Network analysis as a novel tool for assessing SOM stabilization dynamics across soil depths

Understanding the structural organization of organic carbon in soil is essential for predicting stabilization mechanisms and improving management strategies. Network analysis offers a novel approach to reveal complex interactions among soil variables that traditional correlation methods often overlook. We applied network analysis to assess depth-dependent changes in soil-related properties across five soil layers (0–100 cm) in subtropical sandy soils from a Cerrado-Atlantic Forest ecotone in Bauru, São Paulo, Brazil. Thirty samples were characterized using 30 physical, chemical, and spatial attributes. Networks were constructed from significant Spearman correlations (p < 0.05, |r| > 0.3), and centrality metrics identified key regulatory variables. Results revealed pronounced network reorganization with depth. Surface layers (0–20 cm) showed moderate Carbon (C) and Carbon Stock (Cstock) connectivity, while Iron and Aluminum dominated (degree = 14), reflecting mineralization and weathering. Deeper layers (60–100 cm) exhibited high C and Cstock centrality, ranking among top five nodes by degree (18–20) and eigenvector centrality (0.91 – 0.99). This shift coincided with 50% increased network density and strengthened carbon associations with exchangeable acidity (H + Al) and cation exchange capacity (CEC). Magnesium showed elevated betweenness (23.58) in intermediate layers, suggesting a mediating role in carbon – nutrient dynamics. Our findings show that soil organic carbon becomes increasingly related to the chemical properties with depth, supporting physicochemical protection over biochemical recalcitrance. Network analysis provides a robust framework for identifying depth-specific controls on carbon persistence in sandy soils under ecosystem recovery.