Mammalian Biology 92 (2018) 120–128
Contents lists available at ScienceDirect
Mammalian Biology
jou rnal homepage: www.elsevier.com/locate/mambio
Original investigation
Spatiotemporal patterns of Amur leopards in northeast China:
Influence of tigers, prey, and humans
∗
Haitao Yang, Xiaodan Zhao, Boyu Han, Tianming Wang, Pu Mou, Jianping Ge, Limin Feng
Monitoring and Research Center for Amur tiger and Amur leopard, State Forestry Administration, State Key Laboratory of Earth Surface and Resource
Ecology, Ministry of Education Key for Biodiversity Science and Engineering, College of Life sciences, Beijing Normal University, Beijing, 100875, China
a
r t i c l e i n f o a b s t r a c t
Article history: The Amur leopard Panthera pardus orientalis is one of the most endangered cat subspecies in the world.
Received 5 September 2017
The rare leopard is sympatric with Amur tiger Panthera tigris altaica and their prey in human dominated
Accepted 20 March 2018
landscape. To conserve the felid species, it is important to understand the activity patterns of Amur
Available online 22 March 2018
leopards, including its interactions with Amur tigers, prey, and human activities. We used a data set
from 163 camera traps to quantify the spatial-temporal overlap between Amur leopards, Amur tigers,
Handled by Francesco Ferretti
prey species, and human disturbances (e.g., humans presence on foot, vehicles, domestic dogs, and cattle
Keywords: grazing) from January to December 2013 in the Hunchun Nature Reserve, NE China. Our results indicated
that leopards were more active in daytime and twilight; the seasonal spatial-temporal overlaps between
Activity patterns
Amur leopard leopards and tigers were lower than that between leopards and their prey species. Human activities and
camera trap cattle grazing could influence the spatial distribution and activity patterns of the leopards, and therefore,
human disturbance the conservation actions should focus on reduction of human disturbances to minimize the impacts to
spatial overlap
Amur leopard activity patterns.
© 2018 Deutsche Gesellschaft fur¨ Saugetierkunde.¨ Published by Elsevier GmbH. All rights reserved.
Introduction For several years, Amur leopards have been studied using
remote camera-traps in China (Feng et al., 2017, 2011; Wang et al.,
The Amur leopard Panthera pardus orientalis, one of the nine 2016, 2017; Xiao et al., 2014) to estimate population size and spatial
leopard subspecies, is the rarest cat in the world (Nowell and habitat selection across the landscape. Additionally, other studies
Jackson, 1996; Uphyrkina et al., 2001, 2002). Since 1996, it has have focused on the feeding habits of the leopards (Sugimoto et al.,
been classified as Critically Endangered (CR) by the International 2016). However, basic information about the interactions of the
Union for Conservation of Nature (IUCN) (Stein et al., 2017). The cur- Amur leopard with the Amur tiger, their prey, and human dis-
rent population of the Amur leopard is at least 87 individuals (Feng turbances are still lacking, and such information is important for
et al., 2017) and its distribution is confined to approximately 4,000 continued persistence of Amur leopard.
2
km in southwestern Primorsky Krai and adjacent habitats in Jilin We focused on the spatial and temporal dimensions of the eco-
and Heilongjiang Provinces, China (Feng et al., 2017; Hebblewhite logical overlap when investigating the interactions of the leopards
et al., 2011). In this area, there are also >38 Amur tigers Panthera with the Amur tiger, their prey, and human disturbances, given
tigris altaica (Feng et al., 2017) existing in the same habitat of the these factors may change leopard behavior. High spatial overlap
leopards. Habitat isolation, inbreeding, environmental stochastic- between leopards and prey may increase leopard encounter rates
ity, and infectious diseases have increased the stress on the leopards with prey species, while high temporal overlap between leopards
(Sugimoto et al., 2014; Uphyrkina et al., 2002). More knowledge of and human disturbances and/or tigers may change activity patterns
basic ecology and behavior are needed for conservation efforts to of the leopards (Linkie and Ridout, 2011). O’Brien et al. (2003) found
combat the trend towards extinction of the Amur leopard. a significant spatial relationship between the Sumatran tiger (Pan-
thera tigris sumatrae) and wild pig (Sus sp.), suggesting that tigers
preferred areas where the wild pigs are more abundant. In Thailand,
leopards (Panthera pardus) exhibited reduced diurnal activity in
∗ more heavily used areas compared to the areas less used by peo-
Corresponding author at: College of life Sciences, No. 19, Xin Jie Kou Outer Street,
Haidian District, Beijing 100875, China. ple (Ngoprasert et al., 2007). Tigas et al. (2002) found that bobcat
E-mail address: [email protected] (L. Feng).
https://doi.org/10.1016/j.mambio.2018.03.009
1616-5047/© 2018 Deutsche Gesellschaft fur¨ Saugetierkunde.¨ Published by Elsevier GmbH. All rights reserved.
H. Yang et al. / Mammalian Biology 92 (2018) 120–128 121
(Lynx rufus) activity was higher during the diurnal period than in farms), animal husbandry (cattle grazing and frog farming), and
a fragmented study area, suggesting certain degree of avoidance poaching.
of humans. In this study, we used remote camera-trap data from
the Hunchun Nature Reserve in NE China to investigate the sea-
Study design and data collection
sonal spatial-temporal overlap of leopard-tiger, leopard-prey, and
leopard-human disturbances. The two main hypotheses were: (1)
We established 163 camera traps in the gridded study area to
the leopards will have a high spatial or temporal overlap with prey
monitor the Amur leopard, the Amur tiger, and their prey (Fig. 1).
species; and (2) the leopards will avoid human disturbances and
In each grid (3.6 * 3.6 km), mostly 2 camera traps were placed along
tigers spatially or temporally.
road, trail or ridge, which were natural routes for leopards, tigers,
and prey species. The cameras were fastened to trees, 40 - 80 cm
above the ground, and were programmed to shoot 15-sec videos
with a 1-min interval between consecutive events. The camera
Material and methods
traps operated 24-hours per day throughout the year. We visited
each camera monthly to download videos and check batteries.
Study area
We analysed only videos taken at a minimum time interval of
30 minutes (O’Brien et al., 2003). Only videos from the 163 stations
The Hunchun National Reserve (HNR) was located in the east-
were used for analysis. Based on local climate characteristics, we
ern part of Jilin Province, China, bordering Russia and North Korea
◦ ◦ ◦ ◦ defined the seasons as the snow period (winter: Jan-Apr and Nov-
(E 130 14 08 -131 14 44 , N 42 32 40 -43 28 00 ) (Fig. 1). This
Dec) and the snow-free period (summer: May-Oct).
region serves as core habitat for both the Amur tiger and the
Amur leopard in China, tigers and leopards can transfer through the
2
fences on the Sino-Russia border. The 1,087-km HNR was in the Spatial overlap
northern portion of the Changbai Mountains. The major vegetation
types included deciduous birch (Betula linn.) and oak (Quercus mon- To investigate spatial overlap (Pianka, 1973), we calculated the
golica) forests, most of which were secondary deciduous forests, Relative Abundance Index (RAI) at each trap site as the number of
as well as some coniferous forests distributed in the northeastern detections per 100 camera-trap days of every species for the two
region (Tian et al., 2011). The HNR has been exposed to human seasons (O’Brien et al., 2003). Each camera trap was considered an
disturbance for decades, including plantations (crops and ginseng independent spatial point, and the RAI of each site was examined for
Fig. 1. Study area showing locations of remote camera traps in the Hunchun Nature Reserve, northeast China, 2013. Data about the Amur leopard current extant area derived
from Feng et al. (2017).
122 H. Yang et al. / Mammalian Biology 92 (2018) 120–128
Table 1
correlations among the leopards, tigers, prey species, and human
Number of events and Relative Abundance Index (RAI) for the Amur leopard, Amur
disturbances (grazing, domestic dog, human activity (humans pres-
tiger, prey, and various human disturbances (grazing, domestic dog, human activity
ence on foot) and vehicles) using the Spearman Rank Correlation
(humans presence on foot) and vehicles) during summer and winter, NE China, 2013.
Index (Ramesh et al., 2012) and Pianka’s index (Pianka, 1973) which
Species Events (RAI)
can reflect the spatial overlap of leopard-tiger, leopard-prey and
leopard-human disturbances (Ramesh et al., 2012). The Spearman Winter (RAI) Summer (RAI)
rank correlation index and Pianka’s index were calculated using R
Amur leopard 42 (0.23) 79 (0.35)
software (v. 3.1.2)(Team, 2014). Amur tiger 101 (0.53) 152 (0.67)
Wild boar 116 (0.63) 562 (2.46)
Roe deer 213 (1.16) 469 (2.05)
Temporal overlap
Sika deer 362 (1.97) 846 (3.70)
Red fox 158 (0.85) 295 (1.29)
Events were selected, including the date and time of animal Asian badger 49 (0.27) 604 (2.64)
Raccoon dog 19 (0.10) 176 (0.77)
activity, to assess the temporal patterns of leopard-tiger, leopard-
Musk deer 9 (0.05) 14 (0.06)
prey, and leopard-human disturbance. Random samples from the
Manchurian hare 221 (1.21) 343 (1.50)
continuous temporal distribution could reflect the probability of
Human activity 985 (5.31) 4876 (21.35)
the events being recorded in any particular interval during the day Grazing 16 (0.09) 1406 (6.16)
Domestic dog 348 (1.90) 639 (2.80)
(Monterroso et al., 2013; Ridout and Linkie, 2009). We followed the
Vehicle 837 (4.52) 3700 (16.20)
procedures of Ridout and Linkie (2009) to quantify the overlap of
the activity patterns of the leopards with tigers, prey species and
human disturbances. The first step included separately estimat-
Results
ing the probability density function based on the non-parametric
kernel density. We used the distribution function for pairwise com-
Abundance
parisons of the activity patterns of the leopards, tigers, prey species,
and human disturbances. For the second step, the coefficient of
From January 2013 to December 2013, we used 163 camera traps
overlap, , which ranges from 0 (no overlap) to 1 (complete over-
for 41,122 trap-nights. We collected 121 events for leopards, 253
lap), was used to measure the overlap between the two probability
for tigers, 678 for wild boars, 682 for roe deer, 1,208 for sika deer,
distributions of two species (Linkie and Ridout, 2011). The coeffi-
12,807 for human disturbances (humans on foot, vehicles, domestic
cient was obtained from the area under the curve formed by taking
dog and grazing), and 1,888 for small- and medium-sized mammals
the minimum of two density functions at each time point (Linkie
(453 for red fox, 653 for Asian badger, 195 for raccoon dog,23 for
and Ridout, 2011). Ridout and Linkie (2009) developed three ways
musk deer, and 564 for Manchurian hare)(Table 1).
of estimating ; we used 1 for small sample sizes (<75) and 4
The RAI for leopards in the summer was higher than the RAI in
for larger sample sizes (≥75). The 95% confidence interval were
winter; the RAI for tigers was similar to leopard (Table 1). Among
obtained by using 10000 bootstrap samples. Statistical analyses
the three main prey species, the RAI was highest for sika deer in
were completed using the “overlap” package (Meredith and Ridout,
both seasons (Table 1). The RAI of human disturbances in summer
2013) in R software (version v.3.1.2)(Team, 2014).
was higher than winter, and the RAI for cattle were very rare in
The actual time of sunrise and sunset varied in each sample
the winter (Table 1). For small-medium size mammals, the RAI was
period. According to the exact time of sunrise and sunset, the
higher in summer, but only 19 events of raccoon dog were recorded
diurnal cycle was divided into three phases: day, night, and crepus-
in winter (Table 1). For musk deer, we got 23 events during the
cular (1 h before sunrise to 1 h after sunrise and 1 h before sunset
study period. Due to the low detections of raccoon dog and musk
to 1 h after sunset)(Lucherini et al., 2009). The activities of the
deer, we excluded the winter data of raccoon dog and the all data
leopards and their prey species were then classified into the fol-
of musk deer in the next analysis (Table 1).
lowing three categories: diurnal (activity predominantly between
1 h after sunrise and 1 h before sunset), nocturnal (activity predomi-
Spatial overlap
nantly between 1 h after the sunset and 1 h before the sunrise) and
twilight (activity predominantly between ± 1 h from sunrise and
The seasonal spatial overlap between leopards and tigers was
sunset)(Foster et al., 2013). The selection ratios of each species was
low and Spearman’s rank correlation coefficients of spatial pattern
used to determine whether a species’ activity was predominantly
indicated that overlap was not significant (p>0.05) (Table 2). For
classified as twilight, diurnal, or nocturnal (Bu et al., 2016; Manly
three main prey species (wild boar, roe deer, sika deer), spatial over-
et al., 2007):
lap between leopards and wild boars was highest in summer, while
spatial overlap between leopards and sika deer was highest in win-
wi = oi/eˆi
ter (Table 2). However, only the seasonal Spearman rank correlation
w o
where i is the selection of the period i; i is proportion of detec- coefficients of spatial pattern between leopards and wild boars
e
tions in period i; ˆi is the proportion of the length of the period to were significantly positive (p<0.01) (Table 2). Pianka’s index values
w
the length of all periods. When i >1, the time period is highly between leopards and small-medium size mammals, except rac-
w
preferred; when i < 1, the time period is avoided (Bu et al., 2016; coon dog, were relatively higher than the three main prey species
Gerber et al., 2012). and the Spearman rank correlation coefficients of spatial patterns
The synchrony of the times of peak activities for a pair of species were significantly positive (p<0.01) in seasonal periods (Table 2).
can also be an indicator of the activity pattern of the two species The low spatial overlap between leopards and human disturbances
(Ramesh et al., 2012; Ridout and Linkie, 2009). We divided the 24 (Table 2) and correlation coefficients of spatial patterns between
hr of the day into 512 equal intervals (approximately 2.8 min per leopards and human disturbances varied seasonally.
interval)(Ridout and Linkie, 2009; Schmid and Schmidt, 2006), and
the probability density of each time point was estimated via kernel
Temporal overlap
density estimation. Spearman’s rank correlation was used to esti-
mate the degree of synchronization of the temporal peak activities
Leopards were more active in daytime and twilight (Table 3),
among the leopards, tigers, prey species, and human disturbances.
and they exhibited a peak of activity at approximately 9:00 dur-
H. Yang et al. / Mammalian Biology 92 (2018) 120–128 123
Fig. 2. Probability density distribution and activity pattern overlap of Amur leopards and Amur tigers and three main prey species throughout the day for summer and winter.
Note: The solid line represents the activities of the Amur leopard density curve, and the dotted line indicates the activity density curve of the species being compared. The
gray area under the curve represents the degree of overlap between the activity patterns of the two species. The vertical dotted lines indicate the average times of sunrise
and sunset.
124 H. Yang et al. / Mammalian Biology 92 (2018) 120–128
Fig. 3. Probability density distribution and activity pattern overlap of Amur leopards and small-sized animals throughout the day in the summer and winter. Note: The solid
line represents the activities of the Amur leopard density curve, and the dotted line indicates the activity density curve of the species being compared. The gray area under
the curve represents the degree of overlap between the activity patterns of the two species. The vertical dotted lines indicate the average times of sunrise and sunset. There
was no data for raccoon dogs during winter.
H. Yang et al. / Mammalian Biology 92 (2018) 120–128 125
Fig. 4. Probability density distribution and activity pattern overlap of Amur leopards and human disturbances (grazing, domestic dog, human activity (humans presence on
foot) and vehicles) during the day in the summer and winter. Note: The solid line represents the activities of the Amur leopard density curve, and the dotted line indicates
the activity density curve of the disturbance being compared with the Amur leopard. The gray area under the curve represents the degree of overlap between the activity
patterns of the two variables. The vertical dotted lines indicate the average times of sunrise and sunset. There was no grazing during winter.
126 H. Yang et al. / Mammalian Biology 92 (2018) 120–128
Table 2
Spatial overlap index (Pianka’s index (95% confidence interval)) and Spearman rank correlation (SRC) between the Amur leopard and Amur tiger, prey, and various human
disturbances (grazing, domestic dog, human activity (humans presence on foot) and vehicles) during summer and winter, NE China, 2013.
Paired species Spatial overlap
Winter Summer
Pianka’s index (CI) SRC Pianka’s index (CI) SRC
Amur leopard VS Amur tiger 0.081 (0.013-0.189) 0.044 0.109 (0.030-0.230) -0.042
** *
Amur leopard VS Wild boar 0.171 (0.077-0.326) 0.230 0.379 (0.171-0.597) 0.197
Amur leopard VS Roe deer 0.067 (0.021-0.164) -0.055 0.317 (0.171-0.510) 0.04
Amur leopard VS Sika deer 0.208 (0.091-0.402) 0.111 0.301 (0.164-0.453) 0.138
** **
Amur leopard VS Red fox 0.344 (0.208-0.499) 0.329 0.263 (0.130-0.513) 0.285
** **
Amur leopard VS Asian badger 0.441 (0.082-0.680) 0.244 0.372 (0.237-0.537) 0.329
Amur leopard VS Raccoon dog - - 0.094 (0.036-0.198) 0.052
* **
Amur leopard VS Manchurian hare 0.357 (0.108-0.591) 0.201 0.289 (0.142-0.491) 0.272
Amur leopard VS Human activity 0.062 (0.029-0.124) 0.105 0.093 (0.050-0.187) -0.114
Amur leopard VS Grazing - - 0.121 (0.014-0.394) 0.047
Amur leopard VS Domestic dog 0.063 (0.025-0.192) 0.149 0.035 (0.013-0.090) -0.138
Amur leopard VS Vehicle 0.030 (0.003-0.106) 0.002 0.024 (0.008-0.057) -0.122
*
P < 0.05.
**
P < 0.01.
Table 3
Number of events n (selection ratio wi) of leopards, tiger, prey, and various human disturbances (grazing, domestic dog, human activity (humans presence on foot) and
vehicles) in a given time period during summer and winter, NE China, 2013.
Species Season
Winter Summer
n (wi) in given time period n (wi) in given time period
Diurnal Nocturnal Twilight Diurnal Nocturnal Twilight
Amur leopard 28 (1.85) 5 (0.25) 9 (1.29) 48 (1.24) 10 (0.37) 21 (1.59)
Amur tiger 18 (0.49) 58 (1.21) 25 (1.49) 32 (0.43) 65 (1.24) 55 (2.17)
Wild boar 46 (1.10) 41 (0.75) 29 (1.50) 270 (0.98) 153 (0.79) 139 (1.48)
Roe deer 92 (1.20) 71 (0.70) 50 (1.41) 194 (0.85) 127 (0.79) 148 (1.89)
Sika deer 171 (1.31) 115 (0.67) 76 (1.26) 411 (0.99) 196 (0.67) 239 (1.70)
Red fox 11 (0.19) 121 (1.62) 26 (0.99) 13 (0.09) 183 (1.80) 99 (2.01)
Asian badger 14 (0.79) 28 (1.21) 7 (0.14) 187 (0.63) 327 (1.57) 90 (0.899)
Raccoon dog - - - 11 (0.13) 115 (1.90) 50 (1.70)
Manchurian hare 2 (0.03) 212 (2.03) 7 (0.19) 15 (0.09) 266 (2.25) 62 (1.01)
Human activity 886 (2.50) 32 (0.07) 67 (0.41) 4517 (1.89) 68 (0.04) 291 (0.36)
Grazing - - - 942 (1.37) 184 (0.38) 280 (1.19)
Domestic dog 286 (2.28) 26 (0.16) 36 (0.62) 560 (1.79) 26 (0.12) 53 (0.50)
Vehicle 680 (2.25) 52 (0.13) 105 (0.75) 3281 (1.81) 94 (0.07) 325 (0.53)
Table 4
Temporal overlap index ( (95% confidence interval)) and Spearman rank correlation (SRC) between the Amur leopard and Amur tiger, prey species, and various human
disturbances (grazing, domestic dog, human activity (humans presence on foot) and vehicles) during summer and winter, NE China, 2013.
Paired species Seasonally temporal overlap
Winter Summer