Cross-scale predictions allow the identification of local conservation priorities from atlas data.

For planning practical measures aimed at biodiversity protection, conservation priorities must be identified at a local scale. Unfortunately, identifying local conservation priorities requires high-resolution data on species distribution, and these are often unavailable. Atlases of species distribution provide data for several groups of organisms in many different areas but are often too coarse in resolution to provide valuable information. We explored the possibility of cross-scale modelling species distributions and we clarified, for the first time, its effect on prioritization exercises. We used different modelling techniques for scaling down atlas data for Sardinian reptiles, validated the outcomes with detailed, field-sampled data, and compared conservation priorities deriving from atlas maps and downscaled models. Doing this, we obtained as a further result the identification of priority species and areas for future conservation strategies. Our results encourage us to experiment further with this approach. Through the downscaling procedure, we obtain high-resolution models with strong variations in predictive performances, although most of the models show satisfactory/excellent scores. This testifies that low-resolution data can be downscaled maintaining low rates of omission and commission errors. Increasing the resolution of distribution maps used for prioritization influences the spatial patterns of priority but does not modify the evaluation of species representation. Overall, we show that atlases can meet the large demand for distribution data by decision makers if appropriate downscaling procedures are adopted. In addition, we provide practical instruments for the conservation of reptiles in Sardinia by identifying priority species and areas that require strict management.