Designing a Multi-scale, Process-oriented and Covariate-Guided Sampling Strategy for Capturing Spatial Heterogeneity of Soil Organic Carbon in Sahelian Semi-arid Agroecosystems

Abstract

Soil organic carbon (SOC) plays a central role in soil fertility, ecosystem resilience, and climate regulation, particularly in semi-arid drylands where degradation pressures are increasing. In Sahelian sandy agroecosystems, SOC distribution is highly heterogeneous due to interacting influences of sparse vegetation cover, sandy parent materials, biogenic structures, and subtle topographic gradients. Conventional random or systematic grid-based sampling designs often fail to capture short-range variability and localized carbon hotspots, potentially compromising spatial inference and digital soil mapping accuracy. This study proposes a multi-scale, process-oriented, and covariate-guided sampling framework tailored to Sahelian semi-arid environments. The approach integrates the SCORPAN conceptual model with high-density radial transect sampling around biogenic structures at the local scale and landscape-scale environmental stratification based on NDVI classes, soil reflectance indices (BI, BSI, RI), and terrain attributes. Sampling points were proportionally allocated across environmentally homogeneous strata to ensure balanced representation of vegetation gradients, exposed sandy surfaces, and micro-topographic conditions. The proposed design enhances environmental representativeness, reduces sampling bias, and strengthens the ecological coherence of SOC spatial assessment. By explicitly accounting for nested spatial processes, the framework provides a transferable methodological foundation for soil surveys, carbon stock estimation, and digital soil mapping applications in semi-arid drylands facing climate variability and land degradation pressures.