Climatic and land-use variables, present and future scenarios
To avoid possible mismatches between observations and climatic variables, we excluded from the analysis the records before 1970 and those without date (time range: 1970-2018, but for konradini1960-2020 due to the few data available). Three different categories of possible environmental drivers were considered: climate, topography, and land-use/land-cover (LULC). Climatic variables were derived from the database CHELSA V2.1 (Karger et al. 2017, 2021), and were the following ones: mean annual 2-m air temperature, annual range in 2-m air temperature, sum of annual precipitation, precipitation seasonality (Thuiller et al. 2019), all calculated for the period 1981–2010. Topographic variables were computed starting from a fine-scale digital elevation model (25 m-resolution; EU-DEM v1.0, publicly available from the European Environment Agency). Finally, LULC variables were obtained from the CORINE land cover map (Corine Land Cover — European Environment Agency , 2018). All variables were then estimated for 1 × 1 km2 cells, as average values (climate and topograhy), or as proportional cover (LULC). When needed (climatic variables), raster resampling was carried out by bilinear interpolation. LULC categories with negligible cover were excluded, while some other categories poorly represented were merged (Supporting text A2). The variables so worked out showed relatively modest correlations (r < |.7|; Grimmett et al. 2020).
To describe possible alternative future climates on the medium-term, we relied on the downscaled CMIP6 (Coupled Model Intercomparison Project Phase 6) data, choosing the period 2041–2070, and two alternative climate models (a ‘warmer’ one and a ‘colder’ alternative) as provided by ISIMIP (Intersectoral Impact model Intercomparison Project; Warszawski et al., 2014): GFDL-ESM4 and UKESM1-0-LL. Those data are tailored for such a kind of application. For both climate models, we picked the ‘worst case’ scenario SSP585 and the moderate change one SSP370 (Eyring et al. 2016). Therefore, we based our assessment on four alternative climatic conditions for the future, based on the combination between two very different climate models and on two different scenarios. Also those data were retrieved from the CHELSA V 2.1 database.