Search for a command to run...
            Pollen data are among the most abundant and spatially-temporally resolved proxies for quantitatively reconstructing European climate evolution since the Last Glacial Maximum (LGM, ~19-26 kyr, [1]). This period of near-climatic stability, characterized by a climate drastically different from the present, serves as a key reference point for evaluating the reliability/performance of climate models used to project anthropogenic climate change, especially in Paleoclimate Modelling Intercomparison Project (PMIP) ([2]) Pollen assemblages are particularly useful for studying climatic spatial gradients, the proximity effects of ice sheets, seasonality, and environmental changes (e.g., [3]). However, the number of available fossil sites, the quality of the records (i.e., diversity and taxonomic resolution), and the age constraints remain limited. In Europe, various pollen-based transfer functions have been used to provide climate reconstructions (e.g., Modern Analogue Technique (MAT) and Weighted Averaging Partial Least Squares (WAPLs); [4]), each depending differently on sampled modern analog climates. The lack of agreement between inverse methods necessitates the development of new transfer approaches to enable more robust and reliable reconstructions.            Here, we present a new compilation of pollen sequences and revised age models for Europe ([5]) as well as a synthesis of reconstructions based on the most comprehensive European calibration dataset available to date (~8700 spectra): EMPD2 (Eurasian Modern Pollen Database, [6]). Temperatures and precipitation reconstructed using MAT and WAPLs are compared with outputs from the probabilistic CREST method ([7]), which is applied for the first time in Europe. In the studied areas, we demonstrate the value of the Plant Functional Type (PFT) biomization method ([8]) and so-called megabiomization approach ([9]) for quantifying coherent and large-scale environmental and climatic changes. By comparing outputs from these different approaches, we show that CREST (i) is particularly sensitive to detailed pollen (i.e., species), (ii) performs better with more taxonomically detailed pollen data, and (iii) is less dependent on the availability of modern analogues than MAT and WAPLs. Furthermore, reconstructions from CREST exhibit less abrupt variability than those from the other two methods.To assess the significance, sensitivity, and robustness of the cooling and drying trends inferred from pollen data, we present some inter-proxy comparisons and compare these with simulation outputs from the intermediate-complexity climate model iLOVECLIM. [1] - Tarasov, P. E., et al., (2013). https://doi.org/10.1016/j.quaint.2012.04.007[2] - Harrison, S. P., et al., (2014).  https://doi.org/10.1007/s00382- 013-1922-3[3] - Brewer.S, et al., (2008). https://doi.org/10.1016/j.quascirev.2008.08.029[4] - Overpeck, J., Webb, T. et Prentice, I. (1985). https://doi.org/10.1016/0033-5894(85)90074-2[5] - Blaauw.M, (2010). https://doi.org/10.1016/j.quageo.2010.01.002[6] - Davis, B. A. S., et al., (2020). https://doi.org/10.5194/essd-12-2423-2020[7] - Chevalier.M, (2022). https://doi:10.5194/cp-18-821-2022[8] - Prentice, C., et al., (1996). https://doi.org/10.1007/BF00211617[9] - Li, C., et al., (2024). https://doi.org/10.5194/egusphere-2024-1862