Mineral properties identified as most influential drivers of mineral-associated organic carbon formation using a community-based sorption database

Increasing soil organic carbon (SOC) storage has the potential to substantially offset anthropogenic CO2 emissions. Globally, mineral-associated organic carbon (MAOC) constitutes the majority of SOC (~65%) and represents a more persistent SOC pool for sequestration. However, the factors controlling how much organic carbon sorbs to soil minerals remain poorly understood, limiting our ability to predict and optimize SOC sequestration strategies. Here, we compiled a first-of-its-kind community database with sorption experiments of organic compounds on a range of soil minerals (1,673 data points). We found a general, negative pH dependency of the partition coefficient Kd (reflecting the amount of mineral-sorbed organic carbon) across anionic compounds, while the cationic compound showed a positive pH relationship with the partition coefficient. In addition, iron oxides and extracellular polymeric substances had the highest Kd values. Our random forest model reproduces the observed Kd (R2 = 0.77, RMSE = 0.59) and points to mineral type and surface area characteristics as the dominant drivers of MAOC sorption phenomena, as opposed to the minor role of organic compound properties. Our analysis suggests that incorporating more mineral-specific parameters in soil carbon cycling models could inform and improve predictions of MAOC, and in turn, the effectiveness of SOC management strategies.