Assessing the abundance and distribution of mammalian carnivores is vital for understanding their ecology and providing for their long-term conservation. Because of the difficulty of trapping and handling carnivores many studies have relied on abundance indices that may not accurately reflect real abundance and distribution patterns. We developed statistical analyses that detect spatial correlation in visitation data from combined scent station and camera-trap surveys, and we illustrate how to use such data to make inferences about changes in carnivore assemblages. As a case study we compared the carnivore communities of adjacent communal and freehold rangelands in central Namibia. We used an index of overdispersion to test for repeat visits to individual camera-trap scent stations and a bootstrap simulation to test for correlations in visits to camera neighbourhoods. After distilling our presence-absence data to the most defensible spatial scale, we assessed overall carnivore visitation using logistic regression. Our analyses confirmed the expected pattern of a depauparate fauna on the communal rangelands compared to the freehold rangelands. Additionally, the species that were not detected on communal sites were the larger-bodied carnivores. By modelling these rare visits as a Poisson process we illustrate a method of inferring whether or not such patterns are because of local extinction of species or are simply a result of low sample effort. Our Namibian case study indicates that these field methods and analyses can detect meaningful differences in the carnivore communities brought about by anthropogenic influences. ?? 2007 FFI.
Additional publication details
Remote camera-trap methods and analyses reveal impacts of rangeland management on Namibian carnivore communities