42)(90)(1)(148)(28)(10)(9)(13)(65)(3)(3)(1)(5)(14)(6)(7)(17)(1)(32)PercentageofPercentageof Fig.4.Resultsfromstudiesthattestedforbothedgeandareaeffectsonecologicalpatterns,summarizedbytaxonomicgroupfor()non-confoundedanalysesand()confoundedanalyses.Leftpanelsshowthepercentageofresponsevariablesinvestigatedthatwereinfluencedbyarea,edge,bothedgeandarea(both),orhadnoreportedeffect(neutralresponse,NR),andnumbersabovebarsrepresentthenumberofresponsevariablesinvestigated.Notethatresponsevariablesforeachtaxadonotsumtothetotalbecausetherewereothertaxainvestigatedthatarenotshownowingtolimitedsamplesize(e.g.,herptiles).Forthoseresponsesinfluencedbyareaandedge,therightpanelsshowthepercentageofresponsevariablesreportedasbeingmoreinfluencedbyedgeorarea(basedoninferentialstatisticsreported),andthosethatdidnotprovideenoughinformationtodistinguish(unknown).Fortaxonomicgroupings,wedidnotincludestudiesthatfocussedonspeciesinteractionsacrossdifferenttaxonomicgroups.Onlystudieswherethesameresponsevariablewastestedforbothfactorswereincluded.Can.J.Zool.Vol.85,20072007NRCCanada specificcompetitionnearedgesasanexplanationforob-servedareaeffectsinmigratoryforestbirds.Asnotedabove,someedgeeffectscanalsoexplainotherfragmenta-tioneffects,suchastheuseofconservationcorridors(Haddad1999;Leveyetal.2005).Whenpatchsizeeffectsoccur,however,edgeeffectsarenotexplicitlypredicted,ei-therbasedonvariationinresourcesinfluencingpopulationsize(MacArthurandWilson1967)orbasedonvariationinemigrationandimmigrationratesinpatchesthatvaryinsize.Consequently,wheninvestigatorsfindsupportforbothedgeandareaeffectsthatareconcordantintheirdirection-ality,andstudiesareconfoundedintheirdesignand(or)analysis,sucheffectsmaybedrivenentirelybyhabitatedgesalone(Lauranceetal.1998).Whenthestrengthofedgeeffectsvarieswithpatchsize(KiviniemiandEriksson2002),patchsizeeffectscouldaccountfortheadditionalvariation.However,cumulativeeffectsfrommultipleedgescouldalsoproducesimilarpatterns(Fletcher2005),therebycomplicatingtheinterpretationofthearea–edgerelationship.Thestrongeststudydesignforteasingaparttheuniquein-fluenceofareaandedgeoccurswheninvestigatorsmeasureresponsesatsimilardistancesfromthenearestedgeinfrag-mentsofdifferentsizes,withreplicatesforeachpatchsizecategory.Suchadesigncontrolsforpotentialedgeeffects(fromthenearestedge)inareatests,andpotentialareaef-fectsinedgetests,byaddressingvariationinedgeeffectsacrossdifferentpatchsizes(see,e.g.,Nouretal.1993).Ad-ditionally,investigatorscanusestatisticalapproachestofur-therlimitconfoundinginthisdesign(e.g.,distancestomultipleedges),whereasstatisticalapproachesalonemaybefruitlessinlimitingpotentialconfoundingforsomeotherstudydesigns.Anotherapproachtoisolatetheeffectofareaindependentofedgeistocomparepredictionsderivedfromedgere-sponsestoobservedresponsesthatoccurwithvariationinpatchsize.Theeffectiveareamodel(Sisketal.1997;Brandetal.2006)providesoneapproachtodoso.Inthismodel,responsevariablesareestimatedfordifferentpatchesviadatafromedgetransects.Edgeresponsepatternsmeasuredatsurveyplotsatdifferentdistancesfromhabitatedgescanbeextrapolatedtopatchesofdifferentsizestopredicttheaverageresponseinapatchthatwouldresultentirelyfromedgeeffects.Whencontrollingforothersite-leveleffects,consistentdiscrepanciesbetweenpredictionsfromtheeffec-tiveareamodelandobserveddataasafunctionofpatchsizemaybeattributedtopatchsizeandnotedgeeffects.Weemphasizethatfocussingonthedirectinfluencesofhabitatareaandedgeonecologicalprocessesprovidesamorerigorousframeworkwithwhichtounderstandtheim-pactsofhabitatfragmentation.Riesetal.(2004)arguedthatresourcemappingandspeciesinteractionsarethedriversofmostobservededgeeffects,yetthesecanalsovarywithpatchsize(Fig.2).Eventhoughindirecteffectsfrompatchsizeandedgemaybeinevitablyintertwined,makingitdiffi-culttoisolatetheeffectsofeachlandscapeelement,directeffectsaredistinct.Therefore,focussingonresponsevaria-blesthatilluminatethesedirecteffectswillprovidenovelinsightintotheuniquerolesofareaandedge.Ourconcep-tualframeworks(Fig.2;Riesetal.2004)suggestthatstud-iesfocussingprimarilyonmovementinrelationtoedgeand ResponsePositive(+)PatchsizeNeutral(0)Negative(–)Negative(–)DistancefromedgeNeutral(0)Positive(+)(b)Confounded(a)Nonconfounded(c)Allanalyses Fig.6.Concordanceofedgeandareaeffectsinfragmentedlandscapesfor()nonconfoundedanalyses,()confoundedanalyses,and()allstudiescombined.Resultsarefromstudiesthatexplicitlytestedforbothedgeandareaeffectsonecologicalpatterns.Tobeconsistentwiththeliterature,wereportpositiveareaeffectsasthosethatoccurwhentheresponsevariableincreaseswithincreasingpatchsize,whereaspositiveedgeeffectsoccurwhenaresponsevariableincreasesnearedges.Onlystudieswherethesameresponsevariablewastestedforbothfactorswereincluded.Can.J.Zool.Vol.85,20072007NRCCanada arearelationships:therelationshipbetweenanimalpopulationdensityandarea.Ecology,:734–748.Cronin,J.T.2003.Movementandspatialpopulationstructureofaprairieplanthopper.Ecology,:1179–1188.doi:10.1890/0012-9658(2003)084[1179:MASPSO]2.0.CO;2.Cronin,J.T.2003.Matrixheterogeneityandhost–parasitoidinter-actionsinspace.Ecology,:1506–1516.doi:10.1890/0012-9658(2003)084[1506:MHAHII]2.0.CO;2.Davies,K.F.,andMargules,C.R.1998.Effectsofhabitatfragmen-tationoncarabidbeetles:experimentalevidence.J.Anim.Ecol.:460–471.doi:10.1046/j.1365-2656.1998.00210.x.Davies,K.F.,Melbourne,B.A.,andMargules,C.R.2001.Effectsofwithin-andbetween-patchprocessesoncommunitydy-namicsinafragmentationexperiment.Ecology,:1830–Didham,R.K.,Lawton,J.H.,Hammond,P.M.,andEggleton,P..Trophicstructurestabilityandextinctiondynamicsofbeetles(Coleoptera)intropicalforestfragments.Philos.Trans.R.Soc.Lond.BBiol.Sci.:437–451.doi:10.1098/rstb.1998.Didham,R.K.,Hammond,P.M.,Lawton,J.H.,Eggleton,P.,andStork,N.E.1998.Beetlespeciesresponsestotropicalforestfragmentation.Ecol.Monogr.:295–323.Diffendorfer,J.E.,Gaines,M.S.,andHolt,R.D.1995.Habitatfrag-mentationandmovementsofthreesmallmammals(,and).Ecology,:827–839.doi:10.2307/Dijak,W.D.,andThompson,F.R.2000.Landscapeandedgeef-fectsonthedistributionofmammalianpredatorsinMissouri.J.Wildl.Manage.:209–216.doi:10.2307/3802992.Doak,D.F.,andMills,L.S.1994.Ausefulrolefortheoryincon-servation.Ecology,:615–626.doi:10.2307/1941720.Donovan,T.M.,Jones,P.W.,Annand,E.M.,andThompson,F.R.1997.Variationinlocal-scaleedgeeffects:mechanismsandlandscapecontext.Ecology,:2064–2075.Dunford,W.,Burke,D.M.,andNol,E.2002.Assessingedgeavoidanceandareasensitivityofred-eyedvireosinsouthcentralOntario.WilsonBull.:79–86.doi:10.1676/0043-5643(2002)114[0079:AEAAAS]2.0.CO;2.Dunning,J.B.,Jr.,Danielson,B.J.,andPulliam,H.R.1992.Ecolo-gicalprocessesthataffectpopulationsincomplexlandscapes.:169–175.doi:10.2307/3544901.Estades,C.F.2001.Theeffectofbreeding-habitatpatchsizeonbirdpopulationdensity.Landsc.Ecol.:161–173.doi:10.1023/A:1011197432467.Ewers,R.M.,andDidham,R.K.2006.Confoundingfactorsinthedetectionofspeciesresponsestohabitatfragmentation.Biol.Rev.:117–142.doi:10.1017/S1464793105006949.PMID:16318651.Ewers,R.M.,Thorpe,S.,andDidham,R.K.2007.Synergisticin-teractionsbetweenedgeandareaeffectsinaheavilyfragmentedlandscape.Ecology,:96–106.doi:10.1890/0012-9658(2007)88[96:SIBEAA]2.0.CO;2.PMID:17489458.Fagan,W.E.,Cantrell,R.S.,andCosner,C.1999.Howhabitatedgeschangespeciesinteractions.Am.Nat.:165–182.Fahrig,L.1997.Relativeeffectsofhabitatlossandfragmentationonpopulationextinction.J.Wildl.Manage.:603–610.doi:10.Fahrig,L.2003.Effectsofhabitatfragmentationonbiodiversity.Annu.Rev.Ecol.Evol.Syst.:487–515.doi:10.1146/annurev.ecolsys.34.011802.132419.Flather,C.H.,andBevers,M.2002.Patchyreaction-diffusionandpopulationabundance:therelativeimportanceofhabitatamountandarrangement.Am.Nat.:40–56.doi:10.1086/324120.Fletcher,R.J.,Jr.2005.Multipleedgeeffectsandtheirimplicationsinfragmentedlandscapes.J.Anim.Ecol.:342–352.doi:10.1111/j.1365-2656.2005.00930.x.Fletcher,R.J.,Jr.2006.Emergentpropertiesofconspecificattrac-tioninfragmentedlandscapes.Am.Nat.:207–219.doi:10.1086/505764.PMID:16874630.Fletcher,R.J.,Jr.,andKoford,R.R.2002.Habitatandlandscapeassociationsofbreedingbirdsinnativeandrestoredgrasslands.J.Wildl.Manage.:1011–1022.doi:10.2307/3802933.Fletcher,R.J.,Jr.,andKoford,R.R.2003.Spatialresponsesofbo-bolinks(Dolichonyxoryzivorus)neardifferenttypesofedgesinnorthernIowa.Auk,:799–810.doi:10.1642/0004-8038(2003)120[0799:SROBDO]2.0.CO;2.Galetti,M.,Alves-Costa,C.P.,andCazetta,E.2003.Effectsoffor-estfragmentation,anthropogenicedgesandfruitcolourontheconsumptionofornithocoricfruits.Biol.Conserv.:269–273.doi:10.1016/S0006-3207(02)00299-9.Gascon,C.,andLovejoy,T.E.1998.EcologicalimpactsofforestfragmentationincentralAmazonia.Zoology—AnalysisofComplexSystems,:273–280.Gates,J.E.,andGysel,L.W.1978.Aviannestdispersionandfled-gingsuccessinfieldforestecotones.Ecology,:871–883.Gilpin,M.E.,andDiamond,J.M.1976.Calculationofimmigrationandextinctioncurvesfromthespeciesareadistancerelation.Proc.Natl.Acad.Sci.U.S.A.:4130–4134.doi:10.1073/pnas.73.11.4130.PMID:16592364.Haddad,N.M.1999.Corridorusepredictedfrombehaviorsathabitatboundaries.Am.Nat.:215–227.doi:10.1086/Haila,Y.,Hanski,I.K.,andRaivio,S.1993.Turnoverofbreedingbirdsinsmallforestfragments:the‘‘sampling’’colonizationhy-pothesiscorroborated.Ecology,:714–725.doi:10.2307/Hamback,P.A.,andEnglund,G.2005.Patcharea,populationden-sityandthescalingofmigrationrates:theresourceconcentra-tionhypothesisrevisited.Ecol.Lett.:1057–1065.doi:10.1111/j.1461-0248.2005.00811.x.Hanski,I.1998.Metapopulationdynamics.Nature(London),41–49.doi:10.1038/23876.Harrison,S.,andBruna,E.1999.Habitatfragmentationandlarge-scaleconservation:whatdoweknowforsure?Ecography,225–232.doi:10.1111/j.1600-0587.1999.tb00496.x.Hartley,M.J.,andHunter,M.L.1998.Ameta-analysisofforestcover,edgeeffects,andartificialnestpredationrates.Conserv.:465–469.doi:10.1046/j.1523-1739.1998.96373.x.Henle,K.,Davies,K.F.,Kleyer,M.,Margules,C.,andSettele,J.2004.Predictorsofspeciessensitivitytofragmentation.Biodi-vers.Conserv.:207–251.doi:10.1023/B:BIOC.0000004319.Herkert,J.R.1994.Theeffectsofhabitatfragmentationonmid-westerngrasslandbirdcommunities.Ecol.Appl.:461–471.Hickerson,C.A.M.,Anthony,C.D.,andWalton,B.M.2005.Edgeeffectsandintraguildpredationinnativeandintroducedcenti-pedes:evidencefromthefieldandfromlaboratorymicrocosms.Oecologia(Berl.),:110–119.doi:10.1007/s00442-005-0197-y.PMID:16041613.Horn,D.J.,Fletcher,R.J.,Jr.,andKoford,R.R.2000.Detectingareasensitivity:acommentonpreviousstudies.Am.Midl.Nat.:28–35.doi:10.1674/0003-0031(2000)144[0028:DASACO]Jules,E.S.,andRathcke,B.J.1999.Mechanismsofreducedtril-liumrecruitmentalongedgesofold-growthforestfragments.Can.J.Zool.Vol.85,2007#2007NRCCanada bilityexplained.Annu.Rev.Ecol.Evol.Syst.:491–522.doi:10.1146/annurev.ecolsys.35.112202.130148.Risch,S.J.1981.Insectherbivoreabundanceintropicalmonocul-turesandpolycultures:anexperimentaltestoftwohypotheses.:1325–1340.doi:10.2307/1937296.Root,R.B.1973.Organizationofaplantarthropodassociationinsimpleanddiversehabitats:thefaunaofcollards(Brassicaoler-).Ecol.Monogr.:95–124.doi:10.2307/1942161.Schtickzelle,N.,andBaguette,M.2003.Behaviouralresponsestohabitatpatchboundariesrestrictdispersalandgenerateemigra-tion–patcharearelationshipsinfragmentedlandscapes.J.Anim.Ecol.:533–545.doi:10.1046/j.1365-2656.2003.00723.Schultz,C.B.,andCrone,E.E.2001.Edge-mediateddispersalbe-haviorinaprairiebutterfly.Ecology,:1879–1892.Sisk,T.D.,Haddad,N.M.,andEhrlich,P.R.1997.Birdassem-blagesinpatchywoodlands:modelingtheeffectsofedgeandmatrixhabitats.Ecol.Appl.:1170–1180.doi:10.1890/1051-0761(1997)007[1170:BAIPWM]2.0.CO;2.Stamps,J.A.,Buechner,M.,andKrishnan,V.V.1987.Theeffectsofedgepermeabilityandhabitatgeometryonemigrationfrompatchesofhabitat.Am.Nat.:533–552.doi:10.1086/284656.Stephens,D.W.,andKrebs,J.R.1986.Foragingtheory.PrincetonUniversityPress,Princeton,N.J.Stephens,S.E.,Koons,D.N.,Rotella,J.J.,andWilley,D.W.2004.Effectsofhabitatfragmentationonaviannestingsuccess:are-viewoftheevidenceatmultiplespatialscales.Biol.Conserv.:101–110.doi:10.1016/S0006-3207(03)00098-3.Storch,I.1991.Habitatfragmentation,nestsiteselection,andnestpredationriskincapercaillie.OrnisScand.:213–217.doi:10.Stratford,J.A.,andStouffer,P.C.1999.Localextinctionsofterres-trialinsectivorousbirdsinafragmentedlandscapenearManaus,Brazil.Conserv.Biol.:1416–1423.doi:10.1046/j.1523-1739.Strayer,D.L.,Power,M.E.,Fagan,W.F.,Pickett,S.T.A.,andBel-nap,J.2003.Aclassificationofecologicalboundaries.:723–729.doi:10.1641/0006-3568(2003)053[0723:ACOEB]2.0.CO;2.Suarez,A.V.,Bolger,D.T.,andCase,T.J.1998.Effectsoffrag-mentationandinvasiononnativeantcommunitiesincoastalsouthernCalifornia.Ecology,:2041–2056.doi:10.1890/0012-9658(1998)079[2041:EOFAIO]2.0.CO;2.Tewksbury,J.J.,Hejl,S.J.,andMartin,T.E.1998.Breedingpro-ductivitydoesnotdeclinewithincreasingfragmentationinawesternlandscape.Ecology,:2890–2903.Tscharntke,T.,andBrandl,R.2004.Plant–insectinteractionsinfragmentedlandscapes.Annu.Rev.Entomol.:405–430.doi:10.1146/annurev.ento.49.061802.123339.PMID:14651470.Weakland,C.A.,andWood,P.B.2005.Ceruleanwarbler(droicacerulea)microhabitatandlandscape-levelhabitatcharac-teristicsinsouthernWestVirginia.Auk,:497–508.doi:10.1642/0004-8038(2005)122[0497:CWDCMA]2.0.CO;2.Wiens,J.A.1995.Habitatfragmentation:islandvlandscapeperspectivesonbirdconservation.Ibis,(Suppl.1):S97–Wiens,J.A.,Crawford,C.S.,andGosz,J.R.1985.Boundarydy-namics:aconceptualframeworkforstudyinglandscapeecosys-tems.Oikos,:421–427.doi:10.2307/3565577.Wilcove,D.S.,Rothstein,D.,Dubow,J.,Phillips,A.,andLosos,E.1998.QuantifyingthreatstoimperiledspeciesintheUnitedStates.Bioscience,:607–615.doi:10.2307/1313420.Wilder,S.M.,andMeikle,D.B.2005.Reproduction,foragingandthenegativedensity–arearelationshipofageneralistrodent.Oe-cologia(Berl.),:391–398.doi:10.1007/s00442-005-0086-4.Winter,M.,Johnson,D.H.,andFaaborg,J.2000.Evidenceforedgeeffectsonmultiplelevelsintallgrassprairie.Condor,256–266.doi:10.1650/0010-5422(2000)102[0256:EFEEOM]2.0.Can.J.Zool.Vol.85,2007#2007NRCCanada