On-farm assessment of soil C response to regenerative management: integrating erosional processes into daycent

• On-farm measurements can diversify representation in SOC datasets. • Historical erosion appears to be a significant driver of SOC stocks. • RUSLE2 erosion estimations can be integrated into DayCent simulations of SOC. • Incomplete activity data may limit modeling activities or lead to inaccurate assumptions. Soils of the Midwestern United States (U.S.) have lost up to 70% of their original soil organic carbon (SOC) due to decades of conventional agricultural management, but regenerative practices such as no-till and cover cropping may help restore a portion of this lost SOC. Most measurements of SOC change under regenerative practices come from a limited number of long-term field experiments (LTEs), which do not fully represent the range of management practices, soil types, and landscape conditions of real commercial farms. On-farm data collection using observational experiment techniques can supplement LTE results, improve the representativeness of SOC data, and advance process-based model capabilities. We collected a series of across-the-fence soil samples to a depth of one meter from twenty-two paired regenerative and conventional fields across the states of Iowa and Southern Minnesota. Management and yield records collected from farmer participants allowed us to simulate observed SOC stocks using the DayCent process-based ecosystem model, integrated with erosion estimates from the RUSLE2 model. Our measurement and modeling results indicate that accounting for historical sheet and rill erosion and soil displacement, even on fields with seemingly minor slopes (2-4°), can improve simulations of measured SOC stocks. Simulating erosive soil loss improved the weighted R 2 of measured vs. modeled SOC stocks from 0.27 to 0.74, significantly improving estimates for fields with average slopes between 2° and 4°. These findings highlight that incorporating erosion into DayCent can improve simulations of SOC stocks on commercial farms with variable topography, and for extending process-based SOC modeling from experimental settings to real agricultural landscapes.