3.4. Distribution-wide assessment of dhole pack size:
We used average dhole pack size as the response variable and tiger density, ungulate density, area of PA, terrain ruggedness as predictor variables. We scaled predictor variables (Size of PA, elevational heterogeneity and terrain ruggedness) and checked for correlation among all predictor variables. We dropped elevational heterogeneity as the predictor variables because of its high correlation with terrain ruggedness. We ran a total of 10 additive and interactive models (Table 2, Figure 3). The top two models achieved the model selection criterion of ΔAICc <2. Upon model selection we found, additive effect of tiger density and prey density and interactive effect of tiger density and prey density, to be the top two best models (Table 3, Figure 4 & 5). On averaging the two top models (Table 4, Figure 6) we found a negative association of tiger density (-0.89 ± 0.33, p = 0.01) and a positive association of prey density (0.09 ± 0.03, p =0.03) with the pack size and prey*tiger density (0.01 ± 0.0, p =0.18) was not significant but still explained the relation with the response variable.