It is feasible to quantify the effect of agricultural practices on soil carbon stocks through sampling

There is strong disagreement about the potential for regenerative management practices to sequester sufficient soil carbon to help mitigate climate change. Measuring change in carbon stocks following practice adoption at the grain of farm fields, within the extent of regional agriculture, could help resolve this disagreement. Yet sampling demands to quantify change are considered infeasible primarily because within-field variation in stock sizes is thought to obscure accurate quantification of management effects on incremental carbon accrual. We evaluate this 'infeasibility assumption' using high-density, within-field, sampling data from 45 cropland fields inventoried for soil carbon. We explore how within-field sampling density, field numbers, and magnitude of simulated change in soil carbon stocks impacts the ability to accurately quantify management effects on soil carbon change. We find that (1) stock change estimates for individual fields are inaccurate and inconsistent, where marked losses and gains in carbon stocks are frequently estimated even when no change has occurred. Higher sampling densities narrow the range of estimated effect sizes but inaccuracies remain large. Similarly, (2) the accuracy of estimates of mean effects on stock change at the project level (i.e., multiple fields) were sensitive to sampling density and not magnitude of simulated stock change. In contrast to individual fields, however, higher sampling densities (e.g., 1.2 ha sample-1), as well as a greater number of fields (e.g., 30), generated consistent and accurate, mean project-level estimates of carbon accrual, with ~80% of the estimates falling within 20% of the simulated stock change. Yet such monitoring designs do not account for dynamic baselines, which necessitates measurement of stock changes in control, non-regenerative fields. We find (3) that higher sampling densities, field numbers, and magnitudes of simulated soil carbon stock change are then collectively required to make accurate estimates of management effects on stock change at the project level. The simulated effect sizes that could be consistently detected included rates of carbon accrual considered achievable and meaningful for climate mitigation (e.g., 3 t C ha-1 10 y-1), using field numbers and sampling densities that are reasonable given current sampling methods. However, the sampling densities and field numbers we report are not recommendations; they should be tailored to the fields in each project. Nevertheless, our findings reveal the potential to use empirical approaches to accurately quantify soil carbon stock responses to practice change. We provide recommendations for data that government, farmer and corporate entities should measure and share to build confidence in the effects of regenerative practices, freeing the soil carbon debate from overreliance on theory and data collected at scales mismatched with agricultural management.