SIPPOM-WOSR: A Simulator for Integrated Pathogen POpulation Management of phoma stem canker on Winter OilSeed Rape: II. Sensitivity analysis

SIPPOM, a simulator for integrated pathogen population management, has been developed to assess and rank Integrated crop management (ICM) strategies, at the regional scale. The input variables are weather data, soil characteristics, the description of cropping systems (crop sequence and winter oilseed rape crop management) and their spatial distribution, plus the initial size and genetic structure of pathogen populations. Here, we use SIPPOM to simulate phoma stem canker severity, the genetic structure of the pathogen populations, and the yield loss caused by the disease. Sensitivity analysis is conducted to quantify how strongly state variables (sub-model output variables) respond to variations in parameters. The results indicate which parameters need to be more accurately estimated, and it elucidates the steadiness of the rankings of contrasting control strategies under various weather conditions when parameters were varied. Due to the complexity of SIPPOM, the scope of this work was limited to a sensitivity analysis of each sub-model independently. Three values of each parameter were tested under various environmental conditions and crop management according to their expected or known effects on disease and yield. Qualitatively speaking, variations in input variables and parameters provided sub-model output variables that behaved as expected by experts. Parameters with the greatest effect on state variables and that need to be estimated more accurately are for instance those related to pseudothecia maturation and disease severity index estimates. Improvements are foreseen (e.g., the calculation of both the number of phoma leaf spots and the severity disease index). Because the ranking of the simulated control strategies remained steady, despite large variations in the simulated variables linked with variations in parameters, the sensitivity analysis shows that the model, as it stands, can be used to compare and rank ICM strategies with respect to their effectiveness. Possibilities of a sensitivity analysis of the overall model are discussed.