简介：Background:Manytreespeciesintropicalforestshavedistributionstrackinglocalridge-slope-valleytopography.Previousworkina50-haplotinKorupNationalPark,Cameroon,demonstratedthat272species,or63%ofthosetested,weresignificantlyassociatedwithtopography.Methods:Weusedtwocensusesof329,000trees≥1cmdbhtoexaminedemographicvariationatthissitethatwouldaccountforthoseobservedhabitatpreferences.Wetestedtwopredictions.First,withinagiventopographichabitat,speciesspecializingonthathabitat(‘residents')shouldoutperformspeciesthatarespecialistsofotherhabitats(‘foreigners').Second,acrossdifferenttopographichabitats,speciesshouldperformbestinthehabitatonwhichtheyspecialize(‘home')comparedtootherhabitats(‘away').Species'performancewasestimatedusinggrowthandmortalityrates.Results:Inhierarchicalmodelswithspeciesidentityasarandomeffect,wefoundnoevidenceofademographicadvantagetoresidentspecies.Indeed,growthratesweremostoftenhigherforforeignspecies.Similarly,comparisonsofspeciesontheirhomevs.awayhabitatsrevealednosignofaperformanceadvantageonthehomehabitat.Conclusions:Werejectthehypothesisthatspeciesdistributionsalongaridge-valleycatenaatKoruparecausedbyspeciesdifferencesintrees≥1cmdbh.Sincetheremustbeademographiccauseforhabitatspecialization,weofferthreealternatives.First,thedemographicadvantagespecialistshaveathomeoccursatthereproductiveorseedlingstage,insizessmallerthanwecensusintheforestplot.Second,speciesmayhavehigherperformanceontheirpreferredhabitatwhendensityislow,butwhenpopulationbuildsup,therearenegativedensity-dependentfeedbacksthatreduceperformance.Third,demographicfilteringmaybeproducedbyextremeenvironmentalconditionsthatwedidnotobserveduringthecensusinterval.