Impacts of regenerative agriculture on the mechanisms and processes that control soil health in Brazil

We will analyze a standardized dataset with more than 1.8 million observations from 1,906 farms across 15 Brazilian states, sampled in 2021 and resampled in 2024 in paired fields under conventional and regenerative management. Changes in soil organic carbon stocks and a soil health index will be estimated using hierarchical models that account for farm level heterogeneity and uncertainty. In parallel, we will develop and test four soil health indices with increasing complexity, ranging from a survey and management based index to simplified laboratory and field protocols and an ultra detailed diagnostic approach. Concordance among indices will be evaluated to identify indicator sets that retain decision relevant information while remaining feasible for broad adoption. To investigate mechanisms, we will study 20 representative farms using synchrotron X ray microtomography combined with deep learning to quantify pore networks, aggregate architecture, and the three dimensional spatial distribution of organic matter within aggregates. These data will be integrated with physical fractionation of soil organic matter and selected nitrogen and phosphorus pools. Finally, we will compare selected indicators with two long term regenerative experiments in the United Kingdom to test transferability across contrasting soils and climates. Our working hypotheses are that soil organic carbon gains over the 2021 to 2024 window will be detectable but modest at national scale, while improvements in soil structure and nutrient cycling proxies will provide earlier and mechanistically interpretable signals of change.