Abstract
Population monitoring is key to wildlife conservation and management but is challenging at the spatial and temporal extents necessary for understanding changes. Non-invasive survey methods and spatial capture-recapture (SCR) models have revolutionized wildlife monitoring by providing the means to more easily acquire data at large scales and the framework to generate spatially-explicit predictions, respectively. Despite the opportunities for improved monitoring, challenges can remain in the study design and model fitting phases of an SCR approach. Here, we used a search-encounter design with multi-session SCR models to collect spatially-indexed photographs and estimate the changes in density of cheetahs between 2005 and 2013–2016 in the Masai Mara National Reserve (MMNR) in southwestern Kenya. Our SCR models of cheetah encounters suggested little change in cheetah density from 2005 to 2013–2016, though there was some evidence that density fluctuated annually in the MMNR. The sampling period length (5 vs. 10 months) and timing (early, late, full year) over which spatial encounters were included in the modeling did not substantially alter inferences about density when sample sizes were adequate (>20 spatially distinct encounters). We estimated an average cheetah density of ~1.2 cheetahs/100 km2, consistent with the impression that the MMNR provides important habitat for cheetahs in Africa. During most years and seasonal periods, the spatial distribution of vegetation greenness (a proxy for ungulate habitat quality) accounted for important variation in encounter rates. The search-encounter design used here could be applied to other regions for the purposes of cheetah monitoring. While snap-shot estimates of population size across time are useful for wildlife monitoring, open population models could identify the mechanisms behind changes and further facilitate better conservation and management decision making.