Abstract:
Population estimates are a fundamental requirement of ecology and conservation.
While capture–recapture models are an established method for producing such estimates,
their assumption of homogeneous capture probabilities is problematic given
that heterogeneity in individual capture probability is inherent to most species. Such
variation must be accounted for by abundance models; otherwise, biased estimates
are risked. Here, we investigate the performance of four types of heterogeneity
models for estimating abundance of male cheetah Acinonyx jubatus, a species with
two distinct spatial tactics of territorial and nonterritorial (floater) males. The differences
in spatial movements of territory holders and floaters are expected to result in
intrasexual heterogeneous capture probabilities. Four heterogeneity models were
used to model male abundance at five territories in central Namibia; (a) a spatial tactic
model, (b) a finite mixture model, both run in program MARK, (c) a floater-only
model, and (d) a heterogeneity Mh model, both run in the program CAPTURE. Camera
trap data of cheetah, taken at frequently visited marking trees, were used to derive
true abundance. Model results were compared to the true abundance to assess the
accuracy of estimates. Only models (a), (b), and (c) were able to consistently produce
accurate results. Mixture models do not require prior knowledge regarding
spatial tactic of males, which might not always be available. Therefore, we recommend
such models as the preferred model type for cheetahs. Results highlight the
potential for mixture models in overcoming the challenges of capture probability heterogeneity and in particular their use with species where intrasexual behavioral
differences exist.
