Using vegetation attributes to rapidly assess degradation of East African wetlands

Wetlands in East Africa harbor a large biodiversity and provide diverse ecosystem services. Also, wetlands are highly suitable for crop production due to generally fertile soils and water availability. As a result of rising demand for cropland that is driven by population growth, degradation of upland areas and changing food demand patterns, wetlands are increasingly used for agricultural production. Such land use changes can negatively affect biodiversity and the provision of ecosystem services, especially when formerly little-disturbed wetlands are converted into croplands. For evaluating wetlands and developing land use or conservation strategies, we require methods to assess disturbance and degradation of wetlands. For instance, many indices of biological integrity for wetlands using vegetation attributes have been developed especially in the USA, but to date, no comparable assessment schemes are available for East Africa. To develop such a scheme, we sampled four different wetland localities, covering both floodplain and inland valley wetlands in Kenya, Rwanda, Tanzania and Uganda. A total of 198 wetland plots were characterized regarding land use and evaluated in terms of geomorphology, hydrology, vegetation and water quality regarding their deviation from a theoretical natural reference state without human disturbances and given respective impact scores using the WET-Health approach. Additionally, nine variables characterizing vegetation attributes were recorded. The impact scores were rescaled from 0 to 1 and used as response variables in multiple linear regression models. The predictor variables were selected for each regression model out of the nine vegetation attributes in a stepwise process. The regression models for the different response variables differed strongly regarding their accuracy. The root mean square prediction error (RMSPE) ranged between 0.14 for the “vegetation disturbance” and 0.27 for the “hydrological modification” regression model. However, we conclude that vegetation attributes such as “absolute cover of perennial species” or “average height of vegetation” are generally useful to estimate anthropogenic impacts on East African wetlands as assessed with the WET-Health approach. The presented approach can hence be a pragmatic addition to the WET-Health approach as a cost-efficient and rapidly applicable method. Furthermore, it may also be useful for ex-post application to historic data and vegetation surveys.