No Longer Tracking Greenery in High Altitudes: Pastoral Practices of Rupshu Nomads And

No longer tracking greenery in high altitudes: Pastoral practices of Rupshu nomads and their implications for biodiversity conservation

Methods

We applied the ‘Area tool’ in ‘Hawth’s tools extension’ of ArcGIS 9.2 (ESRI Inc.) for estimating the approximate area of seasonal pasture. Areas > 5,300 m in elevation were removed from the analysis as they lie above the vegetation limit.

Argali use

The vantage points for observations of argali were determined during previous reconnaissance surveys to allow for a complete overview of the study area. The observation of argali groups covered the entire daylight period, i.e. between 0600 and 1900 hrs. The observed groups’ locations (n=102) were recorded into a GPS and overlaid on topographical maps and DEM.

Livestock use

Each nomad campsite was assigned a weight (SUD/campsite) which we used to generate the intensity of use for the point using kernel density estimation (KDE) (Figure 2b). We used fixed kernel estimation using the Gaussian kernel function to estimate the density around the point (Silverman, 1986; Seaman & Powell, 1996). The smoothing parameter (h) was chosen using least squares cross validation (LSCV) method (Summer h: 2676.9, Winter h: 3795.5) (Worton 1995; Gitzen and Millspaugh 2003). As in the context of ‘home range’ analysis, the density at any location is also an estimate of time spent there, which we estimated as SUD / campsite. Hence, for camp locations, the 50% volume contour (VC) will show the area with higher pressure and intensity progressively reduced towards the edge (95% VC). In the home range context, it will show that these are the 'core zones' - areas that are important due to some key resource (snow free pastures, lower altitude, and seasonal water).

Argali locations were mainly based on actual observations of the groups. For comparison with livestock activity areas, it was important to estimate utilization distributions (UD) (Worton 1989) and make spatial predictions in terms of area of influence or home ranges for both of the considered groups. We therefore used KDE to delineate spatially the utilizing distribution of argali based on the density of the argali observations. Fixed-kernel estimates for argali were made using Gaussian bivariate normal function and bandwidth selection using LSCV procedure (Summer h-2772.4, Winter h-4703.2). For argali, 50% (VC) will show a boundary of the area that contains core of the probability density distribution (Seaman & Powell 1996). The 95% VC would contain most of the points that were used to generate the kernel density estimate. This density forms a basis for measuring the overlap in area of individuals or species in terms of area and intensity of use (Smith & Dobson 1994).

Each of the volume contours for both the groups were converted into polygons to estimate area enclosed by them. The KDE estimation was carried out using ade4 library in R (Chessel et al. 2004) and areas for the polygons enclosed by VCs were estimated in ArcGIS 9 using area module of Hawth’s tool extension.

To assess the UD overlap, we used the following formula to estimate the overlap index (OI) (Cole’s coefficient of association – (Cole 1949, Kenward 2001):

Where A1 is area of argali UD polygon and A2 is area of livestock UD polygon. We assessed the OI bounded by the 50 % and 95% volume contours of the two UDs. Overlap among the seasonal polygons was estimated using intersect tool in the ArcToolbox in ArcGIS 9. The overlapping polygons were obtained using this method and their area was estimated again using the ‘area’ module (see above). We assessed the OI of 50% and 95% volume contours for argali and livestock during summer and winter.

The overlapping polygons were obtained using this method and their area was estimated again using the ‘area’ module. We assessed the OI of 50% and 95% volume contours for argali and livestock during summer and winter.

To estimate the magnitude of the habitat variables in the utilizing polygons of argali and livestock, we sampled 300 random points, each in the argali and core zones. Terrain variables altitude and slope were extracted from DEM using ArcGIS 9 (ESRI Inc). We observed the distribution of the habitat variables altitude and slope for all the sampled points from the core zones of livestock activity and argali groups.

Results

Table S1. Individual camp sites and the estimated grazing intensity. SU-Sheep units (total number of livestock including Sheep, goat, Yak and horse), SUD- Sheep Unit Days (measurement of time spent by livestock at a camp site).

Overlap Index (OI)

BTable S2. Seasonal range use of the argali (Ovis ammon) and livestock groups and their spatial overlap in the Tso Kar basin of easter Ladakh, India.

OI for 50% Volume contour

The overlap was marginal for all the three tested combinations (Table 2 and Figure 3-a,b and c). The largest value of OI was 12.9 %, which was for the combination of argali and livestock in winter. For other two combinations, argali summer vs livestock summer and argali summer vs livestock winter, the OI values were extremely low or 0, indicating no overlap or very marginal overlap.

OI for 95% Volume contour

OI for 95% volume contours varied for different combinations (Table 2, Figure 3-a,b and c). Highest overlap for 95% VCs was observed for winter whereas the least overlap was observed for summer. The OI values for argali summer vs livestock winter, however, were intermediate.

Figure S1. Maps showing – a) Seasonal pastures in winter, nomads’ camp sites locations, b). Utilizing Distribution of livestock around camp sites, estimated based on grazing pressure (calculated as Sheep unit days (SUD)) along with 50% and 95% Volume contours calculated using Kernel density estimation.

1