The rapid decline of soil fertility of cultivated lands in the sub-Saharan savannas of West Africa is considered to be the main cause of the increasingly severe constraints of food production. The soils in this tropical area are highly fragile, and crop yields are limited by characteristically low levels of available phosphorus. Under such preconditions, the multiple benefits of the arbuscular mycorrhizal (AM) symbiosis are likely to play a pivotal role for maintaining natural soil fertility by enhancing plant nutrient use efficiency, plant health, and stabilization of a favorable soil structure. Thus, it is important to explore the impact of the commonly applied farming practices on the native AM fungal community. In the present study, we determined the AM fungal species composition in three ecological zones differing by an increasingly prolonged dry season from South to North, from the Southern Guinea Savanna (SG), to the Northern Guinea Savanna (NG), to the Sudan Savanna (SU). In each zone, four "natural" and four "cultivated" sites were selected. "Natural" sites were three natural forest savannas (at least 25-30 years old) and a long-term fallow (6-7 years old). "Cultivated" sites comprised a field with yam (Dioscorea spp.) established during the first year after forest clearance, a field under mixed cropping with maize (Zea mays) and peanut (Arachis hypogaea), a field under peanut, and a field under cotton (Gossypium hirsutum) which was the most intensively managed crop. Soil samples were collected towards the end of the wet season in each zone. AM fungal spores were extracted and morphologically identified. Soil subsamples were used to inoculate AM fungal trap cultures using Stylosanthes guianensis and Brachiaria humidicola as host plants to monitor AM root colonization and spore formation over 10 and 24 months, respectively. A total of 60 AM fungal species were detected, with only seven species sporulating in the trap cultures. Spore density and species richness were generally higher in the natural savannas and under yam than at the other cultivated sites and lowest under the intensively managed cotton. In the fallows, species richness was intermediate, indicating that the high richness of the natural savannas was not restored. Surprisingly, higher species richness was observed in the SU than in the SG and NG, mainly due to a high proportion of species in the Gigasporaceae, Acaulosporaceae, and Glomeraceae. We conclude that the West African savannas contain a high natural AM fungal species richness, but that this natural richness is significantly affected by the common agricultural land use practices and appears not to be quickly restored by fallow.