@article { , title = {Water availability is a principal driver of large-scale land cover spatial heterogeneity in sub-Saharan savannahs}, abstract = {© 2018, The Author(s). Context: The heterogeneous mosaic nature of African savannah vegetation is a key aspect of its ecology. This study evaluates mosaic distributions and characteristics across sub-Saharan Africa, investigating the environmental drivers of mosaic formation. Objectives: This study was designed to determine: (1) on a continental scale, how frequent are mosaics in savannahs? and (2) what are the key environmental drivers in the formation of mosaics? Methods: Landsat ETM+ satellite imagery was used to generate land-cover maps for 39 sample areas across sub-Saharan Africa. The spatial complexity of land-cover mosaics at 4628 savannah sub-sites was quantified, and modelled using random forests to identify the relative importance of environmental variables driving mosaic presence. Results: Only six sub-sites constituted a single land-cover class, illustrating that mosaic habitats are abundant at the scale analysed (19.6km 2 ), although mosaic characteristics varied considerably. Results indicate precipitation is most important in influencing mosaic complexity, followed by evapotranspiration, temperature, lithology and distance to rivers. Fire and ecosystem engineer presence are of lesser importance at this study scale. Conclusions: Mosaics are ubiquitous in the African savannahs studied, their presence influenced by multiple environmental drivers, with water being key. The lower importance of fire and large mammal disturbance is likely resulting from these highly individualistic site-based process varying between sites, resulting in no single, coherent, across-Africa disturbance signal, and/or lack of detail in available data at this scale. Therefore, large-scale determinants of savannah mosaics appear climate-driven. Under future global warming scenarios, African savannahs are likely to become more homogenous.}, doi = {10.1007/s10980-018-0750-9}, eissn = {1572-9761}, issn = {0921-2973}, journal = {Landscape Ecology}, note = {OA (CCBY). OL 04.12.2018}, pages = {131-145}, publicationstatus = {Published}, publisher = {Springer Verlag}, url = {https://nottingham-repository.worktribe.com/output/1365505}, volume = {34}, keyword = {Heterogeneity, Land cover, Landsat, Remote sensing, Savannah, Sub-Saharan Africa}, year = {2019}, author = {Marston, Christopher G. and Wilkinson, David M. and Reynolds, Sally C. and Louys, Julien and O’Regan, Hannah J.} }