Savanna soil carbon accrual occurs through particulate organic matter from grass rather than tree biomass, regardless of atmospheric CO ₂

Afforestation schemes in savannas are increasingly promoted as a carbon storage strategy despite threats to biodiversity. We also lack a clear understanding of how trees and grasses differentially contribute to the major carbon store in savannas, that is, soil organic carbon (SOC) and its fractions. Because SOC fractions vary in their persistence, saturation potential, and vulnerability to loss, it is crucial to understand how shifts in tree cover and biomass will influence both the amount and stability of SOC, particularly under disturbance regimes and future climate scenarios such as rising atmospheric CO ₂ .

Using an Open‐Top Chamber system, we examined the responses of five C ₃ leguminous savanna tree species and the C ₄ grass species Themeda triandra to grass‐tree competition under ambient (400 ppm) or elevated (550 ppm) atmospheric CO ₂ . We hypothesised that SOC increases will mostly depend on plant biomass rather than plant life form; and that CO ₂ fertilisation will benefit C ₃ savanna trees more than C ₄ grasses, leading to a loss in labile grass root exudates and thus the mineral‐associated organic carbon (MAOC) that depends on these exudates.

We found that grasses, not trees, were associated with increased SOC in savanna soil to a depth of 0.3 m due to their relatively large biomass, regardless of CO ₂ concentration. Soil planted with both grass and trees had 10% more carbon and 8% more nitrogen compared to trees only. End‐member mixing models indicated that 50% to more than 90% of this carbon was grass‐derived. Savanna SOC accumulation occurred primarily through below‐ground and indirectly through aboveground biomass, which drove the formation of the occluded particulate organic carbon fraction (oPOC) according to a structural equation model. Similar results were observed for soil nitrogen. The responsiveness of oPOC (and not MAOC) in this savanna soil is different to temperate grassland soils, emphasising the need to understand savanna SOC dynamics.

While recognising the limitations of pot culture, our results provide a clear message for policymakers and land managers: Conservation of grassy biomes such as savannas, and not afforestation, aligns with both climate and biodiversity goals.