Estimating lion’s prey species profile in an arid environment M. Beukes 1 , F. G. T. Radloff 1 & S. M. Ferreira 2 1 Department of Conservation and Marine Sciences, Faculty of Applied Sciences, Cape Peninsula University of Technology, Cape Town, South Africa 2 Scientific Services Department, South African National Parks, Skukuza, South Africa Keywords lion; Panthera leo; Kgalagadi Transfrontier Park; diet; GPS cluster; scat analysis; carnivore; arid environment. Correspondence Maya Beukes, Department of Conservation and Marine Sciences, Faculty of Applied Sciences, Cape Peninsula University of Technology, PO Box 652, Cape Town 8000, South Africa. Tel: +27 - 21 - 460 3195; Fax: +27 - 21- 460 3217 Email: mayabeukes@hotmail.com Editor: Matthew Hayward doi:10.1111/jzo.12474 Abstract Information on large carnivore diet composition is important when conservationists seek to address stressors such as habitat encroachment, bush meat trade and retri- bution killing on the persistence of populations. Detailed diet sampling approaches can contribute to effective conservation management. We compare different approaches including GPS cluster analysis, scat analysis and opportunistic observa- tions and describe lion diet profiles in the Kgalagadi Transfrontier Park (KTP). We determined that a sample size of between 65–69 lion feeding events out of 278 found by GPS cluster analysis and 55–59 lion scats out of 189 samples can pro- vide a representative description of lion diet in the KTP. Our opportunistic obser- vations (n= 52) were found to be too few for robust deductions. Scat analysis yielded the highest prey diversity (n= 11) and best representation of prey <50 kg, but required 160 scats at ~ 263 km driven per scat collected. GPS cluster analysis provided good information for large prey, but underrepresented prey <50 kg. Kill remains were found at only 37% of the investigated GPS cluster points (n= 823). Even so, GPS cluster analysis yielded diet descriptions quicker than scat analyses. The use of GPS cluster and scat analysis in tandem is recommended for the esti- mation of lion diet in this arid environment. Introduction Several stressors affect large carnivore populations (Beschta & Ripple 2009). For lions (Panthera leo) these include direct stressors such as habitat degradation (Chardonnet et al., 2010) and human-lion conflict (Bauer & van Der Merwe, 2004) but also indirect stressors that affect the prey base such as the bush meat trade (Lindsey et al., 2013). Lion diet estimates pro- vide important insights into predator-prey dynamics (Owen- Smith & Mills, 2008) and lion population demographics (Becker et al., 2013) and thus provide important information for conservationists. Traditionally the most common methods for defining lion diet estimates include stomach content analysis (Smuts, 1979), scat analysis (Mukherjee, Goyal & Chellam, 1994; Breuer, 2005), spoor tracking (Eloff, 1984), continuous follows (Schal- ler, 1972; Stander, 1992), and opportunistic observations (Mills, 1984). More recently, researchers have used GPS clus- ter analysis (Tambling et al., 2010), isotopic analysis (Hilder- brand et al., 1996; Newsome et al., 2009) and DNA analysis (Reed, 2004; Shehzad, 2011) to investigate the diet of carni- vores. Even so, all techniques have inherent bias (Mills, 1992; Tambling et al., 2010; Shores, Mondol & Wasser, 2015). Scat (faecal) analysis identifies undigested remains of prey that have passed through the digestive system (Mukherjee et al., 1994) and allows estimates of consumed biomass (Ruhe, Ksinsik & Kiffner, 2008). However, the approach, can over- represent smaller prey (Floyd, Mech & Jordan, 1978), is not able to identify age and sex, nor whether the prey was hunted or scavenged (Breuer, 2005). A sufficient number of scats rep- resentative of the population is required to depict the frequency of various prey items in the diet (Trites & Joy, 2005). Other lion dietary studies using scat analysis have used sample sizes of 74 (Davidson et al., 2013), 96 (Ogara et al., 2010) and 119 (Breuer, 2005). GPS cluster analysis identifies kill sites by clustered geo- graphical positions collated from individuals fitted with GPS collars (Tambling et al., 2010; Bacon et al., 2011). The tech- nique is useful as it can provide insights into foraging beha- viour such as the time spent on a kill, duration between kills, landscape utilization and age and sex of prey (Merrill et al., 2010). However, GPS cluster analysis often presents an overes- timation of large prey and may completely miss smaller prey that are consumed either too quickly to be detected using GPS cluster selection criteria, or are consumed whole without leav- ing detectable remains (Bacon et al., 2011; Tambling et al., 2012). Continuous observations provide robust information on lion foraging, and identification of both small and large prey (Stan- der, 1992), but may influence lion behaviour and are depen- dent on the vigilance of the researchers (Mills, 1992). Using opportunistic observations of lion kills as an alternative skews Journal of Zoology  (2017) – ª 2017 The Zoological Society of London 1 Journal of Zoology. Print ISSN 0952-8369