Nitrification, the bacterial oxidation of ammonium to nitrate, is a major process of the nitrogen cycle in most soils and sediments. The lack of nitrification is generally linked with low nutrient availability and is supposed to improve nitrogen conservation by suppressing denitrification and limiting mineral nutrient leaching. Wet tropical savanna soils are typical examples of ecosystems characterised by low nutrient availability and low nitrification capacities. They generally support high plant productivity despite the low nutrient availability, suggesting efficient and conservative nutrient recycling (Abbadie et al. 1992). The wet savanna of the Lamto reserve, West African reference site, is representative of these ecosystems where nitrogen has been shown to be the major factor limiting primary production. Previous studies conducted at Lamto suggest that the expression of the nitrification potential is related to the specific composition of the vegetation cover (Lensi et al. 1992). The grass savanna is characterised by two Poaceae dominating facies: Loudetia simplex and Hyparrhenia diplandra (Menaut & César 1979). Among these facies, the sites dominated by L. simplex generally display a significant nitrification potential while the sites dominated by H. diplandra display extremely low nitrification potential (Lensi et al. 1992). Moreover, inside the H. diplandra sites, the nitrification potential was significantly higher between than under the tufts. An allelopathic hypothesis based on a specific inhibitory effect of nitrification by H. diplandra, was proposed by Meiklejohn (1962, 1968), Munro (1966) and Lensi et al. (1992). Recently, this hypothesis was questioned by the fortuitous discovery of a site dominated by H. diplandra which displayed high rates of nitrification (Le Roux et al. 1995). This observation needs to be confirmed by a systematic experimental design in the soil sampling and nitrification rate assays. The first objective of this work (Lata et al. submitted) was to confirm that one of the sites dominated by H. diplandra effectively displays a high potential of nitrification when compared to standard low nitrification potential sites. The second objective was to investigate if the intensity of nitrification in soil is a permanent (or at least a long-term) characteristic of each site. In order to answer this 2 question, we checked if H. diplandra from the two sites exhibit stable differences related to their nitrate nutrition. Consequently, we assayed nitrate reductase activity (NRA) in plants cultivated under identical conditions from seeds sampled in both sites. The third objective was to investigate if differences in nitrification rate, by changing the nitrate/ammonium ratio and the ecosystem nutrient potential losses, would have significant effects on several H. diplandra characteristics, such as productivity, structure, phenology and physiology. We consequently measured in situ production and phenotypical traits of H. diplandra. The fourth objective was to elucidate the hypothesis of inhibition of nitrification by H. diplandra: in situ construction of crossed plots were used to see if subpopulations of H. diplandra originating from the two sites were able to display a different inhibition capacity, that should explains the differences in nitrification potentials between the two sites.