The development of alternative pest-control strategies based on the spatial design of cropping systems requires a thorough understanding of the spatial links between the pest and its environment. Mechanistic models, especially individual-based models (IBMs), are powerful tools for integrating key behaviours, such as habitat selection and dispersal, with spatial heterogeneity. In this paper, we used an IBM calibrated and evaluated from real data to represent the spatial dynamics of the banana weevil in relation to the cropping system. We considered crop fragmentation and mass trapping as tools for suppressing pest numbers. Our simulation results showed that manipulating crop residues in the area surrounding each pheromone trap greatly improved trap efficiency. For an intensive banana plantation in fallow, traps were most effective when situated at the transition zone between banana area and fallow so as to maximize the trapping of weevils escaping the fallow. The model also showed that weevil numbers decreased when fragmentation of banana plantations was reduced.