Abstract
Aim In a rapidly changing climate, the assumption of a species‐environment equilibrium is increasingly strained. Yet, this equilibrium remains a central assumption of species distribution modelling. Species distributions in disequilibrium with their environment may take different forms depending on past species distribution changes, but for most species, these past changes are unknown. Revealing this uncertainty is essential when constructing and interpreting species distribution models. Location Europe. Taxon Virtual species and bird species. Methods We modelled distributions of virtual species exhibiting different past distribution changes based on simulated presences‐absences records using six models that varied in how climate data were assigned to species records. The first set of models used a fixed climate period ending in either 1979, 2000, or 2020. An additional model matched the climate data with the year of observation for each species record. We further applied the fixed period 1979 model and the matched‐climate model to six bird species to explore real‐world implications. Results For the virtual species, the 1979 model best estimated those niche edges corresponding to range edges not tracking climate change, while the matched‐climate model best estimated the edges that were tracking climate change. The union or intersection of both models encompasses or is encompassed by the niche of a species respectively, even when both models are applied to the bird species with an unknown niche. Main Conclusions A complementary approach using both the fixed‐period climate model and the matched‐climate model reveals the uncertainty in SDM predictions that arises from unknown past species distribution change. Engaging explicitly with this uncertainty can aid the interpretation of SDMs and the implied species niche. Our analysis also revealed that current common practice might lead to inaccurate niche estimates due to the indiscriminate use of standard historical climate periods.