Predictive mapping of season distributions of large mammals using GIS: an application to elephants on Mount Kenya

Abstract

1. Knowledge of elephants in forested environments where visibility is poor is often restricted to elephant density estimates derived from counts of elephant dung piles along line transects. Elephant dung piles are a good indicator of elephant habitat use over a period of time (i.e. a season) because they remain visible for several months. Using Mount Kenya (MK) in Kenya as a case study, we demonstrate how those dung pile counts can additionally be used to develop elephant dung pile explanatory models and distribution maps. 2. The dung pile explanatory models are built from a selected number of line‐transect samples per season using generalized linear model analysis, in which the counted dung of each sample is explained by the combined effects of a series of environmental and anthropomorphic parameters. For the significant explanatory parameters in the models, digital map layers are developed and integrated into a geographic information system. Applying the explanatory models, predictive seasonal elephant dung pile distribution maps are developed. 3. Data preparation and analysis are intensive, but the distribution maps derived from explanatory parameters of distribution are a powerful tool for patrol planning and land‐use management and to locate areas of high elephant density and the habitats they move between. This method is useful for sites where physical, environmental, logistical or other constraints render it impossible to spread line transects evenly over the entire study area, assuming that transects are representative samples of the entire habitat.