Pollen, patterns, processes: Data-driven insights into Holocene vegetation dynamics

With carbon emissions driving rapid climate change and subsequent vegetation shifts, managing landscapes to conserve biodiversity and mitigate climate change is becoming increasingly urgent. However, short observational time series limit our ability to understand long-term vegetation dynamics under shifting climatic conditions. Large-scale vegetation patterns often respond slowly to environmental changes, and the extent to which nonlinearity governs these patterns remains uncertain, as such dynamics cannot be directly observed. Paleoecology offers a valuable perspective by enabling the investigation of vegetation changes over extended timescales. Sedimentary fossil pollen data serves as an exceptional proxy to reconstruct and study past vegetation. However, previous reconstructions have often been temporally or geographically restricted, lacked resolution, or relied on qualitative vegetation estimates, leaving Holocene vegetation patterns uncertain. Furthermore, pollen-based vegetation data is characterized by several biases, leading to uncertainties about the robustness of paleovegetation data in the analysis of complex vegetation dynamics. This thesis explores Holocene vegetation dynamics globally and in the Northern Hemisphere using a synthesized and harmonized pollen data set. A global qualitative (biomization) and a Northern Hemispheric quantitative (REVEALS) reconstruction were performed and validated against modern vegetation data, to reduce compositional biases in pollen data. In addition, a model-data comparison was conducted between simulated and reconstructed biomes. Reconstructed tree cover was used to explore nonlinearities in past forest cover states across the Northern Hemisphere. To assess the robustness of paleovegetation estimates, surrogate data modeling was employed to incorporate paleo-uncertainties. Additionally, taxonomic richness was derived from the pollen dataset, allowing investigation of its drivers using mixed effects models and additional paleo-environmental variables. The reconstructions showed significant improvements over previous reconstructions, enabling detailed quantitative descriptions of spatiotemporal Holocene vegetation dynamics. Surrogate vegetation data revealed that even complex vegetation dynamics, such as nonlinearity, could be inferred from pollen-based data. However, comparisons with reconstructed tree cover indicated a lack of alternative stable states in Northern Hemisphere forest cover during the Holocene, implying a potential absence of tipping points for the boreal forest biome. Analysis of richness dynamics highlighted a significant influence of past forest cover trajectories, pointing to nonlinear effects or a substantial temporal lag in richness response. Additionally, a negative relationship between forest cover and richness during the Holocene suggests a potential trade-off between aboveground carbon storage -critical for climate change mitigation- and plant biodiversity conservation, highlighting the applicability of paleoecology to modern challenges. This thesis enhances the ability of paleoecology to provide long-term insights into vegetation dynamics and their drivers. This deeper understanding of vegetation responses to climate change is crucial for developing effective management strategies that mitigate climate change and conserve biodiversity in the face of anthropogenic climate change.