Geospatial Analysis and Quadratic Regression Modelling of Soil Chemical Properties and Exchangeable Sodium in Aridisols of Sahel Savannah for Sustainable Soil Management

Abstract

Soil degradation, particularly sodicity, constitutes a major constraint to soil fertility and sustainable irrigation agriculture in semi-arid environments. Exchangeable sodium (Exch. Na⁺) is a critical indicator of sodicity because its accumulation can impair soil structure, permeability and agricultural productivity. This study aimed to evaluate the spatial distribution of total exchangeable bases (TEB), soil organic carbon (SOC) and soil pH, and quantify their linear and nonlinear relationships with Exch. Na⁺ in the Ajiwa Irrigation Scheme, Katsina State, Nigeria. Thirty-six surface soil samples (0–30 cm) were collected using a hybrid purposive-systematic sampling technique and analysed for exchangeable cations, TEB, SOC and soil pH. In ArcGIS 10.8, inverse distance weighting (IDW) was employed to generate spatial distribution surfaces, while second-order quadratic regression was used to model Exch. Na⁺ using standardized TEB, SOC and soil pH. Model robustness was evaluated using leave-one-out cross-validation (LOOCV). Results revealed distinct spatial relationships, with high TEB zones [>15 cmol(+)/kg] coinciding with Exch. Na⁺ hotspots [>25 cmol(+)/kg], particularly at sampling points 3, 5, 7 and 36. The quadratic regression explained 78.9% of the variation in Exch. Na⁺ (R² = 0.789; adjusted R² = 0.746) and was highly significant (F = 18.10, p < 0.001). TEB was significant (β = −0.738, p = 0.033), while its quadratic component was highly significant (β = 0.052, p = 0.002), demonstrating a pronounced nonlinear threshold effect. The TEB–Exch. Na⁺ quadratic relationship achieved R² = 0.899, compared with 0.521 for the linear relationship. LOOCV produced a low RMSE of 0.459, supporting predictive robustness.