This is because locating faecal samples is extremely difficult in areas with rugged terrain and dense vegetation. On CX5461 Welgevonden, faecal samples are rarely found on roads or on easily accessible animal paths. Therefore, to collect a faecal sample dataset that is representative of leopard diet on Welgevonden, researchers would have to attempt to locate faecal samples in

areas where leopards frequent the most. These same areas often happen to qualify as GPS clusters using our GPS cluster method. However, we still believe that our approach – the analysis of nested datasets – is valuable to leopard research and dietary analyses. Feeding trial biomass corrections for faecal samples – used to convert prey frequency of occurrence to relative biomass consumed – were click here not applied to the faecal analysis as successive faecal samples were collected infrequently, and we assume that leopards consumed the majority of their prey (Ackerman, Lindzey & Hemaker, 1984; Martins et al., 2011). All statistical analyses were conducted in R v.2.13.0 (R Development Core Team, 2011) using preloaded packages. We investigated 362 GPS clusters and

located 133 leopard feeding sites from August 2010 to March 2011. Kills were distributed in the following order: LF1 = 43, LF2 = 46, LF3 = 40, LM1 = 4. Most GPS clusters (63%) did not have kills associated with them. Prey composition based on 127 kills (six kills were unidentifiable) comprised 64.6% ungulates, 14.2% chacma baboon Papio ursinus, 6.3% lagomorphs, 5.5% mongoose Mungos mungo and African civet Civettictis Rho civetta, 5.5% rock hyrax Procavia capensis, rodents, birds and reptiles, and 3.9% warthog Phacoochoerus africanus and bushpig Potamochoerus larvatus. We collected a total of 62 leopard faecal samples, consisting of 25 faecal samples from GPS clusters with kills, 13 faecal samples from GPS clusters without kills, and 24 faecal samples from locations not associated with GPS clusters. There was no significant difference in dietary composition between faecal samples collected at GPS cluster sites and those collected independently of GPS cluster

investigations (G2 = 1.226, P > 0.5). Faecal samples were distributed in the following order: LF1 = 14, LF2 = 11, LF3 = 10, LM1 = 3. Faecal contents (n = 97 individual prey items) based on percentage occurrence comprised 70.1% ungulates, 10.3% chacma baboon, 8.2% warthog and bushpig, 5.2% mongoose and African civet, 3.1% lagomorphs, and 3.1% rock hyrax, rodents, birds and reptiles. There was no significant difference between estimates of leopard dietary composition (G2 = 4.55, P = 0.1; Fig. 2) and biomass intake (W = 1, P > 0.2; Fig. 3) of small, medium and large prey for ‘GPS cluster analysis’ versus ‘faecal analysis’. Faecal samples collected at a feeding site contained the same prey species as the carcass found there 68% of the time.