Spatial variation in population-density, movement and detectability of snow leopards in a multiple use landscape in Spiti Valley, Trans-Himalaya

The endangered snow leopard Panthera uncia occurs in human use landscapes in the mountains of South and Central Asia. Conservationists generally agree that snow leopards must be conserved through a land-sharing approach, rather than land-sparing in the form of strictly protected areas. Effective conservation through land-sharing requires a good understanding of how snow leopards respond to human use of the landscape. Snow leopard density is expected to show spatial variation within a landscape because of variation in the intensity of human use and the quality of habitat. However, snow leopards have been difficult to enumerate and monitor. Variation in the density of snow leopards remains undocumented, and the impact of human use on their populations is poorly understood. We examined spatial variation in snow leopard density in Spiti Valley, an important snow leopard landscape in India, via spatially explicit capture recapture analysis of camera trap data. We camera trapped an area encompassing a minimum convex polygon of 953 km2. We estimated an overall density of 0.49 (95% CI: 0.39-0.73) adult snow leopards per 100 km2. Using AIC, our best model showed the density of snow leopards to depend on wild prey density, movement about activity centres to depend on altitude, and the expected number of encounters at the activity centre to depend on topography. Models that also used livestock biomass as a density covariate ranked second, but the effect of livestock was weak. Our results highlight the importance of maintaining high density pockets of wild prey populations in multiple use landscapes to enhance snow leopard conservation.


17
The endangered snow leopard Panthera uncia occurs in human use landscapes in the mountains of 18 South and Central Asia. Conservationists generally agree that snow leopards must be conserved 19 through a land-sharing approach, rather than land-sparing in the form of strictly protected areas. 20 Effective conservation through land-sharing requires a good understanding of how snow leopards 21 respond to human use of the landscape. Snow leopard density is expected to show spatial variation 22 within a landscape because of variation in the intensity of human use and the quality of habitat. 23 However, snow leopards have been difficult to enumerate and monitor. Variation in the density of 24 snow leopards remains undocumented, and the impact of human use on their populations is poorly 25 understood. We examined spatial variation in snow leopard density in Spiti Valley, an important 26 snow leopard landscape in India, via spatially explicit capture recapture analysis of camera trap data. 27 We camera trapped an area encompassing a minimum convex polygon of 953 km 2 . We estimated an 28 overall density of 0.49 (95% CI: 0.39-0.73) adult snow leopards per 100 km 2 . Using AIC, our best 29 model showed the density of snow leopards to depend on wild prey density, movement about 30 activity centres to depend on altitude, and the expected number of encounters at the activity centre 31 to depend on topography. Models that also used livestock biomass as a density covariate ranked 32 second, but the effect of livestock was weak. Our results highlight the importance of maintaining by their potential to influence a range of scientific inferences as well as conservation 48 interventions. However, large carnivores in general are difficult to enumerate due to their large 49 ranges (1), naturally low densities (2), and elusive behaviour. 50 The threatened snow leopard Panthera uncia is a typical example of a difficult to sample, 51 elusive carnivore that is reported to occur at relatively low population densities (0.15-3.88/100 52 km 2 ) even in best habitats (5-7). Snow leopards have relatively large home ranges, and of the 53 170 protected areas in the global snow leopard range, 40% are smaller than the home range 54 size of a single adult male (8). The distribution range of the snow leopard across Asia is subject 55 to pervasive human use, predominantly in the form of pastoralism and agro-pastoralism (9). 56 Over the past two decades, snow leopard habitats have also come under the increasing purview 57 of developmental activities and mining (6), commercial livestock rearing such as cashmere 58 goats (10), extraction of Cordyceps, and tourism. 59 Snow leopard habitats represent multiple use landscapes dominated by pastoralism and agro-60 pastoralism. Conservationists generally agree that snow leopards must be conserved amidst 61 people, following a land-sharing approach, rather than too much emphasis on creating strictly 62 protected areas (8 distance assumption can bias estimates of density and they suggest that least-cost distance be 137 tested in highly structured landscapes. 138 We used the maximum likelihood based SCR models (14)   We defined an integration area with spacing of 500 m, assuming that snow leopard density was  The spatial distribution model in SCR is a spatial Poisson process for animal activity centres 155 whose intensity (expected number of animal activity centres per unit area) can be homogeneous 156 (constant over space) or inhomogeneous (varying over space) (Borchers and Efford 2008). We 157 use the notation D(x; ) for density, signifying that density is a function of activity centre 158 location, x, which is a vector representing the x and y coordinates of an activity centre, and of 159 parameters represented by the vector . 160 We fitted SCR models with various combinations of covariates defined a priori. A candidate of an animal at a camera that is a distance d from its activity centre is ( ) = 0 exp { -2 (2 2 180 }.

181
For snow leopard density, we considered models in which the ( )s were wild prey density, 182 livestock density, terrain ruggedness and altitude at . We investigated the effect of 183 topographies (a factor with levels "ridgeline", "cliff" or "gully bed") and different altitudes at 184 the activity centre on the encounter function intercept and range parameters. We also 185 investigated models in which movement cost depends on altitude. 186 We modelled ( ) as a function of six spatial covariates ( ( )s) that could affect snow 187 leopard density (Figure 1). These included terrain ruggedness (typical snow leopard habitats 188 are steep and rugged (18)    Our camera traps spanned the covariate space of the wild prey density reasonably well ( fig. 2).

238
The density estimates from the top model ranged from 0.23 to 1.08 per 100 km 2 across the 239 region of integration (

269
All the top models in our study indicated that conductance is greater at higher altitudes.

270
Ecologically, this can be translated as snow leopards tending to move greater distances at 271 higher altitudes, which matches natural history observations that suggest snow leopards to 272 move along ridgelines (36-38).

273
Our top model showed that the variation in snow leopard density was largely associated with to be lower in areas with high livestock density (39).

288
Although snow leopards are known to prefer rugged terrain (40-42) we did not find much 289 support for snow leopard density to be dependent on ruggedness. This is presumably because 290 our entire study area was reasonably high on typical scales of ruggedness estimates.

291
Human settlements and associated anthropogenic pressures are considered to have a negative 292 influence on carnivore habitat use (43,44). In the case of snow leopards, studies report 293 conflicting results. For instance while one study found human settlements to exert a negative 294 influence on snow leopard habitat use (45), other studies reported no such effect (46,47). In our 295 study area, human density was low (<2 per km 2 ), and livestock grazing the major anthropogenic 296 activity. To some extent, the density of snow leopard activity centres in our study was 297 marginally higher in areas with low livestock biomass. leopard density (Fig. 3). There is evidence that creation of such 'core' landscape units with 304 community support can lead to the recovery of wild prey, and therefore, of snow leopards (49).

305
Such efforts require building long term partnerships with local communities by co-opting them 306 in conservation efforts (50).

307
We suggest that the land sharing approach to snow leopard conservation can be strengthened 308 considerably in snow leopard landscapes of Asia by creating core landscape units that can   Spatial capture recapture models are described using the following notation: "~1" shows the 469 RHS of Equations (1) to (3) to only contain an intercept term; "~x" means that it contains an 470 intercept and covariate "x"; "~x+y" means that it contains an intercept and covariates "x" and 471 "y"; "x*y" indicates an interaction between x and y, npar = number of parameters in the