Using a capture-recapture approach for modelling the detectability and distribution of Houbara Bustard in southern Tunisia

Modelling the distribution of species of conservation concern is an important issue in population ecology. Classically, logistic regression analyses are conducted to estimate species' distributions from detection/non-detection data in a sample of sites and to test for the significance of several environmental variables in predicting the probability of occurrence. These modelling approaches assume that species detection probability is constant and equals one in all sampled sites, which is critical, notably in the case of rare, shy and cryptic species. The capture-recapture-like approach developed by Mackenzie et al. (2002, 2003) provides a reliable tool that accounts for imperfect detection when estimating species occurrence, as well as for assessing the relevance of site features as predictors of species occurrence probability. The aim of this study was to explore the possibility of using this approach in the context of Houbara Bustard Chlamydotis undulata in southern Tunisia. Our results show once more the low detectability of this emblematic species and stress the need to take this factor into account when estimating Houbara spatial distribution. The distribution of Houbara in southern Tunisia is more likely to be shaped by human-related than by habitat factors. In particular, Houbara occurrence was positively associated with site remoteness and camel numbers. Houbara seemed to avoid areas with high human presence and shared the most remote and agriculture-free zones with free-ranging camels.