wileyonlinelibrary.com/journal/ddi Diversity and Distributions. 2017;23:592–603.© 2017 John Wiley & Sons Ltd
Combining landscape suitability and habitat connectivity to
conserve the last surviving population of cheetah in Asia
Mohsen Ahmadi1 | Bagher Nezami Balouchi2,3 | Houman Jowkar3 |
Mahmoud-Reza Hemami1 | Davoud Fadakar1 | Shima Malakouti-Khah1 |
1Department of Natural Resources, Isfahan
University of Technology, Isfahan, Iran
2Department of Natural Resources and
Environment Sciences, University of
Environment, Karaj, Iran
3Conservation of Asiatic Cheetah Project
(CACP), I.R. Iran Department of Environment,
4Wildlife Conservation Society (WCS), Bronx,
Mohsen Ahmadi, Department of Natural
Resources, Isfahan University of Technology,
DoE of the Islamic Republic of Iran; Global
Environmental Facilities (GEF); United Nations
Development Program (UNDP); Wildlife
Conservation Society (WCS)
Editor: Piero Visconti
Aim: The Asiatic cheetah, Acinonyx jubatus venaticus, a critically endangered large felid,
has disappeared from vast tracks of its historical range across south- western Asia. It is
currently confined to the arid ecosystems of central Iran for which little is known
about its distribution and habitat linkages. We proposed the first evaluation of Asiatic
cheetah’s distribution and developed models of landscape suitability and connectivity
to inform future conservation planning.
Location: Central Iran.
Methods: We analysed presence data of a 14- year- long cheetah monitoring pro-
gramme according to environmental and anthropogenic factors, and generated an en-
semble model of habitat suitability based on seven species distribution models (SDMs).
We then used the concept of circuit theory and landscape connectivity prioritization
(LCP) on resultant core habitats and landscape suitability to evaluate potential linkages
between core areas.
Results: Core habitats, that is, the areas hosting the largest continuous suitable habi-
tats for Asiatic cheetahs, covered approximately 49,144 km2 (c. 6.3% of the study
area). Availability of prey species, avoidance of human- dominated areas and their in-
frastructures, and rough landscapes covered with sparse vegetation were the most
predictive factors of the core habitats for the last cheetah population in Asia. Although
relatively vast, the area of potential core habitats available to cheetahs appeared to be
fragmented with limited connectivity between the northern and southern parts of this
Main conclusions: Our approach highlights the importance of distribution models to
recognize, at a coarse- scale level, a spatial population structure and habitat suitability
characteristics for a large carnivore surviving at very low density. We have identified
specific areas of suitable habitat where developing new landscape protection and
adaptive conservation management; and improving the safety of important linkages
between core habitats are likely to promote the conservation of the last surviving
population of cheetah in Asia.
arid environment, cheetah conservation planning, circuit theory, ensemble model, Iran, species
AHMADI et Al.
1 | INTRODUCTION
Apex predators play a fundamental role in many ecosystems as key-
stone species and are also important flagship species for conservation
(Ford et al., 2014; Ripple et al., 2014), but they are among the most
controversial and challenging groups of species to be conserved in the
face of human development in modern world (Chapron et al., 2014).
While conservation of large carnivores seems an effective strategy
for protecting habitat necessary for their prey and associated species
(Kunkel, Atwood, Ruth, Pletscher, & Hornocker, 2013; Sergio et al.,
2008), large carnivores management strategies and their conserva-
tion implications face many challenges. For example, it is difficult to
dedicate to them spatially extensive heterogeneous landscapes to ful-
fil their broad ecological requirements and range- wide home ranges
(Chapron et al., 2014; Ripple et al., 2014; Santini, Boitani, Maiorano,
& Rondinini, 2016). In many cases, they require action on a scale
that is seldom seen in terrestrial conservation, including coordinated
trans- boundary initiatives (Farhadinia et al., 2015; Rabinowitz & Zeller,
2010). Also identifying and preserving connectivity among large car-
nivore’s heterogeneous habitats appears crucial for the maintenance
of functional ecological linkages, and vital to their long- term survival
(Crooks, Burdett, Theobald, Rondinini, & Boitani, 2011; Dickson,
Roemer, McRae, & Rundall, 2013; Santini, Saura, & Rondinini, 2016).
The Asiatic cheetah (Acinonyx jubatus venaticus; Griffith, 1821) is a
critically endangered large feline now confined to the arid landscapes
of central Iran and is thought to number <100 individuals (Hunter et al.,
2007). Similar to the critically endangered Saharan cheetah (A. j. hecki),
also living in desert habitats, the Asiatic cheetah is wide ranging and
occurs at very low density compared to cheetahs in more productive
habitats (Belbachir, Pettorelli, Wacher, Belbachir- Bazi, & Durant, 2015;
Farhadinia et al., 2013). Although the Asiatic cheetah has been regularly
reported from a number of protected areas scattered across central Iran
(Hunter et al., 2007; Moqanaki & Cushman, 2016), it does not seem to
be confined to these sites and has been documented to move long dis-
tances, over large stretches of deserts between distant areas (Farhadinia
et al., 2013). Overall, habitat suitability criteria for the Asiatic cheetah
are poorly understood, and as a corollary, the extent of environmental,
biological and anthropogenic factors affecting the connectivity within
this habitat and the proportion of suitable habitat receiving some level of
protection are unknown. These uncertainties hinder the implementation
of effective land use planning across its vast landscape to maintain con-
nectivity between suitable habitats and mitigate conflicts with humans.
The deteriorating situation of the Asiatic cheetah requires conserva-
tion measures that are supported by accurate information on its distri-
bution patterns and dispersal possibilities (Hunter et al., 2007). Species
distribution models (SDMs) have been used in many studies to better un-
derstand habitat suitability criteria for large carnivores (Almasieh, Kaboli,
& Beier, 2016; Brito, Acosta, Álvares, & Cuzin, 2009; Farhadinia et al.,
2015) and have enabled to prioritize large carnivore’s conservation actions
(Farhadinia et al., 2015; Rabinowitz & Zeller, 2010; Sanderson, Redford,
et al., 2002). Nonetheless most range- wide priority- setting attempts to
achieve conservation goals for carnivores have been confronted with the
difficulty of addressing corridors and habitat connectivity (Rabinowitz &
Zeller, 2010). To ensure that populations of large carnivores are conserved
within a sustainable habitat complex, it is necessary to have a connected
network of protected areas or functional conservation networks (Crooks
et al., 2011), which aim to increase connectivity and promote dispersal of
large mammals between core habitats or/and population units (Almasieh
et al., 2016; Rabinowitz & Zeller, 2010).
Recently, the concept of habitat permeability and landscape
connectivity prioritization (LCP) has proved a powerful approach for
wildlife conservation planning (e.g., Carroll, McRae, & Brookes, 2012;
Dickson et al., 2013; Visconti & Elkin, 2009). Functional connectiv-
ity allows biologists to take into account the effect of compositional
structure of the landscape on ecological and evolutionary processes
of species dispersal, gene flow and population dynamics (Carroll et al.,
2012; McRae & Beier, 2007). Furthermore, identifying patches requir-
ing extra protection improves the maintenance of ecological integrity
and enables conservation planning to prompt long- term population
viability (Saura & Pascual- Hortal, 2007; Visconti & Elkin, 2009). This
approach may prove appropriate for the Asiatic cheetah, which shows
exceptionally high degree of mobility across patchily dispersed strong-
holds, all vulnerable to habitat deterioration (Farhadinia et al., 2013).
This study, which is based on all reliable Asiatic cheetah presence
data compiled over the past 14 years, is the first attempt to under-
stand the global distribution patterns of the species in Iran. Recently,
Moqanaki and Cushman (2016) proposed a landscape connectivity
model among the Iranian conservation areas (CAs) for the Asiatic
cheetah. However, their results were limited by the facts that they
did not use data on cheetah presence, they considered CAs as the
only cheetah strongholds across the landscape and did not use habitat
suitability and patterns of distribution along the environmental gradi-
ents outside CAs. In the current study, we present an approach that
combines SDM, circuit and graph theories to (1) identify habitat suit-
ability and the remaining core habitats for cheetahs, (2) evaluate the
most important environmental factors influencing their distribution,
(3) assess landscape permeability among core habitats and (4) priori-
tize core habitats and linkages based on their contribution to maintain
long- term connectivity.
2 | METHODS
2.1 | Study area
The central plateau of Iran covers approximately 780,000 km2 of
land limited by the Alborz and Zagros mountain chains to the north
and west/south- west, respectively, and the international border with
Afghanistan and the Sistan- Baluchistan desert to the east and south-
east, and is administered by nine provinces (Figure 1). The area is char-
acterized by a warm arid climate and is composed of vast flat drylands
with patchily distributed mountainous areas. It is part of the Irano-
Turanian floristic region in which xerophytic plant taxa of Artemisia
sp., Stipa sp. and Salsola sp. dominate (Manafzadeh, Salvo, & Conti,
2014). Starting in the early 1970s, the Department of Environment
(DoE) of Iran has been creating and administrating an expanding net-
work of CAs with the aim to protect and manage the faunal, floral and
AHMADI et Al.
geological diversity of Iran (Makhdoum, 2008). Ten CAs created in the
central plateau, including three National Parks, five Wildlife Refuges
and two Protected Areas (corresponding to management categories
II, IV, and V, respectively, of the IUCN protected areas categories sys-
tem), have been implementing since 2001 targeted conservation ef-
forts to protect the Asiatic cheetah, its habitat and prey.
2.2 | Asiatic cheetah locations
The “Conservation of the Asiatic Cheetah Project” (CACP) of DoE has
compiled cheetah occurrence information via a country- scale monitoring
programme, extending from 2001 to 2014. The preliminary data subset
used in the study included the totality of the 680 presence locations of
this compilation, corresponding to direct observations (i.e., field patrols
of guards, 485 presence points) and camera trap photographs (195 pres-
ence points) of free- ranging cheetahs, because this information might
have suffered spatial biases in sampling effort, resulting in over- fitting
of spatial models in areas with clumping of presence points (Dormann
et al., 2007; Kramer- Schadt et al., 2013). We performed a spatial filter-
ing procedure to account for spatially biased records (Kramer- Schadt
et al., 2013). We ran a Global Moran’s I test to evaluate the spatial auto-
correlation of the presence data across the study area. We then filtered
presence points to only single points within a 5 km- distance from others,
which resulted in 205 unique locations with decreased spatial autocor-
relation, upon which we based the SDM approach.
2.3 | Environmental variables
We selected 10 environmental and anthropogenic variables likely to
affect the distribution of the cheetah (Pettorelli, Hilborn, Broekhuis, &
Durant, 2009; Farhadinia & Hemami, 2010; Burton, Sam, Balangtaa, &
Brashares, 2012; Andresen, Everatt, & Somers, 2014; Table 1). Land
cover classes including low canopy rangelands, moderate canopy
rangelands, shrublands and barelands (see Table 1 for list of variables
FIGURE1 Documented cheetah presence locations in the nine provinces of the central plateau of Iran between 2001 and 2014. Cheetah
conservation areas are those with specific management aiming at conserving cheetahs
I R A N
Cheetah conservation areas
AHMADI et Al.
and descriptions) were extracted from maps developed by the Iranian
Forests, Ranges and Watershed Management Organization (IFRWO).
To provide continuity for the extracted categories (Franklin, 2010), we
calculated the proportion of each cover type within a 5 × 5 km grid by
running the ArcMap Neighborhood statistic tool.
To take into account the anthropogenic effects in our model-
ling approach, we used the human footprint model developed by
Sanderson, Jaiteh, et al (2002) which integrates data on population
density and the presence of infrastructures including road networks,
land transformation and human access. Because of the coarse preci-
sion of the human footprint model, we also included the density of
villages in the study area estimated from a kernel density function ap-
plied to village point layer obtained from a topographic military map of
Iran (1:25,000). Using the Shuttle Radar Topography Mission (SRTM)
elevation model (http://srtm.csi.cgiar.org), we also considered altitude
and topographic roughness (i.e., standard deviation of altitude of all
raster cells within a grid of 5 × 5 km) in the modelling method as the
most important variables affecting physiographic heterogeneity.
To account for the availability of the main cheetah prey species; the goi-
tered gazelle (Gazella subgutturosa), Jebeer gazelle (G. bennettii), wild sheep
(Ovis orientalis) and Persian ibex (Capra aegagrus) (Farhadinia & Hemami,
2010; Harrison & Bates, 1991), we used distributional data compiled in
the “Atlas of Mammals of Iran” (Karami, Ghadirian, & Feizollahi, 2013) at a
25 × 25 km grid scale. We overlaid shape files of these four preys’ distribu-
tion to obtain a composite map of presence. We then calculated distance
to areas hosting prey species by running ArcMap Spatial Analyst Tools.
All the explanatory variables were prepared in ArcgIs 9.3 (Esri,
2010) at a grid size of 1 × 1 km. Before starting the modelling work,
we calculated Pearson correlation coefficients to test for multicol-
linearity among predictors, but detected no high correlation (more
than 0.7) between any pair of explanatory variables.
2.4 | Distribution modelling approach
To predict cheetah distribution, we used biomod2 package (Thuiller,
Lafourcade, Engler, & Araújo, 2009) in R environment v. 3.1.2
(R Development Core Team, 2014). Because different modelling
methods can yield widely varying results, using this method allowed
us to simultaneously take into account results from multiple model-
ling approaches and build a consensus model called as “ensemble”
model (Araújo & New, 2007; Thuiller et al., 2009). We used three
regression- based methods: generalized linear models (GLM), gen-
eralized additive models (GAM) and multiple adaptive regression
splines (MARS), and four machine learning algorithms: general-
ized boosting model (GBM), random forest (RF), maximum entropy
(MaxEnt) and artificial neural network (ANN) to obtain an integrative
prediction of Asiatic cheetah’s distribution in the study area. As all
these models require background data (e.g., pseudo- absence points),
we generated a randomly drawn sample of 5,000 background points
from the extent of study area excepting occurrence cells. We cali-
brated models using the 75% of occurrence points as training data,
and evaluated models prediction based on the remaining 25% of
data set as test data. Models were evaluated using area under the
curve (AUC) of a receiver operating characteristic (ROC) plot and the
true skill statistic (TSS) because of their independence from preva-
lence in the species data (Allouche, Tsoar, & Kadmon, 2006).
Using BIOMOD framework, we estimated the contribution (i.e.,
importance) of variables in the cheetah’s distribution models, and the
response of the species distribution to the gradient of explanatory
variables was also evaluated based on the response curves derived
from GLM, GBM and RF models.
Finally, we implemented the ensemble model by weighted- averaging
the individual models proportionally to all their evaluation metrics scores
(Thuiller et al., 2009). In addition to obtaining final cheetah’s distribution
model, we also intended to find the most suitable areas as core habitats
for cheetahs and evaluate habitat connectivity among them. For this
reason, instead of a binary classification of presence/absence from the
final ensemble model, we overlaid presence/absence map of the seven
aforementioned models using a map algebra procedure. Using stacked
binary map of presence/absence models, we obtained a composite map
of suitable/unsuitable areas with raster values of 0–7 in which 0 score
indicates areas being unsuitable in all models and 7 represents areas
identified as suitable by all models and categorized as core habitats.
To identify habitat patches with minimum areas capable of sustaining
cheetahs, we focused on patches larger than 1,700 km2. We selected
this threshold value based on preliminary telemetry results (H. Jowkar,
personal communication, 2007) and the estimated mean home- range
size of the Saharan cheetah, a subspecies living in a comparable arid
ecosystem (i.e., 1,583 km2; Belbachir et al., 2015).
2.5 | Cheetah habitat connectivity
To evaluate the connectivity within cheetah’s suitable habitats in the vast
desert of central Iran, we used the concept of circuit theory and cIrcuItscApe
software (McRae & Shah, 2009). Through identifying multiple alternative
Overlaid shape file of the distribution of main prey
Sparse vegetation with density ≤ 25%
Mixture of grassland–scrubland with density ≥ 25%
Shrubland Patches covered by scrubs–shrubs with canopy
cover ≥ 10%
Bareland Uncovered areas including sand dunes and salty lands
Altitude Elevation above sea level
Roughness SD of altitude of all raster cells within a 5 × 5 km grid
Cropland Agricultural properties including dry and irrigated farms
Village density Number of villages within a 5 × 5 km grid
Integrated index of population density, land
transformation, human access and presence of
TABLE1 Environmental variables used in species distribution
modelling for evaluating Asiatic cheetah distribution in arid
ecosystems of central Iran
AHMADI et Al.
pathways, this method provides a detailed exploration of potential linkage
and connectivity variability (Walpole, Bowman, Murray, & Wilson, 2012).
Circuit theory treats cells in a landscape as electrical nodes connected to
neighbouring cells by resistors, with resistance values determined by the
cells’ landscape resistance/conductance values (McRae, Dickson, Keitt, &
Shah, 2008). Furthermore, using this method, we identified “pinch points”
as areas where current densities are high and alternative pathways are not
available (see McRae et al., 2008 for more details). We used core habitats
as source patches, and the ensemble distribution model as a measure of
conductance (i.e., conductance of each raster point for movement).
2.6 | Landscape connectivity prioritization
For LCP, we focused on the probability of connectivity (PC) index,
which is among the most well- performing indices in landscape con-
nectivity analysis (Bodin & Saura, 2010; Saura & Pascual- Hortal,
2007). The characteristic that used to derive PC index can refer to
different attributes such as patch area (i.e., core habitat in this study),
area- weighted habitat quality, carrying capacity or other relevant at-
tributes (Saura & Pascual- Hortal, 2007; Visconti & Elkin, 2009). In this
study in addition to typically used patch area, we tested, as a novel
FIGURE2 (a) Ensemble distribution model of Asiatic cheetah based on weighted- averaging seven species distribution models (SDMs). (b) Stacked
binary prediction as an ensemble model based on overlaid suitable/unsuitable distribution models used to identify Asiatic cheetah’s core habitats.
[Colour figure can be viewed at wileyonlinelibrary.com]
High : 964
Low : 36
Cheetah conservation areas
Overlaid suitability score
AHMADI et Al.
procedure, the current values of cheetah’s core habitats that were de-
rived from circuit theory as the patch quality characteristic. Moreover,
the prioritization of core habitats (i.e., their contribution to overall
habitat connectivity) was calculated from the percentage of the varia-
tion in PC (dPC) caused by the removal of each individual patch from
the landscape (Saura & Pascual- Hortal, 2007), both for patch area
(dPC A) and patch current (dPC C). A description of the dPC calcula-
tions is provided in Appendix S1. We calculated dPC values at two
dispersal distances of 50 km and 150 km using conefor 2.6 software
(Saura & Torné, 2009). conefor needs distance- probability values cor-
responding to dispersal ability of the targeted species. Although there
is little information on Asiatic cheetah movements, we chose 50 km
as a reasonable median dispersal distance and 150 km as a maximum
dispersal distance estimated based on camera trap recapture data
(Farhadinia et al., 2013). Accordingly, we set distance- probability val-
ues of 0.5 and 0.05 for 50 km and 150 km dispersal distances, respec-
tively, as recommended by Saura and Torné (2009).
3 | RESULTS
Our ensemble model indicated a patchily distributed suitable habi-
tat for Asiatic cheetah in the central plateau of Iran (Figure 2). GLM,
MaxEnt, GBM and RF distribution models showed excellent predictive
performance with respect to AUC metric (i.e., model’s discrimination
capacity) and GAM, MARS and ANN good performance (Table 2). The
prediction accuracy was good (e.g., TSS ˃ 0.6) for all models (Table 2).
The average importance of the variables among the models
showed that the prey availability, human footprint, roughness, village
density and low canopy rangeland variables contributed the most to
the cheetah distribution (Table 3). The response curves produced to
evaluate cheetah’s response to environmental gradient revealed an al-
most similar pattern between GLM, GBM and RF (Figure 3), all indicat-
ing that the highest probability of cheetahs’ presence occurs in areas
with highest prey availability. The effect of anthropogenic variables
(i.e., human footprint and village density) indicated that with increasing
human presence cheetahs’ occurrence decreased. Finally taking land-
scape attributes into account, response curves also depicted a high
preference of the Asiatic cheetah for rough landscapes covered by
sparse vegetation with avoidance of bare lands (Figure 3).
Overlaying presence/absence distribution maps to obtain an in-
tegrated suitability map of all models indicated that 40.5% of the
study area was identified as suitable habitat by at least one of the
distribution models (i.e., areas with suitability score of 1–7; Table 4).
Accordingly, 59.5% of the study area was not identified as suitable by
any of the seven distribution models (Table 4). We identified five core
habitats that represented areas of highest environmental suitability
for the species in Iran (Figure 4). We also included an additional core
habitat in North Khorasan Province (see Figure 1). Although this patch
of habitat has an area smaller than the estimated mean home- range
size for Asiatic cheetah, it hosts a documented population of cheetahs,
possibly as a result of an unusually abundant population of gazelles
(Farhadinia et al., 2012). Accordingly, we estimated that the current
core habitats for cheetahs in Iran stretch over 49,144.5 km2 or approx-
imately 6.3% of the central plateau.
Connectivity analysis using a circuit theory based approach re-
vealed that while there is strong permeability within southern and
northern populations, the connectivity between these two distribu-
tion patches is limited (Figure 4), as a result of low landscape suitabil-
ity between them (Figure 2). Moreover, our cumulative current map
highlighted the existence of several pinch points across the predicted
linkages, which, as landscape bottlenecks, may contribute to constrain
Landscape connectivity prioritization analysis showed a differ-
ent pattern of patch prioritization depending on whether using the
extent of core habitats or circuit current as patch characteristics.
Based on the extent of core habitats, core habitats 5, 6 and 4 were
the most important patches for sustaining connectivity, respectively
(Table 5). However, with respect to circuit current as patch charac-
teristics, core habitats ranking was 6, 4 and 5, respectively (Table 5).
Nonetheless, we found that for both patch area and patch circuit
current the connectivity ranking of core habitats correlated posi-
tively with their characteristics, and was independent of the disper-
GLM GAM MARS MaxEnt GBM RF ANN
AUC 0.901 0.837 0.872 0.905 0.913 0.902 0.876
TSS 0.730 0.655 0.652 0.707 0.713 0.727 0.670
AUC, area under the curve of ROC plot; TSS, true statistical skill.
For models description, see “Methods”.
TABLE2 Performance of discrimination
capacity and accuracy of different
algorithms to predict Asiatic cheetah
distribution in central Iran
TABLE3 Mean and standard deviation (SD) of the contribution of
environmental variables in seven Asiatic cheetah’s distribution
models in central Iran. Contribution values were calculated based on
the difference in Pearson correlation scores between general model
and randomized (e.g., permuted) models for each variable
Variables Mean SD
Prey availability 0.521 0.091
Human footprint 0.223 0.056
Roughness 0.169 0.102
Villages density 0.110 0.056
Low canopy rangeland 0.083 0.030
Bareland 0.056 0.048
Altitude 0.033 0.038
Cropland 0.031 0.026
Shrubland 0.016 0.024
Mod canopy rangeland 0.013 0.015
AHMADI et Al.
4 | DISCUSSION
Setting priority actions for species conservation should be primarily
conducted based on reliable, detailed and spatially explicit understand-
ing of the species requirements, and available conservation options
(Rabinowitz & Zeller, 2010; Sanderson, Redford, et al., 2002). In the
present study, we propose the first combined evaluation of habitat
suitability and connectivity for the Asiatic cheetah, an apex predator
that suffered considerable contraction of its distribution range for at
least the past 100 years (Harrison & Bates, 1991). We used the most
complete compilation of recent locations collected for this cheetah
subspecies and a sophisticated ensemble modelling approach that ac-
counts for SDMs- specific uncertainty (Araújo & New, 2007; Thuiller
et al., 2009). The stacked binary prediction as the ensemble model
achieved 100% sensitivity, which means that all cheetah presence lo-
cations were correctly predicted as suitable by the overlaid suitabil-
ity maps. Although such an approach is likely to decrease the model’s
specificity (i.e., correctly predicted pseudo- absence), here estimated at
65%, it minimized the omission of potentially suitable areas for chee-
tahs. Decreasing the omission error compared to the commission error
seems an acceptable choice in the case of a species on the brink of
extinction and requiring generous and immediate conservation efforts.
The comparison of SDMs results revealed that GLM has the highest
TSS value (Table 2), confirming the efficiency of this simple regression-
based model to correctly classify out- of- sample data when doing ex-
trapolations (Franklin, 2010; Merow et al., 2014). However, machine
learning methods (MaxEnt, GBM and RF), that have good propensity
for interpolation, showed the best performance based on discrimina-
tion capacity (i.e., AUC) values (Table 2), a result supported by compar-
ative examinations of SDMs (Elith & Graham, 2009; Elith et al., 2006).
The present study confirms that availability of prey species is a fun-
damental criterion of landscape suitability for cheetah in Iran. Habitat se-
lection for areas with high prey abundance has been largely reported for
other large felids including the African cheetah (A. j. jubatus) in South Africa
(Rostro- García, Kamler, & Hunter, 2015). Relative availability of natural
FIGURE3 Response curves of Asiatic cheetah’s distribution to the gradient of the most important predictors for habitat suitability. Results
shown are for GLM (blue line), GBM (red line) and RF (black line) models. For description of variables, see Table 1. Human footprint (ratio of 100),
low canopy rangeland and bareland (for both ratios of 1) are adimensional variables. [Colour figure can be viewed at wileyonlinelibrary.com]
TABLE4 Surface and proportion of
suitable/unsuitable habitats for the Asiatic
cheetah in central Iran
score Area (km2)
of study area
Protected by all
CAs in km2 (%)
Protected by cheetah’s
CAsa in km2 (%)
0 464,466 59.5
1–7 316175.7 40.5 73,818.86 (23) 51,512.65 (16)
7 49,144.5 6.3 30,759.16 (63) 26,130.45 (53)
aConservation areas (CAs) for cheetahs are protected areas with specific management aiming at con-
AHMADI et Al.
prey versus livestock has also been shown a good predicator of landscape
suitability for cheetahs in Botswana (Winterbach, Winterbach, Boast,
Klein, & Somers, 2015). The Asiatic cheetah mainly relies on medium-
sized ungulate prey, with a preference for the Jebeer and goitered gazelles
(Farhadinia & Hemami, 2010; Farhadinia et al., 2012), two species that
have suffered severe declines in population size and distribution in Iran
since the 1970s (Mallon, 2007). Therefore, the high dependency on prey
identified by our distribution model may provide support for the hypothe-
sis that the Asiatic cheetah is likely to be ecologically constrained in its last
stronghold in Asia due to the decline of its favoured prey species.
Cheetahs in Iran are scattered predominantly through low can-
opy rangelands (e.g., sparse vegetation with density <25%; Table 3) in
relatively rough terrains. The high contribution of topographic rough-
ness in the current distribution of the Asiatic cheetah is in contrast
with what has been documented for sub- Saharan African cheetahs,
which live in flat to undulating grasslands, savannas and shrublands
and only occasionally in montane areas (e.g., see Andresen et al., 2014;
Rostro- García et al., 2015). The widespread use by Asiatic cheetahs
of the rugged parts of the predominantly flat central plateau of Iran
is coherent with the relatively low habitat selectivity of cheetahs
compared to other carnivores (Durant et al., 2010), and may reflect
in Iran a shift in prey selection. Because cheetahs prefer prey within a
body mass range of 23–56 kg (Hayward, Hofmeyr, O’brien, & Kerley,
2006), thus with the decline of Jebeer and goitered gazelles, the wild
sheep and Persian ibex, two mid- sized species inhabiting rough areas
(Esfandabad, Karami, Hemami, Riazi, & Sadough, 2010), have emerged
as the most available wild prey species for cheetahs in Iran (Farhadinia
& Hemami, 2010). This possibly resulted in cheetahs increasingly using
rough habitats. The hypothesis of a contemporary shift in prey selec-
tion is supported by the historical distribution of cheetahs in south-
west Asia, which extended largely in accordance with the presence
of plain- dwelling gazelles (Harrington, 1977; Harrison & Bates, 1991).
Unsurprisingly the Asiatic cheetah prefers areas without humans
and associated activities, supporting the documented trend that conflict
FIGURE4 Habitat permeability
between the six Asiatic cheetah’s core
habitats based on circuit theory. Stepping
stone areas are proposed temporary
strongholds between core areas for moving
cheetahs. Hatched conservation areas (CAs)
are those with specific management aiming
at conserving cheetah. [Colour figure can
be viewed at wileyonlinelibrary.com]
High : 2.355
Low : 00150 30075 km
TABLE5 Patch characteristics and results of landscape
connectivity prioritization (LCP) to identify the most important core
habitats for maintaining landscape connectivity within Asiatic
cheetah distribution range. Patches (or core areas) are identified by
species distribution modelling using BIOMOD method, and the
current is derived from cIrcuItscApe software. The values of 50 km
and 150 km correspond to assumed median and maximum dispersal
distance of the species. dPC A and dPC C are the percentage of PC
index value loss for patch area and patch current, respectively. The
geographical location of patches is shown in Figure 4
50 km 150 km
dPC A dPC C dPC A dPC C
Patch 1 8093.58 2.25 9.65 20.83 7.98 17.95
Patch 2 766.084 1.62 0.57 11.82 0.38 10.63
Patch 3 4560.55 1.43 5.52 9.97 3.73 7.69
Patch 4 7278.62 2.27 25.56 36.21 21.30 34.26
Patch 5 17,491.84 2.07 65.89 35.01 66.17 32.36
Patch 6 10,953.92 2.35 45.26 39.79 45.03 39.81
AHMADI et Al.
with humans is a primary factor decreasing large carnivore survival
(Winterbach, Winterbach, Somers, & Hayward, 2013). Major demo-
graphic changes have occurred in Iran since the 1930s with a 6- fold
increase of the total human population and a doubling of the rural pop-
ulation over the period (Amiraslani & Dragovich, 2011). Even so the last
remaining stronghold for cheetahs in the central plateau, hosts the low-
est human population densities in Iran (NPHC, 2011), these demographic
shifts together with developmental changes have created a situation in
which more anthropogenic pressure is being exerted on the cheetah
habitat, particularly in relation to an increase in infrastructure and live-
stock numbers. Livestock overgrazing and desertification have resulted
in an intensification of food resource degradation for the main cheetah
prey species (Hunter et al., 2007; Karami, Hemami, & Groves, 2002),
while guard dogs accompanying livestock herds have proved dangerous
predators for cheetahs and their prey (CACP database, unpublished).
Landscapes that retain more connections between patches of
otherwise isolated habitat are assumed to be more likely to maintain
dispersal pathways for large mammals and increase demographic and
genetic population size (Mills & Allendorf, 1996). Although one and
half time larger than the Serengeti/Mara/Tsavo landscape in Kenya and
Tanzania, which hosts the largest population of cheetahs in Africa, the
area of potential core habitats remaining available to cheetahs in Iran
appears fragmented with possibly a limited connectivity between the
northern and southern populations. Although cheetahs display a high
mobility and excellent dispersal abilities (Boast, 2014), better conserved
connecting habitats would help the species to persist, recolonize empty
habitat patches and exchange individuals and genes among subpopula-
tions (Hanski & Ovaskainen, 2000; Mech & Hallett, 2001). Our connec-
tivity model predicts that linkages between core areas exist, although
their functionality for dispersal might be to some extent affected by a
lack of protection and the risk of road- kill accidents (Figure 4).
The analysis of patch prioritization revealed that for both patch area
(i.e., extent of the core habitats) and patch current (i.e., circuit flows
through core habitats), patch prioritization was positively correlated
with the patch characteristic regardless of the dispersal ability of the
species. We also found that core habitats will have a lower chance of
being reached by a cheetah when dispersal distances become larger.
These results have been documented in other studies (Saura & Rubio,
2010; Zhao et al., 2014). However, our finding highlights that consider-
ing different patch characteristics might results in different pattern of
patch prioritization. For example, while core habitats 5 and 6 have high-
est dPC value based on patch area characteristic, core habitats 6 and 4
are the most important patch when using patch current. Although spa-
tial aggregation of patches and the variance of the patch characteristics
determines the appropriateness of the application of metrics for LCP
analysis (Visconti & Elkin, 2009), our approach reveals that, using the
same metrics and spatial aggregation, including patch characteristics
that are intrinsically more related to the movement of the species (e.g.,
patch circuit current vs. patch area), might provide better understand-
ings of the patch contribution for maintaining landscape connectivity.
Using circuit theory allows calculating cumulative current (i.e., land-
scape permeability) terminating to each of the patch habitats and could
potentially be used as habitat characteristic for landscape prioritization.
4.1 | Conservation implications
The present study supports that CAs with dedicated resources to chee-
tah conservation currently protect only 53% and 16% of cheetah core
and suitable habitats, respectively (Table 4). Because of its crucial impor-
tance for conservation, the network of CAs and associated conservation
resources for cheetahs would therefore benefit from being expanded to
achieve more effective conservation coverage of cheetah habitats in Iran.
Currently, high priority core habitats 1, 2 and 5 are fairly well covered by
CAs, but the protection coverage would deserve being expanded to the
south- east of core area 3, the north of core area 4 and between existing
CAs in core area 6. In combination, this landscape protection scheme
will strengthen the role of core areas 4, 5 and 6 for cheetah dispersal as
revealed by the patch prioritization analysis. Also, to ensure the cohe-
sion between core habitats in the north and the south we propose to
establish new intermediate CAs located in corridors of suitable habitats
between core areas 1 and 3, and 3 and 4 (Figure 4). As featured for other
large carnivores (Cushman et al., 2012; Riordan et al., 2015) these areas,
as stepping stones, would provide temporary strongholds between core
areas for moving cheetahs and reconnect through suitable landscapes
the northern and southern parts of the cheetah range.
New CAs in the central plateau of Iran as well as their associated
management policies should be developed with the specific objective
of cheetah conservation. They should include large extents of habitats
favoured by gazelles and support in priority their conservation. In ad-
dition, the water and food supplementations actively implemented in
CAs during summers and droughts should be implemented to support
in priority gazelles in their favoured habitats. Hence, this management
policy would less likely support the mountain dwelling Persian leopard
which, besides being a predator to cheetahs (Hayward et al., 2006),
competes with them on prey resources.
Our connectivity approach showed that linkages within and between
adjacent cheetah core habitats still exist across the central plateau of Iran
(Figure 3). Unfortunately, these pathways are nowadays dissected by main
through roads that carry large volume of traffic, putting cheetahs at risk of
car collisions. Of 33 documented cheetah mortalities between 2001 and
2016 due to various causes, at least 14 were killed on roads within or be-
tween core areas, making it the major cause of documented mortality for
cheetahs in Iran (CACP database, unpublished). By providing an applicable
tool to identify with accuracy the most important portions of roads to be
secured against cheetah car collisions, the circuits approach together with
connectivity metrics could help prioritize mitigation measures, increase
their cost- effectiveness and likelihood of success. Our connectivity anal-
ysis supports that securing primary roads within core areas and between
the core areas 1 and 2, and 3 and 4 would be critical to reduce the risk
of cheetah car collisions. Fencing of dangerous stretches of roads partic-
ularly in combination with wildlife passages has indeed been suggested
as one of the most effective methods to minimize car- collision risk with
large mammals (Bissonette & Adair, 2008; Ascensão et al., 2013). Large
carnivores differently respond to crossing structures given to taxon- and/
or habitat- specific factors (Ng et al., 2004; Clevenger & Waltho, 2005).
To our knowledge, anticar- collision methods have not been developed
specifically for cheetahs, and therefore, methods and structures used for
AHMADI et Al.
other large carnivores (e.g., Clevenger & Waltho, 2005) will have to be
tested for the Asiatic cheetah and adjusted to the Iranian context.
The value of identifying core habitats and linkage areas for Asiatic
cheetah at country scale is to inform targeted land planning and better
conservation management. Species distribution modelling in combina-
tion with circuit theory and LCP analysis provide a robust representa-
tion of most suitable habitats for cheetahs in Iran and offers possible
adaptive measures for country- scale management of cheetah habitats.
Extending landscape protection over larger stretches of suitable hab-
itat, developing in parallel cheetah- specific conservation actions aim-
ing at increasing the size and distribution of gazelle populations, and
improving the safety of important linkages between core habitats are
likely to promote the conservation of the last surviving population of
cheetah in Asia. In the future, evaluating the long- term persistence of
the Asiatic cheetah over its last remaining strongholds in Asia would
also require studies on metapopulation dynamics based on patch pop-
ulation growth models.
We are grateful for the financial and technical support of the DoE of the
Islamic Republic of Iran, the Global Environmental Facilities (GEF), the
United Nations Development Program (UNDP) representation in Iran,
and the Wildlife Conservation Society (WCS). The work of WCS in Iran
would not have been possible without the long- standing support of the
Flora Family Foundation. Our special thanks go to all contributors to
the Conservation of the Asiatic Cheetah Project, and particularly suc-
cessive managers, to directors of the DoE operations in the provinces
covered by this study, to managers of CAs and their staff who were the
kingpins of all data collection efforts. We thank anonymous referees
for helping to improve an earlier version of the manuscript.
M.A., B.N.B. and S.O. conceived the study; B.N.B., H.J. and S.O. com-
piled the data; M.A., B.N.B., M.R.H., D.F. and S.M. analysed the data;
and M.A. and S.O. led the writing.
Allouche, O., Tsoar, A., & Kadmon, R. (2006). Assessing the accuracy of spe-
cies distribution models: Prevalence, kappa and the true skill statistic
(TSS). Journal of Applied Ecology, 43, 1223–1232.
Almasieh, K., Kaboli, M., & Beier, P. (2016). Identifying habitat cores and
corridors for the Iranian black bear in Iran. Ursus, 27, 18–30.
Amiraslani, F., & Dragovich, D. (2011). Combating desertification in Iran
over the last 50 years: An overview of changing approaches. Journal of
Environmental Management, 92, 1–13.
Andresen, L., Everatt, K. T., & Somers, M. (2014). Use of site occupancy models
for targeted monitoring of the cheetah. Journal of Zoology, 292, 212–220.
Araújo, M. B., & New, M. (2007). Ensemble forecasting of species distribu-
tions. Trends in Ecology & Evolution, 22, 42–47.
Ascensão, F., Clevenger, A., Santos-Reis, M., Urbano, P., & Jackson, N.
(2013). Wildlife–vehicle collision mitigation: Is partial fencing the
answer? An agent- based model approach. Ecological Modelling, 257,
Belbachir, F., Pettorelli, N., Wacher, T., Belbachir-Bazi, A., & Durant, S. M.
(2015). Monitoring rarity: The critically endangered Saharan cheetah as
a flagship species for a threatened ecosystem. PLoS One, 10, e0115136.
Bissonette, J. A., & Adair, W. (2008). Restoring habitat permeability
to roaded landscapes with isometrically- scaled wildlife crossings.
Biological Conservation, 141, 482–488.
Boast, L. (2014). Exploring the causes of and mitigation options for human-pred-
ator conflict on game ranches in Botswana: How is coexistence possible? D.
Phil thesis, University of Cape Town, Cape Town, South Africa.
Bodin, Ö., & Saura, S. (2010). Ranking individual habitat patches as connec-
tivity providers: Integrating network analysis and patch removal exper-
iments. Ecological Modelling, 221, 2393–2405.
Brito, J. C., Acosta, A. L., Álvares, F., & Cuzin, F. (2009). Biogeography and
conservation of taxa from remote regions: An application of ecological-
niche based models and GIS to North- African Canids. Biological
Conservation, 142, 3020–3029.
Burton, A. C., Sam, M. K., Balangtaa, C., & Brashares, J. S. (2012). Hierarchical
multi- species modeling of carnivore responses to hunting, habitat and
prey in a West African protected area. PLoS One, 7, e38007.
Carroll, C., McRae, B., & Brookes, A. (2012). Use of linkage mapping and central-
ity analysis across habitat gradients to conserve connectivity of gray wolf
populations in western North America. Conservation Biology, 26, 78–87.
Chapron, G., Kaczensky, P., Linnell, J. D. C., von Arx, M., Huber, D., Andrén,
H., … Boitani, L. (2014). Recovery of large carnivores in Europe’s mod-
ern human- dominated landscapes. Science, 346, 1517–1519.
Clevenger, A. P., & Waltho, N. (2005). Performance indices to identify at-
tributes of highway crossing structures facilitating movement of large
mammals. Biological conservation, 121, 453–464.
Crooks, K. R., Burdett, C. L., Theobald, D. M., Rondinini, C., & Boitani, L.
(2011). Global patterns of fragmentation and connectivity of mamma-
lian carnivore habitat. Philosophical Transactions of the Royal Society of
London B: Biological Sciences, 366, 2642–2651.
Cushman, S. A., Landguth, E. L., & Flather, C. H. (2012). Evaluating the suf-
ficiency of protected lands for maintaining wildlife population connec-
tivity in the US northern Rocky Mountains. Diversity and Distributions,
Dickson, B. G., Roemer, G. W., McRae, B. H., & Rundall, J. M. (2013). Models
of regional habitat quality and connectivity for pumas (Puma concolor)
in the southwestern United States. PLoS One, 8, e81898.
Dormann, C. F., McPherson, J. M., Araujo, M. B., Bivand, R., Bolliger, J., Carl, G.,
… Wilson, R. (2007). Methods to account for spatial autocorrelation in the
analysis of species distributional data: A review. Ecography, 30, 609–628.
Durant, S. M., Craft, M. E., Foley, C., Hampson, K., Lobora, A. L., Msuha, M.,
… Pettorelli, N. (2010). Does size matter? An investigation of habitat
use across a carnivore assemblage in the Serengeti, Tanzania. Journal of
Animal Ecology, 79, 1012–1022.
Elith, J., & Graham, C. H. (2009). Do they? How do they? Why do they
differ? on finding reasons for differing performances of species distri-
bution models. Ecography, 32, 66–77.
Elith, J., Graham, H. C., Anderson, P. R., Dudík, M., Ferrier, S., Guisan, A.,
… Zimmermann, E. N. (2006). Novel methods improve prediction of
species’ distributions from occurrence data. Ecography, 29, 129–151.
Esfandabad, B. S., Karami, M., Hemami, M. R., Riazi, B., & Sadough, M. B.
(2010). Habitat associations of wild goat in central Iran: Implications for
conservation. European Journal of Wildlife Research, 56, 883–894.
ESRI (2010). ArcGIS 9.3. Redlands, CA: Environmental Systems Research
Farhadinia, M. S., Ahmadi, M., Sharbafi, E., Khosravi, S., Alinezhad, H., &
Macdonald, D. W. (2015). Leveraging trans- boundary conservation
partnerships: Persistence of Persian leopard (Panthera pardus saxicolor)
in the Iranian Caucasus. Biological Conservation, 191, 770–778.
Farhadinia, M. S., Akbari, H., Mousavi, S. J., Eslami, M., Azizi, M., Shokouhi,
J., … Hosseini-Zavarei, F. (2013). Exceptionally long movements of the
Asiatic cheetah Acinonyx jubatus venaticus across multiple arid reserves
in central Iran. Oryx, 47, 427–430.
AHMADI et Al.
Farhadinia, M., & Hemami, M.-R. (2010). Prey selection by the critically en-
dangered Asiatic cheetah in central Iran. Journal of Natural History, 44,
Farhadinia, M., Hosseini-Zavarei, F., Nezami, B., Harati, H., Absalan, H., Fabiano,
E., & Marker, L. (2012). Feeding ecology of the Asiatic cheetah Acinonyx
jubatus venaticus in low prey habitats in northeastern Iran: Implications for
effective conservation. Journal of Arid Environments, 87, 206–211.
Ford, A. T., Goheen, J. R., Otieno, T. O., Bidner, L., Isbell, L. A., … Pringle,
R. M. (2014). Large carnivores make savanna tree communities less
thorny. Science, 346, 346–349.
Franklin, J. (2010). Mapping species distributions: Spatial inference and predic-
tion. New York: Cambridge University Press.
Hanski, I., & Ovaskainen, O. (2000). The metapopulation capacity of a frag-
mented landscape. Nature, 404, 755–758.
Harrington, F. A. (1977). A guide to the mammals of Iran. Tehran, Iran:
Department of the Environment.
Harrison, D. L., & Bates, P. J. J. (1991). The mammals of Arabia. Kent,
Engeland: Harrison Zoological Museum Sevenoaks.
Hayward, M., Hofmeyr, M., O’brien, J., & Kerley, G. (2006). Prey preferences
of the cheetah (Acinonyx jubatus) (Felidae: Carnivora): Morphological
limitations or the need to capture rapidly consumable prey before klep-
toparasites arrive? Journal of Zoology, 270, 615–627.
Hunter, L., Jowkar, H., Ziaie, H., Schaller, G., Balme, G., Walzer, C., …
Christie, S. (2007). Conserving the Asiatic cheetah in Iran: Launching
the first radio- telemetry study. Cat News, 46, e11.
Karami, M., Ghadirian, T., & Feizollahi, K. (2013). Atlas of the mammals of
Iran. Tehran, Iran: Wildlife Center Publication.
Karami, M., Hemami, M. R., & Groves, C. P. (2002). Taxonomic, distributional
and ecological data on gazelles in Iran. Zoology in the Middle East, 26, 29–36.
Kramer-Schadt, S., Niedballa, J., Pilgrim, J. D., Schröder, B., Lindenborn, J.,
Reinfelder, V., … Wilting, A. (2013). The importance of correcting for
sampling bias in MaxEnt species distribution models. Diversity and
Distributions, 19, 1366–1379.
Kunkel, K. E., Atwood, T., Ruth, T., Pletscher, D. H., & Hornocker, M. (2013).
Assessing wolves and cougars as conservation surrogates. Animal
Conservation, 16, 32–40.
Makhdoum, M. (2008). Management of protected areas and conservation
of biodiversity in Iran. International Journal of Environmental Studies, 65,
Mallon, D. P. (2007). Cheetahs in Central Asia: A historical summary. Cat
News, 46, 4–7.
Manafzadeh, S., Salvo, G., & Conti, E. (2014). A tale of migrations from east
to west: The Irano- Turanian floristic region as a source of Mediterranean
xerophytes. Journal of Biogeography, 41, 366–379.
McRae, B. H., & Beier, P. (2007). Circuit theory predicts gene flow in plant
and animal populations. Proceedings of the National Academy of Sciences
of the United States of America, 104, 19885–19890.
McRae, B. H., Dickson, B. G., Keitt, T. H., & Shah, V. B. (2008). Using circuit
theory to model connectivity in ecology, evolution, and conservation.
Ecology, 89, 2712–2724.
McRae, B., & Shah, V. (2009). Circuitscape user’s guide. Santa Barbara: The
University of California.
Mech, S. G., & Hallett, J. G. (2001). Evaluating the effectiveness of corri-
dors: a genetic approach. Conservation Biology, 15, 467–474.
Merow, C., Smith, M. J., Edwards, T. C., Guisan, A., McMahon, S. M.,
Normand, S., … Elith, J. (2014). What do we gain from simplicity versus
complexity in species distribution models? Ecography, 37, 1267–1281.
Mills, L. S., & Allendorf, F. W. (1996). The one- migrant- per- generation rule in
conservation and management. Conservation Biology, 10, 1509–1518.
Moqanaki, E., & Cushman, S. (2016). All roads lead to Iran: Predicting land-
scape connectivity of the last stronghold for the critically endangered
Asiatic cheetah. Animal Conservation, 20, 29–41.
Ng, S. J., Dole, J. W., Sauvajot, R. M., Riley, S. P., & Valone, T. J. (2004). Use
of highway undercrossings by wildlife in southern California. Biological
Conservation, 115, 499–507.
NPHC (2011). National population and housing census report. Tehran, Iran:
Statistical Centre of Iran Press.
Pettorelli, N., Hilborn, A., Broekhuis, F., & Durant, S. (2009). Exploring hab-
itat use by cheetahs using ecological niche factor analysis. Journal of
Zoology, 277, 141–148.
R Development Core Team(2014). R: A language and environment for statisti-
cal computing. Vienna, Austria: R Foundation for Statistical Computing.
Rabinowitz, A., & Zeller, K. A. (2010). A range- wide model of landscape
connectivity and conservation for the jaguar, Panthera onca. Biological
Conservation, 143, 939–945.
Riordan, P., Cushman, S. A., Mallon, D., Shi, K., & Hughes, J. (2015).
Predicting global population connectivity and targeting conservation
action for snow leopard across its range. Ecography, 39, 419–426.
Ripple, W. J., Estes, J. A., Beschta, R. L., Wilmers, C. C., Ritchie, E. G.,
Hebblewhite, M., … Wirsing, A. J. (2014). Status and ecological effects
of the world’s largest carnivores. Science, 343, 1241484.
Rostro-García, S., Kamler, J. F., & Hunter, L. T. (2015). To kill, stay or flee:
The effects of lions and landscape factors on habitat and kill site selec-
tion of cheetahs in South Africa. PLoS One, 10, e0117743.
Sanderson, E. W., Jaiteh, M., Levy, M. A., Redford, K. H., Wannebo, A. V.,
& Woolmer, G. (2002). The human footprint and the last of the wild:
The human footprint is a global map of human influence on the land
surface, which suggests that human beings are stewards of nature,
whether we like it or not. BioScience, 52, 891–904.
Sanderson, E. W., Redford, K. H., Chetkiewicz, C. L. B., Medellin, R. A.,
Rabinowitz, A. R., Robinson, J. G., & Taber, A. B. (2002). Planning
to save a species: The jaguar as a model. Conservation Biology, 16,
Santini, L., Boitani, L., Maiorano, L., & Rondinini, C. (2016). Effectiveness
of protected areas in conserving large carnivores in Europe. In L. N.
Joppa, J. E. M. Baillie, & J. G. Robinson (Eds.), Protected areas: Are they
safeguarding biodiversity? (pp. 122–133). Oxford: Blackwell Scientific
Santini, L., Saura, S., & Rondinini, C. (2016). Connectivity of the global net-
work of protected areas. Diversity and Distributions, 22, 199–211.
Saura, S., & Pascual-Hortal, L. (2007). A new habitat availability index to
integrate connectivity in landscape conservation planning: Comparison
with existing indices and application to a case study. Landscape and
Urban Planning, 83, 91–103.
Saura, S., & Rubio, L. (2010). A common currency for the different ways in
which patches and links can contribute to habitat availability and con-
nectivity in the landscape. Ecography, 33, 523–537.
Saura, S., & Torné, J. (2009). Conefor Sensinode 2.2: A software package for
quantifying the importance of habitat patches for landscape connectiv-
ity. Environmental Modelling & Software, 24, 135–139.
Sergio, F., Caro, T., Brown, D., Clucas, B., Hunter, J., Ketchum, J., … Hiraldo,
F. (2008). Top predators as conservation tools: Ecological rationale,
assumptions, and efficacy. Annual Review of Ecology, Evolution, and
Systematics, 39, 1–19.
Thuiller, W., Lafourcade, B., Engler, R., & Araújo, M. B. (2009). BIOMOD—A
platform for ensemble forecasting of species distributions. Ecography,
Visconti, P., & Elkin, C. (2009). Using connectivity metrics in conservation
planning—When does habitat quality matter? Diversity and Distributions,
Walpole, A. A., Bowman, J., Murray, D. L., & Wilson, P. J. (2012). Functional
connectivity of lynx at their southern range periphery in Ontario,
Canada. Landscape Ecology, 27, 761–773.
Winterbach, H. E., Winterbach, C. W., Boast, L. K., Klein, R., & Somers, M. J.
(2015). Relative availability of natural prey versus livestock predicts land-
scape suitability for cheetahs Acinonyx jubatus in Botswana. Pe erJ , 3,
Winterbach, H., Winterbach, C., Somers, M., & Hayward, M. (2013). Key
factors and related principles in the conservation of large African carni-
vores. Mammal Review, 43, 89–110.
AHMADI et Al.
Zhao, H., Liu, S., Dong, S., Su, X., Liu, Q., & Deng, L. (2014). Characterizing
the importance of habitat patches in maintaining landscape connec-
tivity for Tibetan antelope in the Altun Mountain National Nature
Reserve, China. Ecological Research, 29, 1065–1075.
Mohsen Ahmadi is a research associate at the Department of Natural
Resources, Isfahan University of Technology, Iran. His primary inter-
est is in methods for evaluating spatial and temporal dynamics of bio-
diversity and species distribution models, with a special interest in the
incorporation of new knowledge into conservation goals.
Additional Supporting Information may be found online in the
supporting information tab for this article.
How to cite this article: Ahmadi M, Balouchi BN,
Jowkar H, et al. Combining landscape suitability and habitat
connectivity to conserve the last surviving population of
cheetah in Asia. Divers. Distrib. 2017;23:592–603.