Welgevonden hosts a number of large carnivores

such as lion, cheetah Acinonyx jubatus, and leopard as well as a large population of brown hyaena Hyaena brunnea. Research was conducted under the University of Pretoria Animal Use and Care Committee ethics clearance protocol A022-06 with all its amendments and the Limpopo province (South Africa) standing permit (No. S13631) for scientific research. During August 2010 to March 2011, we captured four leopards [two adult females (LF1 and LF2); one sub-adult female (LF3); one adult male (LM1)] using soft-hold foot snares (Frank, Simpson & Woodroffe, 2003). Each leopard was immobilized MG 132 using 4–5 mg kg−1 teletamine-zolazepam (Zoletil 100, Virbac RSA, Halfway House, South Africa) and fitted with a remote drop-off, Akt inhibitor ultra-high frequency GPS collar (Followit™ Tellus, Lindesberg, Sweden). Collars were programmed to record a GPS location every 2 h (06:00 h, 08:00 h, 10:00 h; apart from 12:00 h that consistently failed to fix) resulting in 11 GPS locations per day. Collars were released via a remote-controlled drop-off system on completion of the study. GPS data were imported into

ArcGIS v.9.2 (ERSI, Redlands, CA, USA). GPS clusters were classified by a set of decision rules: (1) consecutive GPS locations within 50 m of each other were merged into one cluster; (2) clusters within 100 m and 8 h (closest point to closest point) of other clusters were merged (Pitman et al., 2012). Therefore, the smallest GPS clusters possible were those that consisted of two GPS points (representing a site fidelity of 2 h). A GPS cluster site represents an activity performed by a

leopard in time and space; these activities are not automatically related to predation (other potential activities include resting, mating, territorial disputes, etc.). Clusters not damaged by fire or flood were systematically catalogued and investigated in the field for prey remains (e.g. carcass, hair, bone, blood) fitting the appropriate time frame. Prey remains were photographed and representative material taken for later identification conducted either macroscopically (e.g. carcass, horns, MCE skull) or microscopically (hair cuticle scale patterns and cross sections) using published references (Dreyer, 1966; Keogh, 1979, 1983; Buys & Keogh, 1984; Douglas, 1989), and a reference collection housed at the Centre for Wildlife Management, University of Pretoria, South Africa. Faeces were collected in two ways: (1) at GPS clusters; (2) opportunistically (i.e. independent of cluster investigations) while traversing the home ranges of collared leopards. Only samples with a diameter greater than 20 mm were collected for analysis to minimize the collection of non-leopard faecal samples (Norton, Henley & Avery, 1986). Accurately determining the age of a desiccated faecal sample was not possible.