Résumé
Species distribution changes associated with human-mediated climate change have important consequences on ecosystems and human well-being. Despite mounting evidence, our knowledge of biodiversity redistribution is still incomplete with only 0.6% of the known species for which distribution changes have been documented. Contrary to former meta-analyses on biodiversity redistribution, I will show the extreme variability in the magnitude and direction at which species’ range positions (e.g., the cold/leading edge, the centroid of the range or the warm/trailing edge) are shifting over time. Scientists often use niche models based on where species lived historically to forecast changes in species distributions, but it remains unclear how well these forecasts align with observed changes. By comparing predictions from niche-based models to actual observations of over 9,500 reports of species distribution changes across land and sea, I will show that while niche models tend to get the direction right, especially for marine species, they often fail to predict the pace of species redistribution. One potential reason behind these discrepancies involves the lack of consideration for microclimatic processes when training niche models. Focusing on forest ecosystems, I will finally demonstrate that niche models trained on microclimates can outperform traditional niche models trained solely on macroclimatic grids.
