Predictive models of movement by Serengeti grazers

Many animal species are unevenly distributed across the landscape, in spatial patterns that continually shift over time. Such a shifting mosaic is thought to have profound implications for the persistence and stability of ecosystems. Management and conservation of natural systems would be enhanced if we could accurately predict movement. Such prediction has not yet been possible. Here we use an extensive set of field data on food abundance and quality, combined with experimentally derived measures of nutritional value, to predict the spatial distribution of Thomson's gazelles (Gazella thomsoni thomsoni Gunter) on the Serengeti Plains of East Africa. Twelve plausible models, based on alternate foraging objectives or movement rules, were assessed against field data on food and grazer abundance gathered at biweekly intervals (every two weeks) over the course of the wet seasons in two different years. Nomadic movements of gazelles closely tracked changes in the spatial distribution of short grass swards. Gazelles left short grass patches when local daily energy intake dropped below the expected intake averaged across the landscape. Subsequent redistribution of gazelles among neighboring patches was proportional to daily rates of energy intake in each patch. Thus, nomadic movements by Thomson's gazelles were predictable on the basis of local energy gain. This suggests that adaptive behavioral models can provide useful predictive tools for understanding the dynamics of complex natural systems.