“…Instead of class indices derived from the tree species map, the indices based on the CLC dataset (“Corine Land Cover” in the Appendix ) were integrated. While the application of land cover information (e.g., Amini Tehrani et al., 2020 ; Bandara et al., 2022 ; Bellamy et al., 2020 ; Cable et al., 2021 ; Luo et al., 2020 ; Thomas et al., 2021 ; Wright et al., 2021 ) and/or landscape information via Fragstats or the R package landscape metrics is common in habitat modeling (e.g., Cable et al., 2021 ; Neubaum & Aagaard, 2022 ; Thomas et al., 2021 ), the highest resolution for forest classes typically extends no further than the categories provided by CLC, namely broadleaf, coniferous, or mixed forest (e.g., Amini Tehrani et al., 2020 ; Andersen et al., 2022 ; Bandara et al., 2022 ; Cable et al., 2021 ; Thomas et al., 2021 ; True et al., 2021 ). Therefore, the GVM approach incorporates all indices derived from the CLC dataset, amounting to 133 indices (compared to 265 for the tree species map variables with six classes for forest).…”