ABSTRACT
ABSTRACT: Relationships between patterns of plant species richness and landscape degradation were analysed in the Sahel zone of Burkina Faso, West Africa. To generate predictive models of the distribution of plant species in the study region we used the Genetic Algorithm of Rule-set Production (GARP) modelling system. This system has excellent potential for describing the ecological envelopes and the spatial distribution of species and can be used for many different applications. Furthermore, the spatial pattern of biodiversity can be delineated by combining the distribution maps of many species across different taxonomic and functional groups. In combination with environmental layers we used 12,000 geo-referenced collection localities to model species distributions in the Sahel zone of Burkina Faso. Previous studies have usually been conducted at regional scale using climatic and biomass parameters with a low spatial resolution of environmental variables. However, informed decision-making regarding conservation priorities mainly take place on the local scale. In a new approach we used derivates of high resolution satellite images (LANDSAT ETM+) as environmental input parameters. The potential of these images to differentiate vegetation and soil properties, which determine the distribution of plant species, has been shown in preceding analyses. Modelled distributions were evaluated with independent test data and selected by expert knowledge and statistical analysis. 138 plant species with at least 15 spatially unique occurrence points were modelled with a high accuracy. Consequently we calculated a map of phytodiversity by combining distribution maps of single species. The highest diversities have been predicted for a little disturbed area of dunes and tiger bush vegetation. Our approach has potential for widespread application to assist in conservation priorities and environmental management.
