Assessing the feasibility of density estimation methodologies for African forest elephant at large spatial scales

Considerations on effort, precision and accuracy for long-term monitoring of African lions (Panthera leo), when using Bayesian spatial explicit capture-recapture models, in fenced protected areas

Intensive management is frequently required in fenced wildlife areas to reduce deleterious effects of isolation. Decisions on how best to manage such wildlife are ideally informed by regular and reliable estimates of spatiotemporal fluctuations in population size and structure. However, even in small, fenced areas, it is difficult and costly to regularly monitor key species using advanced methods. This is particularly the case for large carnivores, which typically occur at low density and are elusive yet are central to management decision-making due to their top-down effects in ecosystems and attracting tourism. In this study, we aimed to provide robust estimates of population parameters for African lions (Panthera leo) and use the data to inform a resource-efficient long-term monitoring programme. To achieve this, we used unstructured spatial sampling to collect data on lions in Pilanesberg National Park, a small (~550 km²) fenced protected area in South Africa. We used Bayesian spatial capture-recapture models to estimate density, abundance, sex ratio and home range size of lions over the age of 1 year. Finally, to provide guidance on resource requirements for regular monitoring, we rarefied our empirical data set incrementally and analysed the subsets. Lion density was estimated to be 8.8 per 100 km² (posterior SD = 0.6), which was lower than anticipated by park management. Sex ratio was estimated close to parity (0.9♀:1♂), consistent with emerging evidence in fenced lion populations, yet discordant with unfenced populations, which are usually ~2♀:1♂ in healthy, source populations. Our rarefied data suggest that a minimum of 4000 km search effort needs to be invested in future monitoring to obtain accurate and precise estimates, while assuming similar detection rates. This study demonstrates an important utility of Bayesian spatial explicit capture-recapture methods for obtaining robust estimates of lion densities and other important parameters in fence-protected areas to inform decision-making.