There is strong evidence that the geographic distribution of many species is changing under anthropogenic climate change. Climatic niche models are widely used to predict range shifts, but, as many factors not captured by these models can co-determine a species’ distribution, the accuracy of such forecasts is debated. In a study published in the Proceedings of the National Academy of Science USA, Brunno Oliveira and colleagues tested niche-based predictions against empirical data. Drawing from databases of documented range shifts, they were able to build ensemble climatic niche models for more than 3,500 species of plants and animals. Model predictions and the empirical observations generally agree on the direction of the range shifts, but the former are often inaccurate at predicting the magnitude of change. Niche-based models generally tend to underestimate the pace of species’ redistribution (62% of the cases), and the median documented range shift rate is about 4 times greater than predicted. This might be surprising, as experts often warn that niche-based forecasts risk overestimating range shifts, whereas these findings suggest that they tend to be too conservative; however, the article also reports a substantial proportion of slower-than-expected range shifts (38% of the documented cases). Oliveira and colleagues found that the discrepancies between models and observations depend partly on ecological factors such as dispersal ability and landscape connectivity, and partly on methodological aspects such as duration of the study and spatial resolution of the climate variables. This suggests that increasing model realism and validating predictions against long-term monitoring data are both necessary.
Original reference: Proc. Natl Acad. Sci. USA 123, e2515903123 (2026)
This is a preview of subscription content, access via your institution
