Big cats kill more livestock when wild prey reaches a minimum threshold
Igor Khorozyan ⁎, Arash Ghoddousi, Mahmood Sooﬁ, Matthias Waltert
Workgroup on Endangered Species, J.F. Blumenbach Institute of Zoology and Anthropology, Georg-August University of Göttingen, Bürgerstrasse 50, Göttingen 37073, Germany
Received 22 April 2015
Received in revised form 8 August 2015
Accepted 23 September 2015
Available online xxxx
Livestock predation by big cats, i.e., lion (Panthera leo), tiger (Panthera tigris), leopard (Panthera pardus), jaguar
(Panthera onca), snow leopard (Panthera uncia), puma (Puma concolor), and cheetah (Acinonyx jubatus), creates
conﬂicts with humans which challenge biodiversity conservation and rural development. Deﬁciency of wild prey
biomass is oftendescribed as a driver of such conﬂicts, but the question “at which level of prey density and bio-
mass do big cats begin to kill livestock?”still remains unanswered. We applied logistic regression to meta-data
compiledfrom recent peer-reviewed scientiﬁc publications and showthat cattle predationis high when prey bio-
mass is b812.41 ± 1.26 kg/km
, whereas sheep and goat predation is high at b544.57 ± 1.19 kg/km
of sizes of studyareas and species, bodymasses, and populationdensities of big cats. Through mappingcases with
known prey biomass and case-speciﬁc comparison of actual vs. threshold-predicted livestock predation we con-
ﬁrm the reliability of these thresholds in predicting livestock predation by big cats. The map also demonstrates
that some protected areas of India, Nepal lowlands, and South Africa contain sufﬁcient prey that makes big
cats less likely to kill livestock, but in other sampled areas prey biomass is not high enough and the probabilities
of livestockpredation are moderate to high. We suggest that these thresholds represent important landmarks for
predicting human–felid conﬂicts, identifying conﬂict hotspots, and setting priorities for targeted conservation
actions. It is essential to maintain and restore wild prey to forestall local extinctions of big cats.
© 2015 Elsevier Ltd. All rights reserved.
Mammalian carnivores inﬂict socio-economic losses to rural
livelihoods, mainly due to predation on domestic livestock, and are in-
tensively wiped out because of retaliatory or preventive persecution
(Treves and Karanth, 2003; Baker et al., 2008; Gusset et al., 2009;
Loveridge et al., 2010). These losses and arising human–carnivore
conﬂicts are particularly strong for small-scale households and near
protected areas, thus challenging a synergy between rural development
and biodiversity conservation (Treves and Karanth, 2003; Bauer and de
Iongh, 2005; Namgail et al., 2007; Baker et al., 2008; Lagendijk and
Gusset, 2008; Dar et al., 2009; Loveridge et al., 2010). Encroachment
of carnivore habitats by expanding human populations is a potential
spark for new conﬂicts, which deteriorate the complex functioning of
the environment at all levels, from individuals to ecosystems (Ripple
et al., 2014). Big cats, namely the lion (Panthera leo), tiger (Panthera
tigris), leopard (Panthera pardus), jaguar (Panthera onca), snow leopard
(Panthera uncia), puma (Puma concolor), and cheetah (Acinonyx
jubatus), are amongthe best-known carnivores responsible for conﬂicts
with humans (Inskip and Zimmermann, 2009). Retaliatory killing,
poaching and prey loss are the main threats for these species, of
which six are classiﬁed by the IUCN Red List of Threatened Species as
“Endangered”to “Near Threatened”and only puma is still common
having the “Least Concern”status (Macdonald et al., 2010).
Albeit the density and biomass of livestock exceed those of wild prey
manifold, big cats would prefer to kill wild prey to avoid human retribu-
tion (Loveridge et al., 2010). When prey, especially medium-sized and
large ungulates, becomes scarce due to population declines or seasonal
migrations felids increase predation on livestock to survive (Polisar
et al., 2003; Bauer and de Iongh, 2005; Azevedo and Murray, 2007;
Kumaraguru et al., 2011; Mondal et al., 2011; Amador-Alcalá et al.,
2013; Zhang et al., 2013; Kabir et al., 2014). In some areas, cats kill
livestock mostly during the wet season when prey disperses into lush
vegetation, regains ﬁtness and thus becomes less available, whereas
livestock enters these areas for uncontrolled grazing (Polisar et al.,
2003; Patterson et al., 2004; Kissui, 2008). In other areas, livestock
predation is minimal during winter when prey attains high densities
in certain areas with little snow (Dar et al., 2009) or it is maximal during
the dry season when limited cover decreases hunting success, prey
moves away and livestock concentrates around a few waterholes
(Schiess-Meier et al., 2007). Overall, the relationships between prey
availability and livestock predation by big cats appear to be straightfor-
ward, but some more intricate cause-and-effectpatterns are also possi-
ble. For example, Harihar et al. (2011) found out that the natural
recovery of prey after relocation of local people has led to a sharp rise,
and not a decline as expected, of livestock predation by leopards be-
cause recovering tigers displaced them closer to villages. Moreover,
Biological Conservation 192 (2015) 268–275
E-mail addresses: email@example.com (I. Khorozyan),
firstname.lastname@example.org (A. Ghoddousi),
mahmood.sooﬁ@stud.uni-goettingen.de (M. Sooﬁ), email@example.com (M. Waltert).
0006-3207/© 2015 Elsevier Ltd. All rights reserved.
Contents lists available at ScienceDirect
journal homepage: www.elsevier.com/locate/bioc
Suryawanshi et al. (2013) concluded that snow leopard predation on
livestock may intensify with more abundant prey, presumably because
higher stock of prey supports a greater number of cats. Therefore, these
authors recommend that prey recovery programs should be accompa-
nied by strengthening livestock protection measures.
If the causality between wild prey scarcity and increased livestock
predation is real, then a new question arises: at what threshold levels
of prey density/biomass are attacks on livestock triggered? We did not
ﬁnd such information in the scientiﬁc literature. Such a threshold may
vary with the species, body masses and population densities of big
cats, as well as with size of the study areas. Although being similar in
regard to obligatory meat eating, big cats may differ in livestock
predation patterns due to species-speciﬁc ecological properties. For ex-
ample, snow leopards are known for surplus killing and group-living
lions might be expected to kill more livestock than other cats, which
are solitary (Jackson et al., 2010; Loveridge et al., 2010). Livestock
predation can also be allometric, because large-bodied big cats select
cattle and buffaloes, and smaller species usually prefer sheep, goats,
and juveniles of larger species (Dar et al., 2009; Zarco-González et al.,
2013; Kabir et al., 2014). Population density of felids and other
carnivores is positively related to prey biomass and this relationship is
so strong that it allows estimating carnivore densities and carrying
capacity from current prey resources (Carbone and Gittleman, 2002;
Hayward et al., 2007; Carbone et al., 2011). However, this rule relies
only on bottom–up processes (carnivores controlled by prey) and fails
when ever-increasing top–down processes (carnivores controlled by
humans, e.g., via poaching) limit carnivore numbers while prey remains
sufﬁcient(Khorozyan et al., 2008; Kiffner et al., 2009; Zhang et al., 2013;
Bauer et al., 2014). Unlike other big cats, cheetah density is related more
to competition with larger competitors than to prey availability
(Carbone and Gittleman, 2002; Carbone et al., 2011). Sizes of study
areas are inversely related to carnivore and prey densities, so they can
mediate the strength of predator–prey relationships and livestock
predation (Carbone andGittleman, 2002).For instance, for practical rea-
sons prey populations are often studied in relatively small high-density
enclaves or protected areas, which may represent the areas of low
predation on livestock (Biswas and Sankar, 2002).
In this paper, we (a) study the linkage between livestock predation
by big cats, wild prey biomass and above-mentioned confounders,
(b) identify and estimate the minimum thresholds of prey biomass
that move predation rates up, and (c) discuss these thresholds as a
potentially useful metric for assessing and predicting human–felid
2. Materials and methods
We retrieved peer-reviewed English language scientiﬁcarticlesand
book chapters dated 2000–2014 through the ISI Web of Knowledge
(http://www.webofknowledge.com) and the IUCN/SSC Cat Specialist
Group Digital Library (http://www.catsg.org). Only recent publications
were considered to assure the most accurate and consistent data on
predictors and confounders, especially on prey biomass, prey density,
and cat density, which are particularly demanding for up-to-date
research techniques. As information on wild prey density and its
derivate prey biomass was a priori assumed to be most limited
in livestock predation studies, we used the search words
“panthera*prey density”,“acinonyx*prey density”and “puma*prey
density”. These combinations gave us more output than if we used
narrower options, e.g., “panthera*livestock*prey density”, because the
“*prey density”combination revealed control studies without livestock
predation. In an array of publications that met these criteria, we
selected those which contained at least some of the predictors and
confounders (see below) or held sufﬁcient information so that we
could calculate them (Appendix A). Original data are provided in
Appendix B. Each publication contained one livestock predation/no
predation case (one study area for one big cat species) or more cases
(2–5 study areas for a species, e.g. different protected areas in a puma
study by Donadio et al. (2010) in which each protected area was consid-
ered a separate case). We took the cases as independent if they
described different big cat species, areas and/or study periods in the
same area. Otherwise, we considered the cases as dependent and
lumped them into a single case.
2.2. Input data
The numbers of livestock killed per year, which are reported in
publications, usually do not represent actual livestock losses to
carnivores. They come mostly from interviews and also from carnivore
diets, livestock carcasses, farm reports and authority appraisals for
compensation. Interviews may underestimate losses if remote or less
and accessible villages are under-represented, if villagers forget cases
or if they are reluctant to share information (Holmern et al., 2007;
Kissui, 2008). On the other hand, villagers may overestimate losses if
they assign other mortality causes to carnivore attacks, if they perceive
carnivores as evil disproportionally to actual threat or if they want to
attract attention or get compensation (Holmern et al., 2007; Gusset
et al., 2009; Suryawanshi et al., 2013). Such biases are common, since
in most cases villagers do not get compensations for losses and there-
fore they are not obliged to accurately document them. Although
some authors try to minimize these biases by ﬁeld veriﬁcation of report-
ed losses, it is applicable only to the most recent and identiﬁable cases
and when verifying researchers are available on place (Azevedo, 2008;
Kabir et al., 2014). Feces and livestock records suffer from low detection
probabilities and underestimate livestock losses (Bagchi and Mishra,
2006; Sollmann et al., 2013; but see Wegge et al., 2012). Authority
appraisals also tend to underestimate losses because they record only
the most recent cases conﬁrmed by carcasses or other irrefutable
evidence (Sangay and Vernes, 2008). Farm reports are the most
accurate, but their published data are only few (Patterson et al., 2004;
Schiess-Meier et al., 2007; Wegge et al., 2012). As a result, inaccurate
data on livestock losses may hide a relationship between livestock
losses and predictors, which is present but goes undetected (type II
error of false negatives, non-detections or underestimations;
Zarco-González et al., 2013).
To overcomethese issues, we considered livestock predation rates in
terms of binary response variables: probability of cattle predation (CP)
and probability of sheep and goat predation (SP). We lumped sheep
and goats as “shoats”as they usually graze together and chose cattle
and shoats because of their ubiquitous predation by big cats (Inskip
and Zimmermann, 2009; Loveridge et al., 2010). We coded CP and SP
as 1 if predation was high and 0 if it was none or minimal as described
in references. If a livestock species was not taken, we coded it as 0 only if
that species was bred, i.e. available for predation. Alternatively, we left
CP or SP blank, as in the case of shoats not bred on cattle ranches
(Rosas-Rosas et al., 2008).
Although the published numbers of livestock kills can be inaccurate,
they allow getting an impression of whether livestock losses in a study
are high or low, especially when they are discussed further by the au-
thors. Our recent studies (Khorozyan et al., 2015a, in press)showed
that the numbers of killed livestock are random and unpredictable
while the binary data of high and low predation can be well described
and predicted by variables. Dietary studies routinely use the correction
factors that estimate the proportions of the numbers of livestock
consumed to the numbers of all prey consumed, which are also useful
for classifying livestock predation as high or low (Marker et al., 2003;
Azevedo, 2008; Athreya et al., in press). The main criterion that we
used to separate the cases of high and low predation was whether the
studied big cat species depended on livestock as staple food (high
269I. Khorozyan et al. / Biological Conservation 192 (2015) 268–275
predation) or they did not kill livestock or did that sporadically (low
The following predictors were considered and retrieved from the
references in Appendix A: (1) cattle holdings, individuals/household,
(2) shoats holdings, individuals/household, (3) cattle density,
, (4) shoats density, individuals/km
, (5) cattle biomass,
, (6) shoats biomass, kg/km
, (7) wild prey density, individuals/
, and (8) wild prey biomass, kg/km
. When predictors were not
provided in publications directly, we estimated them from other data
available in these publications, e.g. by estimating prey biomass from
species densities and body masses. Livestock and prey body masses
were reported as either average masses taken by a given big cat species
or as 3/4 of female body mass, to account for juveniles (Zhang et al.,
2013). We took wild prey density and biomass data mostly for ungu-
lates, which are the staple prey for big cats (Inskip and Zimmermann,
2009; Loveridge et al., 2010). However, we also included the capybara
(Hydrochoerus hydrochaeris), caiman (Caiman yacare), nine-banded
armadillo (Dasypus novemcinctus) and giant anteater (Myrmecophaga
tridactyla) for jaguar and the common langur (Semnopithecus entellus),
rhesus monkey (Macaca mulatta) and black-naped hare (Lepus
nigricollis) for leopard, which are intensively hunted by these cats
(Polisar et al., 2003; Azevedo and Murray, 2007; Kumaraguru et al.,
2011; Mondal et al., 2011; Bhattarai and Kindlmann, 2012).
We also considered four confounders which might affect the rela-
tionships between above-mentioned predictors and response variables:
(1) big cat species, (2) big cat body mass, kg, (3) big cat density, individ-
, and (4) sizeof study area, km
(Appendix C). We took the
species-speciﬁc body masses from PanTHERIA database (Jones et al.,
2009). We obtained the sizes of study areas and big cat densities from
the references in Appendix A.
We took the coordinates of livestock predation/no predation cases
from the references (Appendix A) or, if unavailable there, from the
case locality names in Geoplaner v. 2.7 (http://www.geoplaner.com).
When a study included several cases, like Donadio et al. (2010)
mentioned above, we selected a midpoint as a plausible reference. We
took the country midpoints for the nationwide cheetah studies in
Namibia and for the jaguar and puma study in anonymous ranches of
Venezuela (Marker et al., 2003; Hoogesteijn and Hoogesteijn, 2008;
Marker et al., 2010). We mapped the cases in WGS84 georeference
system in QGIS v. 2.8.1 Wien (http://www.qgis.com).
2.3. Data analysis
First, we used chi-square (χ
) test to check the difference between
the actual and expected numbers of publications as the indicators of
their bias and representativeness for each big cat species. Predictors
and confounders were ln-transformed to reduce intrinsic variation
and minimize the effect of outliers. Estimation of lethal dose 50%
), which we borrowed for this study from toxicological practice
(see below), also uses ln-transformation of concentrations as predictors
(Faraggi et al., 2003). We identiﬁed signiﬁcant predictors by comparing
them between predation and no predation cases by Mann–Whitney
test. We tested the relationships between signiﬁcant predictors,
response variables and confounders by means of logistic regression
models. We studied the effect of confounders by comparing the values
and 95% conﬁdence intervals (95% CI) of the odds ratios exp(slope) of
signiﬁcant predictors with and without the confounders in the models
(Yan and Su, 2009). We excluded outlier predictors withCook's distance
~1 and higher and multicollinear predictors with signiﬁcant Spearman's
correlation coefﬁcient r
(Yan and Su, 2009). We applied two-way
ANOVA and r
to check the dependence of big cat density upon density
and biomass of wild prey (Appendix D). The area under curve of
receiver operating characteristic (ROC), denoted as AUC, indicated
strong predictive capacity of logistic models if it exceeded 0.7
(Zarco-González et al., 2013).
We determined the threshold values of predictors from the best
logistic regression models by estimating the values of predictors,
which would cause CP and SP to equal 0.5, i.e. a 50:50 chance of
livestock to be killed by big cats. These threshold values were calculated
as -(intercept β
) of the logistic model (Faraggi et al., 2003).
The standard errors (SE) and 95% CI of these thresholds were estimated
in Bioassay, Dose response and LD
option of Simﬁt 7.0 package
(University of Manchester, UK). This approach is similar to determining
, the dose of an experimentally administered substance, which
kills 50% of subjected individuals. Apart from toxicological applications,
it is also efﬁcient in wildlife ecology, for example in calculating the
critical size of protected areas, which ensures 50% probability of survival
of large carnivores inside these areas (Woodroffe and Ginsberg, 1998;
Woodroffe, 2001). We used Wilcoxon signed-rank test to check how
signiﬁcant, in the same cases, is the difference between the actual
predation/no predation status and the predation/no predation status
predicted by thresholds. All analyses were performed in SPSS 17.0
(IBM Corp., USA) at two-tailed signiﬁcance level P= 0.05.
A total of 99 publications fulﬁlled the search criteria for leopard, 78
for lion, 74 for puma, 73 for tiger, 55 for cheetah, 38 for jaguar and 26
for snow leopard. They represented 315 publications, but only 107
(34.0%) of them contained at least part of required information and
served as references (leopard —38, 38.4%, lion —27, 34.6%, tiger —21,
28.8%, puma —20, 27.0%, jaguar —19,50.0%,cheetah—15, 27.3%,
snow leopard —11, 42.3%; Appendices A and B). The numbers of these
references per species were signiﬁcantly different (χ
= 21.497, df =
6, P= 0.001), implying an objective reality that, for example, leopard
and lion are much better studied and published than snow leopard.
However, the numbers of references did not differ from the expected
ones within the species, given their unequal studiedness (χ
df =6,P= 0.388); therefore, our meta-data were unbiased and repre-
sentative. Our dataset contained 146 geographically diverse livestock
predation/no predation cases, ranging from 11 cases from 5 countries
for cheetah to 33 cases from 12 countries for leopard (Table 1).
There were 39 cases of prey biomass from 13 countries and 63 cases
of prey density from 15 countries (Table 1). The cases with cattle preda-
tion had signiﬁcantly lower wild prey density (Mann–Whitney U=
259.5, P= 0.024) and prey biomass (U=87.0,P= 0.020) than those
without cattle predation. Similarly, the cases with sheep and goat pre-
dation had lower prey density (U=76.5,P= 0.006) and prey biomass
(U=29.5,P= 0.028) than those without it. No other predictors
showed signiﬁcant difference between predation and no predation
cases (PN0.05). The jaguar, lion and snow leopard took mostly cattle
and all big cats, except for tiger, exhibited strong predation on sheep
and goats (Fig. 1).
The signiﬁcant logistic regression models show that the probability
of cattle predation (CP) and the probability of sheep and goat predation
(SP) strongly depended on prey biomass (Table 2;Fig. 2). CP and SP also
depended on prey density (Wald = 4.855, P= 0.028 and Wald = 6.847,
P= 0.009, respectively), but we excluded these models because of low
predictive capacity of the CP model (AUC = 0.662) and the estimation
of SP N1 at low-density values.
The probability of livestock killing by big cats signiﬁcantly increased
when prey biomass fell below certain minimum thresholds. According
to the models in Table 2, the threshold values of ln-transformed prey
biomass were 6.70 ± 0.23 (95% CI =6.21–7.19) for CP and 6.30 ±
0.17 (95% CI =5.94–6.66) for SP (Fig. 2). Back-transformation of these
estimates produced the threshold values of prey biomass to equal
812.41± 1.26 kg/km
(95% CI = 497.70–1326.10 kg/km
) for CP and
544.57 ± 1.19 kg/km
(95% CI = 379.93–780.55 kg/km
(Fig. 2). The map of 39 studied cases of prey biomass in relation to
these thresholds conﬁrmed the areaswith and without known livestock
predation, with a few exceptions (Fig. 3). In Gir Protected Area (India),
270 I. Khorozyan et al. / Biological Conservation 192 (2015) 268–275
the threshold model predicted low conﬂicts while the actual conﬂicts
are high. In the national parks of Emas (Brazil) and Pench (India) and
in the Wanda Mts. (China), the model predicted high predation but it
is not recorded. Despite this, in the same 39 cases the difference be-
tween the actual predation/no predation status and the status predicted
by the thresholds was insigniﬁcant (Wilcoxon Z=−0.577, P= 0.564).
The parameters and performance of the models in Table 2 were not
confounded by big cat species, big cat body mass, big cat density and
size of study area as the odds ratios and their 95% CI from different
models signiﬁcantly overlapped (Table 3). In all models, the odds ratios
were b1 indicating that less prey biomass caused higher CP and SP
Our research shows that big cats are much more likely to kill cattle,
sheep, and goats when the biomass of wild prey is decreased (Table 2
and Fig. 2). This result is in agreement with local studies which
hypothesized a direct causal link between prey biomass and livestock
predation by big cats (Bagchi et al., 2003; Namgail et al., 2007;
Gusset et al., 2009; Harihar et al., 2011; Kumaraguru et al., 2011;
Amador-Alcalá et al., 2013; Kabir et al., 2014). The probability of
livestock killing by big cats signiﬁcantly increases when prey biomass
reaches some minimum thresholds. More speciﬁcally, these
carnivores are more likely to kill cattle when prey biomass is less than
and to kill sheep and goats when prey biomass is
below 544.57 kg/km
(Fig. 2). We suggest that these thresholds may
represent important landmarks for predicting human–felid conﬂicts
and identifying conﬂict hotspots for priority actions in conﬂict
mitigation and species conservation.
The map of cases with known prey biomass and predicted livestock
predation shows that some protected areas of India, Nepal lowlands,
and South Africa contain sufﬁcient prey that makes big cats less likely
to kill livestock. In all other sampled areas, which included mostly
protected areas and also ranches and forestry areas, prey biomass is
insufﬁcient and the probabilities of livestock predation are moderate
to high (Fig. 3). Our threshold model in Table 2 accurately predicted
the areas of high and low predation, but four exceptions were found.
From Banerjee et al. (2013), we have estimated prey biomass in Gir
Protected Area (India) to equal 1984.0 kg/km
, which is well above
the threshold. However, lion attacks on livestock (cattle) are common
in Gir because local people are ofﬁcially permitted to graze livestock in-
side the park and to get compensations to tolerate conﬂicts with lions.
Gir lions kill cattle proportionally to their availability and strongly prefer
wild ungulates (Banerjee et al., 2013). In Pench National Park (India),
Biswas and Sankar (2002) estimated prey biomass as 6013.25 kg/km
in 1998–1999 in a 61.1-km
area, whereas Majumder et al.'s (2013)
data collected in 2007–2010 over 758 km
enabled us to estimate
prey biomass as only 369.54 kg/km
for tiger and 314.03 kg/km
leopard. The discrepancy in prey biomass could ensue from spatial
inconsistency or prey declines over time, but none of these authors indi-
cated livestock predation in Pench. Possibly, this pattern resulted from a
preference for abundant, but smaller, prey like chital (Axis axis) instead
of livestock, which was available only along the park boundaries
(Majumder et al., 2013; also see below). With our threshold model,
we assume a conservative low estimate of prey biomass and predict
moderate to high livestock predation in Pench. The other two
exceptions from Emas National Park (Brazil) and the Wanda Mts.
(China) are described below.
Survival of the largest big cats –tiger, lion and jaguar –is signiﬁcantly
limited by the availability of large-bodied ungulates whose consumption
would offset high energetic costs associated with hunting, maintenance
of vast home ranges, and other activities (Carbone et al., 2011). The
snow leopard also demands high energy intake because of living in
cold, low productive and prey-poor highlands of Central Asia (Namgail
et al., 2007; Jackson et al., 2010). These four big cats are the ﬁrst
candidates to switch to killing the most proﬁtable, large-bodied domestic
animals such as cattle when wild prey biomass becomes insufﬁcient and
drops below ca. 800 kg/km
(Fig. 1). Other big cats, especially leopard,
arealsoabletotakecattle(Loveridge et al., 2010;Fig. 1). When prey bio-
mass falls below ca. 540 kg/km
even cattle cannot compensate for the
lack of food and big cats turn to killing small livestock such as sheep
and goats. This strategy strives to maximize the net energy budget of
The distributionof predictor samplesizes acrossthe big cat speciesand livestock(cattle and shoats= sheep and goats)in this study.Abbreviations:B—biomass, D—density,H—holdings,
hh —household, ind —individuals, N —number of studied cases.
Species N Livestock H(ind/hh) Livestock D(ind/km
) Livestock B(kg/km
) Prey D(ind/km
) Prey B(kg/km
Cattle Shoats Cattle Shoats Cattle Shoats
Cheetah 11 7 5 6 4 1 0 3 1
Jaguar 19 16 1 12 0 4 0 2 3
Leopard 33 17 16 16 9 5 2 17 13
Lion 22 12 10 10 6 3 1 9 5
Puma 29 17 5 19 6 3 1 7 2
Snow leopard 13 1 5 2 6 0 1 11 4
Tiger 19 6 3 7 3 3 1 14 11
Total 146 76 45 72 34 19 6 63 39
Fig. 1. The distribution of publications about predation vs. no predation on cattle (a) and
sheep and goats (b) by cheetah (CH), jaguar (JA), leopard (LE), lion (LI), puma (PU),
snow leopard (SL) and tiger (TI).
271I. Khorozyan et al. / Biological Conservation 192 (2015) 268–275
surviving felids by taking out the best available alternative prey (cattle)
when prey biomass begins to dwindle, and by killing all available
alternative prey (cattle, sheep and goats) when prey resources are too
It is imperative to study, monitor, maintain and restore the popula-
tions of wild prey, especially preferred ungulates, to forestall livestock
predation, human–felid conﬂicts and further escalation of local
extinctions of big cats. Fig. 3 shows that even many protected areas
contain insufﬁcient prey resources, implying an even worse status of
prey in unprotected lands. Suryawanshi et al. (2013) suggest that high
prey abundance may accelerate livestock-taking by supporting high
numbers of carnivores which need more food. Although we did not
ﬁnd support for this opinion, a possibility of bimodal distribution of
livestock predation at low and high levels of prey abundance is interesting
and deserves further investigations.
Apparently, big cats cannot survive when ln-transformed prey
biomass plummets to less than 2, i.e., prey biomass about 7 kg/km
(Fig. 2). The only known exception is the leopard, which can even attain
high densities in some prey-free anthropogenic landscapes by killing
domestic animals (Athreya et al., in press; Shehzad et al., 2015).
Our logistic model of the relationships between livestock predation
and wild prey biomass consists of three zones: high predation risk
zone (low prey and high predation on livestock), low predation risk
zone (high prey and low predation) and uncertainty zone (low prey
and low predation) (Fig. 2). The uncertainty zone is realistic in several
cases. Very small populations of big cats are likely to subsist on limited
prey resources without a need to depend on livestock (Zhang et al.,
2013). In the worst case, these populations may vanish and nullify
livestock predation in spite of existence of wild prey. Also, big cats can
select abundant, but smaller, prey species instead of raiding livestock
(e.g., jaguars specializing on killing giant anteaters —Sollmann et al.,
2013; tigers and leopards selecting chitals —Majumder et al., 2013).
These cases may explain why our threshold model predicted livestock
predation by tigers in the Wanda Mts., China (Zhang et al., 2013), tigers
and leopards in Pench National Park, India (Majumder et al., 2013)and
jaguars in Emas National Park, Brazil (Sollmann et al., 2013), but
actually this predation is none or minimal (Fig. 3). The other option
can take place in situations when low prey biomass drives big cats to
kill domestic animals, which were not considered in this study. In this
case, Fig. 2 would show “false zeroes”(no or low predation) because
our model was developed only for cattle, sheep, and goats. For example,
in areas with insufﬁcient prey snow leopards can take mostly horses,
tigers can kill high numbers of buffaloes and leopards can rely mainly
on domestic dogs (Bagchi et al., 2003; Bagchi and Mishra, 2006;
Athreya et al., in press).
We did not observe a confounding effect of species on the relationship
between prey biomass and livestock predation. Perhaps, this discrepancy
was caused by an insufﬁcient contrast between the species: only three
The logisticregression modelsof the dependence of theprobability of cattlepredation (CP) and theprobability of sheepand goat predation (SP)upon wild prey biomass(preybio,kg/km
Abbreviations: AUC —area under curve of Receiver Operating Characteristic (ROC), P
—signiﬁcance level of the model, P
—signiﬁcance level of the AUC,SE —standard error.
Model Wald statistic P
AUC ± SE P
CP = 1 / [1 + exp(0.729 ∗ln(preybio)−4.885)] 6.889 0.009 0.791 ± 0.081 0.005
SP = 1 / [1 + exp(1.119 ∗ln(preybio)−7.054)] 4.422 0.035 0.885 ± 0.075 0.004
Fig. 2. The logistic regressions describing the effectof prey biomass on (a) the probability of cattle predation (CP) and (b) the probability of sheep and goat predation (SP) by big cats. See
their equations in Ta ble 2. The threshold valuesof ln-transformedprey biomass are provided in boxes. The standard error margins and the 95% conﬁdence intervalsof these thresholds are
marked by dark and light colors, respectively.
272 I. Khorozyan et al. / Biological Conservation 192 (2015) 268–275
species avoided some livestock (cattle —cheetah and puma, sheep and
goats —tiger), while the others took them readily (Fig. 1). Also, the
habit to kill cattle or small livestock is variable within a species depending
on a pressure from larger competitors. For examples, leopards can kill
cattle when they are the top carnivores in the area, but tend to take
more sheep and goats when co-existing with ecologically dominant
lions or tigers (Holmern et al., 2007; Bhattarai and Kindlmann, 2012;
Thorn et al., 2013; Khorozyan et al., 2015b). The same pattern is
documented in pumas living in areas with and without jaguars
(Rominger et al., 2004; Azevedo, 2008; Rosas-Rosas et al., 2008;
Amador-Alcalá et al., 2013).
Likewise, we did not ﬁnd support for thehypothesis that body mass
of big cats would affect the prey biomass–livestock predation relation-
ship by skewing cattle killing to larger cats and sheep and goat killing
Fig. 3. The distribution of studied cases (n = 39) predicting theprobabilities of livestock predation by big cats from currently known preybiomass. Some overlapof cases may take place
when the same area is studied for different big cat species. High probability of predation N0.5 and low probability b0.5.
The odds ratios and their 95% conﬁdence intervals (95% CI) of ln-transformed wild prey biomass (preybio, kg/km
) in livestock predation models with and without the confounders.
Confounders: big cat species (species, dummy variables) and ln-transformed big cat body mass (bodymass, kg), size of study area (studyarea,km
) and big cat density (catdens,individ-
Models Probability of big cat predation on cattle (CP) Probability of big cat predation on shoats (SP)
Odds ratio 95% CI of odds ratio Odds ratio 95% CI of odds ratio
preybio 0.48 0.28–0.83 0.33 0.12–0.93
preybio +species 0.49 0.28–0.87 0.33 0.10–1.02
preybio ∗species 0.61 0.33–1.13 0.45 0.14–1.48
preybio +bodymass 0.47 0.26–0.83 0.42 0.13–1.30
preybio ∗bodymass 0.34 0.10–1.18 0.90 0.13–6.40
preybio +studyarea 0.57 0.33–1.00 0.07 0.00–3.13
preybio ∗studyarea 0.51 0.28–0.94 0.04 0.00–2.78
preybio +catdens 0.58 0.25–1.41 −−
preybio ∗catdens 0.69 0.27–1.75 −−
preybio +bodymass +studyarea 0.55 0.31–1.00 0.15 0.01–2.63
preybio +bodymass +catdens 0.59 0.24–1.44 −−
preybio +studyarea +catdens 0.96 0.33–2.81 −−
preybio +species +bodymass 0.47 0.25–0.87 0.38 0.11–1.29
preybio +species +studyarea 0.56 0.30–1.04 0.07 0.00–1.77
preybio +species +catdens 0.96 0.33–2.81 −−
preybio ∗bodymass ∗studyarea 0.50 0.27–0.93 0.05 0.00–2.88
preybio ∗bodymass ∗catdens 0.63 0.25–1.60 −−
preybio ∗studyarea*catdens 0.60 0.24–1.49 −−
preybio ∗species ∗bodymass 0.57 0.31–1.04 0.50 0.15–1.64
preybio ∗species ∗studyarea 0.63 0.35–1.11 0.07 0.00–1.56
preybio ∗species ∗catdens 0.74 0.31–1.81 −−
preybio +bodymass ∗studyarea 0.57 0.33–1.00 0.07 0.00–3.22
preybio +bodymass ∗catdens 0.56 0.23–1.34 −−
preybio +studyarea ∗catdens 0.57 0.25–1.29 −−
preybio +species ∗bodymass 0.50 0.29–0.88 0.36 0.11–1.17
preybio +species ∗catdens 0.61 0.25–1.47 −−
preybio +species ∗studyarea 0.51 0.28–0.92 0.11 0.01–2.33
273I. Khorozyan et al. / Biological Conservation 192 (2015) 268–275
to smaller cats. There are two possible reasons for this. The snow leop-
ard, which is the smallest and most light-weighted of big cats, is
known to actively kill cattle and also other large-bodied animals like
yaks and horses (Fig. 1;Bagchi and Mishra, 2006; Namgail et al., 2007;
Sangay and Vernes, 2008; Jackson et al., 2010). Also, all the largest big
cats, except for the tiger, are keen to kill sheep and goats (Fig. 1;Bauer
and de Iongh, 2005; Amador-Alcalá et al., 2013). Reliance of livestock
predation upon prey biomass appears to be stable regardless of big cat
densities and sizes of study areas. Higher prey biomass leads to higher
densities of big cats, which are inversely proportional to sizes of study
areas (Appendix D; Carbone and Gittleman, 2002; Carbone et al.,
2011; Zhang et al., 2013). On the other hand, more prey means less
livestock predation (Fig. 2), but we did not ﬁnd differences in big cat
densities or sizes of study areas in predation vs. no predation areas.
Our results did not conﬁrm that high prey abundance may contribute
to higher predation on livestock by supporting more carnivores
(Suryawanshi et al., 2013).
A recent overview by Ripple et al. (2014) claims that expanding
animal husbandry, which strives to meet increasing human demand
for meat, poses a threat to carnivores by intensifying their clashes
with livestock. Machovina and Feeley (2014) further suggest that
reducing livestock numbers and their substituting by alternative
proteins, such as soybeans, would negate human–carnivore conﬂicts.
We do not agree with these views, as we did not ﬁnd the effect of
livestock holdings, density or biomass on its predation by big cats.
Livestock predation depends more on favorable conditions for success-
ful hunting (e.g., dense cover or lax husbandry) rather than on mere
availability of livestock (Bagchi and Mishra, 2006; Sangay and Vernes,
2008; Kabir et al., 2014). Earlier studies have revealed that felid preda-
tion rates can increase with livestock abundance and density, decrease
or remain unaffected by them (Azevedo, 2008; Amador-Alcalá et al.,
2013; Zarco-González et al., 2013).
This study suggests that livestock predation by big cats can be
reliably determined and predicted by biomass of wild prey species.
Predation rates signiﬁcantly increase when prey biomass decreases
below certain minimum thresholds, which are higher for cattle
) than for sheep and goats (544.57 kg/km
). Being opti-
mal for net energy maximization by larger big cats and snow leopards,
cattle are expected to be killed by these species ﬁrst when prey biomass
becomes insufﬁcient. When prey biomass is below ca. 540 kg/km
sheep and goats are more intensively killed along with cattle to
optimize energy intake. These threshold values of wild prey biomass
can be used as important predictors of human–felid conﬂicts allowing
the identiﬁcation of conﬂict hotspots and targeted conservation actions.
Therefore, more efforts are required to study, monitor, maintain, and
restore the populations of wild prey, especially preferred ungulates, to
forestall livestock predation, human–felid conﬂicts and further escala-
tion of local extinctions of big cats.
Supplementary data to this article can be found online at http://dx.
Funding for this work was provided by Alexander von Humboldt
Special Research Fellowship (#1151598), German Academic Exchange
Service/DAAD (#A/11/96604) and Erasmus Mundus/SALAM (#2013-
2437-001-001-EMA2). We appreciate constructive comments made
by M. Lucherini, J. Sanderson and three anonymous reviewers on the
earlier version of the manuscript. The authors declare no conﬂicts of in-
terest. Animal images on Fig. 2 and the map of continents on Fig. 3 were
obtained from the public domains of Rolera LLC, USA (http://www.
clker.com) and ArcGIS, USA (http://www.arcgis.com), respectively.
The funders had no role in study design, data collection and analysis,
decision to publish, or preparation of the manuscript.
Amador-Alcalá, S., Naranjo, E.J., Jiménez-Ferrer, G., 2013. Wildlife predation on livestock
and poultry: implications for predator conservation in the rainforest of south-east
Mexico. Oryx 47, 243–250.
Athreya,V., Odden, M., Linnell,J.D., Krishnaswamy, J., Karanth, K.U.,2014. Acatamongthe
dogs: leopard Panthera pa rdus diet in a human-dominated landscape in western
Maharashtra, India. Oryx (in press). doi:10.1017/S0030605314000106.
Azevedo, F.C.C., 2008. Food habits and livestock depredation of sympatric jaguars and
pumas in the Iguaçu National Park area, south Brazil. Biotropica 40, 494–500.
Azevedo, F.C.C., Murray, D.L., 2007. Spatial organization and food habits of jaguars
(Panthera onca)inaﬂoodplain forest. Biol. Conserv. 137, 391–402.
Bagchi, S., Mishra, C., 2006. Living with large carnivores: predation on livestock by the
snow leopard (Uncia uncia). J. Zool. (Lond.) 268, 217–224.
Bagchi, S., Goyal, S.P., Sankar, K., 2003 . Prey abundance and prey selection by tigers
(Panthera tigris) in a semi-arid, dry deciduous forest in western India. J. Zool .
(Lond.) 260, 285–290.
Baker, P.J., Boitani,L., Harris, S., Saunders, G., White, P.C.L.,2008. Terrestrial carnivores and
human food production: impact and management. Mammal Rev. 38, 123–166.
Banerjee,K., Jhala, Y.V., Chauhan, K.S.,Dave, C.V., 2013.Living with lions: the economics of
coexistence in the Gir forests, India. PLoS One 8, e49457.
Bauer, H., de Iongh, H.H., 2005. Lion (Panthera leo) home ranges and livestock conﬂicts in
Waza National Park, Cameroon. Afr. J. Ecol. 43, 208–214.
Bauer, D., Schiess-Meier, M., Mills, D.R., Gusset, M., 2014. Using spoor and prey counts to
determine temporal and spatial variation in lion (Panthera leo)density.Can.J.Zool.
Bhattarai, B.P., Kindlmann, P., 2012. Interactions between Bengal tiger (Panthera tigris)
and leopard (Panthera pardus): imp lications for their conservation. Biodivers.
Conserv. 21, 2075–2094.
Biswas, S., Sankar, K., 2002. Prey abundance and food habit of tigers (Panthera tigris tigris)
in Pench National Park, Madhya Pradesh, India. J. Zool. (Lond.) 256, 411–420.
Carbone, C., Gittleman, J.L., 2002. A common rule for the scaling of carnivore density.
Science 295, 2273–2276.
Carbone, C., Pettorelli, N., Stephens, P.A.,2011. The bigger they come, the harder they fall:
body size and prey abundance inﬂuence predator–prey ratios. Biol. Lett. 7, 312–315.
Dar, N.I., Minhas, R.A., Zaman, Q., Linkie, M., 2009. Predicting the patterns, perceptions
and causes of hum an–carnivore conﬂict in and around Machiara Nati onal Park,
Pakistan. Biol. Conserv. 142, 2076–2082.
Donadio, E., Novaro, A.J., Buskirk, S.W., Wurstten, A., Vitali, M.S., Monteverde, M.J., 2010.
Evaluating a potentially strong trophic interaction: pumas and wild camelids in
protected areas of Argentina. J. Zool. (Lond.) 280, 33–40.
Faraggi, D., Izikson, P., Reiser, B., 2003. Conﬁdence intervals for the 50 per cent response
dose. Stat. Med. 22, 1977–1988.
Gusset, M., Swarner, M.J., Mponwane, L., Keletile, K., McNutt, J.W., 2009. Human–wildlife
conﬂict in northern Botswana: livestock predation by endangered African wild dog
Lycaon pictus and other carnivores. Oryx 43, 67–72.
Harihar, A., Pandav, B., Goyal, S.P., 2011. Responses of leopard Panthera pardus to the
recovery of a tiger Panthera tigris population. J. Appl. Ecol. 48, 806–814.
Hayward, M.W., O'Brien, J., Kerley, G.I.H., 2007. Carrying capacity of large African
predators: predictions and tests. Biol. Conserv. 139, 219–229.
Holmern,T., Nyahongo, J., Røskaft, E., 2007. Livestockloss caused by predators outside the
Serengeti National Park, Tanzania. Biol. Conserv. 135, 518–526.
Hoogesteijn, R. , Hoogesteijn, A., 2008. Conﬂicts between cattle ranching and large
predators in Venezuela: could use of water buffalo facilitate felid conservati on?
Oryx 42, 132–138.
Inskip,C.,Zimmermann,A.,2009.Human–felid conﬂict: a review of patte rns and
priorities worldwide. Oryx 43, 18–34.
Jackson, R.M., Mishra, C., McCar thy, T.M., Ale, S.B., 2010. Snow leopards: conﬂict and
conservation. In: Macdonald, D.W., Loveridge, A.J. (Eds.), Biology and Conservation
of Wild Felids. Oxford Univ. Press, Oxford, pp. 417–430.
E.H., Carbone, C., Connolly, C., Cutts, M.J., Foster, J.K., Grenyer, R., Habib, M., Plaster, C.A.,
Price, S.A., Rigby, E.A., Rist, J., Teacher, A., Bininda-Emonds, O.R.P., Gittleman, J.L., Mace,
G.M., Purvis, A., 2009. PanTHERIA: a species-level database of life history, ecology, and
geography of extant and recently extinct mammals. Ecology 90, 2648.
Kabir, M., Ghoddousi, A., Awan, M.S., Awan, M.N., 2014. Assessment of human–leopard
conﬂict in Machiara National Park, Azad Jammu and Kashmir, Pakistan. Eur.
J. Wildl. Res. 60, 291–296.
Khorozyan, I.G., Malkhasyan,A.G., Abramov, A.V., 2008. Presence–absence surveys of prey
and their use in predicting leopard (Panthera pardus) densities: a case study from
Armenia. Integr. Zool. 3, 322–332.
Khorozyan, I., Sooﬁ,M., Ghoddousi, A., Hamidi, A.K., Waltert, M., 2015a. The relationship
between climate, diseases of domestic animals and human–carnivore conﬂicts.
Basic Appl. Ecol. (in press). doi:10.1016/j.baae.2015.07.001.
Khorozyan, I., Sooﬁ, M., Hamidi, A.K., Ghoddousi, A., Waltert, M., 2015b. Dissatisfaction
with veterinary services is associated with leopard (Panthera pardus) predation on
domestic animals. PLoS One 10, e0129221.
Kiffner,C.,Meyer,B.,Mühlenberg,M.,Waltert,M.,2009.Plenty of prey, few pred ators: what
limits lions Panthera leo in Katavi National Park, western Tanzania? Oryx 43, 52–59.
Kissui, B.M., 20 08. Livestock predation by lions, leopards, spotted hyenas, and their
vulnerability to retaliatory killing in the Maasai steppe, Tanzania. Anim. Conserv.
274 I. Khorozyan et al. / Biological Conservation 192 (2015) 268–275
Kumaraguru, A., Saravanamuthu, R., Brinda, K., Asokan, S., 2011. Prey preference of large
carnivores in Anamalai Tiger Reserve, India. Eur. J. Wildl. Res. 57, 627–637.
Lagendijk, D.D.G., Gusset, M., 2008. Human–carnivore coexistence on communal land
bordering the greater Kruger area, South Africa. Environ. Manag. 42, 971–976.
Loveridge, A.J., Wang, S.W., Frank, L.G.,Seidensticker, J., 2010. Peopleand wild felids: con-
servation of cats and management of conﬂicts. In: Macdonald, D.W., Loveridge, A.J.
(Eds.), Biology and Conservation of Wild Felids. Oxford Univ. Press, Oxford,
Macdonald, D.W., Loveridge, A.J., Nowell, K., 2010. Dramatis personae: an introduction to
the wild felids. In: Macdonald, D.W., Loveridge, A.J. (Eds.), Biology and Conservation
of Wild Felids. Oxford Univ. Press, Oxford, pp. 3–59.
Machovina, B., Feeley, K.J., 2014. Taking a bite out of biodiversity. Science 343, 838.
Majumder, A., Sankar, K., Qureshi, Q., Basu, S., 2013. Predation ecology of large sympatric
carnivores as inﬂuenced by available wild ungulate prey in a tropical deciduous forest
of central India. J. Trop. Ecol. 29, 417–426.
Marker, L., Dickman, A.J., Mills, M.G.L., Macdonald, D.W., 2010. Cheetahs and ranchers in
Namibia: a case study. In: Macdonald, D.W., Loveridge, A.J. (Eds.), Biology and
Conservation of Wild Felids. Oxford Univ. Press, Oxford, pp. 353–371.
Marker, L.L., Muntifering, J.R., Dickman, A.J., Mills, M.G.L., Macdonald, D.W., 2003.
Quantifying prey preferences of free-ranging Namibian cheetahs. S. Afr. J. Wildl.
Res. 33, 43–53.
Mondal, K., Gupta, S., Qureshi, Q., Sankar, K., 2011. Prey selection and food habits of
leopard (Panthera pardus fusca) in Sariska Tiger Reserve, Rajasthan, India. Mammalia
Namgail, T., Fox, J.L., Bhatnagar, Y.V., 2007. Carnivore-caused livestock mortality in trans-
Himalaya. Environ. Manag. 39, 490–496.
Patterson, B.D., Kasiki, S.M., Selempo, E., Kays, R.W., 2004. Livestock predation by lions
(Panthera leo) and other carnivores on ranches neighboring Tsavo National Parks,
Kenya. Biol. Conserv. 119, 507–516.
Polisar, J., Maxit, I., Scognamillo, D., Farrell, L., Sunquist, M.E.,Eisenberg, J.F., 2003. Jaguars,
pumas, their prey base, and cattle ranching: ecological interpretations of a manage-
ment problem. Biol. Conserv. 109, 297–310.
Ripple, W.J., Estes, J.A., Beschta, R.L., Wilmers, C.C., Ritchie, E.G., Hebblewhite, M., 2014.
Status and ecological effects of the world's largest carnivores. Science 343, 1241484.
Rominger,E.M., Whitlaw, H.A., Weybright, D.L., Dunn, W.C., Ballard, W.B., 2004. The inﬂu-
ence of mountain lion predation on bighorn sheep translocations. J.Wildl. Manag. 68,
Rosas-Rosas, O.C., Bender, L.C., Valdez, R., 2008. Jaguar and puma predation on cattle
calves in northeastern Sonora, Mexico. Rangel. Ecol. Manag. 61, 554–560.
Sangay, T., Vernes, K., 2008. Human–wildlife conﬂict in the Kingdom of Bhutan: patterns
of livestock predation by largemammalian carnivores. Biol.Conserv. 141, 1272–1282.
Schiess-Meier, M., Ramsauer, S., Gabanapelo, T., König, B., 2007. Livestock predation —in-
sights from problem animal control registers in Botswana. J. Wildl. Manag. 71,
Shehzad, W., Nawaz, M.A., Pompanon, F., Coissac, E., Riaz, T., Shah, S.A., Taberlet, P., 2015.
Forest without prey: livestock sustain a leopard Panthera pardus population in
Pakistan. Oryx 49, 248–253.
Sollmann, R., Betsch, J., Malzoni Furtado, M., Hofer, H., Jácomo, A.T.A., Palomares, F.,
Roques, S., Tôrres, N.M., Vynne, C., Silveira, L., 2013. Note on the diet of the jaguar
in central Brazil. Eur. J. Wildl. Res. 59, 445–448.
Suryawanshi, K.R., Bhatnagar, Y.V., Redpath, S., Mishra, C., 2013. People, predators and
perceptions: patterns of livestock depredation by sn ow leopards and wolves.
J. Appl. Ecol. 50, 550–560.
Thorn, M., Green, M., Scott, D., Marnewick, K., 2013. Characteristics and determinants of
human-carnivore conﬂict in South African farmland. Biodivers. Conserv. 22,
Treves, A., Karanth, K.U., 2003. Human–carnivore conﬂict and perspectives on carnivore
management worldwide. Conserv. Biol. 17, 1491–1499.
Wegge, P., Shrestha, R., Flagstad, Ø., 2012. Snow leopard Panthera uncia predation on
livestock and wild prey in a mountain valley in northern Nepal: implications for
conservation management. Wildl. Biol. 18, 131–141.
Woodroffe, R., 2001. Strategies for carnivore conservation: lessons from contemporary
extinctions. In: Gittleman, J.L., Funk, S.M., Macdonald, D.W., Wayne, R.K. (Eds.),
Carnivore Conservation. Cambridge Univ. Press, Cambridge, pp. 61–92.
Woodroffe, R., Ginsberg, J.R., 1998. Edge effects and the extinction of populations inside
protected areas. Science 280, 2126–2128.
Yan, X., Su, X., 2009. Linear Regression Analysis: Theory and Computing. World Sci. Publ.
Zarco-González, M. M., Monroy-Vilchis, O., Alaniz, J., 2013. Spatial model of livestock
predation by jaguar and puma in Mexico: conservation planning. Biol. Conserv. 159,
Zhang, C., Zhang, M., Stott, P., 2013. Does prey density limit Amur tiger Panthera tigris
altaica recovery in northeastern China? Wildl. Biol. 19, 452–461.
275I. Khorozyan et al. / Biological Conservation 192 (2015) 268–275