Reinterpretation of Tswaing Crater palaeoecological evidence reveals the crucial role of fire in shaping savanna

Debates about the role of fire in shaping open savanna ecosystems influence our understanding of environmental changes, human evolution, and ecosystem conservation. Quaternary palaeoenvironmental studies can provide crucial evidence of climate, vegetation, and fire dynamics, offering valuable insights into their interconnections. While droughts have long been considered the primary driver of openness in African savannas, with fire acting as a secondary response, perspectives about fire have evolved. Accordingly, fire frequency and plant tolerance to fire can drive changes in plant growth-forms, vegetation communities, and openness. Flammability, a key trait enabling plants to thrive in frequently burnt environments, is linked to higher fine C4 grass fuel loads but constrained by tree cover. This dynamic interplay results in mosaic landscapes with alternative ecosystem patch states (AES) varying in degree of flammability. This new insight about fire prompted us to re-examine the ~200,000-year (200 Kyr) Tswaing multiple-proxy palaeoenvironmental evidence. We found that vegetation at long timescales was dominated by growth-forms typical of open environments, such as forbs and graminoids (grasses and sedges). Cooler conditions beginning ~90000-18000 Kyr (Kyr)—the last glacial period—triggered the expansion of C4 grasses and savanna at the expense of forest patches. A positive feedback loop between fire, rainfall seasonality, cooler temperatures, and grass fires likely reinforced the subsequent savanna patches. Fire and rainfall amount influenced the main gradient of variation in vegetation from forest to savanna-type patch states. Charcoal production was linked to local-scale Cyperaceae/Poaceae graminoid interactions that affected fire spread and flammability due to differences in seasonality-driven fuel moisture. On the other hand, the association of some Stoebe and Artermisia herbs with Cyperaceae indicates use of disturbance gaps after burning events. The strong climatic forcing signal made it difficult to conclude whether observed forest versus savanna-type patches were fire-driven alternative ecosystem states. However, reinterpreting Tswaing vegetation-fire dynamics was essential for understanding the representation of fire in grassy ecosystems and evaluating climate data, which are invaluable for tracing long-term land use changes and conservation.