Résumé
NO emission rates from soils were measured for twelve major African ecosystems in four countries (Congo, Niger, Ivory Coast, and South Africa) and within four major phytogeographic domains: the Guineo-Congolese, Guinean, Sahelian, and Zambezian domains. Measurements were performed during wet and/or dry seasons. All the measurements were made with the same dynamic chamber device, which allowed true comparisons to be made. This study showed that emission rates strongly differed between ecosystems and exhibited a marked temporal variability. Ecosystem effect was highly significant during both the dry and wet seasons. Emission rates were low (<0.6 ng NO-N m(-2) s(-1)) in Hyparrhenia and Loudetia savannas of the Guinean or Guineo-Congolese domains. Intermediate NO fluxes were obtained in rain forest and gallery forest ecosystems, in a broad-leafed savanna and in a seasonally wetted grassland (sandy soil) of the Zambezian domain, and in a dry fallow savanna of the Sahelian domain. Emission rates were maximum (>7 ng NO-N m(-2) s(-1)) in a seasonally wetted grassland (site 2) and in particular sites subjected to various disturbances, for example soil fauna activity (termite mounds) or past human disturbance (Acacia patches-settlement site). Microbial activity potentials (i.e., carbon mineralization, nitrification, denitrification, and total net N mineralization) were determined for most of the soils where NO fluxes were measured. In some sites, these potential activities were useful to identify the major processes controlling NO emission rates. Denitrification potential was very low and could not explain substantial NO fluxes from broad-and fine-leafed savannas and Hyperthelia savannas of the Zambezian domain. Very low potentials of both nitrification and denitrification could be related to the low NO fluxes for the three Guinean savanna sites studied. NO fluxes were significantly higher during the wet season than the dry season in both savanna and forest ecosystems. Emission rates in savanna ecosystems were significantly increased within a few hours after fire. The measurements presented here provide a unique, consistent database which can be used to further analyze the processes involved in the spatial and temporal variations of NO emissions.