The impact of habitat structure on reptile occurrence in a fragmented tropical landscape

Abstract

Defining the spatial distributions of species with regards to habitat selection and landscape structure is an important part of biogeography, ecology and conservation research. I investigated reptile occurrence and community structure in two patches of dry forest in north western Madagascar using detection/non-detection data collected on repeated transect surveys for four years. A Bayesian hierarchical occupancy model and multispectral satellite imagery were used to assess the effects of vegetation structure, proximity to human development and edge proximity on the site presence of 37 squamate species in the context of taxonomic family and Threat Status. Mean species richness was highest at sites within a forest patch (23 (4, 30)). Sites with dense green vegetation promoted the highest levels of reptile occupancy among the Chamaeleonidae and Gekkoniidae families (with regression coefficient estimates up to 0.75 (0.12, 1.53)) and all species were more likely to occur at sites closer to the forest patch periphery. The Boidae had the widest 95% CRI for the regression coefficient estimates representing the effects of habitat variables on occupancy, indicating that they are highly variable in their habitat use. The regression coefficient estimates of mean reptile occupancy had 95% credible intervals (CRIs) including zero due to high variability between species. The proximity to roads did not have a statistically significant effect on any of the species, and Uroplatus guentheri was the only species which showed a significant preference for being further from a settlement (0.81 (0.05, 1.77)). The findings demonstrated the importance of using caution when assessing Threat Status at a regional level and suggest that the Red List’s measure of Population Trend may be significantly influenced by imperfect detection