Davoli, M.; Jung, M.; Visconti, P.; Rondinini, C.; D'Alessio, A.; Pacifici, M.

Ensuring that species of conservation concern achieve favorable conservation status (FCS) is central to European Union (EU) biodiversity conservation targets. A key criterion for FCS is exceeding the favorable reference range (FRR), i.e., the range extent needed for long-term species stability and full ecological variation. However, due to data limitations, FRRs are often unknown, undermining their applicability. We developed a machine-learning approach to estimate and standardize FRRs across the EU. Applied to amphibians, mammals, and reptiles, our method provided FRRs for 99.5% of species of conservation concern, compared to 17.5% previously available (with satisfactory modelling performance: r-square 0.75). We reassessed conservation status using the estimated FRRs, finding that species in FCS (34.8%) are notably fewer than reported in official documentation (69.1%). The average proportional distance to FRR for species in unfavorable conservation status is -64.4%. Our approach may support periodic FCS reassessments and help refine the targets of EU conservation policies.