Niche Relationships of Carnivores in a Subtropical Primary Forest in
Southern Taiwan
Po-Jen Chiang1,
*, Kurtis Jai-Chyi Pei2
, Michael R. Vaughan1
, and Ching-Feng Li3
1
Department of Fisheries and Wildlife Sciences, Virginia Polytechnic Institute and State Univ., Blacksburg, VA 24061, USA
2
Institute of Wildlife Conservation, National Pingtung Univ. of Science and Technology, Neipu, Pingtung 912, Taiwan
3
Department of Botany and Zoology, Masaryk Univ., 61137 Brno, Czech Republic
(Accepted December 28, 2011)
Po-Jen Chiang, Kurtis Jai-Chyi Pei, Michael R. Vaughan, and Ching-Feng Li (2012) Niche relationships of
carnivores in a subtropical primary forest in southern Taiwan. Zoological Studies 51(4): 500-511. Carnivores
are at the higher trophic levels and have garnered much attention in conservation and management efforts. In
this study, we attempted to understand resource partitioning among sympatric carnivores existing in a primary
forest with minimal human disturbance in southern Taiwan by camera trapping after the disappearance of the
top carnivore, the clouded leopard (Neofelis nebulosa). Niche relationships were studied in terms of habitat,
diet, and time dimensions. Six carnivore species were recorded, but the Asiatic black bear (Ursus thibetanus
formosanus) was very rare. Canonical correspondence analysis of photographic rates and habitat factors of the
other 5 carnivores showed that elevation was the strongest factor explaining the composition of the carnivore
community in the habitat dimension. Carnivores could be divided into 3 groups. The low- to mid-elevation
group consisted of the gem-faced palm civet (Paguma larvata taivana) and crab-eating mongoose (Herpestes
urva formosanus) which had contrasting activity patterns and different diets; the mid- to high-elevation group
consisted of yellow-throated marten (Martes flavigula chrysospila) and Siberian weasel (Mustela sibirica
taivana). These 2 mustelids had similar diets, but Siberian weasels tended to avoid yellow-throated martens
temporally. The Formosan ferret badger (Melogale moschata subaurantiaca) was more widely distributed along
the elevational gradient. Ferret badgers partitioned resource use in either diet, activity patterns, or other habitat
gradients from the other carnivores. Niche segregation and complementary resource use were observed in
these 5 carnivores. http://zoolstud.sinica.edu.tw/Journals/51.4/500.pdf
Key words: Activity pattern, Complementary resource use, Niche segregation, Sympatric carnivores.
*To whom correspondence and reprint requests should be addressed. Tel: 886-4-23595845; 886-919390767.
Fax: 886-4-23595845. E-mail:neckrikulau@gmail.com
When ecologically similar species are in
sympatry, different species often partition the use
of resources, resulting in niche differentiation to
facilitate species coexistence (Schoener 1974,
Gordon 2000). Sympatric carnivores were found
to have different habitat uses (Fedriani et al. 1999,
Loveridge and Macdonald 2003, Vieira and Port
2007), different prey sizes (Karanth and Sunquist
2000, McDonald 2002, Scognamillo et al. 2003),
different prey species (Karanth and Sunquist
2000), or different activity patterns (Fedriani et al.
1999, Loveridge and Macdonald 2003, Vieira and
Port 2007). Schoener (1974) found that resource
partitioning was principally along 3 dimensions
(i.e., habitat, diet, and time) and that the habitat
dimension was the relatively most important
axis followed by food-type and then by temporal
dimensions. When species were similar in 1
dimension, complementary resource use would
imply dissimilarity in other dimensions to reduce
interspecific competition (Schoener 1974 1983).
There are 11 wild carnivore species in
Taiwan, including 2 felids, 5 mustelids, 2 viverrids,
1 herpestid, and 1 ursid. These 11 carnivores
Zoological Studies 51(4): 500-511 (2012)
500
variably occur from coastal to alpine areas in
Taiwan, and most are sympatric. However, the
largest felid, the Formosan clouded leopard
(Neofelis nebulosa) (Chiang 2007), and the largest
mustelid, Eurasian otter (Lutra lutra chinensis) (Lin
2000), are likely extirpated from the wild in Taiwan.
The largest carnivore in Taiwan, the Asiatic
black bear (Ursus thibetanus formosanus), has
also become very rare compared to its historical
distribution and abundance (Hwang et al. 2006).
The remaining larger carnivores in Taiwan are the
leopard cat (Prionailurus bengalensis chinensis)
with a body weight of 2.7-5.5 kg and the yellowthroated
marten (Martes flavigula chrysospila)
with a body weight of up to 3 kg. Disappearance
of top carnivores may result in increased
smaller carnivores, i.e., mesopredator release
(Terborgh et al. 1999). Investigating the resource
utilization of these mesopredators and how they
partition resources could help understand the
mechanisms influencing the vertebrate community,
particularly the current niche relationships of
these mesopredators living in areas without top
carnivores. Such information could be used to
assess reintroduction plans of the disappeared
top carnivores (Terborgh et al. 1999) and also
provide baseline information for conservation and
management of these mesopredators.
However, only a few studies were conducted
on carnivores in Taiwan and most focused on
individual species. Very few studies addressed
niche relationships of sympatric carnivores in
Taiwan. Chuang and Lee (1997) studied food
habits of the Formosan ferret badger (Melogale
moschata subaurantiaca), crab-eating mongoose
(Herpestes urva formosanus), and lesser Oriental
civet (Viverricula indica pallida) in sympatry and
found that these 3 species all mainly fed on
invertebrates but that different invertebrate taxa
had dissimilar relative importance values. Wu
(1999) also found substantial diet overlap, with
some degree of food-type segregation, between
the Siberian weasel (Mustela sibirica taivana) and
ferret badger. On the other hand, scat analysis
revealed that the yellow-throated marten and
Siberian weasel had similar diet niche breadths,
and their diets overlapped to a high degree (Tsai
2007). However, their complementary resource
use in other niche dimensions was not addressed.
The only study involving temporal dimensions
showed that the crab-eating mongoose was diurnal
in contrast to the nocturnal ferret badger, lesser
Oriental civet, and gem-faced palm civet (Paguma
larvata taivana) (Chen et al. 2009). Chen et al.
(2009) also studied habitat characteristics affecting
the presence and absence and found some
differences among these 4 sympatric carnivores in
secondary forests in southern Taiwan.
Taiwan is an orogenic island with the highest
elevation of 3952 m. Rapid development has
encroached upon most lowland areas. The Tawu
Mountain (TWM) area in southeastern Taiwan
contains the largest remaining preserved primary
rainforest with minimum human disturbance.
Thus, it provides a good area to study sympatric
carnivores under the most natural conditions
which are little studied in Taiwan. The objective
of this study was to investigate niche relationships
of carnivores currently living in the TWM area
to provide information for conservation and
management of these carnivores in the changing
landscape of Taiwan. We predicted that niche
segregation and complementary resource use
would be observed for sympatric carnivores.
MATERIALS AND METHODS
Study area
The study site in the TWM area in southern
Taiwan consists of 2 adjacent protected areas, i.e.,
the Tawu Mountain Nature Reserve (48,000 ha)
and Twin Ghost Lake Important Wildlife Area
(45,000 ha). The elevation ranges 130-3100 m,
and the area is primarily comprised of pristine
rainforests, e.g., from tropical, subtropical, to
temperate forest types along the elevational
gradient consisting of the 4 major forest types
of Ficus-Machilus, Machilus-Castanopsis,
Quercus, and Taiwan hemlock (Tsuga chinensis
var. formosana) (Su 1984). All of the forests are
evergreen and have minimal human disturbance
and hunting activities.
Camera trapping
Intensive camera trapping effort with 185
camera trapping sites were conducted in the TWM
area in 2001-2004 (Fig. 1). Camera trap sites
were selected based on stratified sampling of the
4 elevation ranges of < 1200 (72 sites), 1200-
1900 (43 sites), 1900-2500 (49 sites), and 2500-
3100 m (21 sites), corresponding to the 4 major
forest types mentioned above. The photographic
rate of each carnivore species was used as a
surrogate of the relative population abundance
as it was shown to be representative (Carbone et
Chiang et al. – Niche Relationships of Carnivores in Southern Taiwan 501
al. 2001, O’Brien et al. 2003, Liang 2005, Rovero
and Marshall 2009). Higher occurrence rates at
a camera trap site also indicate a higher usage of
that particular habitat. Identifiable individuals were
counted as different occurrence events (O’Brien
et al. 2003). To avoid dependency of consecutive
photos due to lingering animals, consecutive
photographs of the same individual within 1 h
were counted as single occurrence event (Yasuda
2004, Chiang et al. 2007). However, yellowthroated
martens often formed small groups of
2 or 3 individuals. Their photographic events
were counted in groups even though there were
multiple distinguishable individuals in the same or
consecutive photographs.
Habitat-use patterns
A canonical correspondence analysis
(CCA) (Ter Braak 1986) was used to describe
the carnivore community structure and its
relationship with habitat characteristics using
the R software (R Development Core Team
2009) with the “vegan” package (Oksanen et al.
Fig. 1. Locations of camera trap sites in the Tawu Nature Reserve and Twin Ghost Lake Important Wildlife Area (at Tawu Mountain
(TWM)). Four stratified elevation ranges are illustrated in the TWM area, and camera trap sites were sampled in these 4 stratified
elevation ranges corresponding to 4 major vegetation types.
137 - 1200
1200 - 1900
1900 - 2500
2500 - 3091
camera trap site
TWM Boundary
0 5 10 20 km
Elevation (m)
N
502 Zoological Studies 51(4): 500-511 (2012)
2008). Thirty-one habitat variables belonging to 4
categories, of 1) elevation, greenness, wetness,
and terrain shape, 2) ruggedness, 3) forest
understory, and 4) forest structure, were derived
from a 40 × 40-m digital elevation model and
measured in the field within a 17.8-m radius of
the camera trap site (Table 1). A factor analysis
was separately conducted on the 4 categories of
habitat variables based on a correlation matrix
using orthogonal rotation to reduce the number
Table 1. Four categories of habitat variables in the Tawu Nature Reserve and Twin Ghost Lake Important
Wildlife Area (Tawu Mountain), Taiwan (2001-2004). The elevation and terrain categories were derived from
a 40 × 40-m digital elevation model (DEM) as meso-scale habitat variables (except EVEN). EVEN in the
terrain category and other habitat variables were measured within a 17.8-m radius of the camera trap sites
as micro-scale habitat variables
Category Habitat covariate Description
Elevation/greenness/wetness ALT Elevation (m) from a global positioning system and differential post processing
NDVI Normalized difference vegetation index based on SPOT 4 satellite images (at
a 12.5-m resolution).
MOISTGRD Moisture gradient based on DEM aspect and proximity to rivers and valleys,
10 levels: 1 (wettest)- 10 (driest on south-facing slopes)
ASR Annual solar radiation based on Fu and Rich (2002), which considered
atmospheric conditions, elevation, aspect, and influences of the surrounding
topography
RIVERDIST Distance to the nearest rivers or lakes
Terrain shape/ruggedness SLOPEPOS Slope position 0 (valley)- 100 (ridge), ratio of elevation difference to the valley
and ridge
SLOPEDEM Slope from DEM in percentage (ArcGIS 9.2)
SLOPESTD Standard deviation of slopes (in percentage) within neighboring 3 × 3 cells (i.e.,
120 × 120 m)
CLIFFDIST Distance to nearest cliff, defined as a slope of > 45° with an area of > 1.44 ha
(i.e., 3 × 3 cells or 120 × 120 m)
CLIFFCOUNT Number of cliff cells (slope > 45°) within a 25 × 25 cell window (i.e., 1 × 1 km
or 100 ha)
TSI Terrain shape index (Mcnab 1989) based on neighboring 3 × 3 DEM cells (i.e.,
120 × 120 m). Positive values indicate a concave surface, negative values
indicate a convex surface, 0 indicates a linear (not necessarily level) surface
EVEN Terrain evenness: standard deviation of terrain shape indices calculated for 2,
4, and 8 m
Forest understory/ground cover HERBCO Herbaceous cover (%)
SHRUBCO Shrub cover (%)
SHRUBHT Average shrub height (m)
SHRUBDEN Shrub density log(natural)-transformed
SHRUBCV Coefficient of variation (CV) of shrub distances at the 4 quadrants
ROCKCO Rock cover (%)
VO Average visual obscurity of 4 directions at 0.5-2 m (total 32 values) (%)
VOSTD Standard deviation of visual obscurity among 4 directions
Forest structures STREEDEN Tree density (diameter at breast height (DBH) of 1-20 cm), log-transformed
LTREEDEN Tree density (DBH of > 20 cm), log-transformed
TREEDENCV CV of tree densities among 3 size classes (DBHs of 1-5, 5-20, and > 20 cm)
BASAL Total basal area, log-transformed
STREEHT Average tree height (DBH of 1-20 cm)
LTREEHT Average tree height (DBH of > 20 cm)
TREEHTCV CV of tree heights among 3 size classes (DBHs of 1-5, 5-20, and > 20 cm)
STRATUM Number of forest strata (2-5)
CANOPYHT Canopy height (m)
CANOPYCO Average of 8 measurements of canopy cover (%)
CANOPYGAP CV between 8 measurements for canopy patchiness (gaps)
Chiang et al. – Niche Relationships of Carnivores in Southern Taiwan 503
of habitat variables and collinearity. For each
category, principal components were selected so
that these components cumulatively explained at
least 70% of the habitat variation. Components
with eigenvalues of > 1 were also retained, even
though the cumulative variation explained already
exceeded 70%. These principal components
were thereafter orthogonally rotated for ease of
interpretation. Factor scores were used as habitat
variables for each camera trap site in the habitat
matrix of the CCA, and photographic rates of each
carnivore at each camera trap site were used in
the abundance matrix of the CCA.
Temporal patterns
Camera trapping is widely used to study diel
activities of animals (see review in Cutler and
Swann 1999). An index of the activity level for
each hour was calculated by dividing the hourly
number of occurrence events by the total number
of occurrence events for each species. Hourly
activity levels are expressed as percentages and
depicted across 24 h.
Diet patterns
Diet was summarized according to published
studies conducted in Taiwan or nearby Southeast
Asian countries. Food types were divided into 5
categories: invertebrates, smaller mammals, other
smaller vertebrates (amphibians, reptiles, birds,
etc.), larger vertebrates (> 1 kg), and plants. A
food category was assigned as a “major food item”
if the average reported relative importance was
> 40%, a “secondary food item” if it was 15%-40%,
and an “occasional food item” if it was 5%-15%.
Food categories with an average reported relative
importance of < 5% were ignored for simplicity of
comparison.
RESULTS
The total number of camera trap days for
all 185 camera trap sites was 7812. Thirtynine
of 185 camera trap sites documented
no carnivores. Six carnivore species were
photographed by the other 146 camera trap
sites, including Formosan black bear, Formosan
ferret badger, Siberian weasel, yellow-throated
marten, crab-eating mongoose, and gem-faced
palm civet. Of the other existing carnivores
of Taiwan not documented in this study, the
leopard cat and lesser Oriental civet prefer moredisturbed
lowland habitats (Pei and Chen 2007),
and the least weasel (Mustela nivalis formosana)
is currently found only in central and northern
Taiwan (Lin et al. 2010). Thus, all carnivores of
Taiwan that live in primary forests across a large
elevation range were completely documented
in this study. However, the black bear was very
rare and was photographed at only a few sites.
Thus, only the other 5 more widely distributed
species were included in the CCA. In addition,
the 39 camera trap sites at which no carnivores
were documented were excluded from the CCA.
Wilcoxon rank-sum tests showed that there were
no significant differences in photographic rates
between the dry and wet seasons for these 5
smaller carnivores (p = 0.16-0.9) in the study area.
Thus, photographic rates were calculated for each
camera trap site without separating dry and wet
seasons for the analysis.
Habitat patterns
The factor analysis extracted 14 factors
in total from 31 habitat variables belonging to 4
categories (Table 2) to be used in the CCA. For
the CCA, the total inertia was 1.8111, while the
constrained axes explained 23.1% of the total
variability. The 1st 3 axes (with eigenvalues of
0.2193, 0.1136, and 0.0573) accounted for 52.3%,
27.1%, and 13.7% of the total variation explained
(Fig. 2). The yellow-throated marten was located
near the center of the ordination space with nearly
0 correlations with axes 1 and 2 suggesting that it
may be a generalist in the study area.
A permutation test on the trace statistic
showed that the measured habitat factors
significantly explained the species composition
(p < 0.0001 based on 104
permutations). The
elevation(+)/greenness(-)/ASR(+) (A1) factor
had the strongest relationship to the carnivore
community. This factor was likely an overall effect
of elevation change, because higher elevations
tend to receive more solar radiation and are
comprised of more conifers (less leaf area causing
a lower normalized difference vegetation index
(NDVI)). The Siberian weasel occupied the highest
elevation, while gem-faced palm civets preferred
the lowest elevation with higher NDVI and less
solar radiation. For the other 3 carnivores,
yellow-throated marten’s optimum was above the
average along this elevational gradient, while that
of the crab-eating mongoose was lower than the
average. The optimum for the ferret badger was
504 Zoological Studies 51(4): 500-511 (2012)
nearly at the average and was widely distributed
across the elevation range.
The “river distance/slope position” factor
(A2) and “smaller tree density/heterogeneity”
factor (F3) also had strong relationships with the
carnivore community. Gem-faced palm civets
and crab-eating mongooses had a preference
for rivers/valleys, which also usually receive less
solar radiation and are at lower elevations, and
forests with a lower density of trees with diameter
at breast height (DBH) of < 20 cm and more
homogeneous tree densities of different sizes. In
contrast, the 3 mustelids had a converse pattern.
These 2 habitat factors separated ferret badgers
and crab-eating mongooses, which had similar
optima on the elevation gradient (A1).
In terms of the forest structure, gem-faced
palm civets and ferret-badgers occurred more
often in forests with a lower canopy and a lower
tree height of large trees with DBHs of > 20 cm
(F1) and a higher tree height of small trees with
DBHs of < 20 cm (F5). Such a trend may imply
a preference for younger forests instead of oldgrowth
forests which have large trees with high
canopy cover. But the other 3 carnivores exhibited
a contrasting trend. Moreover, ferret-badgers
utilized habitats with more-densely spaced smaller
trees (F3) while gem-faced palm civets preferred
habitats with less-densely spaced smaller trees.
Forest understory and terrain characteristics
generally had lower correlations with the carnivore
community than the major gradients in the
Table 2. Factors extracted by Varimax rotation from the factor analysis of the 31 habitat variables in 4
categories in the Tawu Nature Reserve and Twin Ghost Lake Important Wildlife Area (Tawu Mountain),
Taiwan (2001-2004)
Category Code Factor
High loadings (> 0.5) of
habitat variables
Variation explained for each
category
Elevation/greenness/wetness A1 Elevation 0.74*ALT 0.39
-0.83*NDVI
0.85*ASR
A2 River distance/slope position 0.87*RIVERDIST 0.39
0.91*SLOPEPOS
0.61*MOISTGRD
Terrain shape/ruggedness T1 Cliff/steepness 0.89*CLIFFCOUNT 0.32
-0.89*CLIFFDIST
0.59*SLOPEDEM
T2 Ruggedness/convex terrain 0.8*SLOPESTD 0.22
-0.79*TSI
T3 Uneveness (micro) 0.97*EVEN 0.17
Forest understory U1 Dense shrub/less rocky 0.6*SHRUBCO 0.26
0.81*SHRUBDEN
-0.66*SHRUBCV
-0.63*ROCKCO
U2 Visual obscurity 0.71*VO 0.17
0.85*VOSTD
U3 Shrub height/cover 0.92*SHRUBHT 0.15
0.55*SHRUBCO
U4 Herbaceous cover 0.96*HERBCO 0.13
Forest structure F1 Forest height 0.84*LTREEHT 0.23
0.88*TREEHTCV
0.73*CANOPYHT
F2 Canopy cover 0.91*CANOPYCO 0.16
-0.91*CANOPYGAP
F3 Tree density of DBH < 20 cm and
heterogeneity
0.83*STREEDEN
0.82*TREEDENCV
0.14
F4 Tree density of DBH > 20 cm and
basal area
0.87*LTREEDEN
0.62*BASAL
0.13
F5 Tree height of DBH < 20 cm and
forest stratum
0.88*STREEHT
0.69*STRATUM
0.13
Chiang et al. – Niche Relationships of Carnivores in Southern Taiwan 505
elevation and forest-structure categories. Shrub
height/cover (U3) was slightly stronger and mostly
associated with mustelids in contrast to the gemfaced
palm civet and crab-eating mongoose, which
did not seem to be related to the major food items
(tree fruits and crustaceans, respectively).
Activity patterns
Yellow-throated martens (Fig. 3A) and crabeating
mongooses (Fig. 3B) were basically diurnal
with few nocturnal activities. But crab-eating
mongooses would extend their activities into
nighttime, usually within 2 h of sunrise and sunset.
Nocturnal species included ferret-badgers (Fig.
3A) and gem-faced palm civets (Fig. 3B). They
also extended their activities to the daytime, most
of which were within 0.5 h of sunrise and sunset.
Siberian weasels (Fig. 3A) were active day and
night with a significant tendency for more-nocturnal
activities between 18:00 and 06:00 the following
morning (84.6%, Chi-squared goodness of fit test,
p < 0.0001).
Diet summary
Based on the types of major food items, diets
of the 5 carnivores could be divided into 3 groups:
invertebrates, small mammals, and plants (Table
3). Fruits were the major food items of gem-faced
palm civets. Yellow-throated martens and Siberian
weasels took a lot of small mammals. Ferret
badgers and crab-eating mongooses fed mostly on
invertebrates, including earthworms, crustaceans,
and insects.
Niche relationship
When habitat patterns, activity patterns,
and diet patterns were simultaneously examined,
segregation among the carnivore community
was more pronounced than when considering
habitat patterns alone (Fig. 2). In the TWM area,
when gem-faced palm civets and crab-eating
mongooses shared some similar spatial patterns,
they exhibited completely reversed activity patterns
(i.e., nocturnal vs. diurnal). They also differed in
Fig. 2. Joint biplot of the canonical correspondence analysis based on axes 1 and 2 of the carnivore community (species in ‘+’) of
Tawu Nature Reserve and Twin Ghost Lake Important Wildlife Area (at Tawu Mountain (TWM)), Taiwan in relation to habitat factors
(radiating lines). See table 3 for codes of habitat factors. Data were based on 146 camera trap sites (at an average elevation of 1572 m)
collected in 2001-2004. Ordination scores were optimized for species. Parentheses after a species name gives its percentage of
diurnal (06:00-18:00) activity and major food item type (P, fruits; I, invertebrates; M, small mammals).
+
+
+
+
+
-0.5 0 0.5 1.0
CCA1
A1
A2
T1
T2
T3
U1
U2
U3
U4
F1
F2
F3
F4
F5
(1.5%, P)
Siberian weasel
(15.4%, M)
Crab-eating mongoose
Yellow-throated marten
(87.1%, I)
(0%, I)
(95%, M)
Formosan ferret badger
Gem-faced palm civet
0.2
0
-0.2
-0.4
-0.6
CCA2
506 Zoological Studies 51(4): 500-511 (2012)
Fig. 3. Activity patterns of (A) the 3 mustelids of the ferret-badger (n = 207), yellow-throated marten (n = 80), and Siberian weasel
(n = 91); and (B) the crab-eating mongoose (n = 140) and gem-faced palm civet (n = 132) at Tawu Nature Reserve and Twin Ghost
Lake Important Wildlife Area (Tawu Mountain), Taiwan based on camera trapping data in 2001-2004.
0
2
4
6
8
10
12
14
16
Activitylevel(%)
Hour
crab-eating mongoose
gem-faced palm civet
0
2
4
6
8
10
12
14
16
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
Activitylevel(%)
Hour
ferret badger
Siberian weasel
yellow-throated marten
(A)
(B)
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
the major food items (e.g., plants vs. invertebrates,
Table 3). Gem-faced palm civets also had a
reversed activity pattern with the diurnal yellowthroated
marten and differed in the optimum habitat
along the major A1, A2, and F3 habitat factors
(lower than the average values for those factors for
civets vs. higher than the average values for those
factors for martens). Furthermore, yellow-throated
martens fed on small mammals while gem-faced
palm civets fed on fruits (Table 3).
Similarly, yellow-throated martens and ferret
badgers, which were close to each other in the
ordination space, were segregated in the temporal
and diet dimensions. Although crab-eating
mongooses and ferret badgers occupied similar
optima along the elevation (A1) gradient, they had
contrasting preferences in habitat factors A2 and
F3 and also exhibited reversed activity patterns.
Siberian weasels were active day and night and
were away from the other 4 carnivores in the CCA
Chiang et al. – Niche Relationships of Carnivores in Southern Taiwan 507
ordination community space (Fig. 2). Even with
the more closely related yellow-throated marten
having similar major food item with Siberian
weasels (Table 3), they seemed to be temporally
separated (Fig. 3A). For the nocturnal species
pair (gem-faced palm civet vs. ferret badger) and
the diurnal species pair (crab-eating mongoose vs.
yellow-throated marten) showing less segregation
in time budget and along the elevation gradient
(A1), prey differences (Table 3) seemed to be the
major mechanism of niche segregation (Fig. 2),
while habitat patterns also differed along the A2
and F3 gradients.
DISCUSSION
Analysis of the activity patterns and spatial
patterns by the CCA based on camera trapping
data provided an effective way to study niche
segregation in time and space for carnivores in
Taiwan which has never been explored before.
When diets of these carnivores based on previous
studies in Taiwan were considered for food-type
partitioning, we observed complementary resource
use (Schoener 1974 1983) through partitioning in
terms of space (on microhabitat and landscape
scales), food-type, and temporal dimensions.
Species closer to each other in the ordination
space were more likely to be sympatric and thus
differed in time of activity or prey to facilitate
coexistence.
The CCA revealed that elevation, which is
analogical to latitude with respect to vegetation
and weather patterns, was the strongest gradient
explaining the composition of the carnivore
community under natural conditions. The CCA
biplot generally divided the Mustelidae from the
other 2 lower-elevation species (Fig. 2). This
pattern may reflect some of the phylogenetic
aspects and evolutionary histories of these
species. The Mustelidae is a Holarctic group,
while the Viverridae and Herpestidae originated
from tropical Asia of the Oriental Region (Ewer
1973).
Sympatric populations of Siberian weasels
and yellow-throated martens at an elevation
range of 2700-3700 m were found to have similar
diet niche breadths and a high degree of diet
overlap, mostly on small mammals (Tsai 2007).
Interspecific competition for food therefore may
occur. Within the TWM area where yellow-throated
martens were more abundant at elevations of
< 2500 m, Siberian weasels were significantly
more active at night (nocturnal activities between
18:00 and 06:00 the following morning, 94.2%)
at elevations of < 2500 m than at elevations of
> 2500 m (nocturnal activities between 18:00
and 06:00 the following morning, 71.8%, Chisquared
test of independence, d.f. = 1, χ2
= 6.98,
Table 3. Comparison of food-type dimensions for the 5 carnivores in the Tawu Mountain area of Taiwan.
Summary of diets were based on literature in Taiwan and a few from nearby countries (i.e., for the gemfaced
palm civet). Relative importance levels of food types were averaged across studies and classified
into 3 categories, i.e., M, major food items (> 40%), S, secondary food items (15%-40%), O, occasional food
items (5%-15%). A relative importance level of < 5% was not included in the diet summary
Species Invertebrates Smaller vertebrates Larger vertebrates Plants Source
Others (reptiles,
amphibians, birds)
Small mammals
Yellow-throated marten O M S Tsai 2007
Siberian weasel S O M O (scavenge) Ma 1990, Wu 1999,
Tsai 2007
Formosan ferret badger M O Chuang and Lee 1997,
Wu 1999
Crab-eating mongoose M S O Chuang and Lee 1997
Gem-faced palm civet O O M Wang and Fuller 2003,
Hwang 2008
508 Zoological Studies 51(4): 500-511 (2012)
p < 0.01). Similarly, radio-collared male Siberian
weasels exhibited more diurnal activity (diurnal
activities between 06:00 and 18:00, 71.5%) in a
lower-elevation area (520-1230 m) in northeastern
Taiwan without sympatric yellow-throated martens
(Wong 1997).
Predators separate more often in diel activity
patterns than do other groups of animals (Schoener
1974). Carnivores competing for similar prey were
also observed to exhibit different activity patterns
(Ray 1997, Fedriani et al. 1999, Karanth and
Sunquist 2000, Vieira and Port 2007). The smaller
Siberian weasel may be avoiding the larger, diurnal
yellow-throated marten by being more active at
night, since they are often sympatric and have a
high degree of diet overlap (i.e., small mammals).
St-Pierre et al. (2006) found that the smaller
Mustela erminea avoided the larger M. frenata,
which was related it to intraguild interactions
including interspecific competition and intraguild
predation. Intraguild predation (Polis et al. 1989,
Polis and Holt 1992, Holt and Polis 1997) may
be one of the mechanisms shaping the weasel/
marten community in Taiwan, as Siberian weasels
are much smaller than yellow-throated martens
(with a body weight difference up to 5-10-fold).
Remains of smaller weasel (Mustela) species
were found in the scat of other larger mammalian
carnivores, including larger weasel and marten
(Martes) species (Erlinge 1981, Zielinski et al.
1999, Fedriani et al. 2000, Edwards and Forbes
2003). Although yellow-throated marten’s
predation on Siberian weasels has not yet been
documented, remains of a ferret badger, which is
larger than Siberian weasel, were found in yellowthroated
marten feces (Chung-Yi Lin, personal
communication).
Lower elevations of the TWM area, which
is comprised of primary forest, contained only 1
Niviventer rat species. This may have caused
the yellow-throated marten and Siberian weasel
to inhabit higher elevations in the TWM area (Fig.
2). However, yellow-throated marten’s optimal
habitat being lower than Siberian weasel along
the elevation(+)/greenness(-) gradient within the
TWM area (Fig. 2) suggests that it may be using
its optimum habitat with more-abundant food
including small mammals and larger ungulates
(e.g., muntjacs), which yellow-throated martens
have been observed to hunt (Matyushkin 1987,
Sathyakumar 1999, Koh 2007). As forests at lower
elevations tend to have fewer rodents, competition
between Siberian weasels and yellow-throated
martens for small mammals may be higher. Even
at places where yellow-throated martens were
rare or did not occur and Siberian weasels may
not need to compete for limited small mammals,
Siberian weasels were still found to adjust their
food habits to forage for more invertebrates at
lower elevations in Taiwan, which is likely due to
the scarcity of small mammals (Tatara and Doi
1994, Wu 1999). In other words, Siberian weasels
at low to mid-elevations in Taiwan would need to
compete with yellow-throated martens for scarce
small mammals and possibly compete with the
larger, similarly nocturnal ferret badgers and lesser
Oriental civets for invertebrates. But Wu (1999)
observed that sympatric Siberian weasels and
ferret badgers still differentiated their habitat and
food dimensions to a certain degree. Our biplots
(Fig. 2) also showed habitat segregation between
Siberian weasels and ferret badgers.
Gem-faced palm civets and ferret badgers
seem to be tolerant of human-altered landscapes
to a certain degree and were found to be
omnipresent in many areas across Taiwan, even
very close to human encroachment. In the TWM
area with minimal human disturbance, these 2
carnivores preferred lower forest heights (habitat
factor F1 in Fig. 2), suggesting their adaptation
to secondary forests near human encroachment,
which usually have lower forest heights due to
frequent human disturbances. In addition, they
were both nocturnal and could avoid humans by
being active at night.
Carbone et al. (2001) suggested that 1000
camera trap days could detect tiger (Panthera
tigris) presence at very low densities (0.4-0.7
tigers/100 km2
). Our study area is about 930 km2
.
If there was only 1 tiger in such a large area, the
predicted camera-trapping effort to detect tiger
presence would be around 5000 trap days. The
fact that clouded leopards, Eurasian otters, leopard
cats, and lesser Oriental civets were not detected
in TWM’s contiguous primary forests suggests their
true absence given our extensive camera trapping
effort of 7812 camera trap days. Current habitat
types where lesser Oriental civets and leopard
cats are mostly found in Taiwan are at elevations
of < 1200 m in secondary forest mosaics with
farmlands or adjacent to human encroachment
(Chen 2002, Liu and ChangChien 2004, Pei and
Chen 2006). In contrast to only 1 rat species
occurring in contiguous primary forests at lower
elevations, many other small rodents occur in
lowlands closer to human encroachment (Lin
1982, Lin and Lin 1983). Such places are usually
a mosaic of forests and agricultural lands which
Chiang et al. – Niche Relationships of Carnivores in Southern Taiwan 509
provides suitable prey (e.g., small mammals and
Formosan hares that favor non-forest habitats)
for leopard cats. Scarce small mammals in
contiguous forests at low to mid-elevations would
make competition for small mammal prey even
greater for leopard cats in such pristine habitats
where Siberian weasels and yellow-throated
martens occur. Interspecific competition with other
small mammal predators and a lack of prey are
likely the mechanisms deterring leopard cats from
using primary montane forests in Taiwan.
Our findings support the hypothesis of the
leopard cat and lesser Oriental civet not preferring
primary forests in contrast to populations in other
Southeast Asia countries. Furthermore, of the
3 conservation categories in Taiwan’s Wildlife
Conservation Law, leopard cats are listed in
category I as “endangered”, while lesser Oriental
civets are in category II, “valuable”. Their
habitats in Taiwan are undergoing rapid human
encroachment, fragmentation, and shrinkage.
Conservation actions are urgently needed for
leopard cats and lesser Oriental civets in Taiwan.
In addition, our camera trapping revealed abundant
prey in the study area for clouded leopards. Since
Taiwan is an island, and it is unlikely to be naturally
recolonized by other clouded leopard populations,
the TWM area could be a potential candidate for
the reintroduction of clouded leopards. With the
niche information of meso-carnivores presented in
this paper, a clouded leopard reintroduction plan
could be carefully conducted and monitored to
assess any population or behavioral changes of
other sympatric carnivores.
Acknowledgments: Funding was mostly from the
Forestry Bureau, Council of Agriculture, Executive
Yuan, Taiwan with additional funding from the
Taiwan Ministry of Education, Virginia Polytechnic
Institute and State Univ. (Blacksburg, VA), National
Pingtung Univ. of Science and Technology
(Pingtung, Taiwan), and Research Fellowships
Program from the Wildlife Conservation Society
(Bronx, NY), and personal donations from R.
Tzeng, Y.H. Sun, and C.N. Liu. We are grateful to
2 aborigines C.H. Chen and C.H. Lin for consistent
help with hard fieldwork, and Y.J. Pan, M.T. Chen,
C.C. Chen, Y.J. Liang, C.H. Chiu, H.W. Chang,
and many other field technicians and volunteers in
helping collect field data. Dr. Y.C. Lai generously
provided SPOT images of the study area.
REFERENCES
Carbone C, S Christie, K Conforti, T Coulson, N Franklin, JR
Ginsberg et al. 2001. The use of photographic rates to
estimate densities of tigers and other cryptic mammals.
Anim. Conserv. 4: 75-79.
Chen MT. 2002. Activity patterns and habitat use of sympatric
small carnivores at low elevations in southern Taiwan.
MS thesis, Texas A&M Univ., Kingsville, TX.
Chen MT, ME Tewes, KJ Pei, LI Grassman. 2009. Activity
patterns and habitat use of sympatric small carnivores in
southern Taiwan. Mammalia 73: 20-26.
Chiang PJ. 2007. Ecology and conservation of Formosan
clouded leopard, its prey, and other sympatric carnivores
in southern Taiwan. PhD dissertation, Virginia Polytechnic
Institute and State Univ., Blacksburg, VA.
Chiang PJ, KJC Pei, YJ Pan. 2007. A plea for standardization
in protocols of camera trapping studies in Taiwan. Notes
Newslett. Wildl. 11: 2-12. (in Chinese)
Chuang SA, LL Lee. 1997. Food habits of three carnivore
species (Viverricula indica, Herpestes urva, and Melogale
moschata) in Fushan forest, northern Taiwan. J. Zool.
243: 71-79.
Cutler TL, DE Swann. 1999. Using remote photography in
wildlife ecology: a review. Wildl. Soc. Bull. 27: 571-581.
Edwards MA, GJ Forbes. 2003. Food habits of ermine,
Mustela erminea, in a forested landscape. Can. FieldNat.
117: 245-248.
Erlinge S. 1981. Food preference, optimal diet and reproductive
output in stoats Mustela erminea in Sweden.
Oikos 36: 303-315.
Ewer RF. 1973. The carnivores. Ithaca, NY: Comstock
Publishing Associates. Reprint 1998.
Fedriani JM, TK Fuller, RM Sauvajot, EC York. 2000.
Competition and intraguild predation among three
sympatric carnivores. Oecologia 125: 258-270.
Fedriani JM, F Palomares, M Delibes. 1999. Niche relations
among three sympatric Mediterranean carnivores.
Oecologia 121: 138-148.
Fu PD, PM Rich. 2002. A geometric solar radiation model with
applications in agriculture and forestry. Comput. Electron.
Agric. 37: 25-35.
Gordon CE. 2000. The coexistence of species. Rev. Chil.
Hist. Nat. 73: 175-198.
Holt RD, GA Polis. 1997. A theoretical framework for intraguild
predation. Am. Nat. 149: 745-764.
Hwang JC. 2008. Some notes on the ecology of gem-faced
palm civet: from burrowing behavior to seed dispersal.
Notes Newslett. Wildl. 12: 12-19.
Hwang MH, Y Wang, PF Lee. 2006. Distribution mapping and
conservation status of the Formosan black bear. Taipei,
Taiwan: Taiwan Forestry Bureau. Report Conservation
Research Series no. 94-28.
Karanth KU, ME Sunquist. 2000. Behavioural correlates of
predation by tiger (Panthera tigris), leopard (Panthera
pardus) and dhole (Cuon alpinus) in Nagarahole, India. J.
Zool. 250: 255-265.
Koh CN. 2007. Notes on field survey in Nan-Tz-Shian Stream
Forest Road. For. Res. Newslett. 14: 45-48.
Liang YJ. 2005. Seasonal changes of food quality and
population dynamics of sambar (Cervus unicolor swinhoei)
and muntjac (Muntiacus reevesi micrurus) in Mei-Lan Trail
Area. Master’s thesis, National Pingtung Univ. of Science
and Technology, Pingtung, Taiwan.
510 Zoological Studies 51(4): 500-511 (2012)
Lin JY, LK Lin. 1983. A note on the zoogeography of the
mammals in Taiwan. Ann. Taiwan Mus. 26: 53-62.
Lin LK. 1982. A synopsis of mammals of Taiwan. Master’s
thesis, Tunghai Univ., Taichung, Taiwan.
Lin LK. 2000. The status and conservation of Taiwan’s
mustelids. In HI Griffiths, ed. Mustelids in a modern
world. Management and conservation aspects of small
carnivore: human interactions. Leiden, the Netherlands:
Backhuys Publishers, pp. 331-335.
Lin LK, M Motokawa, M Harada. 2010. A new subspecies of
the least weasel Mustela nivalis (Mammalia, Carnivora)
from Taiwan. Mamm. Study 35: 191-200.
Liu CN, L ChangChien. 2004. The distribution and genetic
variations of Formosa clouded leopard, Neofelis nebulosa
brachyurus, and leopard cat, Felis bengalensis chinensis
(2/3). Chichi, Nantou, Taiwn: Taiwan Endemic Species
Research Institute.
Loveridge AJ, DW Macdonald. 2003. Niche separation in
sympatric jackals (Canis mesomelas and Canis adustus).
J. Zool. 259: 143-153.
Ma XQ. 1990. Ecological study of Siberian weasel (Mustela
sibirica davidiana) at alpine meadow: a basical survey
of diet, habitat and population. Master’s thesis, National
Taiwan Normal Univ., Taipei, Taiwan.
Matyushkin EN. 1987. Hunting the musk deer by the yellowthroated
marten in the Sikhote-Alin’ Mountains. Byull.
Moskovskogo Obshchestva Ispytatelei Prirody Otdel Biol.
92: 28-42.
McDonald RA. 2002. Resource partitioning among British and
Irish mustelids. J. Anim. Ecol. 71: 185-200.
Mcnab WH. 1989. Terrain shape index - quantifying effect of
minor landforms on tree height. For. Sci. 35: 91-104.
O’Brien TG, MF Kinnaird, HT Wibisono. 2003. Crouching
tigers, hidden prey: Sumatran tiger and prey populations
in a tropical forest landscape. Anim. Conserv. 6: 131-139.
Oksanen J, R Kindt, P Legendre, B O’Hara, GL Simpson, M
Henry et al. 2008. vegan: Community Ecology Package.
R package vers. 1.13-2. Available at http://vegan.r-forge.
r-project.org/ Accessed 1 May 2011.
Pei KJC, MT Chen. 2006. Present status and conservation of
small carnivores at low elevation mountains in Shinchu
County and Miaoli County (1/3). Taipei, Taiwan: Forestry
Bureau, Council of Agriculture, Executive Yuan. Report
Taiwan Forestry Bureau Conservation Research Series
no. 94-05.
Pei KJC, MT Chen. 2007. Present status and conservation of
small carnivores at low elevation mountains in Shinchu
County and Miaoli County (2/3). Taipei, Taiwan: Forestry
Bureau, Council of Agriculture, Executive Yuan. Report
Taiwan Forestry Bureau Conservation Research Series
no. 95-03.
Polis GA, RD Holt. 1992. Intraguild predation - the dynamics
of complex trophic interactions. Trends Ecol. Evol. 7:
151-154.
Polis GA, CA Myers, RD Holt. 1989. The ecology and
evolution of intraguild predation - potential competitors
that eat each other. Annu. Rev. Ecol. Systemat. 20: 297-
330.
R Development Core Team. 2009. R: a language and
environment for statistical computing. Vienna, Austria: R
Foundation for Statistical Computing. Vers. 2.10.1. ISBN
3-900051-07-0. Available at http://www.R-project.org
Accessed 1 May 2011.
Ray JC. 1997. Comparative ecology of two African forest
mongooses, Herpestes naso and Atilax paludinosus. Afr.
J. Ecol. 35: 237-253.
Rovero F, AR Marshall. 2009. Camera trapping photographic
rate as an index of density in forest ungulates. J. Appl.
Ecol. 46: 1011-1017.
Sathyakumar S. 1999. Mustelids and viverrids of the
northwestern and western Himalayas. Envis: Wildl.
Protect. Areas 2: 39-42.
Schoener TW. 1974. Resource partitioning in ecological
communities. Science 185: 27-39.
Schoener TW. 1983. Field experiments on interspecific
competition. Am. Nat. 122: 240-285.
Scognamillo D, IE Maxit, M Sunquist, J Polisar. 2003.
Coexistence of jaguar (Panthera onca) and puma (Puma
concolor) in a mosaic landscape in the Venezuelan llanos.
J. Zool. 259: 269-279.
St-Pierre C, JP Ouellet, M Crete. 2006. Do competitive
intraguild interactions affect space and habitat use by
small carnivores in a forested landscape? Ecography 29:
487-496.
Su HJ. 1984. Studies on the climate and vegetation types
of the natural forests in Taiwan (II) altitudinal vegetation
zones in relation to temperature gradient. Q. J. Chin. For.
17: 57-73.
Tatara M, T Doi. 1994. Comparative analyses on food habits
of Japanese marten, Siberian weasel and leopard cat in
the Tsushima islands, Japan. Ecol. Res. 9: 99-107.
Ter Braak CJF. 1986. Canonical correspondence analysis: a
new eigenvector technique for multivariate direct gradient
analysis. Ecology 67: 1167-1179.
Terborgh J, JA Estes, P Paquet, K Ralls, D Boyd-Heger,
BJ Miller, RF Noss. 1999. The role of top carnivores
in regulating terrestrial ecosystems. In ME Soule, J
Terborgh, eds. Continental conservation: scientific
foundations of regional reserve networks. Washington,
DC: Island Press, pp. 39-64.
Tsai CW. 2007. Food resource partitioning between sympatric
yellow-throated marten (Martes flavigula chrysosphila)
and Siberian weasel (Mustela sibirica taivana). Master’s
thesis, National Taiwan Univ., Taipei, Taiwan.
Vieira EM, D Port. 2007. Niche overlap and resource partitioning
between two sympatric fox species in southern
Brazil. J. Zool. 272: 57-63.
Wang H, TK Fuller. 2003. Food habits of four sympatric
carnivores in southeastern China. Mammalia 67: 513-
519.
Wong GJ. 1997. The activity pattern and population dynamics
of Siberian weasel (Mustela sibirica) at Fushan forest.
Master’s thesis, National Taiwan Univ., Taipei, Taiwan.
Wu HY. 1999. Is there current competition between sympatric
Siberian weasels (Mustela sibirica) and ferret badgers
(Melogale moschata) in a subtropical forest ecosystem of
Taiwan? Zool. Stud. 38: 443-451.
Yasuda M. 2004. Monitoring diversity and abundance of
mammals with camera traps: a case study on Mount
Tsukuba, central Japan. Mamm. Study 29: 37-46.
Zielinski WJ, NP Duncan, EC Farmer, RL Truex, AP Clevenger,
RH Barrett. 1999. Diet of fishers (Martes pennanti) at the
southernmost extent of their range. J. Mammal. 80: 961-
971.
Chiang et al. – Niche Relationships of Carnivores in Southern Taiwan 511