Late Pleistocene temperature patterns in the Western Palearctic : insights from rodent associations compared with general circulation models

Abstract

Since rodent fossils are preserved in many low- and high-latitude archaeological and paleontological sites from a wide variety of environments, their associations are a commonly useful proxy for inferring past local climate and environmental conditions. Such a frequent and widespread geographic distribution can help us to better understand past climate evolution by providing access to high spatiotemporal resolution at large geographical scales. The aim of this paper is to develop an approach to generate continental-scale temperature maps based on rodent associations and to assess their reliability compared to state-of-the-art general circulation models (GCMs). We used the Bioclimatic Analysis, based on fossil and modern rodent associations, to infer climate zone distribution and local temperatures (mean annual temperature, mean temperature of the warmest month and mean temperature of the coldest month), at the Western Palearctic (Europe, Middle East and North Africa) for six different periods: the Last Glacial Maximum (LGM), Heinrich Stadial, Bølling, Allerød, Younger Dryas and present-day conditions. The Bioclimatic Analysis is combined with a spatial generalized linear mixed model to interpolate these surface temperatures across the Western Palearctic. We show that the spatial patterns in mean annual temperature and mean temperature of the warmest and coldest months are very similar between our interpolations and GCMs for both present-day and LGM conditions, but the rodent-based approach provides slightly cooler LGM estimations in western Europe and warmer ones in eastern Europe. Throughout the Late Glacial oscillations, the rodent-based model infers globally small variations in mean annual temperature and mean temperature of the warmest months and slightly larger changes in mean temperature of the coldest months. Nonetheless, some events show weak but significant regional variations depending of the events and the climate variable. For instance, the most important shifts in mean annual temperature between the Allerød and Younger Dryas are observed in northwestern regions. Northeastern regions, on the other hand, experienced relatively stable mean annual temperature, although they did experience considerable warming of the warmest month and cooling of the coldest month. Minor discrepancies appear between GCMs and the rodent-based model, the latter showing colder temperature in northwestern Europe and hence a differential west–east gradient in ice-sheet influence. Our results demonstrate that rodent associations are robust proxies for reconstructing and regionalizing past temperatures at broad scales, offering a readily reproducible approach to be reimplemented in future studies incorporating new rodent data.