Continental scale patterns of biodiversity: can higher taxa accurately predict African plant distributions?

The distribution maps of a total of 3563 species, which represent 8.9% of the known African angiosperm. flora, were entered into cells representing a one-degree latitude-longitude grid of Sub-Saharan Africa. The computer programme WORLDMAP was used to explore continental scale patterns of biodiversity. The maps were used to assess the use of higher taxa as a surrogate measure for predicting patterns of species richness. Genera were found to predict species richness distributions most closely, with higher taxa (families, orders, subclasses) exhibiting progressively worse correlations. However in two areas, the Cape Region of South Africa and coastal Cameroon, there was a higher species to genus ratio than in other areas of Africa. In the Cape Region this meant that generic richness failed to predict species richness. Hotspots, defined as the 5% of grid cells with the highest scores for richness and range-size rarity, were identified for species and higher taxa. Whilst a high percentage of species richness hotspots were predicted by higher taxa, there were important exceptions like the Cape Region. Species range-size rarity hotspots were not well predicted by higher taxa. Hotspots of higher taxa (families and orders) do not therefore accurately predict the location of species hotspots. Higher taxa appear to provide a powerful and accurate tool that can be used to predict large scale patterns of species biodiversity in Sub-Saharan Africa. However care must be taken when using taxa higher than genera, especially if selecting areas of highest conservation priority. The special case of the Cape Region indicates the danger of extending predictive generalizations as the ecological mechanisms that promote and retain species may not be the same in all places. (C) 2002 The Linnean Society of London.