Estimating forest elephant numbers with dung counts and a geographic information system

Dung counts are the most practical means of estimating numbers and distribution of forest-dwelling elephants. In the forests of Gabon, which have a sparse human population, the density of elephant dung piles increased with distance from roads. Human influences (roads and density of human habitation) accounted for half the variation in dung-pile density. Ninety-five percent of the dung piles were estimated to occur in the low human density stratum that covers two-thirds of Gabon. We present a new method of estimating the dung-pile population using the gradient of dung density in relation to roads. Estimates of the gradient were combined with estimates from a geographic information system (GIS) of the area of forest in bands at different distances (e.g., 0-5, 5-10 km) from roads. This process gave an estimate of the dung-pile population for the whole forest zone; the standard error and confidence limits were found by bootstrapping. We argue that estimates of forest elephant abundance in central Africa will be both more accurate and more precise if one accounts for the gradient. Simulation showed that the optimum sample for Gabon is 40 transects, each 5 km in length.