Vegetation diversity in East African wetlands

Abstract

Aims: Wetlands in East Africa are important ecosystems for biodiversity conservation and ecosystem service provisioning, yet threatened by degradation and conversion into croplands. Conservation and land use management require data on vegetation structure and dynamics. The presented work is a response to a lacking consistent classification of East African wetland vegetation. Location: Namulonge valley in Uganda and Kilombero floodplain in Tanzania. Methods: We sampled 431 4 m²-plots along land use intensity and flooding duration gradients. A floristic classification using the cocktail method was performed in a two-step approach. We developed definitions for vegetation units, using plot observations from the study sites in a first step and revised them in a second step by adding data from a vegetation-plot database and complied the definitions to an expert system for classification. Resulting vegetation units were analyzed regarding their life form composition, for which we implemented a classification based on life span and growth form. Following a literature review, the identified vegetation units were assigned either to existing phytosociological associations or proposals. Results: We recognize eight units of marsh and reed vegetation (class Phragmito-Magno-Caricetea) and five units of weed and pioneer vegetation under semi-aquatic conditions (class Oryzetea sativae). Five of these associations were previously described in the bibliographic references. The remaining eight are newly described in this work. The associations contrast in their life form composition with the five Oryzetea sativae associations dominated by obligate annuals and the Phragmito-Magno-Caricetea associations dominated by either reed plants or lacking a dominating life form. Conclusions: The developed expert system enables a comparison of wetland vegetation in the East African region and will support vegetation science and informed decision making about land use management and conservation. The two-step approach of revising a classification developed for single wetlands with a database is promising for data-scarce regions. Nomenclature: Haines & Lye (1983); CJBG & SANBI (2012); The Plant List (2013); TNRS (2018). Abbreviations: DCA = Detrended Correspondence Analysis; DRC = Democratic Republic of the Congo; MRPP = Multiple Response Permutation Procedure.