Abstract
Understanding the spatial patterns of herbivore habitat use is important for understanding ecological processes and ecosystem management. Past research has mainly focused on explaining herbivore habitat selection, to the extent that environmental variables driving habitat selection can be considered well researched. However, little is known about how environmental variables contribute to patterns of foraging site revisitation, even though revisited sites might represent sites of ecological significance. The increasing elephant population in South Africa is causing a decline in woody species, necessitating insight into elephant habitat use to improve the management of areas in which they occur. This study aimed to gain insight into the spatial patterns of foraging site revisitations by the African elephant (Loxodonta africana). To achieve this, the non-randomness of the spatial locations of revisited sites in terms of their environmental conditions was assessed. Next, the relations between these environmental variables and the revisitation rate per site were analysed. Lastly, a model was constructed to assess which environmental variables had the largest effect on site revisitations. Ten years of GPS tracking data of elephants in the Greater Kruger Area, South Africa, was used. Environmental variables were obtained from existing data sources. The data were analysed with Kolmogorov tests and linear mixed effects models, using AIC-based model selection. The results indicate that elephants select revisitation sites close to water and at intermediate slopes. Cows also selected revisitation sites with intermediate phosphorus concentrations. Yet, elephants revisited sites at intermediate distances from water and higher phosphorus concentrations more frequently. Moreover, phosphorus is the most important environmental variable influencing the revisitation rate. This study shows that analysing revisitation patterns gives valuable insight into how animals use specific sites within their home range, as the environmental covariates related to the selection of the revisitation sites differ from those related to the revisitation rate.