Modeling carbon budgets in conventional and diverse crop rotations with cover crops using eddy covariance data

Beneficial management practices (BMPs) such as crop diversification can enhance soil organic carbon (SOC) and help mitigate greenhouse gas emissions contributing to global warming. However, documenting carbon (C) exchange under BMPs at scale in agricultural fields is challenging through direct measurements. An alternative approach is to use data with high temporal resolution to calibrate process-based models to predict agroecosystem C dynamics. Using a 6-year dataset of high-frequency eddy-covariance measurements of C exchange over a conventional (corn–soybean) and a diverse (corn–soybean–winter wheat plus cover crops) crop rotation from Ontario, Canada, the main objectives were (1) to evaluate the DNDCv.CAN model performance in simulating C budget components, (2) to examine the impact of crop diversification and cover cropping on C dynamics and (3) to assess the effect of residue retention and 25 years of weather variability on C budgets. DNDCv.CAN reproduced well cash-crop yields, and daily and annual carbon dioxide fluxes, although it showed limited ability to accurately predict cover crop biomass and net ecosystem exchange for winter wheat. DNDCv.CAN generally captured the net ecosystem carbon balance (NECB) directionality but underestimated absolute NECB magnitudes, which was driven mainly by overestimation of C uptake for corn and winter wheat. Measurements and simulations of NECB indicated that both rotations acted as apparent net C sources, even when simulating winter wheat residue retention. Under 25 long-term weather scenarios, modeled C budget components were comparable with the experimental period, with both rotations remaining net C sources overall. Modeled cumulative NECB was stable across extreme climate categories (cooler, warmer & dryer and wetter years), suggesting that the C balance of these systems is more sensitive to management decisions than to the range of climate variability experienced at the study site.