On the Importance of Expert‐Informed Variable Selection in Species Distribution Modelling

ABSTRACT Aim Species Distribution Models (SDMs) are widely used to map species distribution and predict future responses to climate change. While the risk of selecting biologically non‐relevant predictors has been raised, numerous studies continue to fully rely on statistical approaches to automatically select predictors solely focusing on climate variables. Yet, relying only on climate can have important implications for the reliability of the estimated relationships and predictions. Here, we illustrate the implications of ignoring species biology when modelling species distribution. Location West Palearctic. Taxon Boreus westwoodi Hagen, Crucianella maritima L., Rhinolophus hipposideros (Bechstein). Methods We compared a naive, statistically‐informed model selection approach (relying on the full set of 19 bioclimatic variables) with expert‐informed models where we included only variables specifically linked to a species' biology. We tested our approach on three case study species: B. westwoodi , a Palaearctic insect related to snow cover; C. maritima , a Mediterranean coastal dune plant; and R. hipposideros , a forest bat occurring across Europe and the Middle East. Results The two variable selection approaches identified different sets of variables as relevant in all case studies; although the models yielded similar predictive performance. This resulted in consistent differences between predictions under current and novel conditions, along with differences in interpretability of response curves, driven by variables with no clear ecological significance. Main Conclusions We emphasise the importance of a thoughtful selection of predictors based on species biology over naive bioclimatic models. We argue that taxonomy, general adaptations and biome characteristics can guide variable selection when species‐specific information is lacking and that a rational exclusion of non‐relevant variables is often feasible even with limited knowledge. Strengthening the ecological foundation of SDMs will enhance their utility in climate change impact assessments and support biodiversity conservation efforts.