The economics of land use reveals a selection bias in tree species distribution models

Aims In human-dominated ecosystems, the presence of a given species is the result of both the ecological suitability of the site and human impacts such as land-use choices. The influence of land-use choices on the predictions of species distribution models (SDMs) has, however, been often neglected. Here, we provide a theoretical analysis of the land-use selection bias affecting classical SDMs in the case of either presence-only or presence–absence datasets. Land-use selection bias in the predictions of SDMs is then quantified for four widespread European tree species. Location Continental France. Methods We describe a bivariate selection model (BSM) that estimates simultaneously the economics of land-use choices and species responses to bioclimatic variables. The land-use equation, based on an econometric model of landowner choices, is joined to an equation of species responses to bioclimatic variables. Results We found a significant land-use selection bias in all the species studied. The sign and the magnitude of the bias varied among species and were strongly related to the type of dataset used in the SDM calibration (presence-only or presence–absence). In addition, the BSM estimates the spatial covariance between the probability of presence and the presence of compatible land use. We found that, depending on the species, sites with high ecological suitability could present a high probability of compatible land use (positive covariance) or a low probability (negative covariance). Main conclusion We showed that the use of classical SDMs in human-dominated areas can lead to strong miss-estimations of actual species distributions and could therefore prevents sound projections of the effects of climate change. The proposed BSM represents a crucial step to account for the economic forces shaping species distribution in anthropized areas and paves the way for a direct assessment of trade-offs and opportunities that may arise in a context of global change.