Fokker-Planck model for movement of the carabid beetle Pterostichus melanarius in arable land: Model selection and parameterization

Carabid beetles in arable land move between different habitats to exploit resources that vary in time and space. Understanding such movement is key to explaining how the pest control function of carabids in arable crop lands depends on the spatial configuration of crop fields and associated semi-natural habitats, but movement at and beyond field scale is not well understood. Here we use a model selection framework to identify and parameterize a parsimonious movement model, based on mark-release-recapture data in two adjacent arable crop fields, one planted with rye, and the other with oil radish. The simplest model assumes motility of beetles to be the same in the two crops, and it does not consider losses of beetles over time due to death or mark loss. These assumptions are relaxed either separately or together in competing models, resulting in a comparison between four models. All models consider the effect of spatially heterogeneous pitfall trapping on the size of the moving population. Models were fitted to data with Poisson likelihood, and Akaike’s information criterion was then used to rank the models. The model selection showed that including a parameter for loss of beetles due to mortality or mark loss resulted in the best approximation of the observed data. The data did not support the assumption of different motility between the two crops. We conclude that our extended model can be used to simulate beetle recapture in mark-release-recapture experiments, but further refinements to the model are needed. The inverse modeling framework for model identification and parameter estimation that was applied in this study proved effective to select the most promising model and parameter values.