Abstract
Aim To test the hypothesis that macroscale environmental variables explain local taxonomic richness of stream macroinvertebrates, and then to identify the relationships between these variables and benthic fauna richness in North Africa. Location North Africa, from West Morocco to East Tunisia. Methods Using a large‐scale data base made of 211 sampling sites gathered from an area of 500,000 km 2 , an artificial neural network model has been built to identify and predict the influence of macroscale environmental variables on local macroinvertebrate richness. Results The correlation coefficient ( r ) between observed and estimated taxon richness values was 0.75 ( P r 2 = 0.563) of the macroinvertebrate richness variation. Macroinvertebrate richness was, therefore, accurately predicted using only three environmental variables accounting for hydrology (number of rainy days), geographical factors, i.e. connections between European and North African land masses (longitude) and climatic gradient (altitude). Main conclusions Stream macroinvertebrate richness in North Africa results from a combination of climatic, geographical and hydrological parameters. Although consistent with current biogeographic and ecological concepts mainly derived from European and North American streams, this study underlines the specificity of dry Mediterranean ecosystems. The shape of diversity patterns results from climatic conditions and their associated environmental seasonal dynamics, which screens geographical processes.