• Italy
• Maxent
• roost site
• Stone-curlew
• wintering

A thorough knowledge of winter ecology is important for an effective conservation of bird species, and in order to understand how they would cope with ongoing changes due to global warming. This is particularly relevant for short-distance migrants, whose migration strategies could be rather flexible. In recent years these migrants have been observed to change their seasonal movements both in timing and distances. These changes could be closely related to their overwintering strategies. The Stone-curlew is a short-distance migrant whose winter ecology is poorly studied and almost no data are available regarding the way these birds select their roost sites during the non-breeding season. The aim of this study was to identify the environmental factors affecting Stone- curlews roost choice in an area of the Grosseto province at three different space-use scale of the species (local, sight-field and foraging scale). The broader spatial scale considered was derived from the length of foraging flights, between roost and foraging areas, identified by means of 19,295 records, from eight GPS tagged individuals. Stone- curlew surveys provided roost locations. Roost sites were used to develop three models for roost presence at the three spatial scales, using the Maximum Entropy algorithm (MaxEnt), a presence-only data approach. A bias file for modelling the sampling effort in the study area, by fitting a kernel density estimate (KDE) to the roost locations, was calculated and used as probability grid for background points sampling. Land cover, topography variables, farms proximity, hunting disturbance and road density were chosen as predictors. The model at the local scale was the only reliable model emerged and it indicated seven variables as the most important for Stone-curlew winter roost selection: slope, road density and the cover of five land cover categories. The accuracy of the models for the other two spatial scale was not different from a random prediction. Identifying features at different space-use scale which determine winter roosts presence could be essential to improve the species monitoring as well as to help the drawing of habitat management plans addressed to the species usually associated with farmland landscapes, which are among the most threatened species in Europe.Identifying features at different space-use scale which determine winter roosts presence could be essential to improve the species monitoring as well as to help the drawing of habitat management plans addressed to the species usually associated with farmland landscapes, which are among the most threatened species in Europe.