Land-use practices aiming at increasing agro-ecosystem sustainability, e.g. no-till systems and use of temporary grasslands, have been developed in cropping areas, but their environmental benefits could be counterbalanced by increased N2O emissions produced, in particular during denitrification. Modelling denitrification in this context is thus of major importance. However, to what extent can changes in denitrification be predicted by representing the denitrifying community as a black box, i.e. without an adequate representation of the biological characteristics (abundance and composition) of this community, remains unclear. We analysed the effect of changes in land uses on denitrifiers for two different agricultural systems: (i) crop/grassland conversion and (ii) cessation/application of tillage. We surveyed potential denitrification (PD), the abundance and genetic structure of denitrifiers (nitrite reducers), and soil environmental conditions. N 2O emissions were also measured during periods of several days on control plots. Time-integrated N 2O emissions and PD were well correlated among all control plots. Changes in PD were partly due to changes in denitrifier abundance but were not related to changes in the structure of the denitrifier community. Using multiple regression analysis, we showed that changes in PD were more related to changes in soil environmental conditions than in denitrifier abundance. Soil organic carbon explained 81% of the variance observed for PD at the crop/temporary grassland site, whereas soil organic carbon, water-filled pore space and nitrate explained 92% of PD variance at the till/no-till site, without any residual effect of denitrifier abundance. Soil environmental conditions influenced PD by modifying the specific activity of denitrifiers, and to a lesser extent by promoting a build-up of denitrifiers. Our results show that an accurate simulation of carbon, oxygen and nitrate availability to denitrifiers is more important than an accurate simulation of denitrifier abundance and community structure to adequately understand and predict changes in PD in response to land-use changes.
