Environmental and management drivers of topsoil organic carbon variability in Cerrado Mineiro coffee systems, Brazil

Interactions between environment and management in Cerrado Mineiro coffee systems create temporal contrasts in plant vigor, shaping soil organic carbon (SOC) variability. Soils from ten coffee farms distributed across the Cerrado were sampled at a depth of 0–20 cm and analyzed, with NDVI information retrieved from remote sensing for the same points. From a sequence of data mining analyses, we modeled and obtained insights into SOC by integrating Sentinel-2-derived NDVI time-series metrics, including cumulative NDVI (a proxy for biomass inputs) and the peak of NDVI (representing maximum plant vigor), with key environmental factors (soil pH, clay content, sand content, and elevation) and management practices (rainfed versus irrigation). Drivers’ relevance was primarily assessed by Pearson’s correlation and Principal Component Analysis (PCA). SOC showed the strongest and positive correlation with NDVI peak (r = 0.48), cumulative NDVI (r = 0.44), and a negative association with sand content (r = −0.41), whereas relationships with elevation (r = 0.28), soil pH (r = 0.11), and clay content (r = 0.05) were almost negligible. PCA corroborates a strong correlation of SOC with the NDVI vectors. Kruskal–Wallis tests indicated no statistical differences in SOC between rainfed and irrigated (p = 0.76), supporting SOC stability under contrasting water-management strategies. Under intensive management, remotely sensed vigor metrics explained a substantial variability of SOC, highlighting the dominant role of biomass inputs. Remotely sensed vigor metrics, enhanced by soil sand content information, provided a low-cost proxy for monitoring SOC and guiding site-specific management.