Deep SOC stock dynamics under contrasting management systems: Is the EPIC model ready for carbon farming implementation?

Process-based models have been recognized as cost-effective tools to assess carbon farming mechanisms through quantifying the C fluxes in the agroecosystems. A result-based approach is suggested however the wide vari-ability of agricultural environments makes further model implementation necessary to limit the uncertainty of the results, especially on deep soil organic carbon (SOC) stock estimation and stratification and in agro-ecosystems characterized by a shallow water table. In this study, a comprehensive soil and crop dataset collected over a seven-year period from different pedoclimatic conditions across the Veneto Region (NE Italy) was used for EPIC model calibration and validation of SOC stock dynamics. Experimental data included yields from several crops (corn, winter and spring wheat, rapeseed, and soybean), continuous monitoring of soil water content, and SOC stocks (1872 total samples within the 0-50 cm soil profile) under conventional, cover crop and conservation agriculture systems. Modelling was performed by testing two N-C sub-models (CENTURY and PHOENIX), which differentiated in terms of mineralization/immobilization rates.Results showed that the procedure was able to obtain parameters valuable for most of the management system and pedoclimatic condition, reproducing well the tested variables. The EPIC model acceptably captured soil water dynamics (Nash-Sutcliffe coefficient - NSE - was up to 0.26), especially in the topsoil. Furthermore, simulation of weed-crop competition in conservation agriculture strongly contributed to properly explain the variability in crop production among the contrasting agricultural systems (R2 ranged from 0.51 to 0.71). Like-wise, EPIC skillfully simulated SOC stocks within the 0-50-cm profile regardless of the sub-model used (NSE was up to 0.64). Moreover, the model acceptably captured the profile SOC stratification among the different man-agement practices. This study highlights the EPIC robustness for predicting SOC stocks and assessing with high accuracy carbon farming results.