The fate of most of Africa's savanna parks is to become islands in a sea of humanity. The effects of park isolation n community dynamics will depend fundamentally on the degree to which wildlife populations are naturally linked by dispersal and gene flow. I propose to address an issue that will affect the future management of the large herbivores that to a great extent are characterize and justify these protected areas: To what extent are Africa's savanna parks already conserving naturally isolated lineages, and by implication, dynamically independent populations? Since all parks and species cannot be studied, a framework for predicting the gross population structure of a given species is required. This is provided by the 'Resource Use Hypothesis' (RUH) which states that vegetation and geographic discontinuities form barriers to gene flow for species that are habitat 'specialists', but not for species that are habitat 'generalists'. While ecological and macroevolutionary evidence supports the RUH, the microevolutionary component of the hypothesis has not been examined. The RUH will be tested by examining the population structure of six large herbivore species (elephant, buffalo, impala, zebra, giraffe, and wildebeest) that are common to three savanna parks in Tanzania (Serengeti and Tarangire) and Botswana (Chobe). These species were chosen because they have strongly contrasting food and habitat preferences. These savanna parks were chosen because they have similar large mammal communities and, most important, because their geographical attributes create an ideal setting to test the RUH. If the RUh is supported it will provide a mechanistic model for prediction the population structure of species with known habitat requirements within a region of known habitat characteristics.
