Primate communities are structured more by dispersal limitation than by niches.

Abstract

1. A major goal in community ecology is to identify mechanisms that govern the assembly and maintenance of ecological communities. Current models of metacommunity dynamics differ chiefly in the relative emphasis placed on dispersal limitation and niche differentiation as causal mechanisms structuring ecological communities. Herein we investigate the relative roles of these two mechanisms in structuring primate communities in Africa, South America, Madagascar and Borneo. 2. We hypothesized that if dispersal limitation is important in structuring communities, then community similarity should depend on geographical proximity even after controlling for ecological similarity. Conversely, if communities are assembled primarily through niche processes, then community similarity should be determined by ecological similarity regardless of geographical proximity. 3. We performed Mantel and partial Mantel tests to investigate correlations among primate community similarity, ecological distance and geographical distance. Results showed significant and strongly negative relationships between diurnal primate community similarity and both ecological similarity and geographical distance in Madagascar, but significant and stronger negative relationships between community similarity and geographical distance in African, South American and Bornean metacommunities. 4. We conclude that dispersal limitation is an important determinant of primate community structure and may play a stronger role in shaping the structure of some terrestrial vertebrate communities than niche differentiation. These patterns are consistent with neutral theory. We recommend tests of functional equivalence to determine the extent to which neutral theory may explain primate community composition.

Full text

Page 1

Primatecommunitiesarestructuredmorebydispersal
limitationthanbyniches
LydiaH.Beaudrot1*andAndrewJ.Marshall1,2,3
1GraduateGroupinEcology,UniversityofCalifornia,OneShieldsAvenue,Davis,CA95616,USA;2Departmentof
Anthropology,UniversityofCalifornia,OneShieldsAvenue,Davis,CA95616,USA;and3AnimalBehaviorGraduateGroup,
UniversityofCalifornia,OneShieldsAvenue,Davis,CA95616,USA
Summary
1. A major goal in community ecology is to identify mechanisms that govern the assembly and
maintenance of ecological communities. Current models of metacommunity dynamics differ
chiefly in the relative emphasis placed on dispersal limitation and niche differentiation as causal
mechanisms structuring ecological communities. Herein we investigate the relative roles of these
two mechanisms in structuring primate communities in Africa, South America, Madagascar and
Borneo.
2. We hypothesized that if dispersal limitation is important in structuring communities, then com-
munity similarity should depend on geographical proximity even after controlling for ecological
similarity. Conversely, if communities are assembled primarily through niche processes, then com-
munity similarityshould be determined by ecologicalsimilarity regardlessof geographical proxim-
ity.
3. We performed Mantel and partial Mantel tests to investigate correlations among primate com-
munity similarity, ecological distance and geographical distance. Results showed significant and
strongly negative relationships between diurnal primate community similarity and both ecological
similarityand geographical distance in Madagascar, butsignificantand strongernegative relation-
ships between community similarity and geographical distance in African, South American and
Borneanmetacommunities.
4. We concludethat dispersallimitationisan importantdeterminant of primate communitystruc-
ture and may play a stronger role in shaping the structure of some terrestrial vertebrate communi-
ties than niche differentiation. These patterns are consistent with neutral theory. We recommend
testsoffunctionalequivalencetodeterminetheextenttowhichneutraltheorymayexplainprimate
communitycomposition.
Key-words: coexistence, community assembly, continental comparison, neutral theory, verte-
brate
Introduction
A major goal in community ecology is to identify the mecha-
nisms that govern the assembly and maintenance of ecologi-
cal communities. Current models of metacommunity
dynamics differ chiefly in the relative importance they attri-
bute to two mechanisms hypothesized to structure ecological
communities: (i) dispersal limitation and (ii) ecological niche
differentiation through interspecific competition and species
sorting along ecological gradients (Leibold et al. 2004; Holy-
oak, Leibold & Holt 2005). We investigate the relative roles
ofdispersallimitationandnichedifferentiationinstructuring
resident terrestrial vertebrate communities in four biogeo-
graphical regions through analysis of data on primate com-
munitiesinAfrica,SouthAmerica,MadagascarandBorneo.
An ecological community can be defined as a group of tro-
phically similar species that actually or potentially compete
in a local area for the same or similar resources (Hubbell
2001). Community composition refers to the species found at
a site and can be measured with either presence-absence or
abundance data (Legendre, Borcard & Peres-Neto 2005).
Herein we define communities as the primate species present
at a particular locality. We acknowledge that the full set of
species that compete with primates likely includes non-pri-
mate taxa (Estrada & Coates-Estrada 1985; Ganzhorn 1999;
Marshall, Cannon & Leighton 2009). Nevertheless, it has
generally been argued that primate communities within
regions are composed of ecologically similar species and they
*Correspondenceauthor.E-mail:lhbeaudrot@ucdavis.edu
JournalofAnimalEcology2011,80,332–341 doi:10.1111/j.1365-2656.2010.01777.x
?2010TheAuthors.Journalcompilation?2010BritishEcologicalSociety

Page 2

are frequently viewed as self-contained communities (Fleagle
& Reed 1996; Fleagle, Janson & Reed 1999). Moreover, pri-
mates have been well studied at a large number of sites and
their presence or absence can be reliably established through
surveys. Consequently, more comprehensive community
composition data are available for primates than for any
other mammalian taxon, particularly in the tropics. The data
available on contemporary primate community composition
are simply a snapshot in time of continuously changing com-
munities. As we are unable to track long-term fluctuations in
primate community composition, we must rely on the result-
ing present-day patterns to infer the processes that produced
them.
When addressing biogeographical spatial scales and evolu-
tionary time scales, the term metacommunity describes local
communities linked by the dispersal of multiple potentially
interacting species within the regional species pool (Hubbell
2001). Although several metacommunity models predict that
dispersal limitation will affect community structure (Leibold
et al. 2004), neutral models place particular emphasis on the
importance of dispersal limitation as a key mechanism shap-
ing community composition. According to neutral theory
(Hubbell 2001), the three parameters expected to govern the
community structure are metacommunity size, speciation
rate and dispersal rate. The community composition is
expected to drift stochastically over time because dispersal
and recruitment limitation, in which species fail to reach or
establish in all sites favourable to their survival (Hurtt & Pa-
cala 1995), results in competitively inferior species persisting
in sites when the competitively superior species for the sites is
unabletoreachthem(Hubbell2005).
At biogeographical scales, predictions of neutral theory
assume the contiguity of metacommunities so that all species
have the opportunity to disperse to each site within the meta-
community. For this reason, we investigate primate meta-
communities in areas of generally contiguous tropical forest
cover. We limit our study to areas of tropical forest cover in
order to reduce the potential effects of changes in ecosystems
preventing the dispersal of some species (e.g. savanna ecosys-
tems might prevent exclusively arboreal forest-dwelling spe-
cies from dispersing across open areas). We repeated our
analysis using only sites where the most common species in
each region was found in order to test our hypotheses where
the assumption of metacommunity contiguity is demonstra-
bly valid. The presence of the same taxon at all sites demon-
strates that all sites within the metacommunity have been
accessible to dispersing individuals and therefore provides a
moreconservativetest.
Most research in primate community ecology has
focused on the ecological underpinnings of niche differenti-
ation. Previous research has addressed factors that may dif-
ferentiate niches within communities, such as body size,
activity pattern, diet and canopy use (Cannon & Leighton
1994; Marshall, Cannon & Leighton 2009; Schreier et al.
2009). Several regional compilations report the presence or
absence of primate species in particular habitat types
(Fleagle, Janson & Reed 1999), suggesting that the ecological
factors that determine habitat specialization might influ-
ence primate community composition, but the habitat cate-
gories used are typically course-grained and discussions of
the effects of habitat type on primate community structure
have remained largely descriptive. However, cluster analy-
sis of Neotropical primate communities found that commu-
nities were clustered by both forest type and geographical
region (Peres & Janson 1999), suggesting the importance of
both ecological and spatial factors in structuring primate
communities. Only a handful of additional studies have
quantitatively considered factors affecting primate commu-
nity composition across communities (Peres 1997; Ganz-
horn 1998; Lehman 2006). In this study we are interested
in the extent to which variation in community composition
between sites is related to ecological and spatial gradients
and how the relative importance of these factors varies
between regions. We therefore investigate niche differentia-
tion in the form of species sorting along environmental
gradients. Species sorting along environmental gradients
refers to the idea that if species are most competitive under
resource conditions to which they are best adapted, then
any given community will be composed of the individual
species that are best suited for its particular environmental
conditions.
Themostcomprehensivebiogeographicalstudyofprimate
community structure to date found that both ecological and
spatial factors predict community composition (Kamilar
2009). Kamilar (2009) used canonical correspondence analy-
sis (CCA) to evaluate a small set of environmental character-
istics,longitudeandlatitude. Theeffects ofthesefactorswere
measured at a continental scale, however, and it is unclear to
what extent dispersal between all localities was possible. Our
study is betterable to test the relative importance of dispersal
limitation and niche differentiation by restricting the scale of
the analysis to areas of more or less contiguous forest cover
rather than entire continents. Our analysis therefore adheres
more closely than previous ones to the assumption that
speciescandisperseamongallsiteswithin themetacommuni-
ty (although this is less true for Madagascar) and is therefore
better able to distinguish between dispersal limitation and
environmental effects. We also include a more comprehen-
sive set ofecologicalvariables,including net primary produc-
tivity, additional climatic variables and soil characteristics
(Table 1), thereby increasing the likelihood that we incorpo-
rated a biologically realistic characterization of non-human
primatehabitats.
Herein we present a test of dispersal limitation in resi-
dential (i.e. non-migratory) terrestrial vertebrates with data
on primate communities from Africa, South America,
Madagascar and Borneo (Fig. 1). If dispersal limitation
has been important in structuring these metacommunities,
then community similarity will depend predominantly on
geographical distance: the species composition of commu-
nities that are close together will be more similar than that
of communities that are more geographically distant,
regardless of ecological similarity (Chase et al. 2005). Con-
versely, if communities have been primarily assembled
Dispersallimitationstructuresprimatecommunities333
?2010TheAuthors.Journalcompilation?2010BritishEcologicalSociety,JournalofAnimalEcology,80,332–341

Page 3

through niche processes, then community similarity will be
largely determined by ecological similarity: the species
composition of communities inhabiting areas with similar
ecological conditions will be more similar than those
inhabiting more disparate ecological conditions, irrespec-
tive of geographical distance. Therefore, we test predic-
tions emerging from two alternative hypotheses, following
Chase et al. (2005):
H1: Dispersallimitationdeterminesthestructureofprimate
communities
Negative correlation between community similarity
andgeographicaldistance
No (or weak) correlation between community similar-
ityandecologicaldistance
Niche differentiation through species sorting along
environmental gradients determines the structure of
primatecommunities
No (or weak) correlation between community similar-
ityandgeographicaldistance
Negative correlation between community similarity
and ecologicaldistance
P1a:
P1b:
H2:
P2a:
P2b:
Materialsandmethods
COMMUNITY COMPOSITION DATA
Data for 124 sites acrossAfrica (N = 23), South America (N = 45),
Madagascar (N = 28) and Borneo (N = 28) were collected (Fig. 1).
Primate community composition was assessed by compiling
presence-absence data for species from published sources, following
Groves’s (2001) taxonomy. See Appendix S1 for species and site
information.
COMMUNITY SIMILARITY
The Jaccard community similarity index was calculated using the
Vegdist function from the Vegan community ecology package in R
(R Development Core Team 2009). The Jaccard index can be calcu-
lated with presence-absence or abundance data. For presence-
absence data,the Jaccard index isdefined asJI = j⁄(a + b)j) where
j is the number of species occurring in both sites, a is the number of
species occurring in the first site and b is the number of species occur-
ring in the second site (Magurran 1988). High values of the Jaccard
community similarity index mean that two sites have high commu-
nity similarity and thus a high number of species in common and few
speciesfoundonlyatonesite.TheconverseistrueforlowJIvalues.
GEOGRAPHICAL DISTANCE
Geographical coordinates were collected from the community com-
position site reference when available, or otherwise from the UNEP
and IUCN Worldwide Database on Protected Areas (IUCN-UNEP
2009). Geographical distances were calculated using the Pairdist
Function fromtheSpatstat PackageinR. For all pairsof sites within
a region, Pairdist computes the matrix of Euclidean distances
betweenlatitudeandlongitude.
ECOLOGICAL DISTANCE
To determine ecological distance, 41 ecological variables (Table 1)
were collected for each site from publically available datasets using
ArcGIS.Excludinganthropogeniceffects,itislikelythatprimatedis-
tributions are shaped primarily by the distributions of plant species
Table 1. Ecological variables considered. Variables used in the analysis and their transformations are shown. Variables excluded due to high
correlationvalues(|r| ‡ 0Æ8)aredisplayedonthesamerowastheincludedvariablewithwhichtheywerecorrelated
Code Includedvariables TransformationExcludedvariables
NPP
ELEV
BIO1
Netprimateproductivity1
Elevation2
Annual meantemperature2
None
Log
NoneMaxtemperatureofwarmestmonth2,coldestmonth2,wettest
quarter2, driestquarter2, warmestquarter2,coldestquarter2
Temperatureannualrange2
BlO2 Meandiurnaltemperaturerange[meanof
monthly(max)min)]2
Isothermality[(meandiurnal
range⁄temperatureannualrange)*100)2
Annual precipitation2
Precipitationofwettestmonth2
Precipitationofdriestmonth2
Squareroot
BlO3 SquarerootTemperatureseasonality2(SD*100)
BlO12
BlO13
BlO14
None
Squareroot
Squareroot
Precipitationofwettestquarter2
Precipitationseasonality2(coefficientofvariation);precipitation
ofcoldestquarter2,driestquarter2
BlO18
T.SAND
Precipitationofwarmestquarter2
Topsoilsandfraction3
None
SquarerootSubsoilsandfraction3,siltfraction3,cationexchangecapacity
(soil)3;topsoilsiltfraction3,cationexchangecapacity(soil)3
Subsoilclayfraction3,referencebulkdensity3,topsoilreference
bulkdensity3
Subsoilorganiccarbon3
Topsoilbasesaturation3, subsoilbasesaturation3,pH3
Topsoiltotalexchangeablebases3; subsoiltotalexchangeable
bases3,cationexchangecapacity(clay)3
T.CLAYTopsoilclayfraction3
Log
T.OC
T.PH.H20
T.CEC.CL
Topsoilorganiccarbon’
TopsailpH3
Topsoilcationexchangecapacity(clay)3
Log
Squareroot
Log
Sources:1GlobalLandCoverFacility(Carrollet al.2009),2WorldClimGlobalClimateDatabase(Hijmanset al.2005),
SoilDatabase[FAO⁄IIASA⁄ISRIC⁄IISSCAS⁄JRC(2009)].
3HarmonizedWorld
334L.H.Beaudrot&A.J.Marshall
?2010TheAuthors.Journalcompilation?2010BritishEcologicalSociety,JournalofAnimalEcology,80,332–341

Page 4

on which they feed. Although the debate remains ongoing, there is
growing evidence that tropical plant distributions are largely deter-
mined by soil characteristics and topography (Potts et al. 2002;
Russo et al. 2008). We therefore collected information on a wide
range of ecological variables including net primary productivity,
elevation, temperature, precipitation and soil characteristics in an
attempt to assess factors that affect plant distributions (Franklin
1995). Whentwo variables werehighlycorrelated(|r| > 0Æ8)one was
discardedfromthedataset priortoanalysisbyremovingthevariable
that was believed to be less likely to affect primate ecology (Table 1).
The remaining14 ecologicalvariables were used to calculate a matrix
ofecologicaldistancesbetweeneachpairofprimatecommunitysites.
Asecologicalvalueswerecorrelated,weusedaMahalanobisdistance
calculation that would accommodate this correlation (Seber 1984).
The Mahalanobis distance calculation is best performed with vari-
ables that have normal distributions. As ecological data are rarely
normally distributed, the distributions of ecological variables were
transformed when needed to stabilize their variances and make their
distributions more symmetrical and therefore more normally distrib-
uted (Table 1). These analyses were conducted in R 2.8.1 (R Devel-
opmentCoreTeam2009).
Ecologicalvariationwithinandamongregionswascomparedwith
a manova using the 14 transformed ecological variables as response
variables to the four regions. We tested for significant differences
between all regions using six contrasts and a Bonferroni correction
for the a = 0Æ05 level (P = 0Æ008). Ecological variation among
regions was illustrated using a canonical centroid plot and within
regionsbyplotting99%densityellipsesforeachregiononabivariate
plot of canonical axes 1 and 2. These analyses were conducted in
JMP8.0.1(SASInstituteInc.,Cary,NC).
ANALYSES
There has been substantial academic debate over the appropriate
statistical methods, particularly Mantel tests and CCA, for testing
the relative importance of spatial and ecological drivers of commu-
nitycomposition.Thecrucialdistinctionreachedinthisdebateisthat
CCAisappropriateforquestionsconcerningvariationincommunity
composition, also known as beta diversity, whereas Mantel tests are
appropriate for questions concerning variation in beta diversity
between groups of sites (Legendre, Borcard & Peres-Neto 2005,
2008; Tuomisto & Ruokolainen 2006, 2008). As we are interested in
the factors driving the variation in community composition across
regions (i.e. dispersal limitation and species sorting along environ-
mental gradients), we are interested in comparing the variation in
beta diversity across sites; Mantel tests are therefore an appropriate
statistical tool. It is important to note, however, that a major differ-
ence between Mantel tests and CCA is that when distances are calcu-
lated in the Mantel approach, the species identities, the actual values
of the geographical coordinates of the sites and the actual values of
the environmental values at the sites are no longer compared, but
whatiscompared isthemagnitudeofthedifferencesbetween themin
Elevation
High
Low
(a)(b)
(c)(d)
Fig. 1. Locations of primate communities. Map of primate community sites included in this study (N = 124) in (a) Africa (N = 23) (b) South
America (N = 45) (c) Madagascar (N = 28) and d) Borneo (N = 28). Shading indicates elevation with high elevations displayed in dark grey.
Squares indicate communities in which the most commonly found regional species was absent. There were no species found at all sites within
any of the metacommunities. It is unlikely that this pattern is an artefact of the spatial scale at which we conducted this analysis because sites
lacking the most common taxon in the regional metacommunity were not clustered together spatially, although this is less true for Africa where
community composition is likely influenced by the Congo river. After conducting the analysis with all sites, we repeated our analysis using only
siteswherethemostcommonspeciesineachregionwasfound inordertotestourhypotheseswheretheassumptionofmetacommunitycontigu-
ity would be demonstrably valid. The presence of the same taxon at all sites demonstrates that all sites within the metacommunity have been
accessibletodispersingindividualsandthereforeprovidesamoreconservativetestofthedispersallimitationhypothesis.
Dispersallimitationstructuresprimatecommunities335
?2010TheAuthors.Journalcompilation?2010BritishEcologicalSociety,JournalofAnimalEcology,80,332–341

Page 5

theformof‘distances’. CCA,ontheotherhand,quantifiesthe effects
of the specific location on individual species at a site and does not
quantify the effect of distances between sites on variation in commu-
nitycomposition(Tuomisto&Ruokolainen2006).
SimpleMantel testswerethereforeperformedtoinvestigatebivari-
ate correlations of communitysimilaritywith ecological distance and
geographical distance. Partial Mantel tests (Smouse, Long & Sokal
1986) were employed to investigate partial relationships between
these variables. These tests were performed in R using the ‘mantel’
and ‘mantel.partial’ commands, the Pearson method and 10 000 per-
mutations. A similarity matrix was used for community composition
and dissimilarity matrices were used for ecological and geographical
distance.TheRcodefortheseanalysiscanbefoundinAppendix S2.
Added variable plots (Weisberg 1985) were constructed to depict
the correlation values of the partial Mantel tests. To illustrate the
influence of ecological distance on primate community similarity
while controlling for geographical distance, diurnal community simi-
larity was regressed against geographical distance, and ‘community
similarityresiduals’werecalculatedbasedontheregressionline.Eco-
logical distance was regressed against geographical distance, and the
resulting ‘ecological distance residuals’ were calculated. We then
plotted the community residuals against the ecological residuals and
calculated the corresponding regression line for each metacommunity.
Similarly, the effects of ecological distance were partialled out to
illustrate the effects of geographical distance on community similar-
ity. The correlation coefficient for the points in each added variable
plot is equalto the partialcorrelation coefficient having corrected for
the partialled out predictor. The slope of the regression line through
each added variable plot is equal to the coefficient that the respective
predictorwouldhaveinamultipleregressionmodelthatincludedthe
otherpredictor.
Results
COMMUNITY COMPOSITION
There wasconsiderable variation in the totalnumber of diur-
nal species, nocturnal species and genera between the regio-
nalspeciespools(Table 2).Therewerenospeciesfoundatall
sites within any of the metacommunities. It is unlikely that
this pattern is an artefact of the spatial scale at which we con-
ducted this analysis because sites lacking the most common
taxon in the regional metacommunity were not clustered
together spatially (Fig. 1),although this is less true for Africa
where communities are likely affected by the Congo river.
The species most frequently present in each region were Pan
troglodytes and Perodicticus potto present in 70% (16⁄23) of
African sites, Cebus apella present in 93% (42⁄45) of South
American sites, Eulemur fulvus present in 86% (24⁄28) of
Malagasy sites and Hylobates muelleri present in 79% of
(22⁄28) Bornean sites. We list site information and species
foundineachregioninAppendix S1.
ECOLOGICAL DISTANCE
There were significant differences in ecology among all
regions (manova whole model Wilks’ Lambda: DFnum= 42,
DenDF = 318Æ18, P < 0Æ0001; Region contrasts: N = 124,
NumDF = 14, DFden= 107, P < 0Æ0001, Exact F: Mada-
gascar – Borneo = 55Æ95, Madagascar – Africa = 40Æ92,
Madagascar – South America = 39Æ70, South America –
Borneo = 15Æ66, South America – Africa = 13Æ05, Borneo –
Africa = 6Æ23). Madagascar is differentiated from the other
three regions by the first canonical axis, which is largely
explainedbyelevationandprecipitationinthewarmestquar-
ter. South America is differentiated from the other three
regions by the second canonical axis, which is largely
explained by the topsoil properties pH, sand fraction and per
cent organic carbon. AfricaandBorneo are the most ecologi-
cally similar regions (Fig. 2a). Madagascar’s larger density
ellipse illustrates its greater within-region variation than the
other regions (Fig. 2b). Regional summary statistics (min-
ima, maxima and medians) for the 14 ecological variables
includedintheanalysesareprovidedinTable S1.
MANTEL TESTS
The results of the simple Mantel tests show that geographical
distance was a stronger predictor of community similarity
than ecological distance in diurnal primate communities in
Africa, South America and Borneo, but not in Madagascar
(Table 3). Thus, sites that are geographically farther from
each other have fewer species in common than sites that are
geographicallyclosertoeachother.Theseresultssuggestthat
dispersal limitation affects primate community composition
more than niche differentiation affects primate community
composition in Africa, South America and Borneo, but not
in Madagascar. We found the same pattern when we per-
formedtheanalysesusingallspecies.
When we performed the simple Mantel tests using gen-
era, we found that geographical distance was a stronger
predictor of community similarity in Africa and South
America, but not in Madagascar or Borneo. These results
suggest that dispersal limitation is a stronger predictor of
primate community composition at the generic level in
Africa and South America, but not in Madagascar or Bor-
neo. Contrary to predictions of both hypotheses, commu-
nity similarity in Borneo was more strongly predicted by a
positive relationship with ecological distance. This result
suggests that primate communities in Borneo at the genus
level are more similar the more dissimilar the environ-
ments are.
The partial Mantel results were qualitatively the same as
the simple Mantel results (Table 3). Importantly, these pat-
terns remained consistent once the potentially confounding
effects of spatial and environmental autocorrelation were
removed. Geographical distance was a stronger predictor
of community similarity than ecological distance in diurnal
primate communities in Africa, South America and Bor-
neo, but not in Madagascar (Table 3). These results indi-
Table 2. Summarydataforregionalspeciespools
TaxaAfricaSouthAmericaMadagascarBorneo
Diurnalspecies
Nocturnalspecies
Genera
35
9
17
28
3
13
13
16
14
11
2
8
336L.H.Beaudrot&A.J.Marshall
?2010TheAuthors.Journalcompilation?2010BritishEcologicalSociety,JournalofAnimalEcology,80,332–341

Page 6

cate that sites that are geographically farther from each
other have fewer species in common than sites that are geo-
graphically closer to each other even once differences in
ecology are removed. These results suggest that dispersal
limitation affects primate community composition more
than niche differentiation affects primate community com-
position in Africa, South America and Borneo, but not in
Madagascar. We illustrated the partial Mantel results for
diurnal primate communities using added variable plots
(Fig. 3).
The results of the partial Mantel tests differed from the
simple Mantel tests in two ways. First, the correlation values
of the simple tests were consistently higher. This indicates
that removing the spatial and environmental autocorrelation
with the partial Mantel tests weakened the relationships
betweenprimatecommunitysimilarityandecologicaldistance
and between primate community similarity and geographical
distance. Secondly, the correlation between community
similarity and ecological distance was non-significant for
South America after partialling out geographical distance.
(a)(b)
Fig. 2. Ecological variation between and within regions. (a) Canonical centroid plot depicting ecological variation between regions. Regional
centroids are indicated by the · marks within ovals. The biplot rays show the direction and strength of the ecological variables responsible for
thecanonical axes.Variablecodescorrespondtovariablesdescribed inTable 1.Madagascarisdifferentiatedfromtheotherthreeregionsbythe
first canonical axis, which is largely explained by elevation and precipitation in the warmest quarter. South America is differentiated from the
other three regions by the second canonical axis, which is largely explained by the topsoil properties pH, sand fraction and per cent organic car-
bon.AfricaandBorneoarethemostecologicallysimilarregions.(b)Ecologicalvariationwithinregionsasshownby99%densityellipses.Mala-
gasysitesare representedby bluetriangles.South Americansitesare representedby red diamonds. Africansitesare represented byblackcircles.
Bornean sites are represented by green asterisks. Madagascar’s larger density ellipse illustrates its greater within-region variation than the other
regions, which is one possible explanation for why the niche differentiation hypothesis of species sorting along environmental gradients is
stronglysupportedinMadagascar, butnotintheotherthreeregions.
Table 3. SimpleandpartialManteltestsofprimatecommunitysimilarityagainstecologicalandgeographicaldistance
DistanceCommunity
AfricaSouthAmerica MadagascarBorneo
rPrPrPrP
GeographicDiurnalspecies
Allspecies
Genera
Diurnalspecies
Allspecies
Genera
Diurnalspecies
Allspecies
Genera
Diurnalspecies
Allspecies
Genera
)0Æ64
)0Æ65
)0Æ55
)0Æ26
)0Æ30
)0Æ43
)0Æ61
)0Æ61
)0Æ48
)0Æ07
)0Æ13
)0Æ32
<0Æ00l
<0Æ001
<0Æ00l
<0Æ001
<0Æ001
<0Æ001
<0Æ001
<0Æ001
<0Æ00l
0Æ032
0Æ025
0Æ002
)0Æ37
)0Æ36
)0Æ25
)0Æ19
)0Æ21
)0Æ20
)0Æ33
)0Æ30
)0Æ17
+0Æ01
)0Æ02
)0Æ08
<0Æ00l
<0Æ001
0Æ007
0Æ025
0Æ014
0Æ043
<0Æ001
<0Æ001
0Æ029
0Æ533
0Æ37
0Æ232
)0Æ31
)0Æ28
)0Æ27
)0Æ33
)0Æ43
)0Æ32
)0Æ25
)0Æ22
)0Æ22
)0Æ28
)0Æ39
)0Æ27
<0Æ001
<0Æ001
<0Æ001
<0Æ001
<0Æ001
<0Æ001
<0Æ001
0Æ004
0Æ002
<0Æ001
<0Æ001
0Æ005
)0Æ29
)0Æ22
)0Æ04
)0Æ11
)0Æ06
+0Æ07
)0Æ27
)0Æ22
+0Æ01
<)0Æ01
+0Æ04
+0Æ06
0Æ001
0Æ007
0Æ64
0Æ152
0Æ306
0Æ724
<0Æ00l
0Æ002
0Æ543
0Æ524
0Æ638
0Æ709
Ecological
Geographical(Ecological
partialledout)
Ecological(Geographical
partialledout)
Correlationvalues(r)andsignificancevalues(P)aregivenforeachtestforprimatecommunitiesinAfrica,SouthAmerica,Madagascarand
Borneo.Communitiesweredefinedusingdiurnalspecies,allspeciesorgenera.Resultssignificantatthea £ 0Æ05levelareindicatedinbold.The
strongerpredictor(i.e.greaterabsolutecorrelationvalueandequalorsmallerP-value)isshadedingreyforeachcomparisonbetweengeograph-
icalandecologicalresultsfromeachsimpleandpartialtestineachregionateachcommunitylevel.Of24comparisons,16supporthypothesis1
andrejecthypothesis2andnocomparisonssupporthypothesis2fully,suggestingthatneutral mechanismsareimportantdeterminantsofpri-
matecommunitystructureandmayplayastrongerroleinshapingprimatecommunitystructurethannichemechanisms.
Dispersallimitationstructuresprimatecommunities337
?2010TheAuthors.Journalcompilation?2010BritishEcologicalSociety,JournalofAnimalEcology,80,332–341

Page 7

Nevertheless, despite removing the effects of the spatial
autocorrelation, Africa and Madagascar maintained signifi-
cant relationships between primate community similarity and
ecological distance. These results suggest that niche differen-
tiation may significantly affect primate community structure
in Africa and Madagascar, although the strength of the
(a) (e)
(b)(f)
(c)(g)
(d) (h)
Fig. 3. AddedvariableplotsdepictingcorrelationvaluesofpartialManteltestsfordiurnal primatecommunities.Weplottedcommunityresidu-
als against ecological residuals for (a) Africa (b) South America (c) Madagascar and (d) Borneo. We plotted community residuals against geo-
graphical residuals for (e) Africa (f) South America (g) Madagascar and (h) Borneo. When a = 0Æ05, ecological distance is only significantly
correlated with diurnal community similarity in Africa (r = 0Æ03) and Madagascar but geographical distance is significantly negatively corre-
lated withdiurnal communitysimilarityin metacommunities fromall regions (Table 3). Theseresults supportthe hypothesis thatdispersallimi-
tation structures primate communities in all regions. For Madagascar, and very weakly for Africa, there is also support for the hypothesis that
nichedifferentiationthroughspeciessortingalongenvironmentalgradientsstructuresprimatecommunities.
338L.H.Beaudrot&A.J.Marshall
?2010TheAuthors.Journalcompilation?2010BritishEcologicalSociety,JournalofAnimalEcology,80,332–341

Page 8

relationshipinAfrica isvery low(r = 0Æ07). Given the recent
statisticaldebateoverthemeritsofstatisticalsignificanceand
P-values (Hurlbert & Lombardi 2009), we believe that the
effect sizes or in this case the correlation values are better
indicators of the results than the significance values. There-
fore, although there is a significant negative relationship
between community similarity and ecological distance in
Africa,thesmalleffectsizesuggeststhatthebiologicalsignifi-
canceoftherelationshipisminor.
Model summaries for the multiple regression models for
communitysimilaritywithecologicaldistanceandgeographi-
cal distance as predictors are Africa, N = 253, R2= 0Æ41;
South America,N = 990,
N = 378, R2= 0Æ17; Borneo, N = 378, R2= 0Æ08. The
numberofobservationsinthepairwisedistancematrix(N)for
each metacommunity is N = n*(n)1)⁄2 where n is equal to
the number of primate community sites per metacommunity.
The R2values can be interpreted as the percent of variance in
communitysimilarityexplainedbythetwodistancemeasures.
WethereforewereabletoexplainthemostvarianceforAfrica
(41%), intermediate levels for South America (14%) and
Madagascar(17%)andtheleastvarianceforBorneo(8%).
To assess the extent to which our choice of community
similarity index affected the results of the Mantel tests, we
repeated our analyses using an alternative similarity index,
the Sørensen Index. Similarly, to assess the extent to which
our choice of ecological variables affected the results of the
Mantel tests, we repeated our analyses using a set of 19 vari-
ables from the overall database of 41 variables.In both cases,
additional analyses produced qualitatively similar results in
that the relative importance of dispersal limitation and niche
differentiation remained unchanged in each of the regions
(Table S2).
To evaluate whether individual ecological variables
become buried in the ecological distance matrix, we exam-
ined scatter plots of ecological dissimilarities between pairs
of sites vs. raw differences in ecological variables (Appen-
dix S3). Theplotsshow that as pairwisedifferencesin a given
ecological variable deviate from zero (positively or nega-
tively), the corresponding Mahalanobis distances between
pairs of sites tend to increase. Large Mahalanobis distances
occurring near the zero value of the horizontal axis are the
result of differences between sites with respect to other eco-
logicalvariables.Asthe plotsshowthat when pairwisediffer-
ences for each ecological variable are large, the Mahalanobis
distance value is also large, they therefore show that the
effects of individual ecological variables are not buried in the
ecologicalmatrix.
R2= 0Æ14;Madagascar,
Discussion
We found significant and stronger negative relationships
between community similarity and geographical distance
than between community similarity and ecological distance
in Africa, South America and Borneo. In Madagascar, we
found significant negative relationships between community
similarity and geographical distance as well as community
similarity and ecological distance. These results support both
predictions from H1 for Africa, South America and Borneo
and fail to support both predictions from H2 for Madagas-
car. The same pattern occurred when we included nocturnal
species and to a lesser extent when we conducted the tests at
the genericlevel. Overall,16 of24 comparisonsare consistent
with H1 but not H2, and no comparison fully supports H2
(Table 3). These results suggest that dispersal limitation is an
important determinant of primate community structure and
may play a stronger role in shaping primate community
structure than niche mechanisms, such as species sorting
alongenvironmentalgradients.
Although we detected patterns consistent with dispersal
limitation across regions, dispersal limitation explains only a
subset of the variance (Fig. 3). A more refined model would
include relative species abundances, speciation rates, species-
specific maximum dispersal distances, the influence of
geographical barriers on dispersal and indices of human dis-
turbance. Our analyses do not explicitly address phylogeog-
raphy, historical climates, primate diets, non-primate
competitors, or human impacts. Detailed modelling that
parses out the relative influence of such factors would pro-
vide a better understanding of primate community assembly;
however, a lack ofsufficientdataacross species and sites con-
strainedouranalysis.
We also caution that without experimental evidence it is
possible that mechanisms other than dispersal limitation
couldhave produced similar patterns(for review, see Alonso,
Etienne & McKane 2006). In addition, it has previously been
pointedoutthatfailuretofindcorrelationswithenvironmen-
tal variables may simply reflect failure to incorporate the
appropriate environmental variables (Bell, Lechowicz &
Waterway 2006). Moreover, relevant variables may include
biotic rather than abiotic variables. Lastly, we recognize that
ecological factors may vary in their importance for primates
among regions. We used the same ecological distance matrix
across regions so that results would be comparable, but
future tests within regions would benefit from model selec-
tionapproaches.
Although primates are highly mobile and capable of trav-
elling substantial distances, many species restrict their activi-
ties to areas of well-known forest for efficient exploitation
(Fleagle 1999). Moreover, primate dispersal is costly for rea-
sons such as exposure to higher predationrisk, loss of knowl-
edge about local food resources and loss of benefits of
remaining near kin (Isbell & van Vuren 1996). Colonization
of new areas might impose similar costs. Dispersal and
recruitment limitation may therefore influence primate com-
munity assembly, as has been found for sessile taxa (Hubbell
1997).
We detected a stronger signal of dispersal limitation when
nocturnal primate species were excluded from the partial
Mantel analyses. This difference may be due to sampling
biases against detecting cryptic nocturnal species. Alterna-
tively, solitary nocturnal species may experience lower dis-
persal costs and therefore reduced dispersal limitation in
comparison with group living diurnal species. For example,
Dispersallimitationstructuresprimatecommunities339
?2010TheAuthors.Journalcompilation?2010BritishEcologicalSociety,JournalofAnimalEcology,80,332–341

Page 9

their anti-predator strategies involve crypsis and solitary liv-
ing, which may cause them to be less dependent than gregari-
ous diurnal taxa on dispersing with other individuals.
However, despite the large difference in the proportion of
nocturnal primates (Table 2), we do not see a response in the
dispersal limitation signal that is proportional to the number
ofnocturnalprimatesacrossregions.
Like the other three regions, Malagasy community struc-
ture was strongly predicted by geographical distance, which
suggests that dispersal limitation is an important factor.
Unlike the other three regions, however, ecological distance
was a stronger predictor. Interestingly, a recent study of
primate community phylogenetics found that a majority of
primate communities from Africa, South East Asia and the
Neotropics exhibit random phylogenetic structure, but that
Malagasy communities tend to be overdispersed, such that
species within these communities are less closely related to
each other than is expected by chance (Kamilar & Guidi
2010).Thesefindingsare consistentwith the ideathatcompe-
tition between closely related species in the past may have
resulted in their competitive exclusion (Webb et al. 2002).
The results of our study as well as the community phyloge-
netics study (Kamilar & Guidi 2010) suggest that competi-
tion may be a more important mechanism structuring
primate communities in Madagascar than in other regions.
One possible explanation is that primates in Madagascar
represent a larger portion of the appropriate vertebrate
competitor community and therefore the effects of niche
differentiation were more readily detected. In Madagascar,
primates compose 44% of the nonvolant terrestrial mam-
mals, whereas in other regions primates compose 8–12%
(Jernvall & Wright 1998). If non-primate competitors are
more important than primate competitors, the effects of
niche differentiation in other regions may be masked by their
exclusion. Alternatively, the stronger ecological result in
Madagascar may be attributable to greater ecological varia-
tionwithintheregion(Fig. 2b).
Geographical barriers, such as mountain ranges, rivers
and forest discontinuities may have affected primate dis-
persal. It is unlikely that mountain ranges posed major geo-
graphical barriers for these sites (Fig. 1), but major rivers are
found in all four regions. Although the role of rivers as dis-
persal barriers to primates has received support (Ayres &
Clutton-Brock1992;Jalilet al.2008),othershavequestioned
the extent to which rivers are complete barriers (Oates 1988;
Gascon et al. 2000). The presence of a negative non-linear
relationship between African primate community similarity
and geographical distance (Fig. 3e) may be attributable to
the Congo River, which serves as a barrier to the distribution
of some primates. Primate communities to the east and west
of the Congo Basin may contain species that have dispersed
without crossing the Congo River. Past or present forest
cover discontinuities may also have hindered primate dis-
persal (Grubb 1982), but riverine forests may have func-
tioned as dispersal corridors, particularly in Africa and
Madagascar (Colyn, Gautier-Hion & Verheyen 1991; Ganz-
horn et al. 2006). Although geographical barriers may be the
reason why no species was found at all sites within a region,
the high proportion of sites occupied by the most frequently
found species per region suggests that most sites within a
region could potentially be reached. Moreover, when we
repeated analyses using only sites at which the most common
regional species was present, we found the same qualitative
results in that the relative importance of dispersal limitation
and niche differentiation remained unchanged in each of the
regions (Table S2). These findings suggestthat dispersal limi-
tation has shaped primate community structure beyond the
effects of geographical barriers, although detailed modelling
of the effects of geographical barriers on species distributions
would provide further insight into mechanisms of primate
dispersallimitation.
Neutral models have been applied to vertebrate communi-
ties surprisingly rarely and the few studies to date have pro-
duced mixed results (McGill 2003; He 2005; Ostling 2005;
Muneepeerakul et al. 2008). Our findings are consistent with
neutral theory and warrant further exploration. Hubbell’s
(2001)neutral modelconsidersallmembers ofatrophiccom-
munity to be functionally equivalent. Testing the validity of
the functional equivalencehypothesis will be central to deter-
mining the extent to which neutral theory may explain pri-
matecommunitycomposition.
Acknowledgements
We are especially thankful to M. Grote for help on statistical analyses and to
M. Holyoak for comments that greatly improved the article. We thank J.
Chase and S. Hubbell for encouragement; N. Beaudrot for assistance with R
code; T. Weaver for a recommendation on statistical methods; J. Thorne for
suggestingadatasource;A.HarcourtforaccesstoGISresourcesandreference
suggestions; C. Cannon, M. Grote, A. Ostling, J. Sweeney, C. Webb and sev-
eral anonymous reviewers for feedback on earlier versions of the manuscript;
C. Boyko, R. Boyko, L. Camp, K. Feilen, S. Gravem, K. Kroetz, M. Mauritz,
A. Neyer and A. Stephenson for discussion. This work was supported by a
NationalScienceFoundationGraduateResearchFellowshiptoL.H.B.
References
Alonso, D., Etienne, R.S. & McKane, A.J. (2006) The merits of neutral theory.
TrendsinEcology&Evolution,21,451–457.
Ayres, J.M. & Clutton-Brock, T.H. (1992) River boundaries and species range
sizeinAmazonianprimates.AmericanNaturalist,140,531–537.
Bell, G., Lechowicz, M.J. & Waterway, M.J. (2006) The comparative evidence
relating to functional and neutral interpretations of biological communities.
Ecology,87,1378–1386.
Cannon, C.H. & Leighton, M. (1994) Comparative locomotor ecology of gib-
bons and macaques: selection of canopy elements for crossing gaps. Ameri-
canJournalofPhysicalAnthropology,93,505–524.
Chase, J.M., Amarasekare, P., Cottenie, K., Gonzalesz, A., Holt, R.D., Holy-
oak, M., Hoopes, M.F., Leibold, M.A., Loreau, M., Mouquet, N., Shurin,
J.B.&Tilman,D.(2005)Competingtheoriesforcompetitivemetacommuni-
ties. Metacommunities: Spatial Dynamics and Ecological Communities (eds
M. Holyoak, M.A. Leibold & R.D. Holt), pp. 334–354. University of Chi-
cagoPress,Chicago.
Colyn, M., Gautier-Hion, A. & Verheyen, W. (1991) A reappraisal of paleoen-
vironmental history in central Africa: evidence for a major fluvial refuge in
theZairebasin.JournalofBiogeography,18,403–407.
Estrada, A. & Coates-Estrada, R. (1985) A preliminary study of resource over-
lap between howling monkeys (Alouatta palliata) and other arboreal mam-
mals in the tropical rainforest of Los Tuxtlas, Mexico. American Journal of
Primatology,9,27–37.
Fleagle, J.G. (1999) Chapter 3: Primate Lives. Primate Adaptation and Evolu-
tion.AcademicPress,London.
340L.H.Beaudrot&A.J.Marshall
?2010TheAuthors.Journalcompilation?2010BritishEcologicalSociety,JournalofAnimalEcology,80,332–341

Page 10

Fleagle, J.G., Janson, C.H. & Reed, K.E. (1999) Primate Communities. Cam-
bridgeUniversityPress,NewYork.
Fleagle, J.G. & Reed, K.E. (1996) Comparing primate communities: a multi-
variateapproach.JournalofHumanEvolution,30,489–510.
Franklin, J. (1995) Predictive vegetation mapping: geographic modelling of
biospatial patterns in relation to environmental gradients.Progress in Physi-
calGeography,19,474–499.
Ganzhorn, J.U. (1998) Nested patterns of species composition and their impli-
cations for lemur biogeography in Madagascar. Folia Primatologica, 69,
332–341.
Ganzhorn, J.U. (1999) Body mass, competition and the structure of primate
communities. Primate Communities (eds J.G. Fleagle, C. Janson & K.E.
Reed),pp.141–157.CambridgeUniversityPress,Cambridge.
Ganzhorn,J.U.,Goodman, S.M., Nash,S.& Thalmann,U. (2006) Lemurbio-
geography. Primate Biogeography (eds S.M. Lehman & J.G. Fleagle), pp.
229–254.SpringerScience,NewYork.
Gascon, C., Malcolm, J.R., Patton, J.L., da Silva, M.N.F., Bogart, J.P., Loug-
heed, S.C.,Peres,C.A., Neckel, S. & Boag, P.T. (2000) Riverinebarriers and
the geographic distribution of Amazonian species. Proceedings of the
National Academy of Sciences of the United States of America, 97, 13672–
13677.
Grubb,P.(1982)RefugesanddispersalinthespeciationofAfricanforestmam-
mals. Biological Diversification inthe Tropics(ed. G.T. Prance), pp.537–553.
ColumbiaUniversityPress,NewYork.
He, F. (2005)Derivinganeutralmodelofspeciesabundancefromfundamental
mechanismsofpopulationdynamics.FunctionalEcology,19,187–193.
Holyoak, M., Leibold, M.A. & Holt, R.D. (2005) Metacommunities: Spatial
DynamicsandEcologicalCommunities.UniversityofChicago,Chicago.
Hubbell, S.P. (1997) A unified theory of biogeography and relative species
abundance and its application to tropical rain forests and coral reefs. Coral
Reefs,16,S9–S21.
Hubbell, S.P. (2001) The Unified Neutral Theory of Biodiversity and Biogeogra-
phy.PrincetonUniversityPress,Princeton,NJ.
Hubbell, S.P. (2005) Neutral theory in community ecology and the hypothesis
offunctionalequivalence.FunctionalEcology,19,166–172.
Hurlbert, S.H. & Lombardi, C.M. (2009) Final collapse of the Neyman-Pear-
son decision theoretic framework and rise of theneoFisherian.Annales Zoo-
logiciFennici,46,311–349.
Hurtt, G.C. & Pacala, S.W. (1995) The consequences of recruitment limitation
– reconciling chance, history and competitive differences between plants.
JournalofTheoreticalBiology,176,1–12.
Isbell, L.A. & van Vuren, D. (1996) Differential costs of locational and social
dispersal and their consequences for female group-living primates. Behav-
iour,133,1–36.
IUCN-UNEP (2009) World Database on Protected Areas (WDPA). UNEP-
WCMC,Cambridge,UK.
Jalil, M.F., Cable, J., Inyor, J.S., Lackman-Ancrenaz, I., Ancrenaz, M., Bru-
ford, M.W. & Goossens, B. (2008) Riverine effects on mitochondrial struc-
ture of Bornean orang-utans (Pongo pygmaeus) at two spatial scales.
MolecularEcology,17,2898–2909.
Jernvall, J. & Wright, P.C. (1998) Diversity components of impending primate
extinctions. Proceedings of the National Academy of Sciences of the United
StatesofAmerica,95,11279–11283.
Kamilar, J.M. (2009) Environmental and geographic correlates of the taxo-
nomic structure of primate communities. American Journal of Physical
Anthropology,139,382–393.
Kamilar, J.M. & Guidi, L.M. (2010) The phylogenetic structure of primate
communities: variation within and across continents. Journal of Biogeogra-
phy,37,801–813.
Legendre, P., Borcard, D. & Peres-Neto, P.R. (2005) Analyzing beta diversity:
partitioningthespatialvariationofcommunitycompositiondata.Ecological
Monographs,75,435–450.
Legendre, P., Borcard, D. & Peres-Neto, P.R. (2008) Analyzing or explaining
betadiversity?Comment.Ecology,89,3238–3244.
Lehman,S.M.(2006)Nesteddistributionpatternsandthehistoricalbiogeogra-
phy of the primates of Guyana. Primate Biogeography (eds S.M. Lehman &
J.G.Fleagle),pp.63–80.Springer,NewYork.
Leibold, M.A., Holyoak, M., Mouquet, N., Amarasekare, P., Chase, J.M.,
Hoopes, M.F., Holt, R.D., Shurin, J.B., Law, R., Tilman, D., Loreau, M. &
Gonzalez, A. (2004) The metacommunity concept: a framework for multi-
scalecommunityecology.EcologyLetters,7,601–613.
Magurran, A.E. (1988) Ecological Diversity and its Measurements. Princeton
UniversityPress,Princeton.
Marshall, A.J., Cannon, C.H. & Leighton, M. (2009) Competition and niche
overlap between gibbons (Hylobates albibarbis) and other frugivorous verte-
brates in Gunung Palung National Park, West Kalimantan, Indonesia. The
Gibbons: New Perspectives on Small Ape Socioecology and Population Biol-
ogy (eds S. Lappan & D.J. Whittaker), pp. 161–188. University of Chicago,
Chicago.
McGill, B.J. (2003) A test of the unified neutral theory of biodiversity. Nature,
422,881–885.
Muneepeerakul, R., Bertuzzo, E., Lynch, H.J., Fagan, W.F., Rinaldo, A. &
Rodriguez-Iturbe, I. (2008) Neutral metacommunity models predict fish
diversitypatternsinMississippi-Missouribasin.Nature,453,220-U9.
Oates, J.F. (1988) The distribution of Cercopithecus monkeys in west African
forests. A Primate Radiation: Evolutionary Biology of the African Guenons
(eds A. Gautier-Hion, F. Bouliere & J.P. Gautier), pp. 79–103. Cambridge
UniversityPress,Cambridge.
Ostling,A.(2005)Neutraltheorytestedbybirds.Nature,436,635–636.
Peres, C.A. (1997) Primate community structure at twenty western Amazonian
floodedandunfloodedforests.JournalofTropicalEcology,13,381–405.
Peres, C.A. & Janson, C. (1999) Species coexistence, distribution and environ-
mental determinants of neotropical primate richness: a community-level
zoogeographicanalysis.PrimateCommunities(edsJ.G.Fleagle, C.Janson&
K.E.Reed),pp.55–74.CambridgeUniversityPress,Cambridge.
Potts, M.D., Ashton, P.S., Kaufman, L.S. & Plotkin, J.B. (2002) Habitat pat-
terns in tropical rain forests: a comparison of105 plotsin northwest Borneo.
Ecology,83,2782–2797.
RDevelopmentCoreTeam (2009)R:A LanguageandEnvironmentfor Statisti-
calComputing.RFoundationforStatisticalComputing,Vienna,Austria.
Russo, S.E.,Brown, P.,Tan, S.&Davies,S.J.(2008)Interspecificdemographic
trade-offs and soil-related habitat associations of tree species along resource
gradients.JournalofEcology,96,192–203.
Schreier, B.M., Harcourt, A.H., Coppeto, S.A. & Somi, M.F. (2009) Inter-
specific competition and niche separation in primates: a global analysis.Bio-
tropica,41,283–291.
Seber, G.A.F. (1984) Multivariate Observations. John Wiley and Sons, New
York.
Smouse,P.E.,Long,J.C.&Sokal,R.R.(1986)Multipleregressionandcorrela-
tion extensions of the mantel test of matrix correspondence. Systematic
Zoology,35,627–632.
Tuomisto, H. & Ruokolainen, K. (2006) Analyzing or explaining beta diver-
sity? Understanding the targets of different methods of analysis Ecology, 87,
2697–2708.
Tuomisto, H. & Ruokolainen, K. (2008) Analyzing or explaining beta diver-
sity?Reply.Ecology,89,3244–3256.
Webb, C.O., Ackerly, D.D., McPeek, M.A. & Donoghue, M.J. (2002) Phylo-
genies and community ecology. Annual Review of Ecology and Systematics,
33,475–505.
Weisberg,S.(1985)AppliedLinearRegression.JohnWileyandSons,NewYork.
Received8June2010;accepted21October2010
HandlingEditor:JamesH.Jones
SupportingInformation
Additional Supporting Information may be found in the online ver-
sionofthisarticle.
Appendix S1. Site information and regional species pools for all
regions.
AppendixS2.Rcodeforstatisticalanalyses.
Appendix S3. Effects of individual ecological variables on ecological
distancematrixcalculatedusingMahalanobisdistance.
Table S1. Summary statistics for ecological variables used in the
analyses.
Table S2. Mantel results using (a) Jaccard Index and 19 ecological
variables (b) Sørensen Index and 14 ecological variables (c) sites at
whichthemostcommonregionalspeciesisfound.
As a service to our authors and readers, this journal provides
supporting information supplied by the authors. Such materials may
be re-organized for online delivery, but are not copy-edited or type-
set. Technical support issues arising from supporting information
(otherthanmissingfiles)shouldbeaddressedtotheauthors.
Dispersallimitationstructuresprimatecommunities341
?2010TheAuthors.Journalcompilation?2010BritishEcologicalSociety,JournalofAnimalEcology,80,332–341