Comparison of methods to model species habitat networks for decision-making in nature conservation: The case of the wildcat in southern Belgium

Facing the loss of biodiversity caused by landscape fragmentation, implementation of ecological networks to connect habitats is an important biodiversity conservation issue. It is necessary to develop easily reproducible methods to identify and prioritize actions to maintain or restore ecological corridors. To date, several competing methods are used with recurrent debate on which is best and if expert-based approaches can replace data-driven models. We compared three methods: knowledge-driven (expert based), data-driven (based on species distribution model), and a mixed approach. We quantified their differences in habitat and corridor mapping, and prioritizations of landscape elements in terms of importance for connectivity. Key parameters generating these differences were identified. To put this into practice, the case study of the wildcat (Felis silvestris Schreber, 1777) was chosen. The results highlighted differences and similarities between approaches used. The data-driven approach was more successful in identifying the suitable habitat with regard to wildcat ecology, while the knowledge-driven approach was better able to account for obstacles to wildcat movements in the landscape matrix. However, these two methods converged for the identification of patterns of habitat patches and corridors that are important for global landscape connectivity. For both methods, we identified adjustments that can improve the outcome. The mixed approach largely differed in that it required more inputs to be performed. In the end, conservation actions were identified and could guide nature conservation practitioners in their efforts to restore landscape connectivity.