Abstract
Soil is a major terrestrial carbon reservoir, and enhancing its carbon stock is a central strategy to mitigate climate change. Earth system models project a net soil carbon sink by 2100, the magnitude of which is still under debate, differing significantly between approaches. Radiocarbon-based studies often suggest a limited soil carbon accumulation capacity, but these estimates are biased by the presence of ancient, radiocarbon-free, organic carbon (aOC). This carbon no longer contributes to soil carbon dynamics and increases the average (14)C age of soil carbon because it is radiocarbon-depleted. This known radiocarbon caveat can be overcome with a better understanding of the aOC (ancient radiocarbon-free OC) distribution in the world's soils. Here we apply a mixing linear equation to 313 soils worldwide from radiocarbon databases to estimate the aOC contained in soils. The aOC contained in soils has different origins, from rock-derived to old biospheric C strongly associated with mineral particles during pedogenesis. Our findings show a mean aOC content of 2.4 mg/g ±3.2 SD with an aOC contribution up to 11% of the soil organic carbon in topsoils (0-30 cm depth), reaching 25% in subsoils (30-100 cm depth) and more than half in deep soil (> 100 cm depth). We demonstrate that the aOC content is particularly high in Andosols and Cryosols. We subtracted the aOC contributions to calculate a global mean corrected age of non-aOC carbon to 1 m depth of 290 years, contrasting sharply with previously reported values of 3100 to 4830 years. This corrected estimate aligns more closely with independent isotopic proxies ((13)C and (36)Cl) of soil carbon dynamics. These results also reconcile empirical data with the parameterization of Earth system models.