Evaluating proxy-based vegetation reconstructions against model-based approaches: A case study from Europe during the late Last Glacial period

In recent years, there have been considerable improvements in reconstructing past environments. However, the majority of studies focuses either on the Holocene period (ca. 12 ka BP until present) or the Last Glacial Maximum (LGM, ca. 21 ka BP). Since going further back in time encompasses additional challenges, we aim to assess the capabilities and robustness of methods that are currently in use to reconstruct the paleovegetation during the late Last Glacial period (ca. 60–20 ka BP). We compare four different methods of reconstructing past vegetation cover in Europe during the Last Glacial and highlight the strengths and limitations of each method: 1) The classical biomisation approach using fossil pollen data that assigns taxa into plant functional types (PFTs) and PFTs into biomes based on ecological traits and climatic preferences. 2) The REVEALS (Regional Vegetation Abundance from Large Sites) algorithm, which utilizes fossil pollen data in conjunction with taxon-specific parameters (e.g., relative pollen productivity) to estimate the regional plant cover. 3) The dominant biomes derived from the Biome4 global vegetation model using bioclimatic variables from a global climate model output (HadAMH3 and HadCM3). 4) A dedicated vegetation model that statistically reconstructs land-cover from the output of a global climate model (HadAMH3 and HadCM3) using the present climate-vegetation relationship and a CO₂ correction factor. Our results show that all methods reconstruct a glacial vegetation dominated by open landscapes (e.g., tundra and steppes) and coniferous forests to various degrees. The existence of transient local patches of mixed and temperate forests is consistent with the general interpretation of glacial landscapes in Europe in the literature. However, regional and chronological discrepancies as well methodological challenges render it difficult to decipher which method most closely represents the actual paleovegetation. Nonetheless, exhausting qualitative and quantitative comparisons across different methods using different approaches allow us to limit the ecological range of the potential vegetation. Such a better comprehension of glacial environments has major implications for our understanding of human (Neanderthals and anatomically modern humans) and faunal population dynamics of in Europe, particularly in response to climatic transitions.