Ocelot ecology and its effect on the small-felid guild in the lowland neotropics

Full text

CHAPTER 27
Ocelot ecology and its effect on the
small-felid guild in the lowland neotropics
Tadeu G. de Oliveira, Marcos A. Tortato, Leandro Silveira, Carlos
Benhur Kasper, Fa
´bio D. Mazim, Mauro Lucherini, Anah T. Ja
´como,
Jose
´Bonifa
´cio G. Soares, Rosane V. Marques, and Melvin Sunquist
Introduction
The ocelot (Leopardus pardalis, 6.6–18.6 kg) is the
third largest of the 10 felid species found in the
Neotropical region (Sunquist and Sunquist 2002;
Oliveira and Cassaro 2005). Ocelots extend from
south Texas through to northern Argentina, and
have been the subject of several studies on feeding
ecology, home range, activity patterns and density
(e.g. Ludlow and Sunquist 1987; Emmons 1988;
Crawshaw 1995; Jacob 2002; Di Bitetti et al. 2006;
Moreno et al. 2006). The ocelot appears to be strong-
ly associated with dense habitat cover and preys pre-
dominantly on small rodents (see Sunquist and
Sunquist 2002). However, the extent to which envi-
ronmental or biological variables constrain ocelot
presence and distribution remains poorly under-
stood. Throughout their broad geographic range oce-
lots live in sympatry with the smaller margays
(Leopardus wiedii, 2.3–4.9 kg), little spotted cats (Leo-
pardus tigrinus, 1.7–3.5 kg), jaguarundis (Puma ya-
gouaroundi, 3.0–7.6 kg) and, to a lesser extent, with
pampas cats (Leopardus colocolo, 2.5–3.8 kg) and
Ocelot Leopardus pardalis from eastern Amazonia. #‘Projeto Gatos do
Mato—Brasil’.
OUP CORRECTED PROOF – FINAL, 1/5/2010, SPi

Geoffroy’s cats (Leopardus geoffroyi, 2.8–7.4 kg) (Oli-
veira and Cassaro 2005; Lucherini et al. 2006). They
also coexist with pumas and jaguars through most of
their range. There is considerable overlap among
these sympatric small felids, not only in geographic
range but also in habitat use, feeding ecology, activ-
ity patterns, and body size, suggesting the potential
for interspecific competition (Oliveira 1994; Nowell
and Jackson 1996; Sunquist and Sunquist 2002). Ad-
ditionally, competitive intraguild interactions are
now acknowledged as important in influencing ani-
mal communities (Dayan et al. 1990; Creel and Creel
1996; Donadio and Buskirk 2006; St-Pierre et al.
2006). In this regard, the mesopredator and niche
release theories are of special interest (Crooks and
Soule
´1999; Moreno et al. 2006). The mesopredator
release theory predicts that with the decline of top
predators, mesocarnivore numbers would rise
(Crooks and Soule
´1999), whereas the disappearance
of a potential competitor would broaden the niche of
its subordinate competitors (Moreno et al. 2006).
However, it is unknown to what extent these inter-
actions influence felid dynamics in the neotropics.
Here we tackle these issues by, first, characterizing
ocelot ecology across its broad geographic range in
an effort to understand the extent to which environ-
mental variables influence ocelot demography and
ecology; we then compare ocelot ecology to that of
the other sympatric small cats with a view to reveal-
ing their ecological differences. We also assess the
potential for competition between ocelot and other
felids, combined with testing the competitive niche
release hypothesis. Finally, we consider the potential
role that ocelots might play as mesopredators in the
dynamics of the small-felid assemblage of the low-
land tropical neotropics.
Methods
To describe ocelot ecology and the environmental vari-
ables that might influence it, we compiled information
from 35 sources detailing home range, habitat use,
diet, and density across the species’ full geographic
range, including sites from the northernmost (south
Texas) and southernmost (north-eastern Argentina
and southern Brazil) extremities of its range (Fig.
27.1). In our analyses we consider small felids as all
those smaller than the ocelot; sympatric species to
include the ocelot, jaguarundi, margay, and little
spotted cat; and mesopredators as all small- to medi-
um-sized lowland felids (i.e. ocelot, jaguarundi, Geof-
froy’s cat, pampas cat, margay, and little spotted cat).
Habitat was grouped into rainforest (Panama,
Peru, Brazil-Para
´[PA], Parana
´[PR]), semi-arid habi-
tats, which include scrub, chaparral, and dry forest
(Texas and Mexico), and open flood plains/savan-
nahs (Venezuela, Brazil-Mato Grosso do Sul-MS,
and Goia
´s/Emas-GO) according to ecoregional simi-
larities (Olson et al. 2001). Dense cover is considered
here as either closed canopy or dense understorey, as
described in their original studies. Body mass of adult
male and female ocelots are combined for each site.
Differences in ocelot home range size and habitat
use among sites and between sexes in different areas
were compared using mean values. For comparing
home ranges, whenever possible we used minimum
convex polygons (MCP) incorporating 95% of fixes
(using sources listed in Table 27.1). For habitat use
and preference (but not for home range-estimates)
for Emas National Park (ENP) (Brazil-GO) we only
used 95% fixed kernel estimates. Preference/avoid-
ance for each habitat type was measured using Ivlev’s
index of selectivity or preference (1961) (1, com-
plete avoidance, to þ1, complete preference), on sec-
ond and third order selection (Aebischer et al. 1993),
as described in Manfredi et al. (2006) and Rocha
(2006). Habitat categories for Emas include grassland
savannah, scrub/woodland savannah, gallery forest,
flooded grassland, and pasture/agriculture.
Ocelot dietwas assessed by the analysis of droppings
from study sites across the species range. Dietary re-
sults are presented as percent occurrence (number of
times an item was found as a percentage of total num-
ber of prey items in droppings). We quantified diet
using both percent occurrence and biomass contri-
bution. For biomass calculations only mammalian
prey were included. Prey were classed as very small
mammals (<100 g: marsupials, rodents, others), small
mammals (0.1–0.699 kg: marsupials, rodents), medi-
um-sized mammals (0.7–1.5 kg: marsupials, rodents,
rabbits, primates), large mammals (1.51–10 kg: ro-
dents, xenarthrans, primates, carnivores, others),
very large mammals (>10 kg: ungulates), birds, squa-
mates (lizards and snakes), and other vertebrates.
Maximum prey mass is defined as the mass of the
560 Biology and Conservation of Wild Felids
OUP CORRECTED PROOF – FINAL, 1/5/2010, SPi

three largest species taken by each felid. Diets of sym-
patric species are compared using mean mammalian
prey mass and niche overlap. To test for competitive
niche release, we compared ocelot diet in areas with
and without jaguars following Moreno et al.’s calcula-
tions (2006), and little spotted cat diet in the presence
and absence of ocelots.
Mean prey body mass is often used as a metric in
the assessment of trophic separation among carni-
vores (e.g. Oliveira 1994, 2002a; Ray and Sunquist
2001). The mean body mass of mammalian prey
(MWMP) is determined as the arithmetic mean of
the body mass of all mammalian prey consumed
(geometric means tend to underestimate the mean
mass of prey Walker et al. 2007). Body masses of
prey species are from original studies, or from sum-
maries in Fonseca et al. (1996) or Eisenberg and Red-
ford (1999). All prey items were assumed to be adult
size, unless otherwise noted. However, to avoid over-
representation, prey with a body mass larger than 1.5
times the felid’s body mass were assumed to be juve-
nile and assigned 40% of the adult mass. Canine
diameter, a metric used to evaluate carnivore segre-
gation and character displacement (Dayan et al.
1990), was assessed for Brazilian specimens of sym-
patric species. Mean values were correlated with felid
body mass, MWMP, and maximum prey mass.
Diet niche overlap is calculated using Pianka’s
index: O
jk
¼PP
ij
P
ik
ffip(PP
ij2
PP
ij2
), where P
ij
and P
ik
are the proportions of the ith resource used by the
jth and the kth species, respectively (Pianka 1973).
Values range from 0 (no similarity) to 1 (complete
similarity). Calculations are made at the lowest taxo-
nomic category of prey possible.
Figure 27.1 Ocelot distribution range showing the locations of study sites listed in the text. 1, S-Texas/United States
(Lower Rio Grande Valley); 2, Tamaulipas/Mexico (Los Ebanos Ranch); 3, Barro Colorado Island/Panama; 4, Central Llanos-
Guarico/Venezuela (Hato Masaguaral/Hato Flores Morada); 5, Caraja
´s-Para
´/Brazil; 6, Cocha Cashu Biological Station/Peru;
7, Tocantins/Brazil (Lajeado Dam); 8, ENP-GO/Brazil; 9, S-Pantanal-MS/Brazil; 10, Linhares-ES/Brazil (Vale do Rio Doce
Natural Reserve); 11, Caratinga Biological Station-MG/Brazil; 12, Iguac¸u National Park-PR/Brazil; 13, Serra do Tabuleiro
State Park-SC/Brazil; 14, Sa
˜o Francisco de Paula National Forest-RS/Brazil. (Modified from Oliveira and Cassaro 2005.)
Ocelot ecology and its effect on the small-felid guild in the lowland neotropics 561
OUP CORRECTED PROOF – FINAL, 1/5/2010, SPi

Table 27.1 Average home range (km
2
) of adult lowland Neotropical felids in various habitats.
Species Male nFemale nHabitat Location Source
Ocelot 12.3 5 7.0 3 Thorn scrub/oak forest S. Texas, United States Tewes (1986)
6.3 3 2.9 3 Thorn scrub/oak forest S. Texas, Unites States Laack (1991)
8.1 2 9.6 2 Low tropical forest/pasture Taumalipas, Mexico Caso (1994)
31.2 1 14.7 1 Central American rainforest Cockscomb, Belize Konecny (1989)
30.8 2 30 3 Central American rainforest Ciquibul, Belize Dillon (2005)
10.6 2 3.4 6 Gallery forest/flood plains Llanos, Venezuela Ludlow and Sunquist (1987)
7.0 2 1.8 3 Amazon rainforest Cocha Cashu, Peru Emmons (1988)
19.2 1 Savannah Tocantins, Brazil Trovati (2004)
90.5 1 75 1 Savannah ENP, Brazil Silveira (unpublished data)
5.4 1 2.6 2 Deciduous-gallery forests/flood
plains
Pantanal, Brazil Rocha (2006)
1.3 3 Deciduous-gallery forests/flood
plains
Pantanal, Brazil Crawshaw and Quigley (1989)
11.7 4 7.2 4 Tropical forest Morro do Diabo, Brazil Jacob (2002)
38.8 6 17.4 5 Subtropical forest Iguac¸u, Brazil/Argentina Crawshaw (1995)
Jaguarundi 9.6 8 8.9 5 Low tropical forest/pasture Taumalipas, Mexico Caso (1994, personal
communication)
94.1 2 20.1 1 Central American rainforest Cockscomb, Belize Konecny (1989)
25.3 1 18.0 1 Savannah Tocantins, Brazil Trovati (2004)
40.2 1 Savannah Emas Nat. Park, Brazil Silveira (unpublished data)
8.5 1 1.4 1 Forest/savannah/eucalyptus
plantation
Sa
˜o Paulo, Brazil Michalski et al. (2006b)
17.6 1 6.8 1 Subtropical forest Iguac¸u, Brazil/Argentina Crawshaw (1995)
23.4 3 Subtropical forest fragments/
agriculture
Taquari, Brazil Oliveira et al. (2008)
Margay 4.0 3 0.9 1 Low tropical forest/pasture Mexico Caso (personal
communication)
11.0 1 Central American rainforest Cockscomb, Belize Konecny (1989)
15.9 1 Subtropical forest Iguac¸u, Brazil Crawshaw (1995)
20 1 Subtropical forest fragments/
agriculture
Taquari, Brazil Oliveira et al. (2008)
OUP CORRECTED PROOF – FINAL, 1/5/2010, SPi

Little spotted
cat
4.8 2 Savannah Tocantins, Brazil Trovati (2004)
17.1 1 0.9 1 Savannah Serra da Mesa, Brazil Rodrigues and Marinho-Filho (1999)
25.0 1 Savannah ENP, Brazil Silveira (unpublished data)
8.0 2 2.0 1 Subtropical forest fragments/
agriculture
Taquari, Brazil Oliveira et al. (2008)
Pampas cat 27.7 4 Savannah ENP, Brazil Silveira (unpublished data)
Geoffroy’s cat 3.1 2 Swamp forest/wet grassland/
agriculture
Arroio Grande, Brazil Oliveira et al. (2008)
5.0 2 1.9 2 Wet pampas grassland Campos del Tuyu
´,
Argentina
Manfredi et al. (2006
8.9 2 5.1 1 Pampas grassland/agriculture Tornquist Park,
Argentina
Manfredi et al. (in preparation)
2.2 4 1.1 2 Argentine Monte Laguna de Chasico
´Manfredi et al. (in preparation)
2.0 3 0.3 1 Scrubland Lihue Calel National
Park, Argentina
Pereira et al. (2006)
9.2 5 3.7 2 Patagonian woodland/steppe Torres del Paine, Chile Johnson and Franklin (1991)
OUP CORRECTED PROOF – FINAL, 1/5/2010, SPi

Estimates of felid abundance are based on records
of tracks, photographs, or live captures. At each site
felid species were ranked based on abundance
indices, which varied from 1 (most abundant) to 5
(least abundant) (Oliveira et al. 2008, submitted).
Normalitytests were conducted prior todetermining
the use of parametric or non-parametric statistics. Data
in the form of percentages are converted into arcsin
(y
t
¼arcsin ffipy) for statistical analyses (Zar 1999).
Ocelot home range and habitat use
Home range
Variation in home range size is expected to reflect
differences in resource availability between areas
(e.g. Sandell 1989; Gompper and Gittleman 1991).
Ocelot home range size varies considerably between
sites and between sexes across the continent (1.3–
90.5 km
2
, Table 27.1). The smallest ranges for males
and females are in Brazil’s Pantanal flood plains (5.4
km
2
and 1.3 km
2
, respectively) and the largest in the
savannahs of ENP (90.5 km
2
and 75 km
2
, respective-
ly) also in Brazil (Crawshaw and Quigley 1989; Rocha
2006; Silveira, unpublished data). Both are open
areas, but the former is seasonally flooded and the
latter is dry. Although sample sizes for Emas included
only one male and one female, both were non-dis-
persing adults. At all sites, mean home range sizes of
adult males (mean standard deviation, 22.6 km
2

24.2, range 5.4–90.5 km
2
,n¼30 animals from 12
sites) are significantly larger than those of adult fe-
males (14.4 km
2
20.8, range 1.3–75 km
2
,n¼36
individuals from 12 sites) (t¼3.386, d.f. ¼10,
P¼0.007) in accordance with Crawshaw (1995).
Relationships between home range, body
size, and environmental variables
Ocelot body mass (6.6–18.6 kg, n¼112) differs
among regions (F¼5.835, P<0.001). However,
these differences do not correlate with latitude (r¼
0.385, P¼0.306, Fig. 27.2), but varied with habitat
types grouped as rainforests (mean ¼11.1 kg 2.2, n
¼56), open savannahs/flood plains (mean ¼10.0 kg
2.4, n¼29), and semi-arid habitats (scrub, chapar-
ral, and dry forest; mean ¼8.7 kg 1.4, n¼27) (H¼
21.72, d.f. ¼2, P<0.001). Pairwise comparisons
show that rainforest ocelots are significantly larger
than those of savannahs/flood plains (t¼2.085, d.
f. ¼83, P¼0.040) and scrub/chaparral/dry forest (t¼
5.004, d.f. ¼81, P<0.001), but those of semi-arid
habitats and savannahs/flood plains do not differ
significantly (t¼664.5, P¼0.087). Ocelot body
mass did not differ among evergreen forest sites (F
¼1.741, P¼0.170). The pattern in ocelots contrasts
with that for jaguar, which are larger in flood plains
than in rainforests, due to larger size and greater
abundance of large prey (>15 kg) in flood plains
(Hoogesteijn and Mondolfi 1996). A possible, but
untested, explanation for the greater mass of ocelots
in rainforest than elsewhere is that larger ocelot prey
may be more readily available in rainforests (e.g.
Oliveira 1994; Nowell and Jackson 1996; Sunquist
and Sunquist 2002). Agoutis (Dasyprocta spp.),
pacas (Agouti paca), armadillos (Dasypodidae), large
marsupials, monkeys, and sloths (Bradypus, Choloe-
pus), which are important prey for ocelots, generally
reach their highest densities in rainforests (see Glanz
1982; Emmons 1984).
When data from Emas are excluded due to their
extremely large values, there is an intriguing, but not
significant,tendency for average ocelot body mass and
mean home range to be positively correlated (r¼
0.767, P¼0.075, n¼6, Fig. 27.3). There are no differ-
ences in mean ocelot home range size by habitat type
across all habitat categories (thorn scrub/dry forests,
flood plains, savannahs, and forests) (F¼2.438, P¼
0.157), suggesting that variations among areas are
related to availability of prey and not to habitat struc-
ture per se. No correlation exists between mean annual
precipitation and mean adult ocelot home range per
site (r
2
¼0.079, P¼0.500); either for males (r
2
¼0.128,
P¼0.384) or females (r
2
¼0.022, P¼0.729). In fact, in
the Texas chaparral, where average annual rainfall is
only 640 mm and habitat is considered suboptimal for
the species, ocelot home range sizes are only 25% of
those in wetter (1500 to 1700 mm/year) and presum-
ably prime rainforest habitats of Belize and southern
Brazil (Tewes 1986; Laack 1991; Crawshaw 1995; Dil-
lon 2005). While precipitation in Texas is rather low,
rodents and rabbits are abundant (Tewes 1986). These
facts suggest, unexpectedly, that abiotic variables have
little detectable direct influence on ocelots’ home
range size and, consequently, their density.
564 Biology and Conservation of Wild Felids
OUP CORRECTED PROOF – FINAL, 1/5/2010, SPi

Habitat use
Several studies highlight the importance of dense hab-
itat cover for conservation of ocelots (e.g. Tewes 1986;
Ludlow and Sunquist 1987). In South Texas, 51% of
ocelot locations were in dense habitat cover, which
comprised only 1% of the study area (Harveson et al.
2004). In northern Mexico, 97.6% were in dense brush
(i.e. dense woody cover) (Caso 1994). The strong selec-
tion of ocelots for dense cover in south Texas could
also be related to avoidance of the similar-sized and
sympatric bobcat (Lynx rufus) (Tewes 1986; Fisher
1998). In the open savannahs of ENP, ocelots favoured
areas of dense forest and savannah, but they also used
pastoral/agricultural land, and avoided grassland
savannah and flooded grasslands. Forest cover
30
25
20
15
10
5
08.0 8.5 9.0 9.5 10.0
Ocelot mass (k
g
)
Home range (km
2
)
10.5 11.0 11.5 12.0
Figure 27.3 Correlation of mean ocelot body size and
mean home range in the Americas.
13
12
11
10
Mean weight (kg)
9
8
7
Site (latitude, habitat type)
26.20°N Texas semi-arid
23.27°N Mexico semi-arid
09.05°N Panama rainforest
06.10°S Brazil-PA rainforest
11.22°S Peru rainforest
25.20°S Brazil-PR rainforest
08.34°N Venezuela savanna
17.50°S Brazil-GO savanna
18.50°S Brazil-MS flood plains
Figure 27.2 Ocelot mass (mean ± standard error) in the Americas, grouped by habitat similarity and latitude. (Data
sources: Tewes 1986; Ludlow and Sunquist 1987; Emmons 1988; Laack 1991; Crawshaw 1995; Rocha 2006; Companhia
Vale do Rio Doce, unpublished data; Crawshaw and Quigley, unpublished data; Moreno and Kays, unpublished data; Silveira,
unpublished data.)
Ocelot ecology and its effect on the small-felid guild in the lowland neotropics 565
OUP CORRECTED PROOF – FINAL, 1/5/2010, SPi

comprised only 1.8% of the total area (Table 27.2).
Ocelots varied individually in their habitat use in
Emas but, on average, scrub/woodland savannah
and pasture/agriculture were used in proportion to
their availabilities (Ivlev’s third order values: 0.09 and
0.08, respectively), grassland savannah and flooded
grassland were avoided (0.6, 0.5, respectively), and
forest was positively selected (0.3). In the Pantanal
flood plains, too, ocelot home ranges were configured
to encompass forests and scrub/woodland savannahs,
and under-represented open grassland savannahs
(Rocha 2006).
One suggestion is that, due to their frequent asso-
ciation with dense cover, ocelots might occupy a
narrower range of microhabitats than expected
from their wide geographic distribution (Emmons
1988). However, throughout Brazil, ocelots com-
monly use heavily disturbed areas in the Amazon,
Cerrado, Atlantic Forest, and Pantanal biomes (Oli-
veira, personal observation; R. Boulhosa, personal
communication; E. Payan, personal communication;
Silveira, unpublished data). Indeed, they have been
found in habitats varying from heavily logged and
fragmented, to early and late successional forests, the
outskirts of major cities and towns, disturbed scrub/
woodland savannah and mosaic of forest (disturbed
or not), and savannah with eucalyptus plantations
and agricultural areas. In short, despite being asso-
ciated with dense cover, ocelots show a higher level
of habitat plasticity than previously thought.
Comparative ecology of ocelots and other
sympatric cats
Comparative home range
The limited data on adult home range sizes of the
smaller, sympatric lowland felids also show consid-
erable variation (Table 27.1). Jaguarundi (mean body
mass 5 kg) averages 24.2 20.4 SD km
2
, margay (3.3
kg) 12.3 7.6 SD km
2
, and little spotted cat (2.4 kg)
10.9 9.6 SD km
2
. The expected home range sizes
based on their body mass and metabolic needs (Lind-
stedt et al. 1986) would be 8.9 km
2
, 5.8 km
2
, and 4.2
km
2
, respectively. For ocelots (11 kg) the average and
expected home range sizes are similar, at 20.0 25.2
km
2
and 20.1 km
2
, respectively. Thus, the average
home range sizes of the smaller lowland felids are
2.5 (2.1–2.7) times larger than what would be ex-
pected based on body size. Pairwise comparisons of
mean home range values of male and female of the
smaller species show that male ranges are significantly
larger than those of females (W¼36.00, Tþ¼ 0.00,
T¼ 36.00, P¼0.008), as was noted for ocelots.
There is considerable inter- and intraspecific varia-
tion in home range size among the sympatric felids,
resulting in extensive overlap in size and no signifi-
cant differences among mean adult home ranges (F¼
0.749, P¼0.529, Table 27.1).
Home range size in solitary male carnivores is, on
average, 2.5times that of females, and this value is also
significantly larger than expected based on energy
Table 27.2 Summary of habitat use versus availability of sympatric felids at ENP (Brazil).
Home range within each habitat type (usage within home range, %)
Grassland
savannah
Scrub
savannah Gallery forest
Flooded
grassland
Pasture/
agriculture
Availability 42.50 23.22 1.83 2.61 29.83
Ocelot M1 16.67 (0) 41.06 (55.17) 5.42 (6.90) 0.90 (0) 35.95 (37.93)
Ocelot M2 8.86 (2.70) 28.22 (29.73) 4.01 (27.03) 1.32 (5.41) 57.58 (35.14)
Ocelot F1 6.91 (4.58) 37.39 (45.04) 5.65 (4.58) 1.35 (0) 48.70 (45.80)
Jaguarundi F1 22.25 (22.37) 38.38 (42.11) 2.33 (1.32) 0.45 (0) 36.59 (34.21)
Little spotted cat F1 30.80 (29.41) 40.75 (39.71) 1.30 (0) 0.20 (2.94) 29.95 (27.94)
Pampas cat M1 100.00 (100.00) 0 0 0 0
Pampas cat M2 59.70 (36.74) 3.84 (23.53) 0 (0) 0.04 (0) 36.43 (36.74)
Pampas cat M3 54.49 (58.33) 12.58 (8.33) 2.70 (16.67) 0.16 (0) 30.07 (16.67)
566 Biology and Conservation of Wild Felids
OUP CORRECTED PROOF – FINAL, 1/5/2010, SPi

requirements, suggesting that access to several females
plays an important role in determining male territory
size (Sandell 1989). Thus, the larger home ranges of
male lowland Neotropical felids should be linked to
the distribution of females, rather than to the avail-
ability of food resources (Ludlow and Sunquist 1987;
Manfredi et al. 2006), a pattern typical in the Felidae.
Average home range:felid body mass ratio among
the sympatric assemblage is smallest for ocelots (1.9
km
2
per kg of body mass) and larger for jaguarundis,
little spotted cats, and margays (5.1, 5.0, 3.7 km
2
per
kg of body mass, respectively; Fig. 27.4). Hypotheses
to explain the larger than expected home range sizes
of the smaller species include avoidance of larger
guild-members through the use of areas of lower
prey density, living along the periphery of the domi-
nant species range, or use of different habitats where
they live in sympatry (Creel and Creel 1996; Durant
1998; Palomares et al. 1998). Through such mechan-
isms, larger, dominant species may attain higher
densities and have smaller home range sizes than
do some smaller, subordinate species due to avoid-
ance of the former by the latter (Creel and Creel
1996; Woodroffe and Ginsberg 2005).
Comparative habitat use
Although most felids use open habitats, all lowland
species, except pampas cats, are closely associatedwith
forest cover. At localities with a mosaic of open and
closed environments, including agricultural areas, the
use of closed habitats by ocelots, jaguarundis, mar-
gays, and little spotted cats was similar, and onaverage
74% (40–98%) of the location records, despite the fact
that dense cover sometimes comprised only 11% of
the total vegetation (Ludlow and Sunquist 1987; Caso
1994; Harveson et al. 2004; Trovati 2004; Michalski
et al. 2006b; Oliveira et al. 2008; Silveira, unpublished
data). Margays seem even more strongly associated
with dense cover than are ocelots (Oliveira 1998b,
personal observation). Even jaguarundis, which are
typically associated with open habitats (Oliveira
1998a), use forested habitats part of the time. It
seems that only pampas cats are a predominantly
open-habitat specialist, but they too can be found in
forests (Oliveira 1994; Sunquist and Sunquist 2002).
In the savannahs of Emas, pampas cats prefer
open grassland savannahs, and also use pastoral/
agricultural lands, but avoid forest and scrub/wood-
land savannah. One female jaguarundi favoured
scrub savannah and used forest and pastoral/agri-
cultural lands slightly more than its availability,
whereas one female little spotted cat used forest
and pastoral/agricultural lands less than its avail-
ability; both species avoided the open grasslands.
They used scrub/woodland savannah, open grass-
land, and pastoral/agricultural lands in proportion
to availability. Unexpectedly, jaguarundis and little
spotted cats both avoided forest (Ivlev’s ¼0.3 and
1, respectively), and differed only in that flooded
grassland was strongly avoided by the former and
selected by the latter. Otherwise, all species strongly
avoided flooded grasslands (Table 27.2). The spatial
separation among small felids in Emas could be due
to avoidance of ocelots. In contrast to Emas, 92.3%
and 70% of locations of two male little spotted cats
in the savannah biome of Tocantins State (central
Brazil) were in forests, and only 7.7% and 30% in
savannah. In contrast, jaguarundis in Tocantins
were located 17% of the time in forests and 83% in
savannahs (Trovati 2004). Additionally, in an agri-
culture-forest patch mosaic, both species were
strongly associated with forest cover (Oliveira et al.
2008).
There appear to be parallels within the Neotropical
small felid guild to the differential habitat use by
jackal species in areas of sympatry and allopatry. In
that case, the dominant species narrows the number
of habitats used by the subordinate species, which
avoids the more aggressive competitor (Loveridge
and Macdonald 2002; Macdonald et al. 2004c).
8
7
6
5
Home range : body mass ratio
4
3
2
1
0Ocelot Jaguarundi Margay Little
spotted cat
Figure 27.4 Home range size:body mass ratio (mean ±
standard error) of sympatric lowland felids. (Data sources:
Table 27.1.)
Ocelot ecology and its effect on the small-felid guild in the lowland neotropics 567
OUP CORRECTED PROOF – FINAL, 1/5/2010, SPi

In agricultural landscapes in Brazil, radio-tracked
ocelots, jaguarundis, margays, and little spotted cats
only rarely used open fields, but preferred field bor-
ders near forest cover (Oliveira et al. 2008; Silveira,
unpublished data). These edge habitats may offer
both high rodent abundance and cover. While Geof-
froy’s cats tended to spend more time in densely
vegetated habitat (Manfredi et al. 2006; Lucherini
et al., unpublished data) they may also use open
agricultural fields (Oliveira et al. 2008). Occasional
use of open areas is not restricted to jaguarundis,
pampas cats, and Geoffroy’s cats. Faecal analysis of
little spotted cat diet in secondary Atlantic rainforest
revealed prey from both forest and field/forest edges,
but not from agricultural fields (Facure-Giaretta
2002). In summary, ocelots, margays, little spotted
cats, and Geoffroy’s cats favour dense cover, jaguar-
undis use both open and closed habitats, whereas
pampas cats use predominantly open habitats.
Ocelot foraging ecology, niche
release, and the sympatry issue
Ocelot feeding ecology
Ocelot diet varies throughout its range (Fig. 27.5).
Small mammals (rodents and marsupials <600 g) are
generally the numerically pre-eminent prey (61.3%
28), but their biomass contribution is less (10.7%
11.1). Conversely, large rodents (>1 kg), armadil-
los, ungulates, and possibly sloths and monkeys,
contribute much more in terms of biomass (Table
27.3). Large rodents, especially paca and agouti, are
important in terms both of biomass (27.1% 15.3),
and the frequency with which they are eaten
(11.4% 8.1) (e.g. Bianchi 2001; Aliaga-Rossel
et al. 2006; Moreno et al. 2006), and occurred in
ocelot diets at 11 of 13 study sites. This revises the
earlier perception that small rodents are the major
prey of ocelots (reviewed by Oliveira 1994; Nowell
and Jackson 1996; Sunquist and Sunquist 2002). At
some sites (e.g. Panama and Caratinga-Brazil) sloths
and monkeys also make a considerable contribution
(18.8–26.9%) (Moreno et al. 2006; Bianchi and
Mendes 2007). Ungulates, especially deer (Mazama
spp.), are infrequently taken (3.7% 3.4), but when
they are, can contribute substantially to biomass
intake (29% 26.7).
The MWMP taken by ocelots, while highly vari-
able, averages 1.5 1.1 kg, and is highest in Atlantic
rainforest (Linhares, Espirito Santo State, in South-
east Brazil) at 3.3 kg, and lowest in the Llanos
flood plains of Venezuela, at 0.3 kg (Ludlow and
Sunquist 1987; Bianchi 2001) (Table 27.4). In Lin-
hares this reflects the prevalence of armadillos and
100
80
Total occurrence (%)
60
40
20
0
Site
Mexico
Belize
Costa Rica
Panama-mainland
Panama-BCI
Venezuela
Peru
Linhares-BR
Caratinga-BR
Sta Virginia-BR
Atibaia-BR
Guaraqueçaba-BR
Iguaçu-BR
Other vertebrates
Squamates
Birds
Very large mammals >10 kg
Large mammals 1.5–10 kg
Medium mammals 0.7–1.5 kg
Small mammals 0.1–0.7 kg
Very small mammals <0.1 kg
Figure 27.5 Ocelot diet (total occurrence, %) throughout the Americas. (Data sources: Table 27.3.)
568 Biology and Conservation of Wild Felids
OUP CORRECTED PROOF – FINAL, 1/5/2010, SPi

Table 27.3 Contribution of main prey items by percentage of occurrence (#) and biomass to the diet of ocelots throughout the Americas (values in
percentage). Percentage of occurrence is derived from the number of occurrences of each diet item as a percentage of all occurrences of all diet
items.
Site Large rodents Armadillos Sloths Monkeys Ungulates Small mammals
# Biomass # Biomass # Biomass # Biomass # Biomass # Biomass
Mexico
1
9.43 90.86 90.57 9.14
Belize
2
9.23 33.00 15.38 24.06 4.62 21.45 40.00 5.00
Costa Rica
3
9.09 47.56 81.82 25.68
Panama-
mainland
4
17.31 18.15 11.54 14.52 26.92 41.40 15.38 1.52
Panama-BCI
4
25.00 26.96 2.71 3.43 23.37 40.83 7.07 6.56 1.09 3.19 35.88 3.59
Venezuela
5
1.88 12.34 0.94 8.59 0.94 33.81 87.32 23.25
Peru
6
8.33 54.25 1.67 2.37 85.83 34.9
Linhares-BR
7
23.81 38.53 27.38 26.59 4.76 2.78 8.33 20.74 19.05 1.11
Caratinga-BR
7
15.29 20.68 8.24 12.52 18.82 61.94 49.41 1.09
Sta Virginia-
BR
8
5.26 43.78 5.26 46.78 84.21 9.43
Atibaia-BR
9
1.82 9.34 3.64 13.63 1.82 22.04 56.36 3.96
Guaraquec¸aba-
BR
10
3.23 8.30 3.23 26.02 1.61 28.74 95.16 17.15
Iguac¸u-BR
11
10.94 28.68 10.94 29.18 1.56 11.37 56.25 2.74
Mean
Standard
deviation
11.45 27.07 ±
15.27
8.11
9.56 ±
8.17
19.59 ±
12.45
18.52 ±
11.62
43.00 ±
3.28
7.11 ±
6.84
19.93 ±
25.41
3.68 ±
3.42
29.03 ±
26.73
61.33 ±
28.01
10.66 ±
11.08
Paired t-test 4.114 2.927 3.901 10.834
d.f. 10 8 7 12
P0.002 0.019 0.006 <0.001
Notes:
1
Villa Meza et al. 2002;
2
Konecny 1989;
3
Chinchilla 1997;
4
Moreno et al. 2006;
5
Ludlow and Sunquist 1987;
6
Emmons 1987;
7
Bianchi 2001;
8
Wang 2002;
9
Facure-Giaretta 2002;
10
Vidolin 2004;
11
Crawshaw 1995.
BR, Brazil.
OUP CORRECTED PROOF – FINAL, 1/5/2010, SPi

Table 27.4 Primary ocelot prey by percentage of occurrence (vertebrates) and biomass contribution (mammals), and mean weight of mammalian prey
(MWMP) throughout the Americas, all weights in grams (g).
Habitat/location Main vertebrate prey (occurrence, %) Main mammalian prey (biomass) MWMP n
Species
Prey
mass
All
items
(%) Species
Prey
mass
Mammals
(%) (g)
Tropical dry forest/
Mexico
1
1. Ctenosaura pectinata
2. Liomys pictus
3. Marmosa canescens
800
50
30
36.4
32.3
9.1
1. Odocoileus virginianus
12
2. Liomys pictus
3. Sigmodon mascotensis
15,000
50
180
90.9
2.1
0.7
1465.74 51
Tropical rainforest/
Belize
2
1. Didelphis marsupialis
2. Philander opossum
3. Dasypus novemcinctus
1000
400
3500
25.7
20.3
13.5
1. Agouti paca
2. Dasypus novemcinctus
2. Mazama americana
12
8000
3500
10,400
33.0
24.1
21.4
2238.00 49
Tropical rainforest/Costa
Rica
3
1. Proechimys semispinosus
2. Heteromys desmarestianus
3. Tylomys watsoni
3. Dasyprocta punctata
3. Penelope purpurescens
3. Iguana iguana
375
75
240
4000
1700
3000
34.6
15.4
7.7
7.7
7.7
7.7
1. Dasyprocta punctata
2. Proechimys semispinosus
3. Potos flavus
4000
375
2500
47.6
20.1
14.9
764,55 23
Tropical rainforest/
mainland-Panama
4
1. Bradypus variegatus
2. Dasyprocta punctata
3. Dasypus novemcinctus
3. Iguana iguana
4300
3000
3600
3000
18.8
13.0
8.7
8.7
1. Bradypus variegatus
2. Dasyprocta punctata
3. Nasua nasua
4300
3000
4500
37.6
18.1
15.1
2861.15 49
Tropical rainforest/BCI-
Panama
4
1. Dasyprocta punctata
2. Proechimys semispinosus
3. Bradypus variegatus
3000
375
4300
16.2
15.4
8.5
1. Dasyprocta punctata
2. Choloepus hoffmanni
3. Bradypus variegatus
3000
5700
4300
24.0
22.8
18.0
2852.38 190
Llanos gallery forest/
floodplains/
Venezuela
5
1. Zygodontomys brevicauda
2. Sigmomys alstoni
3. Holochilus brasiliensis
50
40
300
37.6
18.8
14.6
1. Odocoileus virginianus
12
2. Sylvilagus floridanus
3. Holochilus brasiliensis
16000
800
300
21.5
14.0
12.5
349.91 160
OUP CORRECTED PROOF – FINAL, 1/5/2010, SPi

Amazon rainforest/Peru
6
1. Proechimys spp.
2. Oryzomys spp.
3. Birds
280
70
—
31.6
21.5
10.7
1. Proechimys spp.
2. Dasyprocta variegata
3. Agouti paca
280
4000
8000
26.6
20.3
13.6
491.58 62
Tropical rainforest/
Linhares—Brazil
7
1. Dasypus spp.
2. Tupinambis merianae
3. Agouti paca
3200
1200
8000
15.5
8.8
7.4
1. Agouti paca
2. Dasypus spp.
3. Tayassu tajacu
12
8000
3200
7600
31.8
26.6
11.0
3294.94 77
Tropical rainforest/
Caratinga—Brazil
7
1. Calomys sp.
2. Birds
3. Alouatta guariba
25.5
—
5650
12.9
12.1
9.7
1. Alouatta guariba
2. Brachyteles hypoxanthus
3. Agouti paca
5650
13,500
8000
37.9
22.6
13.4
2104.64 60
Tropical rainforest/Sta.
Virgı
´nia—Brazil
8
1. Akodon sp.
2. Monodelphis sp.
3. Colubrid snakes
28.7
48
—
23.3
20.0
13.3
1. Bradypus variegatus
2. Dasypus novemcinctus
3. Akodon sp.
3900
3650
28.7
46.8
43.8
2.4
438,76 17
Tropical rainforest/
Atibaia—Brazil
9
1. Sphiggurus villosus
2. Oligoryzomys nigripes
3. Didelphis spp.
3. Akodon sp.
3. Birds
1325
16.7
1500
28.7
—
19.7
14.8
9.8
9.8
9.8
1. Sphiggurus villosus
2. Mazama americana
12
3. Didelphis spp.
1325
11,560
1500
30.3
22.0
17.2
953.74 34
Tropical rainforest/
Guaraquec¸aba—
Brazil
10
1. Cricetidae
2. Birds
3. Tupinambis merianae
50
—
1200
49.5
21.5
5.4
1. Mazama spp.
2. Cebus/Alouatta
3. Tamandua tetradactyla
9000
4075
5200
28.7
26.0
16.6
505.16 60
Subtropical rainforest/
Iguac¸u—Brazil
11
1. Small rodents
2. Didelphis aurita
2. Dasypus novemcinctus
4. Dasyprocta azarae
50
1500
3300
3200
37.8
8.5
8.5
7.3
1. Dasypus novemcinctus
2. Dasyprocta azarae
3. Didelphis aurita
3300
3200
1500
29.2
24.3
13.3
1236.86 56
1
Villa Meza et al. 2002;
2
Konecny 1989;
3
Chinchilla 1997;
4
Moreno et al. 2006;
5
Ludlow and Sunquist 1987;
6
Emmons 1987;
7
Bianchi 2001;
8
Wang 2002;
9
Facure-Giaretta 2002;
10
Vidolin
2004;
11
Crawshaw 1995;
12
40% of body mass.
n, sample size (number of scats).
OUP CORRECTED PROOF – FINAL, 1/5/2010, SPi

pacas (average body mass of 3.2 kg and 8 kg, respec-
tively), which comprise 40.5% of mammalian items
and 23% of total vertebrates taken (Bianchi 2001).
Conversely, the mean body mass of all vertebrate
prey taken in the Llanos increases to 0.6 kg when
iguanas (Iguana iguana, c. 2.7–3.0 kg) are included
(Ludlow and Sunquist 1987). The figures presented
here double the previously reported (Oliveira 1994)
mean size of ocelot mammalian prey, due to includ-
ing a larger number of studies (e.g. Crawshaw 1995;
Bianchi 2001; Moreno et al. 2006).
Emmons (1987) predicted that the average adult
ocelot would eat 0.56–0.84 kg of meat per day, and
the figures reported in this review either fall within
this range or above it. Given that the average car-
nivore biomass is typically 1–3% of the biomass of
their prey (Ve
´zina 1985), the prey biomass pre-
dicted to sustain one ocelot for a year would
range from 366 to 1100 kg, or 33–100 kg of prey
per kilogram of ocelot, which is below the ratio
reported for the larger jaguars and pumas (Emmons
1987). At all sites where MWMP was >1 kg, prey
mass of at least one of the main vertebrate species
was 0.8 kg. At 10 of 13 sites (76.9%), at least one
main prey of ocelots weighed 0.8 kg (Table 27.4).
However, there was no statistically significant rela-
tionship between MWMP with the body mass of
the three main (in terms of biomass consumed)
mammalian prey (r
2
¼0.344, P¼0.262), but a
significant positive correlation existed between
MWMP and the body mass of the three most nu-
merically prominent vertebrate prey (r
2
¼0.904,
P¼0.006).
Carnivore density is known to be constrained
by metabolic needs and prey abundance (Carbone
and Gittleman 2002). Sunquist (1992) argued that
it would be virtually impossible for a lactating
female to raise cubs on a diet of small mice
(c. 50 g, e.g. Oryzomys, Zygodontomys spp.). Emmons
(1988) reported that a lactating female ocelot
increased her activity from 12–14 h/day to almost
23 h/day, and still was unsuccessful in rearing her
kittens. This prompts the hypothesis that the limit-
ing factor for ocelot persistence may be the avail-
ability of larger sized prey (0.8 kg), such as agouti
and armadillos, to complement the intake of small
mammals.
Dietary comparisons within the
Neotropical felid assemblage
The ocelot’s fundamental niche is much broader
than those of the smaller species, with prey ranging
from small mice to collared-peccary (Tayassu tajacu),
brocket (Mazama spp.), and white-tailed deer (Odo-
coileus virginianus) (Konecny 1989; Bianchi 2001;
Villa Meza et al. 2002). Jaguarundis, margays, and
little spotted cats prey predominantly on small
mammals (<1 kg) (Fig. 27.6), but, like ocelots, they
also prey on birds, squamates, and occasionally lar-
ger mammals. This raises the possibility that exploit-
ative competition occurs between ocelots and
smaller cats as well as among the smaller felids, as
has been noted for larger carnivores (Woodroffe and
Ginsberg 2005). However, among the array of prey,
predators tend to specialize on those that they can
capture most easily, effectively, and profitably (Sin-
clair et al. 2003).
The ocelot is the only felid within its guild that
consistently takes prey weighing 1 kg. The smallest
prey size for all sympatric felid species is in the range of
10–20 g (bats, small mice, and marsupials). However,
maximum prey masses diverge considerably, but are
positively correlated with mean felid mass (r¼0.884,
P¼0.0001, see also Oliveira 1994, 2002a; Oliveira
and Paula, unpublished data). Competition within
carnivore guilds is probably asymmetrical. Larger spe-
cies, whilst supported by larger prey not available to
their smallercompetitors, may deplete thelatter’s food
supply by opportunistic predation on small prey
(Woodroffe and Ginsberg 2005). The diets of sympat-
ric felids (Fig. 27.5 and Fig. 27.6) suggest competitive
pressure from the larger ocelots upon the smaller jag-
uarundis, margays, and little spotted cats.
Canine diameter, a useful measure of ecological seg-
regation among carnivores (Dayan et al. 1990), differs
significantly between sympatric Neotropical small fel-
ids, excepting margays and jaguarundis (Oliveira, un-
published data). Mean upper canine diameters
correlate with felid mean body mass (r¼0.980, P¼
0.0199, Fig. 27.7) and average MWMP (r¼0.987, P¼
0.0125). Conversely, there is no correlation between
canine diameter and maximum prey mass. Thus, ca-
nine diameter seems to reflect the modal, rather than
the maximal, size of prey.
572 Biology and Conservation of Wild Felids
OUP CORRECTED PROOF – FINAL, 1/5/2010, SPi

Food niche overlap values between ocelots and
margays average 0.41 (0.68–0.19, n¼3), between
ocelots and little spotted cats it is 0.66 (0.81–0.45, n
¼3), and between ocelots and jaguarundis it is 0.49.
Given the differences in body size and canine dimen-
sions, these values are high. The highest overlap for
ocelots is with the smallest species, the little spotted
cat, but small sample sizes (and scarcity of studies of
sympatric felids) may distort these comparisons.
A pairwise comparison of the food niches of the
smaller species shows an average overlap of 0.62
(0.253, range 0.18–0.91) (Konecny 1989; Olmos
1993; Facure and Giaretta 1996; Facure-Giaretta
2002; Nakano-Oliveira 2002; Wang 2002; Felipe
2003; Rocha-Mendes 2005). Therefore, the available
datasets provide no evidence of any significant dif-
ference in overlap in food niche between ocelots
and other sympatric felids (t¼0.757, d.f. ¼14,
P¼0.461) or among the smaller species (F¼0.841,
P¼0.484).
The obvious potential for competition amongst
these felids may be offset by the considerable differ-
ence in MWMP between ocelots and the smaller
species, and despite their overlap on more abundant
prey, each may be selecting specific prey (Pimm
1991). Dietary overlaps varying from very high to
very low have been reported for Neotropical felids
(Oliveira 2002a; Moreno et al. 2006). In the high
10
9
8
7
Canine diameter (mm)
6
5
4
3Little spotted
cat (2.4 kg)
Margay
(3.3 kg)
Species body mass
Jaguarundi
(5 kg)
Ocelot
(11 kg)
Figure 27.7 Correlation of mean upper canine diameter
(±SD) and mean body mass for little spotted cats (n= 28),
margays (n= 34), jaguarundis (n= 24), and ocelots (n= 47)
in Brazil (r= 0.980, P= 0.0199). (Data source: Oliveira,
unpublished data.)
100
80
Total occurrence (%)
60
40
20
0
Species/sites
Py-Mexico
Py-Belize
Py-Venezuela
Py-ES/Br
Py-SP/Br
Py-PR/Br
Lw-Belize
Lw-Venezuela
Lw-ES/Br
Lw-SP/Br
Lw-PR/Br
Lt-PI/Br
Lt-ES/Br
Lt-SP1/Br
Lt-SP2/Br
Lt-SP3/Br
Lt-PR/Br
Lt-SC/Br
Other vertebrates
Squamates
Birds
Large mammals 1.5–10 kg
Medium mammals 0.7–1.5 kg
Small mammals 0.1–0.7 kg
Very small mammals <0.1 kg
Figure 27.6 Jaguarundi (Py), margay (Lw), and little spotted cat (Lt) main prey items (total occurrence, %) in the Americas.
(Data sources: Py: Bisbal 1986; Mondolfi 1986; Konecny 1989; Facure and Giaretta 1996; Felipe 2003; Guerrero et al.
2002; Oliveira 2002a; Rocha-Mendes 2005; Lw: Mondolfi 1986; Konecny 1989; Wang 2002; Felipe 2003; Rocha-Mendes
2005; Lt: Olmos 1993; Facure and Giaretta 1996; Facure-Giaretta 2002; Nakano-Oliveira 2002; Wang 2002; Rocha-Mendes
2005; M. Tortato, unpublished data.)
Ocelot ecology and its effect on the small-felid guild in the lowland neotropics 573
OUP CORRECTED PROOF – FINAL, 1/5/2010, SPi

Andes dietary overlap of pampas cats and Andean
cats (Leopardus jacobita) was very high (82%), but
they nonetheless emphasized different principal
prey (Walker et al. 2007; see Marino et al., chapter
28, this volume). Furthermore, high overlap in one
niche dimension might be compensated by low
overlap in other axes, such as space use and time of
activity, or even seasonal temporal variation in hunt-
ing of prey (Fig. 27.8), thus facilitating the sympatry
(Konecny 1989, Oliveira et al. 2008; Silveira, unpub-
lished data).
MWMP differs significantly between sympatric
Neotropical felids (H¼20.362, d.f. ¼3, P<0.001;
Table 27.4, Fig. 27.9). Pairwise comparisons are also
significantly different (P<0.05) for every combina-
tion except for margay–little spotted cat and mar-
gay–jaguarundi. Margays are intermediate between
the other two species, in so far as they are closer in
size and appearance to little spotted cats, but closer
to jaguarundis in feeding morphology (e.g. canine
diameter, jaw length; Kiltie 1984; Oliveira and Cas-
saro 2005; Oliveira, unpublished data). Not surpris-
ingly, the largest difference between the smallest and
largest MWMP is for the ocelot (9.4-fold), whereas
the range of prey sizes taken by jaguarundi, margay,
and little spotted cats is 2.3-, 3.9-, 4.9-fold, respec-
tively. Mean MWMP represents 13.7% of ocelot mass
(3.8–30%), 6.2% of jaguarundi mass (4–9.1%), 7.3%
of margay mass (3.1–12.2%), and 6.3% of little
spotted cat mass (2.4–11.9%).
Assessing the potential for competitive
release
While ungulates, armadillos, and large rodents are
taken by pumas and jaguars (see Oliveira 2002a),
and are favoured by humans as well (Jorgenson and
Redford 1993; Oliveira 1994), these prey are also
taken by ocelots (Table 27.3). Thus, there is potential
for niche overlap between ocelots and big cats. Alter-
natively, ocelots could be taking larger prey only in
areas where jaguars are absent (Moreno et al. 2006).
To test for a jaguar dietary release on mesopredators,
as suggested by Moreno et al. (2006), we compared
ocelot MWMP from areas with robust jaguar popula-
tions (n¼8) to those areas with marginal occurrence
or absence (n¼5) of jaguars, and found no dietary
differences (t¼0.951, d.f. ¼11, P¼0.362). In fact,
the highest MWMP was from an area with a substan-
tial jaguar population (Bianchi 2001; Garla et al.
2001). MWMP for ocelots was 11.8 11.0% of the
jaguar’s and 19.6 13.4% of the puma’s mean prey
body mass where all three were sympatric (Fig. 27.10;
Emmons 1987; Crawshaw 1995; Chinchilla 1997;
Brito 2000; Bianchi 2001; Garla et al. 2001). There-
fore, it seems unlikely, on the basis of their MWMP,
that ocelot compete substantially with the big cats.
To test for dietary release in the absence of ocelots,
we compared the MWMP of little spotted cats
from areas with moderate to robust ocelot popula-
tions (n¼2) to those where ocelots were rare or
absent (n¼3), and found no dietary differences
(t¼1.627, d.f. ¼3, P¼0.202). Although sample
sizes were small, this suggests that, just as jaguars do
not constrain ocelot diets, ocelots do not appear to
be constraining the diets of little spotted cats. Fur-
thermore, the average MWMP of little spotted cats is
9.7% of the ocelot’s (references in Table 27.3 and Fig.
27.6), suggesting considerable scope for resource par-
titioning between them.
Factors affecting ocelot density and
dynamics
A species’ abundance is influenced by proximate fac-
tors such as prey density, habitat, presence of poten-
tial competitors and predators, and ultimately by
environmental variables, especially those that influ-
ence prey productivity.
Abundance rank
The larger ocelot seems to be the most versatile felid
within the lowland, small-medium Neotropical
assemblage. In Brazil, regardless of habitat type
(n¼24), ocelots were the most abundant cat in
84.2% of all areas examined (Fig. 27.11). Of the
smaller sympatric species, margays usually ranked
second in abundance in rainforests and last in savan-
nahs, where jaguarundis were most abundant. Little
spotted cats usually ranked second or first only in
areas where ocelots were absent or rare (Oliveira et al.
OUP CORRECTED PROOF – FINAL, 1/5/2010, SPi
574 Biology and Conservation of Wild Felids

30
(a)
25
20
Pampas cat
Jaguarundi
Ocelot
Activity (%)
15
10
5
0
0001–0200
0201–0400
0401–0600
0601–0800
0801–1000
1001–1200
Time (h)
1201–1400
1401–1600
1601–1800
1801–2000
2001–2200
2201–2400
25
(b)
20
15
Ocelot absent
Ocelot present
Activity (%)
10
5
0
0001–0200
0201–0400
0401–0600
0601–0800
0801–1000
1001–1200
Time (h)
1201–1400
1401–1600
1601–1800
1801–2000
2001–2200
2201–2400
Figure 27.8 Activity patterns of (a) ocelots, jaguarundis, and pampas cats in the savannahs of ENP, central Brazil; and (b)
little spotted cats in mixed broadleaf–pine forests in the presence (Sa
˜o Francisco de Paula National Forest) and absence
(Serra do Tabuleiro State Park) of ocelots in southern Brazil, determined by photographic records. In Emas ocelots are
nocturnal, whereas jaguarundis and pampas cats are predominantly diurnal. Differences in little spotted cat activity between
areas in Fig. 27.8b could be related to avoidance of predominantly nocturnal ocelots (Data sources: Oliveira et al. 2008;
Silveira, unpublished data.)
Ocelot ecology and its effect on the small-felid guild in the lowland neotropics 575
OUP CORRECTED PROOF – FINAL, 1/5/2010, SPi

2008, submitted). In the temperate Southern Cone,
where the ocelot is absent, Geoffroy’s cats might take
its ecological role. Geoffroy’s cat abundance-ranked
first in four locations where this species was sympat-
ric with other small cats and ocelots were absent
(Lucherini and Luengos Vidal 2003; Cuellar et al.
2006; Lucherini et al., unpublished data; Pereira
et al., unpublished data).
Brazil-SC
Brazil-PR
Brazil-SP3
Brazil-SP2
Brazil-SP1
Brazil-ES
Venezuela
Belize
0 500 1000 1500
MWMP (
g
)
2000 2500
Jaguarundi
Margay
Little spotted cat
Ocelot
3000 3500
Sites
Figure 27.9 Mean weight of mammalian prey (MWMP, g) of jaguarundis, margays, little spotted cats, and of ocelots in
areas of sympatric occurrence with the smaller species. (Data sources: same as Figs. 27.5 and 27.6.)
16
14
12
10
Ocelot
Jaguar
Puma
Mean prey weight (kg)
8
6
4
2
0
Costa Rica Peru
Site
Linhares
-Brazil
Iguaçu-Brazil
Figure 27.10 Mean weight of
ocelet mammalian prey, and mean
weight of jaguar and puma
vertebrate prey in areas of sympatry.
(Data source: Emmons 1987;
Crawshaw 1995; Chinchilla 1987;
Brito 2000; Bianchi 2001; Garla
et al. 2001.)
576 Biology and Conservation of Wild Felids
OUP CORRECTED PROOF – FINAL, 1/5/2010, SPi

The effect of environmental variables
Maffei et al. (2005) reported a positive but insignifi-
cant correlation (r
2
¼0.332) between ocelot density
and precipitation. Expanding their dataset and
grouping estimates from the same vegetation type
of the same region led to a significant and stronger
relationship between these two variables (r
2
¼0.415,
P¼0.032, n¼11). However, if the data were not
grouped in this way, the correlation disappeared (r
2
¼
0.114, P¼0.145, n¼20), as did the correlation
between precipitation and home range size. Ocelot
density and home range size did not correlate with
habitat types (grouped in forest, transitional, and
open formations) (F¼0.309, P¼0.738).
The influence of prey
Prey biomass is known to influence carnivore space
use and density (Litvaitis et al. 1986). Carbone and
Gittleman (2002) highlighted the remarkable consis-
tency in average population density of carnivores in
general (despite their ecological variations) to that of
prey biomass.
Analysis of food habits clearly indicates that larger
prey (0.8 kg) is especially important for ocelots. The
only study on predator–prey relations for a small
Neotropical felid that combines data on predator
density and prey density or biomass is of Geoffroy’s
cats in Argentina. The cat’s density declined from
0.29 km
2
to 0.03 km
2
following the numerical
decline of its main prey from 0.56 km
2
to 0.06
km
2
(Pereira et al. 2006).
A test of MWMP against ocelot density at five sites
(using Konecny [1989] MWMP for Dillon [2005]
density estimation; Emmons 1987, 1988; Ludlow
and Sunquist 1987; Crawshaw 1995; Moreno et al.
2006), revealed no correlation (r¼0.247, P¼0.688).
Although the dataset is very limited, it does suggest
that mean prey mass is not a predictor of ocelot
density, although the density of main prey by per-
centage of occurrence might be.
Ocelot density and comparisons with the
smaller species
Ocelot density varies considerably between areas,
from 0.08 to 1.0 individuals per km
2
, and averaged
0.31 0.22 (n¼22) (Ludlow and Sunquist 1987;
Emmons 1988; Jacob 2002; Trolle and Ke
´ry 2003,
2005; Dillon 2005; Maffei et al. 2005; Cuellar et al.
2006; Di Bitetti et al. 2006; Rocha 2006; Oliveira et al.
2008, submitted; Moreno and Kays, unpublished
data). Ocelot densities were higher than those of
the smaller jaguarundi, Geoffroy’s cat, margay, and
100
80
60
Areas where species is found (%)
40
20
0
1st 2nd 3rd 4th 5th
Ocelot
Margay
Little spotted cat
Jaguarundi
Abundance rank
Figure 27.11 Lowland
sympatric felid species ranked
according to their order of
abundance (1, most abundant;
5, least abundant) at each site
where they occur in forests (n=
12 areas), savannahs (n= 5),
transitional areas/mosaic (n=
5), semi-arid scrub (n= 1), and
Pantanal flood plains (n=1)in
Brazil. (Adapted from Oliveira
et al. 2008.)
Ocelot ecology and its effect on the small-felid guild in the lowland neotropics 577
OUP CORRECTED PROOF – FINAL, 1/5/2010, SPi

little spotted cat combined (mean, 0.187 km
2

0.159 km
2
,n¼22; Cuellar et al. 2006; Oliveira
et al. 2008, submitted; Caso, unpublished data).
Thus, density estimates of ocelots and those of the
smaller species are significantly different (t¼2.161,
d.f. ¼42, P¼0.036). Density estimates of the smaller
Neotropical felids (Oliveira et al. 2008, submitted)
would suggest that they should, according to Yu
and Dobson (2000), be categorized as ‘widespread,
but everywhere small population size’.
Carbone and Gittleman (2002) report strong nega-
tive correlation in average population density of
carnivores in relation to carnivore mass and strong
positive correlation with prey biomass, despite varia-
tion in species’ ecology. However, in Brazil, felid
density estimates are not correlated with body mass
(Oliveira et al., submitted). Based on their body size,
the small cats should have smaller home ranges and
higher population densities than ocelots. In fact, the
smaller felids have larger than expected home ranges
and lower than expected population densities. Their
expected population densities, based on Carbone
and Gittleman (2002), would be 0.91 km
2
for little
spotted cats, 0.69 km
2
for margays, 0.48 km
2
for
jaguarundis, and 0.54 km
2
for Geoffroy’s cats,
which are, on average, 3.5 times larger than their
observed mean density. Conversely, the expected
density of ocelots (0.24 km
2
) is 1.3 times less than
its observed mean. This suggests that the smaller
sympatric felids may deviate from expected densities
due to the effect of the larger ocelots.
Interestingly, the smaller felids reach higher densities
only in areas where ocelots are either absent or in low
numbers. Ocelot densities, on the other hand, are all
from areas where both larger and smaller felid species
are found. Additionally, even the highest densities of
jaguarundis, margays, little spotted cats, and Geoffroy’s
cats (0.2–0.4 individuals per km
2
) are lower than that of
the ocelot (0.5–1.0 individuals per km
2
) (Emmons
1988; Cuellar et al. 2006; Oliveira et al. 2008, submitted;
Caso, unpublished data; Moreno and Kays, unpub-
lished data). The density of smaller cats in areas where
ocelot density was <0.1 km
2
or absent was significant-
ly higher than where ocelot density was >0.1 km
2
(t¼
5.564, d.f. ¼19, P<0.001). Interspecific predation
and competition, major factors influencing carnivore
population density (Fuller and Sievert 2001; Donadio
and Buskirk 2006), seem to strongly influence the pop-
ulation dynamics of lowland Neotropical felids. The
larger and dominant species, contrary to expectation,
attain higher densities than smaller subordinate spe-
cies, which range far more widely than the larger guild
members, presumably to avoid intraguild predation
(Palomares and Caro 1999; Woodroffe and Ginsberg
2005). Dominance rank is highly correlated with
body mass and has been detected in other assemblages
(Brown and Maurer 1986; Buskirk 1999; French and
Smith 2005). Body-size differences associated with in-
terference competition normally vary by a factor of
1.5–4, but could be as high as 7 (Buskirk 1999). The
frequency and intensity of carnivore interspecific kill-
ing reaches its maximum when the larger species is 2–
5.4 times larger than the other guild members, as at this
size range there is a higher potential for high diet over-
lap and, in this way, a higher benefit for the larger
species to eliminate the smaller potential competitors
(Donadio and Buskirk 2006). The ocelot is 2.2–4.6
times larger than the jaguarundi, margay, and little
spotted cat. Dietary overlap of ocelots with these smal-
ler sympatric species is also high (as shown above) and,
as dietary overlap seems to be a factor likely to motivate
interspecific killing (Donadio and Buskirk 2006), we
speculate that intraguild killing, or its potential (Woo-
droffe and Ginsberg 2005), might be the mechanism
by which ocelots affect small cat dynamics in the
neotropics.
Ocelot density does not appear to be impacted by
the presence of the larger puma and jaguar (Oliveira et
al., unpublished data), as predicted by the mesopreda-
tor release theory of Crooks and Soule
´(1999). Con-
versely, ocelot density appears to impact negatively
the numbers of smaller little spotted cat, margay, jag-
uarundi, and Geoffroy’s cat (Fig. 27.12) in what has
been called the ‘ocelot effect’ (or the ‘pardalis effect’;
Oliveira et al. 2008, unpublished data). If true, then
this dominant mid-sized carnivore might determine
the dynamics of the mesopredator community in the
neotropics, rather than the top predators, as predicted
by the mesopredator release theory. In this case, the
prediction is that as ocelot numbers decrease small
felid numbers should rise due to reduced intraguild
predation by ocelots (Fig. 27.13). This is what has
been observed over the past 8 years (2000–07) at a
site in southern Brazil (Oliveira et al. 2008).
The emerging pattern within the lowland felid guild
is that the density of ocelots is not influenced by that
578 Biology and Conservation of Wild Felids
OUP CORRECTED PROOF – FINAL, 1/5/2010, SPi

of other felids, but rather by availability of prey and
adequate habitat. In turn, the density of the smaller
sympatric jaguarundis, Geoffroy’s cats, margays, and
little spotted cats seem to be influenced first by ocelot
numbers and then by other environmental factors
typically associated with habitat and prey base (Oli-
veira et al. 2008, unpublished data). Carnivore assem-
blages elsewhere show similar trends (see Macdonald
et al., Chapter 1, this volume). Cheetahs (Acynonyx
jubatus) and African wild dogs (Lycaon pictus)avoid
areas with high prey densities because these are areas
where their intraguild competitors and predators such
as lions (Panthera leo) and spotted hyenas (Crocuta
crocuta) reach high population densities (Laurenson
et al. 1995; Creel and Creel 1996; Mills and Gorman
1997; Palomares and Caro 1999). A similar relation-
ship between the density of ocelots and the smaller
felid species is expected and would have conservation
implications for the latter species. Even the largest
protected areas, that typically harbour ocelot popula-
tion densities >0.1 km
2
, may not hold viable popula-
tions of the smaller species, whose conservation
would be dependent upon their protection outside
reserves, in ever dwindling natural habitats (Oliveira
et al. 2008).
Concluding remarks
Ocelots are generalists, highly adaptable, and the
dominant felid in the lowland mesopredator assem-
blage of the neotropics. According to first principles,
we had expected environmental variables to play a
key role in determining ocelot numbers through pri-
mary productivity, and thus the influence of prey
abundance on their home range and population
density. However, although rainfall was correlated
with ocelot density, it was not correlated with
home range size, so these relationships cannot be
assumed and merit further exploration.
Among the sympatric small cats, small rodents
were the prey whose undigested remains were most
commonly found in droppings; however, ocelots
were the only species that consistently took larger
prey (>1 kg), such as agoutis, pacas, armadillos,
sloths, and monkeys. Small rodents contributed rela-
tively little to the ocelot’s intake of biomass, and the
abundance of larger prey emerges as the likely limit-
ing factor on their persistence. Furthermore, we de-
tected no evidence of competitive dietary niche
release within the lowland Neotropical felid assem-
blage, at least in so far as the absence of larger species
had no observable effect on the diet of the smaller
species. Body mass of main vertebrate prey (as de-
fined by occurrence in diet) emerged as a useful pre-
dictor of the average mass of prey taken by ocelots.
The patterns of ocelot habitat use revealed by our
review indicate a greater degree of plasticity than
often thought. Ocelots have been associated not
only with dense cover (from pristine to highly dis-
turbed), but also use adjacent open areas, including
boundary areas of agricultural fields. This versatility
0.12
(a)
0.10
0.08
0.06
0.04
0.02
0.00 0.0 0.2 0.4 0.6 0.8
Ocelot density (km2)
Jaguar density (km
2
)
12
(b)
10
8
6
4
2
0
–2 0510
Ocelot (number of individuals)
Little spotted cat (number of individuals)
15 20 25
Figure 27.12 Demographic interactions of ocelots with
(a) potential predators, jaguars; and (b) potential
competitors, little spotted cats. (Modified from Oliveira
et al. 2008, unpublished data.)
Ocelot ecology and its effect on the small-felid guild in the lowland neotropics 579
OUP CORRECTED PROOF – FINAL, 1/5/2010, SPi

may explain why ocelots are generally the most
abundant species in most of the habitats in which
they occur.
Our synthesis of available data also suggests
that ocelot numbers have not been detectably
influenced by either the larger jaguar and puma
(potential predators) or by the smaller species (po-
tential competitors). Density estimates of ocelots
are significantly higher than those of the smaller
jaguarundis, Geoffroy’s cats, margays, and little
spotted cats. The pattern that emerges is apparent-
ly one of coexistence between ocelots and the
smaller species, but the lower densities of the
smaller felids may reflect intraguild predation by
ocelots, or the threat of it. If these premises hold
true, then the ‘ocelot effect’ may be a key factor
shaping the dynamics of the small-felid commu-
nity of the lowland neotropics.
Acknowledgements
Project ‘Gatos do Mato—Brasil’ (Wild Cats of Brazil
Project) was funded by Brazil’s National Environ-
mental Fund (FNMA conv. 001/04), with
additional support from Fundac¸a
˜o O Botica
´rio de
Protec¸a
˜oa
`Natureza, Conservation International-
Brazil, FATMA, FAPEMA, ISEC-Canada, and Instituto
Pro
´-Carnı
´voros, which we acknowledge. We also
thank all 12 partner institutions and all project
members. Emas National Park Carnivore Project
was funded by Memphis Zoo, FNMA, Earthwatch
Institute, and Jaguar Conservation Fund (JCF). The
invaluable comments of Renata Leite-Pitman, An-
drew Loveridge, David Macdonald, three anony-
mous reviewers, and the help of Nata
´lia Torres
(JCF), as well as the unpublished data provided by
several researchers, are deeply appreciated.
margay
ocelot
jaguarundi
little spotted cat
Figure 27.13 The ‘ocelot effect’ (or the ‘pardalis effect’) occurs when the dominant mid-sized carnivore (ocelots) impact
the dynamics of the mesopredator community in tropical America. As ocelot numbers decline, smaller felids numbers
increase due to reduced intraguild predation (pictures not to scale). #‘Projeto Gatos do Mato—Brasil’.
580 Biology and Conservation of Wild Felids
OUP CORRECTED PROOF – FINAL, 1/5/2010, SPi