Occupancy patterns of prey species in a biological corridor and inferences for tiger population connectivity between national parks in Bhutan
L ETRO L ETRO,KLAUS F ISCHER,DORJI D UBA and T ANDIN T ANDIN
Abstract Site occupancy models, accounting for imperfect Supplementary material for this article is available at detection and the influence of anthropogenic and ecological doi.org/./S covariates, can indicate the status of species populations. They may thus be useful for exploring the suitability of land- scapes such as biological corridors, to ensure population Introduction dispersal and connectivity. Using occupancy probability arge ungulates influence forest structure and function- models of its principal prey species, we make inferences ing (Gopalaswamy et al., ), and the density and spa- on landscape connectivity for the movement of the tiger L tial distribution of apex predators such as the tiger Panthera Panthera tigris between protected areas in Bhutan. We tigris. The occurrence of tigers is primarily determined used camera-trap data to assess the probability of site occu- by the abundance of ungulate prey (Karanth et al., ; pancy (Ψ) of the sambar Rusa unicolor, wild boar Sus scrofa Harihar & Pandav, ), and by levels of human distur- and barking deer Muntiacus muntjak in biological corridor bance (Linkie et al., ). In the Himalayan region, ungu- no. , which connects two national parks in central Bhutan. lates such as the sambar Rusa unicolor, wild boar Sus scrofa, At least one prey species was recorded at out of trap- barking deer Muntiacus muntjak, chital Axis axis and gaur ping locations. The probability of site occupancy was high- Bos gaurus are the principal prey of tigers, with some varia- est for the barking deer (Ψ = . ± SE .) followed by tions between localities and habitat types (Karanth et al., sambar (Ψ = . ± SE .) and wild boar (Ψ = . ± SE ; Harihar & Pandav, ; Hayward et al., ; .). All three species had higher occupancy probability Tempa, ). These prey species constitute an important at lower altitudes. Sambar occupancy was greater farther structural component of tiger habitats and movement corri- from settlements and on steeper and/or south-facing slopes. dors, and their occupancy is influenced by environmental Barking deer also had higher occupancy on south-facing parameters such as elevation, forest coverage, and distance slopes, and wild boar occurred mainly close to rivers. Our from water, and by levels of human interference (Tempa, findings suggest that this biological corridor could facilitate ). dispersal of tigers. Protecting prey species, and minimizing Bhutan is a hotspot of wild felid diversity (Tempa et al., anthropogenic disturbance and habitat fragmentation, are ; Dhendup & Dorji, ), with the tiger being of vital for tiger dispersal and thus functional connectivity particular conservation concern (DoFPS, a). Tigers are amongst populations in this area. widely distributed in Bhutan from subtropical to alpine re- Keywords Bhutan, biological corridor, dispersal, landscape gions (Tempa, ). The first nationwide tiger survey, using connectivity, occupancy modelling, Panthera tigris, ungu- camera traps, was completed in , with the number of lates tigers estimated to be , at a density of . tigers per km (DoFPS, a). Tiger density was high (– tigers per km ) in protected areas in central Bhutan such as the Royal Manas and Jigme Singye Wangchuck National Parks (Tempa, ). In Bhutan, tigers successfully breed
LETRO LETRO* (Corresponding author, orcid.org/0000-0003-2042-4030) and at altitudes from near sea level to . , m (Jigme & KLAUS FISCHER† ( orcid.org/0000-0002-2871-246X) Zoological Institute and Tharchen, ). Provided that corridors between tiger ha- Museum, University of Greifswald, Loitzer Straße 26, D-17489 Greifswald, Germany. E-mail [email protected] bitats ensure structural and functional connectivity, Bhutan could make a significant contribution towards global tiger DORJI DUBA Department of Forests and Park Services, Bhutan Tiger Centre, Gelephu, Bhutan conservation goals such as doubling global tiger numbers TANDIN TANDIN Department of Forests & Park Services, Nature Conservation by (compared to ; GTRP, ). The functional Division, Taba, Thimphu, Bhutan connectivity of a landscape is determined by its structure *Current address: Department of Forests & Park Services, Nature Conservation and composition (Rudnick et al., ), and the term struc- Division, Taba, Thimphu, Bhutan tural connectivity describes the physical characteristics of †Current address: Institute for Integrated Natural Sciences, University of Koblenz-Landau, Koblenz, Germany a landscape that facilitate or hamper wildlife movement, in- Received April . Revision requested June . cluding topography, hydrology, vegetative cover and human Accepted September . land-use patterns (Metzger & Décamps, ).
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, Downloadeddistribution, from https://www.cambridge.org/core and reproduction in any medium,. IP address: provided 170.106.202.226 the original work, ison properly 30 Sep 2021 cited. at 23:32:35, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/termsOryx, Page 1 of 8 © The Author(s),. https://doi.org/10.1017/S0030605320000976 2021. Published by Cambridge University Press on behalf of Fauna & Flora International doi:10.1017/S0030605320000976 2 L. Letro et al.
Bhutan has protected areas, including five national parks, four wildlife sanctuaries and one strict nature reserve (DoFPS, b). Eight biological corridors, numbered –, were established in , to connect protected areas and fa- cilitate wildlife movement and dispersal (Sherpa & Norbu, ; Wangchuk, ). The biological corridors were de- signated based on vegetation cover derived from Landsat images, land-use maps and evidence of tiger occurrence. Evidence suggests corridors facilitate dispersal of smaller mammals and birds elsewhere (Littlewood et al., ), but the biological corridors in Bhutan were not validated with fieldwork, mainly because of a lack of trained person- nel and resources (Thinley, ). To assess the functional connectivity of these biological corridors for the movement of large carnivores such as the tiger, it is important to docu- ment the occurrence of prey species, as a proxy for structural connectivity. Well managed biological corridors are likely to enhance dispersal of tigers and other carnivores, thus pro- viding connectivity between habitats (Harihar & Pandav, ; Letro & Duba, ). Species distribution models can be used to assess the im- pacts of ecological and anthropogenic covariates on the dis- tribution and habitat use of species of conservation concern (Elith & Leathwick, ). Environmental variables affect species distributions and can thus be used to predict distri- butions across landscapes. In addition, environmental vari- ables such as habitat type or proximity to water influence FIG. 1 (a) Biological corridor no. in central Bhutan, connecting fine-scale habitat use (Harihar & Pandav, ), and can Jigme Singye Wangchuck National Park (JSWNP) and be used to identify essential conservation areas (Penjor Wangchuck Centennial National Park (WCNP) and et al., , ) and guide conservation actions at the (b) location of camera traps in biological corridor no. . local level (Sunarto et al., ; Srivathsa et al., ). Here, we use detection/non-detection data of prey spe- cies collected through camera trapping in Bhutan’s biologi- (Fig. ). The area is at an altitude of ,–, m, with a cal corridor no. for occupancy modelling (Mackenzie et al., warm-temperate climate, mean annual temperature of ), to assess the probability of site occupancy of the ti- °C and a mean total annual rainfall of , mm ger’s principal prey species, accounting for imperfect detec- (NCHM, ). The landscape is undulating and mainly tion. Our objectives were to ascertain occupancy probability covered with various forest types, including broadleaved, and its determinants for three principal prey species (sam- fir, blue pine Pinus wallichiana and mixed conifer forests, bar, wild boar and barking deer), and to develop predictive and alpine scrub. Past surveys documented the presence distribution maps for the study area based on spatially expli- of apex predators such as the tiger, the common leopard cit occupancy models for the three prey species, to assess Panthera pardus and the Asiatic black bear Ursus thibeta- landscape connectivity for tiger dispersal. No studies have nus, and several prey species including the sambar, wild previously been carried out on prey species distribution boar, Himalayan serow Capricornis thar, barking deer, and habitat use for any of the biological corridors in Himalayan goral Naemorhedus goral, musk deer Moschus Bhutan. Our study thus contributes to a better understand- chrysogaster and takin Budorcas taxicolor whitei (DoFPS, ing of the role that biological corridors play in enhancing a). Human land-use types in the study area comprise landscape connectivity in Bhutan. pastures and built-up areas (roads and permanent human settlements). Seasonal livestock grazing and collection of forest resources in and around the study area exert an- Study area thropogenic pressures on the biological corridor (Letro & Fischer, ). The eastern and western sections of the cor- We conducted this study in the parts of biological corridor ridor differ in the level of human interference, with more no. that connect Jigme Singye Wangchuck and Wang- human settlements and roads, and more grazing pressure chuck Centennial National Parks, covering c. km from domestic animals, in the western section.
Oryx, Page 2 of 8 © The Author(s), 2021. Published by Cambridge University Press on behalf of Fauna & Flora International doi:10.1017/S0030605320000976 Downloaded from https://www.cambridge.org/core. IP address: 170.106.202.226, on 30 Sep 2021 at 23:32:35, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0030605320000976 Occupancy of tiger prey species in Bhutan 3
TABLE 1 The characteristics of the three ungulates that are the principal prey species of the tiger Panthera tigris in biological corridor no. (Fig. ), adapted from McShea et al. ().
Sambar Wild boar Barking deer Attribute Rusa unicolor Sus scrofa Muntjacus muntjak Body mass (kg) 180–260 25–40 20–30 Home range (km2) 0.1–1.0 2.0–2.2 1.5–2.5 Group size 3–46–10 1–2 Birth season May–Aug. Unknown Nov.–May Diet Browse, ferns, grass, forbs Browse, graze, rooting, omnivore Browse, forbs Range in South Asia 4,388,000 km2 Unknown 4,157,000 km2 % of range protected 0.09 Unknown 0.09 IUCN Red List category Vulnerable Least Concern Least Concern
Methods and human settlement. We selected these variables based on prior studies of the tiger and its prey in this region Data collection (Kushwaha et al., ; Linkie et al., ; Kanagaraj et al., ; Harihar & Pandav, ; Tempa, ). To estimate the occupancy of tiger prey species, we con- Altitude, aspect, slope and distance to river characterize – ducted a camera-trap survey during March June .We physical properties of the environment, whereas land-use divided the study area into grid cells of . × . km, with type and distances to the nearest protected area, road and grid cell size chosen to account for the expected daily move- settlement describe human impact. We determined the ments of prey species, based on their known home range value of each site covariate as the mean of raster cells with- sizes (Table ) and movement rates (Odden & Wegge, in circles of m radius around each camera-trap sta- ; Gopalaswamy et al., ; Thapa & Kelly, ). tion, using the zonal statistics tool in QGIS .. (QGIS Using a sufficiently large grid cell size is important, to min- Development Team, ). We extracted altitude, slope, imize the risk of violating the assumption of closure when and aspect values from digital elevation raster data (USGS, estimating occupancy (Mackenzie et al., ). We set up ). We categorized land-use types as built-up area, farm- camera-trap stations in of the grid cells ( %), land, broadleaved forest, conifer forest, mixed conifer forest, and in the eastern and western sections, respectively, meadow, alpine scrub, and rocky outcrop. To determine – at , , m altitude (Fig. ). To increase the probability land-use types and locations of roads, protected areas and of photo-capturing prey species and tigers, we installed rivers, we used the Bhutan land-use and land-cover data camera traps along game trails leading to streams and salt (FRMD, ). We measured distances using the licks, and at trail junctions and ridge tops. We placed Euclidean distance tool in QGIS. one automated HCO-ScoutGuard (Professional Trapping Occupancy models using detection/non-detection data Supplies, Molendinar, Australia) camera-trap unit at each have two inferences: the probability of detection (P; the station, fastened on a wooden post cm above ground probability that a species appears during count statistics) – level, and at a distance of c. . . m from the centre of and the probability of occupancy (Ψ; the probability that – the trail. Camera traps were active during March the species occupies a random site in a given time period; – June , for days per location. We recorded the geo- MacKenzie et al., ). We generated detection histories graphical coordinates of the location of each camera-trap ( for detection, for non-detection) for each species at station. We defined independent images of a species as each camera trap over the respective trapping period, those that were taken at least minutes apart. using ReNamer . (Sanderson & Harris, ). For oc- cupancy modelling, we selected the optimal number of Data analysis trap-days (i.e. sampling period) for each prey species based on the χ² goodness-of-fit test of the multivariate Occupancy data may be affected by the detectability of model (MacKenzie & Bailey, ): days for sambar, target species. We examined the impact of trap effort (the for wild boar and for barking deer (Supplementary number of days the camera trap was active) and study site Table ). (eastern vs western section), as we expected these covariates For each prey species, we imported the detection history to influence detection probability of prey species. With re- over the optimal sampling period and the various site co- gard to occupancy probability, we tested the impact of vari- variates into PRESENCE .. (Hines, ), to run single- ous site covariates, including altitude, slope, aspect, land-use season single-species occupancy models (Mackenzie et al., type, and distances to the nearest river, protected area, road ). We z-standardized all continuous covariates to a
Oryx, Page 3 of 8 © The Author(s), 2021. Published by Cambridge University Press on behalf of Fauna & Flora International doi:10.1017/S0030605320000976 Downloaded from https://www.cambridge.org/core. IP address: 170.106.202.226, on 30 Sep 2021 at 23:32:35, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0030605320000976 4 L. Letro et al.
TABLE 2 Detection probability (P) models, with the study site (east- mean of zero and one standard deviation using the formula ern vs western section of the corridor) and trap effort (total num- z =(x − x’)/SD , where x is the individual value, x’ the mean ber of active camera-trap days) as detection covariates, and with of the factor, and SD the standard deviation, to improve the Akaike information criterion (AIC), difference between AIC Δ model convergence (Sunarto et al., ). We tested covari- and the best-performing model ( AIC), model weight and likeli- hood, number of parameters (K), and twice the negative log- ates for collinearity using Pearson correlations in R . . (R . likelihood (−LogLik) for each model. Occupancy probability (Ψ) Core Team, ), and for correlated covariates (|r| . ) was held constant. removed one of the two covariates, to avoid biased estimates (Dormann et al., ). There were high correlations be- Model Model tween land-use type and elevation, distance to protected Model AIC ΔAIC weight likelihood K −2LogLik area and distance to river, and distance to road and distance Sambar to settlement (Supplementary Table ). We retained the co- P(Site 76.58 0.00 0.33 1.00 4 68.58 variates that performed better in univariate occupancy +Effort) P(Effort) 77.17 0.59 0.24 0.74 2 73.17 models, based on a lower value of the Akaike information P(Site) 77.27 0.69 0.24 0.71 3 71.27 criterion (AIC), for multivariate modelling (Supplementary P(.) 77.42 0.84 0.21 0.66 2 73.42 Table ), and thus excluded land-use type, and distances Barking deer from protected area and road. P(Effort) 88.00 0.00 0.40 1.00 2 84.00 We used a two-step approach (Penjor et al., ) to es- P(.) 88.09 0.09 0.38 0.96 2 84.09 timate probability of detection (P) and probability of species P(Site) 89.91 1.91 0.15 0.38 3 83.91 occupancy (Ψ). Firstly, we modelled detection probability P(Site 91.85 3.85 0.06 0.15 4 83.85 +Effort) (P) for each principal prey species in relation to study site Wild boar (eastern vs western section) and trap effort. To identify P(Effort) 83.24 0.00 0.53 1.00 2 79.24 the covariates that significantly affected detection proba- P(Site) 84.97 1.73 0.22 0.42 3 78.97 bility, we used both univariate and multivariate detection P(Site 85.89 2.65 0.15 0.27 4 77.89 probability models, keeping site occupancy (Ψ) constant. +Effort) Secondly, we modelled the probability of site occupancy P(.) 86.58 3.34 0.10 0.19 2 82.58 (Ψ) for each species in relation to the covariates elevation, slope, aspect, distance to settlement and distance to river. In all univariate occupancy models, detection probability TABLE 3 Estimates of β-coefficient values for covariates influenc- was kept constant such as Ψ(covariate), P(.). Subsequently, ing the detection probability (P) of prey species in the biological we built best models based on AIC values (Burnham & corridor in the top-ranking models (Table ). Anderson, ), considering all multivariate models with Δ , β ± SE β ± SE AIC to be strongly supported by the data. We used (eastern (western β ± SE model averaging techniques to determine the grid cell specific Model section) section) (Effort) probabilities of occupancy considering all competing mod- Sambar els. We examined model fit using goodness-of-fit tests with P(Site+Effort) −0.93 ± 0.58 0.66 ± 0.57 0.85 ± 0.43 , bootstrapping steps. We used the mean untrans- Barking deer formed beta coefficient estimate from the best multivariate ± P(Effort) 0.41 0.37 model to predict site occupancy of the three prey species. Wild boar The value of untransformed coefficients reflects the mag- P(Effort) 0.66 ± 0.35 nitude and direction of their influence on occupancy
TABLE 4 Estimates of β-coefficient values for covariates influencing the occupancy of prey species (Ψ) in the biological corridor based on the top-ranking models (Table ).
β ± SE β ± SE β ± SE β ± SE β ± SE Model1 (Elevation) (Distance to settlement) (Aspect) (Distance to river) (Slope) Sambar Ψ(Slope+Aspect+Settlement), 0.20 ± 0.64 −0.02 ± 0.57 1.28 ± 0.74 P(Site+Effort) Barking deer Ψ(Elevation+Aspect), P(Effort) −1.54 ± 0.96 −0.59 ± 0.58 Wild boar Ψ(Elevation+River), P(Effort) −2.64 ± 1.60 −0.73 ± 0.83
River, distance to nearest river; Settlement, distance to nearest settlement.
Oryx, Page 4 of 8 © The Author(s), 2021. Published by Cambridge University Press on behalf of Fauna & Flora International doi:10.1017/S0030605320000976 Downloaded from https://www.cambridge.org/core. IP address: 170.106.202.226, on 30 Sep 2021 at 23:32:35, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0030605320000976 Occupancy of tiger prey species in Bhutan 5
probability. We produced predictive maps for the entire study area using QGIS, based on inferences from the model-averaged outputs for the sampled grid cells.
Results
Of the camera traps deployed, we retrieved (one was lost); from the eastern and from the western section. The total trap effort was , trap-days. At least one prey species was recorded at out of camera-trap locations, yielding independent images. Sambar was recorded at nine, barking deer at and wild boar at locations, resulting in naïve (survey-based) occupancy probabilities of ., . and ., respectively. Tigers were detected at five camera- trap locations. The detection probability (P) of all three prey species was positively affected by trap effort (Tables & ). The likelihood-based analysis of sambar occupancy showed the highest support for the model that included slope, aspect and distance to the nearest settlement (Table ), with occu- pancy being higher in areas that were further away from settlements, and on steeper and/or south-facing slopes. For barking deer, the model that included altitude and aspect performed best, and for wild boar, the model that included altitude and distance to the next river had the highest sup- port. Occupancy of both barking deer and wild boar was higher at lower elevations. Barking deer occupancy was higher on south-facing slopes, and wild boar occurred main- ly closer to rivers. Occupancy probability of barking deer was higher in the eastern (. ± SE .) than in the west- ern (. ± SE .) section (Fig. b). This was also the case for wild boar (eastern . ± SE ., western . ± SE .; Fig. c). Sambar occupancy was higher in the western (. ± SE .) than in the eastern section (. ± SE .; Fig. a) The model-averaged probability of occupancy (Ψ) derived from top-ranked models (Table )was. ± SE ± ± . for sambar, . SE . for barking deer and . FIG. 2 Predicted occupancy probability of the principal prey SE . for wild boar. In all cases, this was higher than species of the tiger Panthera tigris: (a) sambar Rusa unicolor, the naïve occupancy (Fig. ). (b) barking deer Muntjacus muntjak and (c) wild boar Sus scrofa in biological corridor no. , which connects Jigme Singye Wangchuck National Park (JSWNP) and Wangchuck Centennial Discussion National Park (WCNP).
Our findings on the occurrence of the principal prey species species suggests niche partitioning between these three of the tiger suggest that biological corridor no. provides sympatric ungulates. This is important for maintaining a sufficient structural connectivity between Jigme Singye balance between predator species abundance and resource Wangchuck National Park and Wangchuck Centennial availability (Cooke et al., ). National Park to facilitate tiger movements between these Barking deer and wild boar, and to a lesser extent sambar, areas. Overall, prey occupancy was relatively high in the preferred lower elevations (for sambar, see models in Table study area, with at least one prey species photo-trapped at with ΔAIC , ). This corroborates Tempa’s() findings out of grid cells, and naïve occupancy for each species for these species across Bhutan, showing a negative influ- .–.. Occupancy estimates were highest for barking ence of elevation on occupancy. All three species have a deer, followed by sambar and wild boar. The fact that habi- wide altitudinal range, but with few records above , m tat covariates retained in the best models differed between for barking deer and wild boar (Oliver & Leus, ;
Oryx, Page 5 of 8 © The Author(s), 2021. Published by Cambridge University Press on behalf of Fauna & Flora International doi:10.1017/S0030605320000976 Downloaded from https://www.cambridge.org/core. IP address: 170.106.202.226, on 30 Sep 2021 at 23:32:35, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0030605320000976 6 L. Letro et al.
TABLE 5 Multivariate model selection results for the effects of various covariates on the occupancy probability (Ψ) of principal prey species. The table shows the Akaike information criterion (AIC), difference between AIC and the best-performing model (ΔAIC), model weight and likelihood, number of parameters (K), and twice the negative log-likelihood (−LogLik) for each model. The top three models for each species are shown.
Model1 AIC ΔAIC Model weight Model likelihood K −2LogLik Sambar Ψ(Slope+Aspect+Settlement), P(Site+Effort) 75.73 0.00 0.39 1.00 6 63.73 Ψ(Elevation+Aspect), P(Site+Effort) 76.21 0.48 0.31 0.78 5 66.21 Ψ(Elevation, Settlement), P(Site+Effort) 76.31 0.58 0.29 0.75 5 66.31 Barking deer Ψ(Elevation+Aspect), P(Effort) 83.64 0.00 0.44 1.00 3 77.64 Ψ(Elevation+Road), P(Effort) 84.48 0.84 0.29 0.66 3 78.48 Ψ(Elevation+River), P(Effort) 84.59 0.95 0.27 0.62 3 78.59 Wild boar Ψ(Elevation+River), P(Effort) 72.98 0.00 0.25 1.00 3 66.98 Ψ(Elevation+Slope), P(Effort) 73.17 0.19 0.23 0.91 3 67.17 Ψ(Elevation+Road), P(Effort) 73.60 0.62 0.18 0.73 3 67.60
River, distance to nearest river; Road, distance to nearest road; Settlement, distance to nearest settlement.
fields, to which wild boars are often attracted, are generally located near rivers in Bhutan. Elevation had a stronger effect on the occurrence of all three species than land-use or forest type, which is surprising as vegetation type is expected to be an important factor affecting the availability of shelter and food (e.g. Moore, ). However, this finding agrees with an earlier study on sambar (Forsyth et al., ), and can probably be explained by the fact that all three species are opportunistic generalists and able to thrive in a broad range of forest types and environmental conditions (Oliver &Leus,; Timmins et al., , ). Sambar showed a slightly higher occupancy in the west- ern section of the corridor, whereas barking deer and wild boar were more common in the east. The overall higher prey occupancy in the east is probably related to its lower FIG. 3 Naïve (dark grey) and modelled (light grey) occupancy for mean elevation and proximity to a river and agricultural the principal prey species of the tiger Panthera tigris in biological fields (the latter located outside the study area). In Bhutan, corridor no. . the relative abundance of wild boar and barking deer is highest in areas close to croplands (Thinley et al., ). Timmins et al., ). Sambar may occur up to , m Both species regularly damage agricultural fields (Wang (Timmins et al., ). We recorded sambar, barking deer et al., ) and are relatively tolerant of human disturbance and wild boar at a maximum of ,, ,, and , m, (Tempa, ). In addition, the high intensity of seasonal respectively. Sambar and barking deer occupancies were grazing in the western section (Letro & Fischer, )may higher on south-facing slopes (see also Forsyth et al., negatively affect prey species abundance. for sambar deer in Australia). It is unknown whether south- The occupancy patterns of these three principal prey spe- facing slopes are preferred for thermal reasons (tempera- cies of the tiger suggest that biological corridor no. is po- ture, wind shelter; Moore, ) or associated vegetation tentially important for facilitating tiger movements between structure and composition. Sambar deer were recorded the two adjoining national parks. The eastern section of the more frequently on steeper slopes, supporting the findings corridor may be more suitable for tiger movements because of Johnsingh et al. () and Kushwaha et al. (), of higher probability of prey occupancy. However, tigers who documented a preference for hilly terrain. This could were recorded at three and two locations in the eastern be because such areas are less accessible for people. Wild and western sections, respectively. This could be because boar occupancy was positively affected by rivers, reflecting of variation in other factors such as the availability of do- the species’ preference for moist habitats such as riverine mestic prey, which also influence tiger movements (Bagchi forests or marshes (Graves, ). In addition, agricultural et al., ; Sunarto et al., ). Letro & Fischer ()
Oryx, Page 6 of 8 © The Author(s), 2021. Published by Cambridge University Press on behalf of Fauna & Flora International doi:10.1017/S0030605320000976 Downloaded from https://www.cambridge.org/core. IP address: 170.106.202.226, on 30 Sep 2021 at 23:32:35, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0030605320000976 Occupancy of tiger prey species in Bhutan 7
reported frequent incidences of livestock depredation by ti- FORSYTH, D.M., MCLEOD, S.R., SCROGGIE, M.P. & WHITE, M.D. gers in the corridor. To ensure the connectivity of tiger po- () Modelling the abundance of wildlife using field surveys and pulations in the two national parks through this biological GIS: Non-native sambar deer (Cervus unicolor) in the Yarra Ranges, south-eastern Australia. Wildlife Research, , –. corridor, we recommend assessing prey species density, FRMD () Land Use and Land Cover of Bhutan , Maps and and protection of prey species through patrolling. The man- Statistics. Forester Resources Management Division, Department of agement of the biological corridor should be enhanced Forests and Park Service, Thimphu, Bhutan. through the development of conservation management GOPALASWAMY, A.M., KARANTH, K.U., KUMAR, N.S. & plans. These should include adequate land-use planning MACDONALD, D.W. ( ) Estimating tropical forest ungulate densities from sign surveys using abundance models of occupancy. and local forest management plans, to reduce anthropogen- Animal Conservation, , –. ic disturbances and minimize habitat fragmentation. GRAVES, H.B. () Behavior and ecology of wild and feral swine (Sus scrofa). Journal of Animal Science, , . Acknowledgements We thank the management of Jigme Singye GTRP () Global Tiger Recovery Program (–). Global Tiger Wangchuck National Park and the Nature Conservation Division, Initiative Secretariat, The World Bank, Washington, DC, USA. Department of Forests and Park Services, for their support; the documents.worldbank.org/curated/en//pdf/ National Geographic Society, the Mohamed Bin Zayed Species WPBoxFinalVersionEng.pdf [accessed August Conservation Fund and The Rufford Foundation for funding; ]. Wangchuk Dorji, Tshering Dorji and Thinley Lhendup for assistance HARIHAR,A.&PANDAV,B.() Influence of connectivity, wild prey with data collection; and the anonymous reviewers and the editors and disturbance on occupancy of tigers in the human-dominated for their help with improving the article. western Terai Arc landscape. PLOS ONE, ,e. HAYWARD, M.W., JEDRZEJEWSKI,W.&JEDRZEWSKA,B.() Prey Author contributions Study conception and design: LL, KF; secur- preferences of the tiger Panthera tigris. Journal of Zoology, ing research funds: LL; data collection: LL, DD; data analysis: LL, KF, , –. TT; writing: all authors. HINES, J.E. () PRESENCE - Software to Estimate Patch Occupancy and Related Parameters.U.S.GeologicalSurvey,Reston,Virginia,USA. Conflict of interest None. mbr-pwrc.usgs.gov/software/presence.html [accessed August ]. JIGME,K.&THARCHEN,L.() Camera-trap records of tigers at Ethical Standards This research abided by the Oryx guidelines on high altitudes in Bhutan. CATnews, , –. ethical standards and was approved by the Research Steering JOHNSINGH, A.J.T., QURESHI, Q., GOYAL, S.P., RAWAT, G.S., RAMESH, Committee of the Department of Forests and Park Services, Ministry K., DAVID,A.etal.() Conservation Status of Tiger and of Agriculture and Forests of the Royal Government of Bhutan (ap- Associated Species in the Terai Arc Landscape, India. RR-/, plication ID: 5841017285A0C0133DD581). The fieldwork was con- Wildlife Institute of India, Dehradun, India. ducted with approval of the Honourable Director, Department of KANAGARAJ, R., WIEGAND, T., KRAMER-SCHADT, S., ANWAR,M.& Forests and Park Services, Thimphu, Bhutan (no. DoFPS/NCD/ GOYAL, S.P. () Assessing habitat suitability for tiger in the Adm-01/2018/8269). fragmented Terai Arc Landscape of India and Nepal. Ecography, , –. KARANTH, K.U., NICHOLS, J.D., KUMAR, N.S., LINK, W.A. & HINES, J.E. () Tigers and their prey: predicting carnivore densities from References prey abundance. Proceedings of the National Academy of Sciences of the United States of America, , –. BAGCHI, S., GOYAL, S.P. & SANKAR,K.() Prey abundance and KUSHWAHA, S.P.S., KHAN, A., HABIB, B., QUADRI,A.&SINGH,A. prey selection by tigers (Panthera tigris) in a semi-arid, dry () Evaluation of sambar and muntjak habitats using deciduous forest in western India. Journal of Zoology, , –. geostatistical modelling. Current Science, , –. BURNHAM, K.P. & ANDERSON, D.R. () Multimodel inference: LETRO,L.&DUBA,D.() Clouded leopard in biological corridor understanding AIC and BIC in model selection. Sociological of Bhutan, an opportunity for population dispersal. CATnews, Methods and Research, , –. , –. COOKE, R.S.C., WOODFINE, T., PETRETTO,M.&EZARD, T.H.G. LETRO,L.&FISCHER,K.() Livestock depredation by tigers and () Resource partitioning between ungulate populations in arid people’s perception towards conservation in a biological corridor environments. Ecology and Evolution, , –. of Bhutan and its conservation implications. Wildlife Research, DHENDUP,T.&DORJI,R.() Occurrence of six felid species , –. outside protected areas in Bhutan. Cat News, , –. LINKIE, M., CHAPRON, G., MARTYR, D.J., HOLDEN,J.&LEADER- DOFPS (a) Counting the Tigers in Bhutan: Report on the National WILLIAMS,N.() Assessing the viability of tiger subpopulations Tiger Survey of Bhutan –. Department of Forests and Park in a fragmented landscape. Journal of Applied Ecology, , –. Services, Thimphu, Bhutan. LITTLEWOOD, N.A., ROCHA, R., SMITH, R.K., MARTIN, P.A., DOFPS (b) Protected Areas of Bhutan. st edition. Department of LOCKHART, S.L., SCHOONOVER, R.F. et al. () Terrestrial Forests and Park Services, Thimphu, Bhutan. Mammal Conservation: Global Evidence for the Effects of DORMANN, C.F., ELITH, J., BACHER, S., BUCHMANN, C., CARL, G., Interventions for Terrestrial Mammals Excluding Bats and Primates. CARRÉ, G. et al. () Collinearity: A review of methods to deal Synopses of Conservation Evidence Series. University of Cambridge, with it and a simulation study evaluating their performance. Cambridge, UK. conservationevidence.com/actions/ Ecography, , –. [accessed July ]. ELITH,J.&LEATHWICK, J.R. () Species distribution models: MACKENZIE, D.I. & BAILEY, L.L. () Assessing the fit of ecological explanation and prediction across space and time. Annual site-occupancy models. Journal of Agricultural, Biological, and Review of Ecology, Evolution, and Systematics, , –. Environmental Statistics, , –.
Oryx, Page 7 of 8 © The Author(s), 2021. Published by Cambridge University Press on behalf of Fauna & Flora International doi:10.1017/S0030605320000976 Downloaded from https://www.cambridge.org/core. IP address: 170.106.202.226, on 30 Sep 2021 at 23:32:35, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0030605320000976 8 L. Letro et al.
MACKENZIE, D.I., NICHOLS, J.D., LACHMAN, G.B., DROEGE, S., SHERPA, M.N. & NORBU, U.P. () Linking protected areas ANDREW,J.&LANGTIMM, C.A. () Estimating site occupancy for ecosystem conservation: a case study from Bhutan. Parks, rates when detection probabilities are less than one. Ecology, , –. , –. SRIVATHSA, A., KARANTH, K.K., JATHANNA, D., KUMAR, N.S. & MACKENZIE, D.I., NICHOLS, J.D., ROYLE, J.A., POLLOCK, K.H., KARANTH, K.U. () On a dhole trail: examining ecological and BAILEY, L.L. & HINES, J.E. () Occupancy Estimation and anthropogenic correlates of dhole habitat occupancy in the Western Modeling: Inferring Patterns and Dynamics of Species Occurrence. Ghats of India. PLOS ONE, ,e. Academic Press, Burlington, USA. SUNARTO, S., KELLY, M.J., PARAKKASI, K., KLENZENDORF, S., MCSHEA, W.J., DAVIES, S.J. & BHUMPAKPHAN,N.() The Ecology SEPTAYUDA,E.&KURNIAWAN,H.() Tigers need cover: and Conservation of Seasonally Dry Forests in Asia. Smithsonian multi-scale occupancy study of the big cat in Sumatran forest Institution Scholarly Press, Washington, DC, USA. and plantation landscapes. PLOS ONE, ,e. METZGER,J.&DÉCAMPS,H.() The structural connectivity TEMPA,T.() The Ecology of Montane Bengal Tigers (Panthera threshold: an hypothesis in conservation biology at the landscape tigris tigris) in the Himalayan Kingdom of Bhutan. PhD thesis, scale. Acta Oecologica, , –. University of Montana, Missoula, USA. MOORE,I.() Habitat use and activity patterns of sambar deer, TEMPA, T., HEBBLEWHITE, M., MILLS, L.S., WANGCHUK, T.R., Cervus unicolor, in the Bunyip Sambar Enclosure. MSc thesis, NORBU, N., WANGCHUK, T. et al. () Royal Manas National The University of Melbourne, Melbourne, Australia. Park, Bhutan: a hot spot for wild felids. Oryx, , –. NCHM () Climate Data Book of Bhutan . National Center for THAPA,K.&KELLY, M.J. () Prey and tigers on the forgotten trail: Hydrology and Meteorology, Department of Energy, Thimphu, high prey occupancy and tiger habitat use reveal the importance Bhutan. of the understudied Churia habitat of Nepal. Biodiversity and ODDEN,M.&WEGGE,P.() Predicting spacing behavior and Conservation, , –. mating systems of solitary cervids: A study of hog deer and Indian THINLEY,P.() Technical comments on the design and muntjac. Zoology, , –. designation of biological corridors in Bhutan: global to national OLIVER,W.&LEUS,K.() Sus scrofa.InThe IUCN Red List of perspectives. Journal of Renewable Natural Resources, Bhutan, Threatened Species :e.TA. dx.doi.org/./ , –. IUCN.UK.-.RLTS.TA.en [accessed April ]. THINLEY, P., LASSOIE, J.P., MORREALE, S.J., CURTIS, P.D., PENJOR, U., MACDONALD, D.W., WANGCHUK, S., TANDIN,T.&TAN, RAJARATNAM, R., VERNES, K. et al. () High relative abundance C.K.W. () Identifying important conservation areas for the of wild ungulates near agricultural croplands in a livestock- clouded leopard Neofelis nebulosa in a mountainous landscape: dominated landscape in Western Bhutan: implications for crop inference from spatial modeling techniques. Ecology and Evolution, damage and protection. Agriculture, Ecosystems and Environment, , –. , –. PENJOR, U., TAN, C.K.W., WANGDI,S.&MACDONALD, D.W. () TIMMINS, R., KAWANISHI, K., GIMAN, B., LYNAM, A., CHAN, B., Understanding the environmental and anthropogenic correlates of STEINMETZ, R. et al. () Rusa unicolor (errata version published tiger presence in a montane conservation landscape. Biological in ). In The IUCN Red List of Threatened Species : Conservation, , . e.TA. dx.doi.org/./IUCN.UK.-.RLTS. QGIS DEVELOPMENT TEAM () QGIS Geographic Information TA.en [accessed January ]. System. Open Source Geospatial Foundation Project. qgis.osgeo.org TIMMINS, R.J., DUCKWORTH,J.&HEDGES,S.() Muntiacus [accessed August ]. muntjak.InThe IUCN Red List of Threatened Species : RCORE TEAM () R: A Language and Environment for Statistical e.TA. dx.doi.org/./IUCN.UK.-.RLTS. Computing. R Foundation for Statistical Computing, Vienna, TA.en [accessed April ]. Austria. r-project.org [accessed July ]. USGS () Shuttle Radar Topography Mission. U.S. Geological RUDNICK, D., RYAN, M.G., HARMON, M.E., BIRDSEY, R.A., Survey, Reston, USA. usgs.gov [accessed January ]. GIARDINA, C.P., HEATH, L.S. et al. () The role of landscape WANG, S.W., CURTIS, P.D. & LASSOIE, J.P. () Farmer perceptions connectivity in planning and implementating conservation and of crop damage by wildlife in Jigme Singye Wangchuck National restoration priorities. Issue in Ecology, , –. Park, Bhutan. Wildlife Society Bulletin, , –. SANDERSON, J.J.G. & HARRIS,G.() Automatic data organization, WANGCHUK,S.() Maintaining ecological resilience by linking storage, and analysis of camera trap pictures. Journal of Indonesian protected areas through biological corridors in Bhutan. Natural History, , . Tropical Ecology, , –.
Oryx, Page 8 of 8 © The Author(s), 2021. Published by Cambridge University Press on behalf of Fauna & Flora International doi:10.1017/S0030605320000976 Downloaded from https://www.cambridge.org/core. IP address: 170.106.202.226, on 30 Sep 2021 at 23:32:35, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0030605320000976