Evaluation of the ability of the crop model STICS to recommend nitrogen fertilisation rates according to agro-environmental criteria

The use of crop models for nitrogen fertiliser management raises several issues. A first problem is to define suitable criteria for optimising nitrogen fertilisation in function of economic, quality and environmental objectives. A second issue is to assess the capacity of the crop model to predict the variables involved in the calculation of the criteria such as grain yield, grain protein content, residual soil mineral nitrogen or nitrogen balance. A third issue is to evaluate the results obtained by applying the decision rules selected by the crop model. The three problems are considered in this paper in the case of winter wheat and the STICS model. Fourteen field experiments with various N fertilisation strategies were used for evaluating the model. STICS predicted grain yield and nitrogen balance more accurately than protein content and soil mineral N at harvest. Among the eight criteria tested for optimising N fertilisation, those using a maximal threshold on nitrogen balance seem to be the most valuable for satisfying agricultural and environmental objectives. Under conditions of environmental constraint, STICS was more efficient than the reference method (AZOBIL) at selecting the optimal nitrogen fertilisation scenario.