{Modelling the environmental niche of a declining farmland bird species}

Species distribution models are used increasingly in both applied and theoretical research to understand attributes of species' environmental requirements and to predict how species are distributed. However, recent advances in species distribution modelling have shown that the use of a single modelling method can lead to biased conclusions about the determinants of species' distributions. Here we illustrate the application of an alternative modelling framework whereby multiple methods are employed for both making inference and predicting species distributions.We used six modelling techniques to quantify the environmental niche of the Red-backed shrike Lanius collurio, a threatened farmland bird species. Our results show that, while the importance and the direction of the effect of the enviromental variables are broadly concordant with what is known about habitat selection in the Red-backed shrike, there was variation in the predictive performance and the variables judged as important by the models. We conclude that the use of an analytical framework based on multiple modelling methods could be useful, not only to highlight the uncertainties derived from the use of a single best model, but also to make more robust inferences on species distributions and make more accurate predictions.