Stef Haesen , Jonathan Lenoir , Eva Gril, Pieter De Frenne, Jonas J Lembrechts , Martin Kopecký, Martin Macek, Matěj Man, Jan Wild, Koenraad Van Meerbeek

Species’ environmental niches are conventionally modelled using coarse-grained macroclimate data. These data are known to deviate substantially from local, near-ground and proximal conditions (i.e., the microclimate), especially so below forests canopies. Here we aimed to assess the impact of using gridded microclimate data instead of gridded macroclimate data on the performance of species distribution models (SDMs), as well as on the predicted geographical distribution and the derived species response curves of 140 forest specialist plant species across Europe over the 2000-2020 period. We performed a comparative study between SDMs constructed with different sets of bioclimatic predictors to separately test the effect of using (i) proximal climate data instead of conventional macroclimatic data and (ii) high-resolution proximal climate data rather than coarse-gridded macroclimatic data. Therefore, we challenged SDMs with: (1) a macroclimatic dataset at a spatial resolution of 1 km × 1 km; (2) an aggregated microclimatic dataset matching the same resolution of 1 km × 1 km; and (3) a microclimatic dataset at a much finer spatial resolution of 25 m × 25 m. We found significant differences in model performance, indicating that microclimate-based SDMs outperform both their macroclimatic and aggregated counterparts. Most importantly, this study makes clear that macroclimate-based SDMs tend to introduce a systematic bias into the perceived species response curves. Additionally, macroclimatic data is unable to identify warm and cold refugia beyond the range edges of species’ distributions. We thus conclude that microclimate-based SDMs are a crucial tool to gain peculiar insights regarding biodiversity conservation, which is needed to align management actions and prioritize conservation efforts.