Holocene potential natural vegetation in Europe: Evaluating the model spread with three dynamical vegetation models

The period of the early Holocene in Europe is marked by climate warming as Earth comes out of the last glacial period and is followed by the emergence of agriculture and animal husbandry in the second half of the period. Increased human influence had profound impacts on the land surface, but the Holocene climate evolution also drove some changes that are intertwined with it. Deciphering the role of each in the vegetation evolution is becoming more difficult as one progresses to the earlier parts of the Holocene here human induced impacts were fainter. Within this general context, we aim at understanding how much Dynamical Vegetation Models (DGVMs) differ in their representation of Potential Natural Vegetation (PNV) in Europe during the mid-to late Holocene (8.5 k.a. BP to 1900 A. D.). We ran three different DGVMs, SEIB-DGVM, ORCHIDEE-DGVM and CARAIB, in Europe, for six time-slices and forced them with identical climatic imputs obtained from the iLOVECLIM Earth system model (downscaled and bias-corrected). Results are then compared with pollen-based reconstructions from the TERRANOVA database. Overall, the three models have a similar performance in representing the pollen-derived vegetation cover at the european scale. However, their results are largely different at regional scales, particularly in mountainous areas and in boreal regions. They also show a very large spread in simulated PFT diversity at the grid cell scale, highlighting the impact of each model's internal dynamics on the results. On a global scale, they all agree on a decreasing match of their results with pollen base reconstructions over time, indicating - as expected - an increase of the human pressure on the landscape.