A two‐stage framework linking wildlife space use and crop damage occurrence in agricultural landscapes

Human–wildlife conflicts in agricultural landscapes are a major challenge for wildlife management and conservation, yet crop damage remain difficult to predict without a mechanistic understanding of both use and damage processes. Here, we modelled the crop damage process as a sequence of two distinct stages (i) wild boar selection of agricultural parcels, estimated using a resource selection function, and (ii) the crop damage occurrence, modelled with a binomial generalised linear model. This decomposition allows us to distinguish non‐damaged parcels that were not visited from those that were visited but not damaged, thus identifying ecological drivers specific to each stage of the damage process. Applied to wild boar ( Sus scrofa ), one of the most damaging species to crops, our results reveal contrasting crop vulnerability profiles emerging from the interaction between relative intensity of use and damage. Maize and legumes appeared as the most sensitive crops, exhibiting high relative damage probability despite low relative intensity of use. In contrast, cereals and oilseed showed intermediate and more variable vulnerability, with damage risk more linked to the relative intensity of use. Meadows were associated with low relative probabilities of both use and damage, indicating low vulnerability. Proximity to forest edges were generally associated with both greater crop selection and higher damage occurrence, although this effect varied depending on crop type. Our models, validated on an independent dataset, demonstrate reasonably good predictive accuracy and enable spatial identification of the most vulnerable parcels. Synthesis and applications . This framework provides a valuable tool for developing targeted mitigation strategies. By classifying crops according to their vulnerability profiles, it enables prioritisation of management actions and development of preventive measures that are tailored to specific crop types and landscape configurations. The approach can also be extended to other species and contexts and can contribute to early‐warning systems to help farmers anticipate and minimise wildlife‐induced crop damage.