Data-Driven Delineation of Production Environments to Assess Soil Organic Matter and Crop Yield Stability

Understanding soil support capacity is essential for sustainable soil management and appropriate land-use allocation. Production Environment group soil attributes into a limited number of classes representing the support potential of soils and help in study their management, especially regarding soil organic carbon (SOC). We delineate production environments in a soybean-growing region using soil data and relate them to spatial and temporal yield patterns. Unsupervised cluster classification was applied using 42 mapped soil attributes (from the 0–0.2 m and 0.8–1.0 m layers). The mapped attributes included texture, cation exchange capacity (CEC), and SOC. The study area comprises soybean-producing zones in São Paulo State and Paraná State, Brazil. Soybean yield from 2010 to 2023 was used to validate the spatial and temporal behavior of the clusters. Annual statistics of yield were calculated to quantify cluster stability and resilience. Three distinct clusters were identified. Cluster 1 exhibited the highest values of SOC, clay, and CTC in both layers, and showed the highest soybean yields, averaging approximately 3 t ha⁻¹. In contrast, Cluster 2 presented the lowest fertility and SOC levels and consistently displayed the lowest yields and highest coefficients of variation across years, indicating production instability. During climatic stress years, areas with higher SOC and clay content showed yield stability. Spatial analysis further revealed that, despite high potential in some areas, productivity gaps persist, suggesting opportunities for improved management. Conversely, Cluster 2 areas require priority conservation and site-specific management due to their higher susceptibility to nutrient depletion and soil organic carbon losses.