Simulating farming practices within a region using a stochastic bio-decisional model: Application to irrigated maize in south-western France

Abstract

Abstract Agricultural is a major contributor to environmental resource management problems. Modelling the distribution of agricultural land use to evaluate current situations or scenarios is an important issue for policy-makers and natural resource managers. A promising approach is the use of bio-decisional models based on decision rules. However, at the regional scale, the large number of farmers makes it difficult to identify decision rules, and the diversity of farmers' decisions is rarely considered. To this end, we developed SIMITKO, a spatialised and stochastic bio-decisional model, able to simulate the spatial and temporal variability in farming practices. We focused on the choice of varietal earliness and sowing practices of maize ( Zea mays L.) in the Baise sector (south-western France). Model development was based on statistical analyses of surveys of farmers’ practices to identify their current strategies, the best variables to describe the practices and the probabilities associated with the values of the variables for each strategy. We tested SIMITKO by simulating the dynamics of areas sown with maize. Comparing model predictions of practices to observed data showed generally good predictions of sowing dynamics but less satisfactory predictions of varietal earliness choices. Possible improvements are suggested.