Controls on soil organic carbon across soil depths in tropical and temperate non-volcanic regions

Soil organic carbon (SOC) is fundamental for climate regulation, soil fertility, biodiversity, and healthy terrestrial ecosystems. Understanding the key controllers and their pathways is required to estimate SOC distribution and predict C sequestration potential. Previous research emphasized the significant impact of active Al/Fe (acid-oxalate extractable) on SOC content, especially in volcanic soils, yet gaps persist in understanding the interactions among SOC, active Al/Fe, climate, and non-volcanic parent materials across different soil depths. Herein, we explore these dynamics using random forest regression (RFR) and structural equation modeling (SEM). Our analysis included 17 SOC-related physicochemical soil variables and 4 climatic properties across topsoil and subsoil horizons at 211 sites (2 depths). The study covers 178 tropical and 33 temperate sites from sub-humid to humid non-volcanic regions, predominantly with acidic to neutral soil pH. We found that SOC variance explained by SEMs (54 % for topsoil, 75 % for subsoil) closely matched RFR outcomes (61 % for topsoil, 72 % for subsoil), highlighting the efficacy of our SEMs in identifying key SOC controllers: mean annual temperature (MAT) and excess precipitation (moisture index: precipitation – potential evapotranspiration) for climate, Al₂O₃ + Fe₂O₃ (total Al and Fe contents expressed as oxides) for parent material, and active Al/Fe and pH for soil properties. Partial dependence in RFRs and path coefficients in SEM indicated MAT, active Al/Fe, and pH directly contribute to topsoil SOC, whereas active Al/Fe directly controls SOC in subsoil. Furthermore, SEMs indicated bidirectional interaction between SOC and active Al/Fe in topsoil, where active Al/Fe increased SOC and vice versa, while in subsoil, active Al/Fe has substantial unidirectional control over SOC. Importantly, both climate and parent material indirectly affected SOC by regulating active Al/Fe contents, particularly in subsoil. These findings enable predicting SOC distribution and refining SOC dynamics models by focusing on the direct and indirect impacts of active Al/Fe, climate and parent material across different soil depths and climatic zones in non-volcanic regions.