In addition to space use, we estimated leopard density across the study site for 2010 and 2011. To estimate density we
used spatially-explicit capture–recapture (SECR) models, which account for imperfect detection. The SECR models combine
leopard capture histories and the spatial coordinates of the captures to estimate density (Royle et al., 2009a). SECR models
were used because they calculate density without relying on an arbitrary buffer zone (e.g., minimum convex polygon), which
has been shown to inflate density estimates by between 20% and 200% (Obbard et al., 2010). The SECR model we used also
relaxes the assumption of geographic closure, but instead accounts for animal movement and detection probability based
on activity centers (Gardner et al., 2009).