Use of RothC to simulate changes of organic carbon stock in the arable layer of a Mollisol of the southeastern Pampas under continuous cropping

Soil organic matter (SOM) is a key soil component that determines soil quality and health, but its dynamics is driven by complex processes. RothC is a simple model that can be used to simulate SOM content changes in order to plan and recommend soil management practices. The aim of this work was to validate RothC with SOM stock information from two long term experiments carried out at Balcarce, Argentina (37°45′S, 58°18′W) on a complex of fine, mixed, thermic Typic Argiudoll and fine, illitic, thermic Petrocalcic Paleudoll. Yearly soil organic C (SOC) stocks of the arable layer under nineteen different crop sequences with and without N fertilizer and under conventional tillage, were used. Carbon input through crop residues was estimated from grain yield or aboveground biomass production. Model performance was evaluated through several statistical tools: simple least square linear regression of observed vs. simulated values, several indicators based on the difference between observed and simulated values, and the fitting of observed and simulated values to an exponential model as a function of time under cropping. The overall performance of RothC was very good since mean difference between observed and simulated was lower than 2 Mg C ha−1 (2% of the mean observed value), and the root mean square error (RMSE) was lower than RMSE at 95% probability (RMSE95%). Approximately 70% of the relative differences between observed and simulated (RE) were within the range −5% ≤ RE ≤ 5% without identifiable trends associated with the sequences. Besides, all RMSE's corresponding to the crop sequences were lower than RMSE95%. However, RothC tended to underestimate SOC stocks especially when N had been applied and at high levels of SOC stock. Despite there was no statistical difference (P > 0.05) between the models resulting from fitting observed and simulated values to the exponential model, estimated rates of change of SOC stock along time were different between models. This indicates that simulation was poorer in the short term after a pasture (higher SOC stocks) than in the long run. Inaccurate estimation of C input could have been the cause of most of the differences observed between simulated and observed values. We concluded that RothC can be used as a tool to contribute to make recommendations of crop rotations to manage SOC stock for Mollisols of temperate regions under conventional tillage, provided more careful and precise C input estimations were available.

► The overall performance of RothC was very good.
► There were not identifiable trends among crop sequences.
► RothC tended to underestimate SOC stocks.
► Misestimations of C input could have been the cause of the deviations.
► RothC can be used to simulate SOC, provided careful C input estimations are done.