Aim: In tropical Africa, savannas cover huge areas, have high plant species richness and are considered as a major natural resource for most countries. There is, however, little information available on their floristics and biogeography at the continental scale, despite the importance of such information for our understanding of the drivers of species diversity at various scales and for effective conservation and management. Here, we collated and analysed floristic data from across the continent in order to propose a biogeographical regionalization for African savannas. Location: We collated floristic information (specifically woody species lists) for 298 samples of savanna vegetation across Africa, extending from 18° N to 33° S and from 17° W to 48° E. Taxa: We focused on native woody species. Methods: We used ordination and clustering to identify the floristic discontinuities and gradual transitions across African savannas. Floristic relationships, specificity and turnover, within and between floristic clusters, were analysed using a (dis-)similarity-based approach. Results: We identified eight floristic clusters across African savannas which in turn were grouped into two larger macro-units. Ordinations at species and genus levels showed a clear differentiation in woody species composition between the North/West macro-unit and the South/East macro-unit. This floristic discontinuity matches to the High (i.e. N&W) and Low (S&E) division of Africa previously proposed by White () and which tracks climatic and topographical variation. In the N&W savannas, the floristic gradient determined by rainfall was partitioned into the Sudanian (drier) and Guinean (wetter) clusters. Within the highly heterogeneous S&E savannas and woodlands, six clusters were identified: Ugandan, Ethiopian, Mozambican, Zambezian, Namibian and South African. Main conclusions: The proposed pan-African classification of savannas and woodlands might assist the development of coordinated management and conservation policies.
Bibliographical noteFunding Information:
The authors have no potential conflicts of interest to disclose. The authors would like to acknowledge the HERBAXYLAREDD project founded by BELSPO (BR/143/A3/HERBAXYLAREDD) for the opportunity to compile species lists across tropical Africa, and to the SEOSAW project founded by NERC (NE/P008755/1) for the founding of plot-based data compilation in Southern Africa, and meetings organization. Acknowledgements are also due to several monitoring programmes in protected areas and research projects including the ErA Net BiodivERsA CoForChange project (http://www.coforchange.fr), the ACES project (NE/K010395/1, https://miomboaces.wordpress.com/) with support from the Ecosystems Services for Poverty alleviation (ESPA) programme, the SAEON Tree Demography project, the Kruger Experimental Burn Plots and the BMBF-funded projects, including The Future Okavango, SASSCAL and BIOTA Africa for contributing plot data from Angola, Namibia and Botswana. The data from Hluhluwe-iMfolozi Park originate from its Elephant Impact Monitoring programme, funded by the Earthwatch Institute, and originally developed by Dave Balfour and further refined and continued by Susan Janse van Rensburg. We thank Ezemvelo KZN Wildlife and Dave Druce for providing access to this dataset.
© 2018 John Wiley & Sons Ltd
- biogeographical regions
- correspondence analysis
- distance decay in similarity
- floristic clusters
- indicator species
- rainfall and altitude/temperature gradients