Evolution of global vegetation patterns since the last glacial maximum

Vegetation evolution is a crucial component of global change research, providing a foundation for understanding the interactions between vegetation dynamics and climate changes. However, current vegetation reconstructions since the Last Glacial Maximum (LGM) have not yet provided a complete and continuous presentation of global-scale vegetation evolution. In this study, we systematically compiled 3286 pollen records and applied the biomization method to reconstruct global vegetation patterns on a millennium resolution since the LGM. Our results show that during the LGM, the mid- and high-latitudes of the Northern Hemisphere were primarily covered by tundra, with steppe and open coniferous forests in western North America, taiga in eastern North America, and steppe dominating much of Eurasia. In the low latitudes, tropical rainforests retreated significantly compared to the present, while arid shrubland expanded across much of Africa. Since the deglaciation, forests have gradually expanded, with cold-adapted biomes shifting to higher latitudes. In the mid-latitudes, mixed forests and deciduous broadleaf forests increased in North America, Europe, and China. During the early to mid-Holocene, forests dominated in the mid- and low-latitudes, while tundra and taiga dominated in the high-latitudes. By the late Holocene, steppe and desert expanded in central North America and northern China, while tropical rainforests flourished in South America. The results further reveal that the global forest cover increased by ∼31 % from the LGM to the mid-Holocene, then decreased by ∼5 % at the late Holocene. In the mid-to high-latitudes of the Northern Hemisphere, forest cover varied significantly, rising notably from the LGM to the early and mid-Holocene, peaking around ∼7–5 ka BP, and declining after ∼5 ka BP. Conversely, the mid-latitudes of the Southern Hemisphere exhibited a more gradual pattern, with forest percentage peaking earlier ∼12–8 ka BP, and maintaining relatively stable afterward. Tropical regions had more erratic changes, initially fluctuating between 18 and 15 ka BP, followed by steady growth peaking ∼4–6 ka BP, and continued fluctuations thereafter. Our results reveal asynchronous vegetation changes across different global regions and provide a biome-level perspective, offering a finer classification of vegetation dynamics compared to previous global-scale studies. This dataset has the potential for comparison with palaeovegetation simulations, promoting better integration with different vegetation classification schemes. Moving forward, increasing the data density and integrating simulations will be crucial for a deeper exploration of the feedback mechanisms between vegetation evolution and climate change, thereby enhancing the accuracy of predictions of future vegetation dynamics.