SUMMARY
AbstractThis study classified the vegetation of the Savuti-Mababe-Linyanti ecosystem (SMLE), northern Botswana and developed a detailed map that provides a reliable habitat template of the SMLE for future wildlife habitat use studies. The major vegetation units of the SMLE were determined from satellite imagery and field visits and then mapped using Landsat 8 and RapidEye imagery and maximum likelihood classifier. These units were sampled using 40 m x 20 m (800 m²) plots in which the coverage of all plant species was estimated. Non-metric multidimensional scaling (NMS) demonstrated that plant communities were determined by gradients in soil texture or fertility and wetness. NMS 1 represented a gradient of soil texture with seven woodland communities on sandy soils (sandveld communities and Baikiaea forest) dominated by Baikiaea plurijuga in Baikiaea forest and Terminalia sericea and Philenoptera nelsii in sandveld, with various indicator species differentiating the various sandveld community types. Mopane woodland further from and riparian woodland adjacent to permanent water was common on less sandy alluvial soils. Mineral-rich, heavy clay soils in the sump of a large paleolake system support open grassland and mixed Senegalia/Vachellia (Acacia) savanna, with the mineral-rich soils supporting grasses high in minerals such as phosphorus, calcium, sodium and potassium, and thus this region is a critical wet season range for migratory zebra. Taller, high-quality grasses in the mosaic of sandveld and mopane woodland communities provide critical grazing for taller grass grazers such as buffalo, roan and sable antelope, whereas wetland communities provide reliable green forage during the dry season for a variety of herbivores, including elephant. This study has demonstrated how large-scale environmental gradients determine functional habitat heterogeneity for wildlife.Conservation implications: Our study demonstrated that the functionality of protected areas is determined by large-scale environmental gradients. Thus conservation science must aim to ensure that protected areas cover the full range of key environmental gradients in a region (soil texture and wetness in our study). Our habitat map provides a data base for wildlife habitat use studies in the region.