LETTER Ecological correlates of the spatial co-occurrence of sympatric
mammalian carnivores worldwide
Courtney L. Davis,
Lindsey N. Rich,
Zach J. Farris,
Marcella J. Kelly,
Mario S. Di Bitetti,
Yamil Di Blanco,
Mohammad S. Farhadinia,
Bart J. Harmsen,
Mamadou D. Kane,
Asia J. Murphy,
Jody M. Tucker,
Febri A. Widodo,
Nigel G. Yoccoz
David A.W. Miller
The composition of local mammalian carnivore communities has far-reaching effects on terrestrial
ecosystems worldwide. To better understand how carnivore communities are structured, we anal-
ysed camera trap data for 108 087 trap days across 12 countries spanning ﬁve continents. We esti-
mate local probabilities of co-occurrence among 768 species pairs from the order Carnivora and
evaluate how shared ecological traits correlate with probabilities of co-occurrence. Within individ-
ual study areas, species pairs co-occurred more frequently than expected at random. Co-occur-
rence probabilities were greatest for species pairs that shared ecological traits including similar
body size, temporal activity pattern and diet. However, co-occurrence decreased as compared to
other species pairs when the pair included a large-bodied carnivore. Our results suggest that a
combination of shared traits and top-down regulation by large carnivores shape local carnivore
Camera trap, ecological traits, global assessment, interspeciﬁc interactions, local community
structure, spatial co-occurrence.
Ecology Letters (2018)
The composition of ecological communities is shaped by
interspeciﬁc interactions (Birch 1957; Hardin 1960; Rosen-
zweig 1966). Hutchinson’s (1957) theory of a realised vs. fun-
damental niche was one of the ﬁrst to evaluate species
interactions and how they may cause an individual to occupy
areas smaller than the area they would reside in the absence
of competition and predation. Since then, area-speciﬁc assess-
ments of species interactions have illuminated behavioural
responses such as spatial partitioning between apex and
mesocarnivores (Ritchie & Johnson 2009; Brook et al. 2012),
temporal or spatial partitioning between predators and their
prey (Miller et al. 2012; Davis et al. 2017) or between poten-
tially competing carnivores (Di Bitetti et al. 2009, 2010) and
local extinctions resulting from native species being
Department of Ecosystem Science and Management, Pennsylvania State
University, University Park, PA 16802, USA
Intercollege Degree Program in Ecology, Pennsylvania State University,
University Park, PA 16802, USA
Department of Environmental Science, Policy and Management, University
of California, Berkeley, CA 94720, USA
Department of Fish and Wildlife Conservation, Virginia Tech, Blacksburg, VA
Department of Health and Exercise Science, Appalachian State University,
Boone, NC 28608, USA
Instituto de Biolog
ıa Subtropical (IBS) –nodo Iguaz
u, Universidad Nacional
de Misiones and CONICET, Bertoni 85, 3370 Puerto Iguaz
on Civil Centro de Investigaciones del Bosque Atl
Bertoni 85, 3370 Puerto Iguaz
u, Misiones, Argentina
Facultad de Ciencias Forestales, Universidad Nacional de Misiones, Bertoni
124, 3380 Eldorado, Misiones, Argentina
on ProYungas, Per
u 1180, (4107), Yerba BuenaTucum
Wildlife Conservation Research Unit, Department of Zoology, University of
Oxford, The Recanati-Kaplan Centre, Tubney, Abingdon OX13 5QL, UK
Future4Leopards Foundation, No.4, Nour 2, Mahallati, Tehran, Iran
Iranian Cheetah Society, PO Box 14155-8549, Tehran, Iran
Department of Arctic and Marine Biology, Faculty of Biosciences, Fisheries
and Economics, UiT The Arctic University of Norway, 9037 Tromsø, Norway
Panthera,New York, NY 10018, USA
University of Belize, Environmental Research Institute (ERI), Price Centre
Road, PO box 340, Belmopan, Belize
Sackler Institute for Comparative Genomics, American Museum of Natural
History, New York, NY 10024, USA
Senegalese National Parks Directorate, Dakar, Senegal
The Cape Leopard Trust, Cape Town, South Africa
Audubon Canyon Ranch, PO Box 1195,Glen Ellen, CA, USA
Species at Risk, Resource Management, Alberta Environment and Parks,
Grande Prairie, AB, Canada
World Wildlife Fund, Jakarta, Indonesia
World Wildlife Fund, Conservation Science Unit, Baluwatar, Nepal
U.S. Forest Service, Sequoia National Forest, Porterville, CA 93257, USA
Parks Canada, Banff National Park Resource Conservation, Banff, AB,
*Correspondence: E-mail: firstname.lastname@example.org
©2018 John Wiley & Sons Ltd/CNRS
Ecology Letters, (2018) doi: 10.1111/ele.13124
outcompeted by exotics (Bailey et al. 2009; Farris et al.
2015a). As such, the concept of interspeciﬁc interactions has
been, and is still, a central theme of ecological investigations
(Wisz et al. 2013).
One of the primary ways in which interspeciﬁc interactions
are evaluated is by assessing species’ patterns of co-occurrence
(i.e. species asymmetrical spatial distributions; Mackenzie
et al. 2004; Richmond et al. 2010; Waddle et al. 2010). Co-
occurring species often display niche segregation as it serves
to reduce resource competition, promoting coexistence (Brown
& Wilson 1956; Hutchinson 1959; P
eriquet et al. 2015). Niche
segregation should occur when species directly compete for
resources, and competition should be strongest in cases where
species share similar life history traits (Brown & Wilson
1956). Alternatively, if competition and niche segregation are
not the primary drivers of local species distributions, trait sim-
ilarities should lead to greater co-occurrence because of shared
environmental and resource afﬁnities (i.e. habitat or environ-
mental ﬁltering; Van der Valk 1981; Keddy 1992; Weiher &
Keddy 1999; Di
az et al. 1998; Weiher et al. 1998). Attempts
to explain patterns of co-occurrence tend to focus on species’
dietary and habitat preferences because the partitioning of
resources can inﬂuence the degree to which competition
occurs (Donadio & Buskirk 2006; Hayward & Kerley 2008;
Yackulic et al. 2014).
Behaviour, morphology, and phylogenetic proximity also
can play pivotal roles in inﬂuencing the strength and direction
of interspeciﬁc interactions at local scales (Kronfeld-Schor &
Dayan 2003; Donadio & Buskirk 2006; Davies et al. 2007;
Yackulic et al. 2014). Species that exhibit different temporal
activity patterns (e.g. diurnal vs. nocturnal) may be more likely
to co-occur as they have a lower probability of direct competi-
tion compared to species which are active at similar times of
the day (Kronfeld-Schor & Dayan 2003; Hayward & Slotow
2009; Bischof et al. 2014; P
eriquet et al. 2015). Alternatively, a
species’ overarching social structure (i.e. group, pair or soli-
tary) can inﬂuence their resource requirements, detectability by
other species and ability to outcompete interspeciﬁc competi-
tors (Palomares & Caro 1999; de Oliveira & Pereira 2014). In
turn, social structure could inﬂuence the likelihood that species
co-occur. Body size may also inﬂuence co-occurrence via com-
petition (e.g. Dayan et al. 1989, 1990; McDonald 2002) or
direct aggression (e.g. Sidorovich et al. 1999) of similar-sized
carnivores (Rosenzweig 1966), as well as top-down pressures
of larger carnivores (Palomares & Caro 1999; Saether 1999;
Terborgh et al. 1999; Elmhagen & Rushton 2007). Lastly, phy-
logenetic proximity may also shape patterns of co-occurrence
because closely related species (e.g. within a family) are often
similar in their resource requirements, thereby leading to
greater competition among closely related taxa (e.g. within,
rather than among, taxonomic groups; Gittleman 1985; Van
Valkenburgh 1989; Donadio & Buskirk 2006).
The inﬂuence of interspeciﬁc interactions is particularly
widespread within carnivore guilds (Rosenzweig 1966; Palo-
mares & Caro 1999). Elucidating how carnivore interactions
inﬂuence patterns of co-occurrence, and the ecological traits
driving these interactions, is key to our understanding of
niche dynamics, interspecies competition, mesopredator
release (Dayan et al. 1989; Estes et al. 1998; Berger et al.
2001) and carnivore population dynamics (Robinson et al.
eriquet et al. 2015). Interactions between carnivore
species can also inﬂuence human perception and tolerance of
carnivores, thereby affecting human-predator coexistence (e.g.
Farhadinia et al. 2017). Despite the availability of detailed
information on intraguild interactions at the site-speciﬁc
levels, we have a poor understanding of global patterns in car-
nivore co-occurrence (Linnell & Strand 2000; Elmhagen &
Rushton 2007; P
eriquet et al. 2015). Improving this under-
standing requires local occurrence data for carnivore commu-
nities across large spatial or temporal scales. Historically,
resource constraints have limited our ability to collect such
data sets for wide-ranging and often elusive species. In the last
decade, however, the exponential increase in the use of camera
trap surveys has opened the door to studying mammalian car-
nivore species in remote areas across the world (Rich et al.
2017; Steenweg et al. 2017). Collaborative research efforts and
the aggregation of data collected across large spatial scales
and international borders allow us to draw conclusions about
patterns of spatial interactions across ecosystems rather than
solely within a single study area, thus providing new and
important insights into the underlying processes of community
structure that are consistent across global scales (Steenweg
et al. 2017).
Our goal was to investigate co-occurrence within the order
Carnivora and determine which ecological traits inﬂuence the
spatial distributions of sympatric species (i.e. the overlap or
avoidance of two species in habitat use). To accomplish this
goal, we used a pre-existing dataset (see Rich et al. 2017)
consisting of remote camera trap data from surveys in 13
study areas in 12 countries, which included observations of
86 mammalian carnivore species in four of the ﬁve major
biomes worldwide. We approached the analysis as a two-step
process. First, we analysed these data using a pair-wise co-
occurrence estimator to quantify relative co-occurrence of
sympatric species while accounting for imperfect detection
(Mackenzie et al. 2004; Richmond et al. 2010; Waddle et al.
2010). We then used estimates of co-occurrence (i.e. species
interaction factor) to determine how shared ecological traits,
including diet, body size, temporal activity patterns, social
structure and phylogenetic proximity, correlated with co-
occurrence probabilities. We predicted that species pairs with
shared ecological traits (e.g. similar in body size or dietary
preferences) would be more likely to compete for resources,
and hence more likely to display spatial avoidance. Our anal-
ysis provides the ﬁrst global assessment of carnivore spatial
co-occurrence patterns, exemplifying a framework for other
collaborative, global-scale studies on species interactions.
MATERIAL AND METHODS
We used a pre-existing data set consisting of camera trap sur-
vey data (raw data previously published in Rich et al. 2017)
from 13 study areas spanning 12 countries and 5 continents
(Fig. 1). Study area size ranged from 42 to 18 714 km
within each study area, between 22 and 319 (
SD =85.5) camera stations (Table 1) were deployed. We only
©2018 John Wiley & Sons Ltd/CNRS
2C. L. Davis et al. Letter
included study areas with >1000 trap days, with realised
effort ranging from 1170 to 35 441 (
x=9007; SD =8891) trap
days (Table 1; Appendix S1).
North and Central American study areas included ﬁve
national parks in western Canada (Steenweg et al. 2016), the
Sierra Nevada Mountains of California, USA (Tucker et al.
(a) (b) (c) (d) (e) (f)
(g) (h) (i) (j) (k) (l)
Figure 1 Locations of the 13 study areas, which include remote camera trap surveys conducted in 12 countries, spanning 5 continents and 4 of the 5 major
biomes worldwide. Examples of co-occurring species pairs include: (a) Nasua nasua (South American coati) and Eira barabara (tayra) in Argentina, ©M.
Di Bitetti; (b) Puma concolor (puma) and Panthera onca (jaguar) in Belize, ©Belize Jaguar Project/Virginia Tech; (c) Urocyon cinereoargenteus (grey fox)
and Martes pennanti (ﬁsher) in United States, ©J. Tucker/US Forest Service; (d) Ursus arctos (grizzly bear) and Canis lupus (grey wolf) in Canada, ©Park
Canada; (e) Gulo gulo (wolverine) and Vulpes vulpes (red fox) in Norway, ©S. Killengreen; (f) Acinonyx jubatus venaticus (Asiatic cheetah) and Canis lupus
(grey wolf) in Iran, ©Iranian Cheetah Society/CACP/DoE/Panthera; (g) Leptailurus serval (Serval) and Panthera leo (lion) in Senegal, ©M. Kane; (h)
Panthera pardus (leopard) and Vulpes chama (cape fox) in South Africa, ©Q. Martins/Cape Leopard Trust; (i) Proteles cristata (aardwolf) and Otocyon
megalotis (bat-eared fox) in Botswana, ©L. Rich/Panthera; (j) Eupleres goudotii (falanouc) and Cryptoprocta ferox (fosa) in Madagascar, ©Z. Farris; (k)
Hemigalus derbyanus (banded palm civet) and Paradoxurus hermaphroditus (common palm civet) in Sumatra, ©F. Widodo/WWF; (l) Panthera tigris tigris
(Bengal tiger) and Pantherus pardus fusca (Indian leopard) in Nepal, ©K. Thapa/WWF.
©2018 John Wiley & Sons Ltd/CNRS
Letter Global patterns in Carnivora co-occurrence 3
2014) and the Mayan Forest in Belize (Wultsch et al. 2016).
We included two study areas in South America; the ﬁrst from
northeastern Argentina, in the Atlantic Forest of Misiones
Province (Di Bitetti et al. 2006), and the second in the Yungas
ecoregion in northwestern Argentina (Di Bitetti et al. 2011).
In Africa, we included studies conducted in the Ngamiland
District of northern Botswana (Rich et al. 2016), Niokolo
Koba National Park in Senegal (Kane et al. 2015), Cederberg
mountains of South Africa (Martins 2010) and Madagascar’s
Masoala-Makira protected area (Farris et al. 2015b). In Asia,
we included the southern Riau landscape of central Sumatra
in Indonesia (Sunarto et al. 2015), the Chitwan National Park
in Nepal (Thapa & Kelly 2017) and several reserves across
central Iran (Farhadinia et al. 2014). We also included a sin-
gle European study area, located in northern Norway (Hamel
et al. 2013; Henden et al. 2014).
We estimated co-occurrence probabilities for 768 species
pairs. The number of species pairs ranged from 3 in Norway
to 210 in Botswana (Table 1; see Appendix S2 for species
list). In estimating patterns of co-occurrence, we were inter-
ested in determining whether species occurred at a site more
or less often than expected under a hypothesis of indepen-
dence (Mackenzie et al. 2004). Deviations from independence
occur when distributions are non-random with respect to
each other. To quantify these deviations, we used two-species
occupancy models (Mackenzie et al. 2004; Richmond et al.
2010) to understand pair-wise carnivore co-occurrence pat-
terns within each study area. This method allowed us to
account for imperfect detection at a camera station (i.e.
when a species is present but not photographed) by treating
each trap day (i.e. 24-h. period) as a repeat survey (Dorazio
& Royle 2005; Rich et al. 2016). These detection/non-detec-
tion data allowed us to estimate occupancy and detection
probabilities for every combination of co-occurring species.
Species detected on <3 occasions were not included in our
We ﬁt models using a Bayesian formulation of the single-
season two-species model parameterisation presented by
Richmond et al. (2010). This parameterisation estimates con-
ditional probabilities for both occupancy and detection (e.g.
the probability species B is present given species A is present
or absent and vice versa) and improves model convergence
(Richmond et al. 2010). We were able to derive unconditional
probabilities of species occupancy and detection from the con-
ditional probabilities. We did not investigate covariate rela-
tionships on occupancy or detection, which allowed us to
avoid classifying dominant/subordinate relationships between
co-occurring species (Richmond et al. 2010).
We modelled latent occurrence of species A and B at
camera station jas Bernoulli random variables,
zB;jBernoulliðWBa ð1WAÞþWBA WAÞ;
was the probability species A occurred in the study
was the occupancy probability of species B given
species A was present, and w
was the occupancy probability
of species B given species A was absent.w
species B) was a derived quantity, given by w
. We estimated the probability of observing species
A and species B at camera station jas:
The probability of detecting either species A or B at camera
station jduring a trapping session was a function of the prob-
ability of detecting the species (p
) given it occurs at
). Our study focuses on the spatial rather
than temporal co-occurrence between species pairs. In other
are the number of 24-h. time periods dur-
ing which the respective species was photographed at site j
and were modelled using a Binomial (rather than a Bernoulli)
Table 1 List of camera trap surveys included in our analysis and corresponding reference, detailing the number of camera stations, number of trap days,
study area size, number of carnivore species detected and number of detected species pairs in each study area location
Misiones 103 5104 1547 10 45 Di Bitetti et al. (2006)
Yungas 46 1258 376 8 28 Di Bitetti et al. (2011)
Belize 213 12 437 1030 11 55 Wultsch et al. (2016)
Botswana 179 5345 1154 21 210 Rich et al. (2016)
Canada 167 35 441 14 628 12 66 Steenweg et al. (2016)
Iran 220 12 768 4749 10 45 Farhadinia et al. (2014)
Madagascar 151 8795 42 6 15 Farris et al. (2015b)
Nepal 78 1170 509 7 21 Thapa & Kelly (2017)
Norway 66 1832 18 714 3 3 Hamel et al. (2013), Henden et al.
Senegal 58 3721 525 13 78 Kane et al. (2015)
South Africa 22 5077 2476 14 91 Martins (2010)
Sumatra 92 8009 524 12 66 Sunarto et al. (2015)
319 7130 8453 10 45 Tucker et al. (2014)
The number of species detected per study area that were included in our analysis, which does not include species with <3 detections.
©2018 John Wiley & Sons Ltd/CNRS
4C. L. Davis et al. Letter
distribution to speed up computation (K
ery 2010). This means
we evaluated whether species occurred at the same site, but
not necessarily at the same time. Co-occurrence here can be
deﬁned as overlap in the use of sites between species.
We then derived the species interaction factor (SIF; Rich-
mond et al. 2010), or probability of co-occurrence, for each
species pair as
SIF ¼WAWBA=ðWAðWAWBA þð1WAÞWBaÞÞ
When the occurrence of one species is independent of the
other, SIF =1.0. When two species co-occur more frequently
than would be expected under a hypothesis of independence,
SIF and its credible interval will be >1.0. When species occur
less often than expected, SIF and its credible interval will be
<1.0. Note that SIF values <1.0 indicate potential spatial
avoidance, which may result from a habitat-mediated relation-
ship or changes in the behaviour of one or both species. We
were not able to disentangle these two mechanisms using our
sampling framework. Instead, we use the term ‘spatial avoid-
ance’ in situations where SIF and 95% credible intervals are
<1.0 to indicate that two species simply do not co-occur at
the spatial scale we examine.
We estimated posterior distributions in R (R Core Develop-
ment Team 2016) using the package R2Jags (Plummer 2011)
to call JAGS (version 4.2.0). Estimates were generated from 3
chains of 50 000 iterations after a burn-in of 10 000 iterations.
We drew uninformative priors from a uniform distribution of
0 to 1 for all parameters. We assumed model convergence
when values of the Gelman-Rubin statistic were <1.1 (Gelman
et al. 2004). R code is provided in Appendix S3.
Drivers of global co-occurrence
Using our pair-wise SIF estimates, we investigated global car-
nivore co-occurrence patterns to determine which ecological
traits explain the spatial distributions of sympatric carnivores
worldwide. We log-transformed mean SIF estimates for all
species pairs. The log-transformed values provided a more
symmetrical distribution and standardised deviations for cases
where interactions were more and less likely (SIF is con-
strained between 0 and ∞, while ln[SIF] can range from –∞
to ∞). We then ﬁt regression models to ln[SIF] that tested
hypotheses about the drivers of mammalian carnivore co-
Our primary prediction was that similarity of species pairs
would affect the amount of spatial overlap. To assess similar-
ity, we included information on species’ temporal activity pat-
terns, dietary habits, social structure, body size (both
categorically and as a ratio), taxonomy at the family level (as
a proxy for relatedness) and information on the study areas’
species diversity and climate. Categorisation of species’ tempo-
ral activity pattern, dietary habits, social structure and body
size was based on a review of peer-reviewed literature, IUCN
red-list species accounts (IUCN 2016, Appendix S2), and
when necessary, expert knowledge of principal investigators
from individual study areas. We assigned the temporal activity
pattern for a species to be diurnal, nocturnal, crepuscular (i.e.
active primarily at twilight) or cathemeral (i.e. irregularly
active at any time of day or night). We designated the dietary
habit for each species as strict carnivore, omnivore or insecti-
vore. We used a species’ tendency to exhibit grouping (i.e. >2
individuals of the same species) vs. pairing and solitary beha-
viour to assign social structure. To account for body size, we
characterised the mean weight ratio (heavier:lighter species)
between two species, including it as a log-transformed contin-
uous variable. We used the mean weight to accommodate for
body size differences between sexes of the same species. We
also assigned species to a body size group, with species cate-
gorised as small (<2 kg), small–medium (2–5 kg), medium–
large (5–15 kg) or large (>15 kg) body size. Our categorical
and continuous variables characterising body size were never
included in the same model. Finally, to account for related-
ness, we categorised all Carnivora species included in this
study based on their taxonomy at the family level. Our data
included representative species of the families Canidae, Feli-
dae, Herpestidae, Mustelidae, Procyonidae and Viverridae.
Families represented by few species (Ursidae, Mephitidae,
Eupleridae and Hyaenidae) were grouped together into a sin-
gle category and classiﬁed as ‘Other’.
We summarised the categorical variables of diet, temporal
activity pattern, social structure, body size and taxonomic
similarity in two ways. For the ﬁrst coarse comparison
method, we compared species with differing trait values (e.g.
when species A |B are strict carnivore |omnivore) to those
where pairs shared the trait value (e.g. strict carnivore |strict
carnivore). In other words, species pairs were either labelled
the ‘same’ or ‘different’ for all categorical variables of interest.
For the ﬁne-scale trait comparison, species pairs were categor-
ically valued for all combinations of a trait (e.g. strict carni-
vore |strict carnivore =1, strict carnivore |omnivore =2,
strict carnivore |insectivore =3, etc.). By allowing for both
types of comparisons, we were able to explore whether co-
occurrence was driven primarily by coarse- (same vs. differ-
ent) or ﬁne-scale trait combinations. Lastly, to account for
differences among study areas, we included covariates repre-
senting species diversity and climate. Speciﬁcally, we included
ﬁxed effects for the observed number of carnivore species in
each study area and the study area’s climate as determined by
oppen-Geiger climate classiﬁcation system (Kottek et al.
2006). Study areas were categorised into equatorial (n=4),
arid (n=3), warm temperature (n=4), snow (n=1) or
polar (n=1) regions.
We had no a priori hypothesis on the combined inﬂuence of
these variables, so we explored 864 model combinations. We
estimated posterior distributions using R2Jags to call JAGS in
R. Estimates were generated from 3 chains of 20 000 itera-
tions after a burn-in of 5000 iterations (Appendix S3). We
drew uninformative priors from a uniform distribution of 0 to
1 for all parameters. We used Deviance Information Criterion
(DIC; Speigelhalter et al. 2002) to select among our compet-
ing models and present the subset of competitive models (i.e.
DDIC <5) in Appendix S4. Models were considered equivalent
with DDIC <2 and according to the parsimony principle, we
chose the best model as the model with the lowest number of
parameters (Burnham & Anderson 2002; Speigelhalter et al.
2002). We report coefﬁcient estimates from this model in
Appendix S5 and used these estimates to predict SIF (
and 95% credible intervals (CI) for all species pairs.
©2018 John Wiley & Sons Ltd/CNRS
Letter Global patterns in Carnivora co-occurrence 5
We determined how sensitive our results were to changes in
how the data were analysed and which data were included in
our analysis. First, we compared our results to a Bayesian
weighted regression approach to determine how parameter
estimates for the three variables included in all competing
models (i.e. body size, temporal activity pattern and diet cate-
gory) were affected by the estimated uncertainty in mean SIF
for each species pair. In addition, we assessed whether data-
rich study areas, Botswana and South Africa, may have been
driving the observed relationships for our global analysis. A
full description of the methods for these sensitivity analyses
can be found in Appendix S7.
The mean SIF value across all 768 species pairs was ^
CI] =1.24[1.19, 1.28], indicating that on average, species were
more likely to co-occur than expected under a hypothesis of
independence. Nonetheless, SIF variation was large within
study areas, with some species pairs showing strong overlap in
site use and others strong avoidance (Fig. 2). The largest SIF
(4.84[2.25, 7.97]) was estimated for Galerella sanguinea
uppell (slender mongoose) and Cynictis penicillata (G.
[Baron] Cuvier) (yellow mongoose) in Botswana. Other species
pairs exhibiting large spatial overlap included: Vulpes ruepellii
uppel’s fox) and Vulpes cana Blanford (Blanford’s
fox) in Iran (4.25[2.22, 6.77]), and Paradoxurus hermaphroditus
(Pallas) (common palm civet) and Hemigalus derbyanus (Gray)
(banded civet) in Sumatra (3.60[1.53, 6.63]). Strong overlap in
site use (mean SIF and 95% CI >1) was found for 91 species
pairs (Fig. 2).
Spatial avoidance (mean SIF and 95% CI <1) was found
for only 13 of our 768 species pairs (Fig. 2). Examples
include: Panthera pardus saxicolor Pocock (Persian leopard)
and Hyaena hyaena (L.) (striped hyena) in Iran (0.24[0.01;
0.83]), Panthera leo (L.) (lion) and Ictonyx striatus (Perry)
(striped polecat) in Botswana (0.39[0.05, 0.94]) and Urocyon
cinereoargenteus (Schreber) (grey fox) and Conepatus semis-
triatus (Boddaert) (striped hog-nosed skunk) in Belize (0.67
[0.39, 0.97]). The smallest SIF of 0.16[0.00, 0.53] was esti-
mated for Lycalopex gymnocercus G. Fischer (pampas fox)
and Eira Barbara (L.) (tayra) from the Yungas study area in
Argentina. We provide species-speciﬁc estimates of mean
occupancy and detection probabilities in Appendix S6.
Drivers of global co-occurrence
The predominant drivers of species’ co-occurrence were body
size, temporal activity pattern and diet category, being the three
variables included in all competing models (Appendix S4). With
the exception of large species, species pairs with similar body
sizes occurred at the same sites more often than expected under
independence (Fig. 3a). For species pairs categorised as differ-
ent in size, small-sized species occurred more often than
expected at the same sites as small-medium species (Fig. 3a).
Species pairs that included a large-bodied carnivore exhibited
depressed SIF values relative to all other pairs and to the global
mean, yet still indicated an overall independence (SIF =1.0) in
co-occurrence because credible intervals overlapped 1.0
(Fig. 3a). This pattern is particularly evident in large |large spe-
cies pairs, where the mean predicted value was the only body
size grouping with SIF <1.0, indicating that species of large
body size tend to spatially avoid one another, though the 95%
CI for this estimate slightly overlapped 1.0.
With respect to diet, same diet species pairs co-occurred at
the same sites more frequently than expected for strict carni-
vores and particularly for insectivores, but not for omnivores
(Fig. 3b). Species pairs with differing diets showed indepen-
dent occurrence in general, with the exception of strict carni-
vore |omnivore species pairs that showed overlap in site use
(Fig. 3b). Contrary to expectations, carnivores with similar
temporal activity patterns co-occurred disproportionately
more often than pairs that differed in temporal activity pat-
tern (Fig. 3c). Within species pairs with similar temporal
activity patterns, cathemeral, crepuscular and diurnal species
showed the greatest overlap in site use. Species pairs with dif-
ferent temporal activity patterns showed an overall indepen-
dence in spatial site occurrence (SIF =1.0), with the exception
of crepuscular |cathemeral species pairs showing a slight spa-
tial overlap (Fig. 3c).
We found that parameter estimates were comparable (i.e.
overlapping 95% CI) between the unweighted and weighted
regression approaches, suggesting that our results were not sen-
sitive to uncertainties in mean SIF (Appendix S7). Our results
were also robust to whether data-rich study areas, Botswana
and South Africa, were included in the analysis (Appendix S7).
The composition and structure of mammalian carnivore com-
munities have far-reaching effects on the structure and func-
tion of terrestrial ecosystems (Roemer et al. 2009; Estes et al.
2011; Ripple et al. 2014). Local community structure depends
on a combination of species-speciﬁc environmental afﬁnities
(i.e. habitat preferences or selection) and interactions among
species (Leibold et al. 2004). Our study provides important
insights into the role of each in shaping local carnivore com-
munities across ecosystems. Overall, mammalian carnivores
tended to overlap spatially, but there was wide heterogeneity
across species pairs with some showing large spatial overlap
and others showing large spatial avoidance. Speciﬁcally, we
found that body size, temporal activity patterns and dietary
habits were related to co-occurrence patterns, where species
that shared similar ecological traits generally had greater over-
lap in site use. These results suggest that at the spatial scale of
our study, shared ecological traits are not leading to competi-
tive exclusion, but are rather causing species to select sites
where resource availability is likely similar, and thus tend to
co-occur. The overall trend of spatial overlap may be attribu-
ted to our coarse-scale analyses as we were unable to account
for ﬁne-scale differences in species’ spatial and temporal activ-
ity patterns. Regardless, our study provides an important ﬁrst
step in understanding the drivers of carnivore co-occurrence
at a global scale, and a foundation from which future studies
interested in more ﬁne-scale assessments of species-pair rela-
tionships can build.
©2018 John Wiley & Sons Ltd/CNRS
6C. L. Davis et al. Letter
Large carnivores reduce the abundance of co-occurring spe-
cies through direct predation (Hakkarainen & Korpimaki
1996; Salo et al. 2008; Krauze-Gryz et al. 2012) and incite
changes in behaviour (Creel et al. 2001; Ritchie & Johnson
2009) and resource use (P
eriquet et al. 2015), thus shifting the
role that smaller carnivores play in ecological communities
(Bischof et al. 2014; de Oliveira & Pereira 2014). Large
carnivores can also promote the abundance of small species
by reducing the abundance of mesopredators (Estes et al.
1998; Ripple et al. 2014). The cascading effects of mesopreda-
tor release have been documented in a variety of systems and
taxa worldwide (Terborgh et al. 1999, 2001; Brashares et al.
2010), often resulting in the widespread loss of biodiversity
and ecosystem collapse (Estes et al. 1998; Berger et al. 2001;
Figure 2 Estimated carnivore co-occurrence (SIF) for all 768 sympatric species pairs across our 13 study areas (and 95% CI). The red line indicates
independently occurring species (SIF =1) and the blue dotted line represents the estimated mean for each study area, with the light blue zone representing
the 95% CI. SIF values >1 indicate co-occurrence between two species, while SIF values <1 indicate lack of co-occurrence.
©2018 John Wiley & Sons Ltd/CNRS
Letter Global patterns in Carnivora co-occurrence 7
Ripple & Beschta 2006; Prugh et al. 2009). While our results
do not suggest spatial avoidance among species of any body
size, depressed SIF values in pairs that include a large or med-
ium-large species supports the notion that large-bodied carni-
vores inﬂuence local community structure, particularly
through the effects of large-bodied species on one another
(Palomares & Caro 1999; Saether 1999; Terborgh et al. 1999,
2001; Elmhagen & Rushton 2007; Ripple et al. 2014; Swanson
et al. 2016).
Species that overlap in space may reduce competition by
exhibiting different activity patterns (e.g. diurnal vs. noctur-
nal; Kronfeld-Schor & Dayan 2003; Hayward & Slotow 2009;
Di Bitetti et al. 2009, 2010; Bischof et al. 2014; P
eriquet et al.
2015). In our study, however, species sharing similar temporal
activity patterns showed the strongest overlap in site use. For
example, Puma yagouaroundi (
E. Geoffroy Saint-Hilaire)
(jaguarundi) and Nasua nasua (L.) (South American coati) in
Argentina are both medium-large, diurnal species that show
strong co-occurrence but diverge in dietary habits (Appendix S2).
Spatial coexistence also may be maintained through differentia-
tion on another niche axis (e.g. vertical habitat partitioning,
resource partitioning in prey size) or through ﬁne-scale parti-
tioning of temporal activity (i.e. differences in the time of peak
activity; Farris et al. 2015c; Hayward & Slotow 2009; Sunarto
et al. 2015). Again, there is wide heterogeneity in spatial over-
lap between species pairs within study areas, and similar species
that show strong spatial overlap might also display temporal
Differences in dietary niche breadths among species also
inﬂuence the degree to which competition occurs (e.g. Hayward
& Kerley 2008). We report high spatial overlap between species
categorised as insectivores. In this case, heterogeneity in the
Figure 3 Predicted SIF and 95% CI) for each species trait combination of (a) body size, (b) diet, and (c) temporal activity pattern across the 13 study
©2018 John Wiley & Sons Ltd/CNRS
8C. L. Davis et al. Letter
spatial and temporal availability of insect prey (e.g. termites)
likely induces co-occurrence despite the strong overlap in diet-
ary preferences (Pringle et al. 2010). In addition, species that
both scavenge and actively hunt can exploit an ephemeral but
consistent resource, thus reducing the reliance on a particular
prey source during times of low prey abundance, unfavourable
environmental conditions, or high competition (Devault et al.
2003; Selva & Fortuna 2007). Scavenging carnivores, such as
Crocuta crocuta (Erxleben) (spotted hyena), Panthera leo (lion)
and Panthera pardus (L.) (common leopard) in Botswana and
Senegal occurred independently of one another, despite a high
degree of overlap in dietary habit, temporal activity pattern and
body size (Appendix S2). Previous studies have indicated that
oner et al. 2002) and changes in the scaveng-
ing behaviour (e.g. increased consumption of unfavourable ele-
ments, such as bones) of C. crocuta may alleviate strong
competition with P. leo during times of low prey abundance
(Kruuk 1972; P
eriquet et al. 2015). Coexistence can also be
facilitated between species of similar dietary preferences
through partitioned selection (e.g. by prey age or size) and use
of food resources. For example, Panthera onca (L.) (jaguar) and
Puma concolor (L.) (puma) in Belize partition their diet by prey-
ing on different species according to body plan. Panthera onca,
as a stronger predator, prefers the slower but armoured Dasy-
pus novemcinctus L. (nine banded armadillo), while the faster
and more agile P. concolor preys on the more vulnerable but
fast Cuniculus paca (L.) (paca; Foster et al. 2010).
Spatial avoidance can occur at various scales, but it can be
challenging to differentiate microhabitat vs. macrohabitat
resource partitioning (Bischof et al. 2014). When using data
from remote camera trap surveys, assessments of co-occur-
rence patterns are often limited to the macrohabitat scale.
Carnivore movements and responses to other carnivore spe-
cies may occur at a ﬁner spatio-temporal scale than we
assessed in our analysis (Swanson et al. 2016; Dr€
oge et al.
2017). Furthermore, factors such as ecosystem productivity,
topographic features (e.g. mountains vs. open terrains), habi-
tat patch size and quality (e.g. protected vs. degraded or frag-
mented), vegetation structure (e.g. grassland vs. rainforest),
carnivore densities and resource availability (e.g. prey densi-
ties, water), or type of camera trap site (e.g. random vs. trails,
baited vs. not baited), which we did not account for, also play
an important role in determining carnivore co-occurrence at
the landscape level (Elmhagen & Rushton 2007; Hoeinghaus
et al. 2007; Bischof et al. 2014; Peoples & Frimpong 2015;
eriquet et al. 2015; Hernandez-Santin et al. 2016).
Our analyses examined one key axis of co-occurrence, spatial
overlap, but did not allow us to differentiate between a spe-
cies’ presence and the explicit use of resources. Additionally,
while we assigned species according to their general temporal
activity patterns, dietary habits and social structure, this cate-
gorisation may not adequately capture a species’ behaviour at
a particular study area or trap location. Temporal activity pat-
terns, for example, are ﬂuid and can be altered according to
resource availability (Loveridge & Macdonald 2002; Hernan-
dez-Santin et al. 2016), changes in the abundance or behaviour
of co-occurring species (Creel et al. 2001; Ritchie & Johnson
2009) and human activity (e.g. McVittie 1979; Kitchen et al.
2000). Similarly, diet is closely tied to the availability of
resources at the local level (H€
oner et al. 2002; Ramesh et al.
2012). Co-occurrence at higher order interactions (i.e. among
3+species) may also differ from the pair-wise interactions we
examined, but computing SIF values based on multi-species
occupancy modelling remains challenging. Nevertheless, our
examination of pair-wise interactions allowed us to explicitly
test, at the course spatial scale we examined, whether shared
ecological traits affected co-occurrence by increasing competi-
tion due to niche overlap or by similar species sharing similar
environmental and resource afﬁnities.
Our study is the ﬁrst global-scale assessment of species co-
occurrence patterns and provides novel insights into the
macro-ecological processes that inﬂuence the spatial distribu-
tions of sympatric mammalian carnivores worldwide. We
demonstrated that species with similar ecological traits were
often more likely to overlap spatially, suggesting that shared
habitat afﬁnities may inﬂuence occurrence patterns at coarse
spatial scales. Therefore, competition and niche segregation
were not the primary drivers of local species occurrence,
though these patterns may change when considering the ﬁne-
scale differences in species’ spatial and temporal activity
patterns that we were unable to account for. We found a dif-
ferent pattern with respect to body size, with species tending
to have a lower co-occurrence when paired with a larger car-
nivore. These results suggest that top-down processes may
also be important in structuring carnivore communities.
Moreover, the methods we employed highlight the utility of
remote camera trap survey data to non-invasively study inter-
actions among elusive species and offer a starting point for
other collaborative, global-scale assessments (Butchart et al.
2010; Rich et al. 2017; Steenweg et al. 2017).
We thank the Ministry of the Environment, Wildlife and Tour-
ism, the Department of Wildlife and National Parks, and the
Botswana Predator Conservation Trust for permission to con-
duct the study in Botswana; the Ministry of Environment,
Water, Forest and Tourism and Wildlife Conservation Society
in Madagascar; the Department of National Parks and United
States Agency for International Development/Wula Nafaa
Project in Senegal; and The Cederberg Conservancy and Cape-
Nature in South Africa for permission and/or supporting the
research in Africa. We thank Parks Canada staff and volun-
teers for collecting data in Canada, the US Forest Service for
ﬁnancing and collecting data in the USA along with volunteers
from the Student Conservation Association, and the Belize
Forest Department, Belize Audubon Society, Programme for
Belize, Las Cuevas Research Station, Bull Run Farm, Gallon
Jug Estate, and Yalbac Ranch and Cattle Company for per-
mission and support in conducting research in Belize. Funding
for camera trap surveys in Canada was provided in part by
NSF LTREB Grant 1556248. We thank the Ministry of Ecol-
ogy and Natural Resources of Misiones, the National Park
Administration of Argentina, Ledesma S.A. and Arauco SA
for permissions and support to conduct camera trap surveys.
We thank the Iran Department of Environment for permission
to work within the reserves in Iran, Department of National
Parks and Wildlife Conservation in Nepal for permission to
©2018 John Wiley & Sons Ltd/CNRS
Letter Global patterns in Carnivora co-occurrence 9
conduct surveys in Chitwan National Park, and in Indonesia,
WWF Networks, US Fish & Wildlife Service and the Hurvis
Family for ﬁnancially supporting the research, the Indonesian
Ministry of Forestry for permission to conduct the study, and
the WWF Team for their support. We also thank the Direc-
torate for Nature Management and The Norwegian Research
Council for ﬁnancing camera trap surveys in Norway.
Data associated with this paper have been deposited in
CLD and DAWM analysed the data and prepared the manu-
script; LNR and ZJF initiated this global effort and coordi-
nated the consolidation and management of data from all
study areas; study design and data collection was performed
by: MSD, YD and SA in Argentina; MJK, CW and BJH in
Belize; JMT in United States; JW and RS in Canada; SH and
NGY in Norway; MSF, NG and AT in Iran; KT and MJK in
Nepal; SS, FAW and MJK in Sumatra; ZJF, AJM and MJK
in Madagascar; QM in South Africa; LNR, DAWM and MJK
in Botswana, and MDK and MJK in Senegal. All authors con-
tributed input into the design and interpretation of the analysis
and contributed to writing the ﬁnal manuscript.
DATA ACCESSIBILITY STATEMENT
Data available from the Dryad Digital Repository: http://doi.
Bailey, L., Reid, J.A., Forsman, E.D. & Nichols, J.D. (2009).
Modeling co-occurrence of northern spotted and barred owls:
accounting for detection probability differences. Biol. Cons., 142,
Berger, J., Stacey, P.B., Bellis, L. & Johnson, M. (2001). A mammalian
predator-prey imbalance: grizzly bear and wolf extinction affect avian
Neotropical migrants. Ecol. Appl., 11, 947–960.
Birch, L.C. (1957). The meanings of competition. Am. Nat., 91, 5–18.
Bischof, R., Ali, H., Kabir, M., Hameed, S. & Nawaz, M.A. (2014).
Being the underdog; an elusive small carnivore uses space with prey
and time without enemies. J. Zool., 293, 40–48.
Brashares, J.S., Epps, C.W. & Stoner, C.J. (2010). Ecological and
Conservation Implications of Mesopredator Release. Island Press, In J.
Terborgh & J. Estes. Trophic Cascades.
Brook, L.A., Johnson, C.N. & Ritchie, E.G. (2012). Effects of predator
control on behaviour of an apex predator and indirect consequences
for mesopredator suppression. J. Appl. Ecol., 49, 1278–1286.
Brown, W.L. & Wilson, E.O. (1956). Character displacement. Syst. Zool.,
Burnham, K.P. & Anderson, D.R. (2002). Model Selection and
Multimodel Inference: A Practical Information-Theoretic Approach.
Springer Science & Business Media, New York, NY.
Butchart, S.H.M., Walpole, M., Collen, B., van Strien, A., Scharlemann,
J.P.W., Rosamunde, E.A.A. et al. (2010). Global biodiversity:
indicators of recent declines. Science, 328, 1164–1168.
Creel, S., Spong, G. & Creel, N. (2001). Interspeciﬁc competition and the
population biology of extinction-prone carnivores. In Carnivore
Conservation (eds Gittleman, J.L., Funk, S.M., Macdonald, D.W.,
Wayne, R.K.). Cambridge University Press, Cambridge, pp. 35–60.
Davies, T.J., Meiri, S., Barraclough, T.G. & Gittleman, J.L. (2007).
Species co-existence and character divergence across carnivores. Ecol.
Lett., 10, 146–152.
Davis, C.L., Miller, D.A.W., Walls, S.C., Barichivich, W.J., Riley, J.W. &
Brown, M.E. (2017). Species interactions and the effects of climate
variability on a wetland amphibian metacommunity. Ecol. Appl., 27,
Dayan, T., Simberloff, D., Tchernov, E. & Yom-Tov, Y. (1989). Inter-
and intraspeciﬁc character displacement in mustelids. Ecology, 70,
Dayan, T., Simberloff, D., Tchernov, E. & Yom-Tov, Y. (1990). Feline
canines: community-wide character displacement in the small cats of
Israel. Am. Nat., 136, 39–60.
Devault, T.L., Rhodes, O.E. Jr & Shivik, J.A. (2003). Scavenging by
vertebrates: behavioral, ecological, and evolutionary persepectives on
an important energy transfer pathway in terrestrial ecosystems. Oikos,
Di Bitetti, M.S., Paviolo, A. & De Angelo, C. (2006). Density, habitat use
and activity patterns of ocelots (Leopardus pardalis) in the Atlantic
Forest of Misiones. Argentina. J. Zool., 270, 153–163.
Di Bitetti, M.S., Di Blanco, Y.E., Pereira, J.A., Paviolo, A. & Jim
erez, I. (2009). Time partitioning favors the coexistence of sympatric
crab-eating foxes (Cerdocyon thous) and pampas foxes (Lycalopex
gymnocercus). J. Mammal., 90, 479–490.
Di Bitetti, M.S., De Angelo, C.D., Di Blanco, Y.E. & Paviolo, A. (2010).
Niche partitioning and species coexistence in a Neotropical felid
assemblage. Acta Oecol., 36, 403–412.
Di Bitetti, M.S., Albanesi, S., Foguet, M.J., Cuyckens, G.A.E. & Brown,
A. (2011). The Yungas biosphere reserve of argentina: a hot spot of
South American wild cats. CAT News, 54, 25–29.
az, S., Cabido, M. & Casanoves, F. (1998). Plant functional traits and
environmental ﬁlters at a regional scale. J. Veg. Sci., 9, 113–122.
Donadio, E. & Buskirk, S.W. (2006). Diet, morphology, and interspeciﬁc
killing in Carnivora. Am. Nat., 167, 524–536.
Dorazio, R.M. & Royle, J.A. (2005). Estimating size and composition of
biological communities by modeling the occurrence of species. J. Am.
Stat. Assoc., 100, 389–398.
oge, E., Creel, S., Becker, M.S. & M’Soka, J. (2017). Spatial and
temporal avoidance of risk within a large carnivore guild. Eco. Evol.,7,
Elmhagen, B. & Rushton, S.P. (2007). Trophic control of mesopredators in
terrestrial ecosystems: top-down or bottom-up? Ecol. Lett., 10, 197–206.
Estes, J.A., Tinker, M.T., Williams, T.M. & Doak, D.F. (1998). Killer
whale predation on sea otters linking oceanic and nearshore
ecosystems. Science, 282, 473–476.
Estes, J.A., Terborgh, J., Brashares, J.S., Power, M.E., Berger, J., Bond, W.J.
et al. (2011). Trophic downgrading of planet Earth. Science,333,301–306.
Farhadinia, M.S., Eslami, M., Hobeali, K., Hosseini-Zavarei, F.,
Gholikhani, N. & Taktehrani, A. (2014). Status of Asiatic cheetah in
Iran: a country-scale assessment. Project Final Report, Iranian Cheetah
Society (ICS), Tehran.
Farhadinia, M.S., Johnson, P.J., Hunter, L.T. & Macdonald, D.W.
(2017). Wolves can suppress goodwill for leopards: patterns of human-
predator coexistence in northeastern Iran. Biol. Cons., 213, 210–217.
Farris, Z.J., Kelly, M.J., Karpanty, S. & Ratelolahy, F. (2015a). Patterns
of spatial co-occurrence among native and exotic carnivores in north-
eastern Madagascar. Anim. Cons., 19, 189–198.
Farris, Z.J., Golden, C.D., Karpanty, S., Murphy, A., Stauffer, D.,
Ratelolahy, F. et al. (2015b). Hunting, exotic carnivores, and habitat
loss: anthropogenic effects on a native carnivore community,
Madagascar. PLoS ONE, 10, e0136456.
Farris, Z.J., Gerner, B.D., Karpanty, S., Murphy, A., Andrianjakarivelo,
V., Ratelolahy, F. et al. (2015c). When carnivores roam: temporal
patterns and overlap among Madagascar’s native and exotic carnivores.
J. Zool., 296, 45–57.
©2018 John Wiley & Sons Ltd/CNRS
10 C. L. Davis et al. Letter
Foster, R.J., Harmsen, B.J., Valdes, B., Pomilla, C. & Doncaster, C.P.
(2010). Food habits of sympatric jaguars and pumas across a gradient
of human disturbance. J. Zool., 280, 309–318.
Gelman, A., Carlin, J.B., Stern, H.S. & Rubin, D.B. (2004). Bayesian
Data Analysis. Chapman and Hall, Boca Raton, FL.
Gittleman, J.L. (1985). Carnivore body size: ecological and taxonomic
correlates. Oecologia, 67, 540–554.
Hakkarainen, H. & Korpimaki, E. (1996). Competitive and predatory
interactions among raptors: an observational and experimental study.
Ecology, 77, 1134–1142.
Hamel, S., Killengreen, S.T., Henden, J.A., Yoccoz, N.G. & Ims, R.A.
(2013). Disentangling the importance of interspeciﬁc competition, food
availability, and habitat in species occupancy: recolonization of the
endangered Fennoscandian arctic fox. Biol. Cons., 160, 114–120.
Hardin, G. (1960). The competitive exclusion principle. Science, 29, 1292–
Hayward, M.W. & Kerley, G.I.H. (2008). Prey preferences and dietary
overlap amongst Africa’s large predators. South African J. Wild. Res.,
Hayward, M.W. & Slotow, R. (2009). Temporal partitioning of activity in
large African carnivores: tests of multiple hypotheses. South African J.
Wild. Res., 39, 109–125.
Henden, J.A., Stien, A., B
ardsen, B.J., Yoccoz, N.G. & Ims, R.A. (2014).
Community-wide mesocarnivore response to partial ungulate migration.
J. Appl. Ecol., 51, 1525–1533.
Hernandez-Santin, L., Goldizen, A.W. & Fisher, D.O. (2016). Introduced
predators and habitat structure inﬂuence range contraction of an
endangered native predator, the northern quoll. Biol. Cons., 203, 160–
Hoeinghaus, D.J., Winemiller, K.O. & Birnbaum, J.S. (2007). Local and
regional determinants of stream ﬁsh assemblage structure: inferences
based on taxonomic vs. functional groups. J. Biogeogr., 34, 324–338.
oner, O.P., Wachter, B., East, M.L. & Hofer, H. (2002). The response of
spotted hyaenas to long-term changes in prey populations: functional
response and interspeciﬁc kleptoparasitism. J. Anim. Ecol., 71, 236–246.
Hutchinson, G.E. (1957). Concluding remarks. Cold Spring Harbor
Symp., 22, 415–417.
Hutchinson, G.E. (1959). Homage to Santa Rosalia or why are there so
many kinds of animals?. Am. Nat., 92, 145–159.
IUCN (2016). The IUCN Red List of Threatened Species. Version 2016-3.
Available at: http://www.iucnredlist.org. Last accessed August 31, 2017.
Kane, M.D., Morin, D.J. & Kelly, M.J. (2015). Potential for camera-
traps and spatial mark-resight models to improve monitoring of the
critically endangered West African lion (Panthera leo). Biodivers.
Conserv., 24, 3527–3541.
Keddy, P.A. (1992). Assembly and response rules–2 goals for predictive
community ecology. J. Veg. Sci., 3, 157–164.
ery, M. (2010). Introduction to WinBUGS for Ecologists: Bayesian
approach to regression, ANOVA, mixed models and related analyses.
Academic Press, Burlington, MA.
Kitchen, A.M., Gese, E.M. & Schauster, E.R. (2000). Changes in coyote
activity patterns due to reduced exposure to human persecution. USDA
National Wildlife Research Center –Staff Publications. Paper 658.
Kottek, M., Grieser, J., Beck, C., Rudolf, B. & Rubel, F. (2006). World
Map of the K€
oppen-Geiger climate classiﬁcation updated. Meteorol. Z.,
15, 259–263. https://doi.org/10.1127/0941-2948/2006/0130.
Krauze-Gryz, D., Gryz, J.B., Goszczy
nski, J., Chylarecki, P. &
Zmihorski, M. (2012). The good, the bad, and the ugly: space use and
intraguild interactions among three opportunistic predators –cat (Felis
catus), dog (Canis lupis familiaris), and red fox (Vulpes vulpes)–under
human pressure. Can. J. Zool., 90, 1402–1413.
Kronfeld-Schor, N. & Dayan, R. (2003). Partitioning of time as an
ecological resource. Annu. Rev. Ecol. Evol. Syst., 34, 153–181.
Kruuk, H. (1972). The Spotted Hyena: A Study of Predation and Social
Behavior. University of Chicago Press, Chicago, IL.
Leibold, M.A., Holyoak, M., Mouquet, N., Amarasekare, P., Chase,
J.M., Hoopes, M.F. et al. (2004). The metacommunity concept: a
framework for multi-scale community ecology. Ecol. Lett., 7, 601–613.
Linnell, J.D. & Strand, O. (2000). Interference interactions, co-existence
and conservation of mammalian carnivores. Divers. Distrib., 6, 169–176.
Loveridge, A.J. & Macdonald, D.W. (2002). Habitat ecology of two
sympatric species of jackals in Zimbabwe. J. Mammal., 83, 599–607.
Mackenzie, D.I., Bailey, L.L. & Nichols, J.D. (2004). Investigating species
co-occurrence patterns when species are detected imperfectly. J. Appl.
Ecol., 73, 546–555.
Martins, Q.E. (2010). The ecology of the leopard Panthera pardus in the
Cederberg Mountains. Dissertation, University of Bristol, Bristol, UK.
McDonald, R.A. (2002). Resource partitioning among British and Irish
mustelids. J. Anim. Ecol., 71, 185–200.
McVittie, R. (1979). Changes in the social behaviour of South West
African cheetah. Modoqua, 2, 171–184.
Miller, D.A.W., Brehme, C.S., Hines, J.E., Nichols, J.D. & Fisher, R.N.
(2012). Joint estimation of habitat dynamics and species interactions:
disturbance reduces co-occurrence of non-native predators with an
endangered toad. J. Anim. Ecol., 81, 1288–1297.
de Oliveira, T.G. & Pereira, J.A. (2014). Intraguild predation and
interspeciﬁc killing as structuring forces of carnivoran communities in
South America. J. Mammal Evol., 21, 427–436.
Palomares, F. & Caro, T.M. (1999). Interspeciﬁc killing among
mammalian carnivores. Am. Nat., 153, 492–508.
Peoples B.K., Frimpong E.A (2015). Biotic interactions and habitat drives
positive co-occurrence between facilitating and beneﬁciary stream
ﬁshes. J. Biogeogr., 43, 923–931.
eriquet, S., Fritz, H. & Revilla, E. (2015). The Lion King and the
Hyaena Queen: large carnivore interactions and coexistence. Biol. Rev.,
Plummer, M. (2011). JAGS: a program for the statistical analysis of
Bayesian hierarchical models by Markov Chain Monte Carlo.
Pringle, R.M., Doak, D.F., Brody, A.L., Jocqu
e, R. & Palmer, T.M.
(2010). Spatial pattern enhances ecosystem functioning in an African
savannah. PLoS Biol., 8, e1000377.
Prugh, L.R., Stoner, C.J., Epps, C.W., Bean, W.T., Ripple, W.J.,
Laliberte, A.S. et al. (2009). The rise of the mesopredator. Bioscience,
R Core Development Team (2016). R: A language and environment for
statistical computing. R Foundation for Statistical Computing, Vienna,
Austria. Available at: http://www.R-project.org/. Last accessed February
Ramesh, R., Kalle, R., Sankar, K. & Qureshi, Q. (2012). Dietary
partitioning in sympatric large carnivores in a tropical forest of
Western Ghats, India. Mammal Study, 37, 313–321.
Rich, L.N., Miller, D.A.W., Robinson, H.S., McNutt, J.W. & Kelly, M.J.
(2016). Using camera trapping and hierarchical occupancy modeling to
evaluate the spatial ecology of an African mammal and bird
community. J. Appl. Ecol., 53, 1225–1235.
Rich, L.N., Davis, C.L., Farris, Z.J., Miller, D.A.W., Tucker, J.M.,
Hamel, S. et al. (2017). Assessing global patterns in mammalian
carnivore occupancy and richness by integrating local camera trap
surveys. Global Ecol. Biogeogr., 26, 918–929.
Richmond, O.M.W., Hines, J.E. & Beissinger, S.R. (2010). Two species
occupancy models: a new parameterization applied to co-occurrence of
secretive rails. Ecol. Appl., 20, 2036–2046.
Ripple, W.J. & Beschta, R.L. (2006). Linking a cougar decline, trophic
cascade, and catastrophic regime shift in Zion National Park. Biol.
Cons., 133, 397–408.
Ripple, W.J., Estes, J.A., Beschta, R.L., Wilmers, C.C., Richie, E.G.,
Hebblewhite, M. et al. (2014). Status and ecological effects of the
world’s largest carnivores. Science, 343, 151–162.
Ritchie, E.G. & Johnson, C.N. (2009). Predator interactions, mesopredator
release and biodiversity conservation. Ecol. Lett., 12, 1–18.
©2018 John Wiley & Sons Ltd/CNRS
Letter Global patterns in Carnivora co-occurrence 11
Robinson, Q.H., Bustos, D. & Roemer, G.W. (2014). The application of
occupancy modeling to evaluate intraguild predation in a model
carnivore system. Ecology, 95, 3112–3123.
Roemer, G.W., Gompper, M.E. & Van Valkenburgh, B. (2009). The
ecological role of the mammalian mesocarnivore. Bioscience, 59, 165–
Rosenzweig, M.L. (1966). Community structure in sympatric Carnivora.
J. Mammal., 47, 602–612.
Saether, B.E. (1999). Top dogs maintain diversity. Nature, 400, 510–511.
Salo, P., Nordstrom, M., Thomson, R.L. & Korpimaki, E. (2008). Risk
induced by a native top predator reduces alien mink movements.
J. Anim. Ecol., 77, 1092–1098.
Selva, N. & Fortuna, M.A. (2007). The nested structure of a scavenger
community. Proc. Biol. Sci., 274, 1101–1108.
Sidorovich, V., Kruuk, H. & MacDonald, D.W. (1999). Body size and
interactions between European and American mink (Mustel lutreola
and M. vison) in Eastern Europe. J. Zool., 248, 521–527.
Speigelhalter, D.J., Best, N.G., Carlin, B.P. & Van Der Linde, A. (2002).
Bayesian measures of model complexity and ﬁt. J. R. Statist. Soc. B,
Steenweg, R., Whittington, J., Hebblewhite, M., Forshner, A., Johnston,
B., Petersen, D. et al. (2016). Remote-camera-based occupancy
monitoring at large scales: power to detect trends in grizzly bears
across the Canadian Rockies. Biol. Conserv., 201, 192–200.
Steenweg, R., Hebblewhite, M., Kays, R., Ahumada, J., Fisher, J.T.,
Burton, A.C. et al. (2017). Scaling up camera traps —monitoring the
planet’s biodiversity with networks of remote sensors. Front. in Ecol.
Environ., 15, 26–34.
Sunarto, S., Kelly, M.J., Parakkasi, K. & Hutajulu, M.B. (2015). Cat
coexistence in central Sumatra: ecological characteristics, spatial and
temporal overlap, and implications for management. J. Zool.,296,104–115.
Swanson, A., Arnold, T., Kosmala, M., Forester, J. & Packer, C. (2016).
In the absence of a “landscape of fear”: how lions, hyenas, and
cheetahs coexist. Ecol. Evol., 6, 8534–8545.
Terborgh, J., Estes, J.A., Paquet, P., Ralls, K., Boyd-Heger, D., Miller,
B.J. et al. (1999). The role of top carnivores in regulating terrestrial
ecosystems. In: Continental conservation: design and management
principles for long-term, regional conservation networks (eds Soul
e, M. &
Terborgh, J.). Island Press, Covelo, CA; Washington DC. pp.39–64.
Terborgh, J., Lopex, L., Nu~
nez, P.V., Rao, M., Shahhabuddin, G.,
Orihuela, G. et al. (2001). Ecological meltdown in predator-free forest
fragments. Science, 294, 1923–1926.
Thapa, K. & Kelly, M.J. (2017). Density and carrying capacity in the
forgotten tigerland: tigers in understudied Nepalese Churia. Integrative
Zoology, 12, 211–227.
Tucker, J.M., Schwartz, M.K., Truex, R.L., Wisely, S.M. & Allendorf,
F.W. (2014). Sampling affects the detection of genetic subdivision and
conservation implications for ﬁsher in the Sierra Nevada. Conserv.
Genet., 15, 123–136.
Van der Valk, A.G. (1981). Succession in wetlands–a Gleasonian
approach. Ecology, 62, 688–696.
Van Valkenburgh, B. (1989). Carnivore dental adaptations and diet: a
study of trophic diversity within guilds. In: Carnivore Behavior,
Ecology, and Evolution (ed Gittleman, J.L.). Vol 1. Cornell University
Press, Ithaca, NY.
Waddle, J.H., Dorazio, R.M., Walls, S.C., Rice, K.G., Beauchamp, J.,
Schuman, M.J. et al. (2010). A new parameterization for estimating co-
occurrence of interacting species. Ecol. Appl., 20, 1467–1475.
Weiher, E. & Keddy, P.A. (1999). Ecological Assembly Rules: Perspectives,
Advances, Retreats. Cambridge University Press, Cambridge, UK.
Weiher, E., Clarke, G.D.P. & Keddy, P.A. (1998). Community assembly
rules, morphological dispersion, and the coexistence of plant species.
Oikos, 81, 309–322.
Wisz, M.S., Pottier, J., Kissling, W.D., Pellissier, L., Lenoir, J.,
Damgaard, C.F. et al. (2013). The role of biotic interactions in shaping
distributions and realised assemblages of species: implications for
species distribution modelling. Biol. Rev., 88, 15–30.
Wultsch, C., Waits, L.P. & Kelly, M.J. (2016). A comparative analysis of
genetic diversity and structure in jaguars (Panthera onca), pumas (Puma
concolor), and ocelots (Leopardus pardalis) in fragmented landscapes of
a critical Mesoamerican linkage zone. PLoS ONE, 11, e0151043.
Yackulic, C.B., Reid, J., Nichols, J.D., Hines, J.E., Davis, R. & Forsman,
E. (2014). The roles of competition and habitat in the dynamics of
populations and species distributions. Ecology, 95, 265–279.
Additional supporting information may be found online in
the Supporting Information section at the end of the article.
Editor, Jonathan Davies
Manuscript received 28 November 2017
First decision made 18 January 2018
Second decision made 9 April 2018
Third decision made 28 May 2018
Manuscript accepted 6 June 2018
©2018 John Wiley & Sons Ltd/CNRS
12 C. L. Davis et al. Letter