Estimating abundance of striped hyenas (Hyaena hyaena) in the Negev Desert of Israel using camera traps and closed capture-recapture models

Estimating key state variables (e.g., abundance/density) of threatened/endangered species is a difficult yet crucial task. These variables are essential for wildlife conservationists/managers to assess the current states and trends of target populations and make state-dependent management decisions. We estimated, using camera traps and two different methodologies, the abundance of the near-threatened striped hyena (Hyaena hyaena) in the Negev Highlands, Israel. Both the traditional closed capture-recapture (closed-CR) and the census/near-census frameworks (both rely on closed CR models) were employed. For attaining a census/near census, we calculated a range of plausible capture/detection probabilities and their combinations with sampling durations that would result in detection of >= 95% of the target population. We then covered a subsection (circa 400 km(2)) of the Negev Highlands using 15 camera traps for 83 days. This targeted sampling provides a strong indication for attaining a census/near census for capture/detection probabilities as low as p = 0.05. The closed CR models (M-0 and M-b) yielded a model averaged estimates of abundance (N) over bar = 7 (SE 1.94E-05, PLCI 7-7) and capture/detection probability (p) over bar = 0.09 (SE 0.033, 95% CI 0.05-0.18). Using the estimated (p) over bar, we calculated that during the sampling, we (ostensibly) captured the entire (99.9%) target population. Additionally, we used spatially explicit capture-recapture (SECR) approach to estimate the density of the hyenas in a subsection of our study area where hyenas were spotted and got an estimate of D = 0.085 individuals/km(2) (SE 0.0334, 95% CI 0.040-0.178) or 8.5 individuals/100 km(2). Many studies estimate abundance/density of threatened/endangered species using closed CR models; however, for small populations of elusive animals, these models often yield less accurate estimates exactly where accuracy is needed the most. While it likely increases the costs and time, the census/near-census framework provides a more accurate solution for such cases.