Geological and morphotectonic controls on soil erosion risk in the Central Eastern Cape, South Africa

Abstract

Soil erosion by water poses a persistent challenge to sustainable land management, particularly in geologically complex terrains where conventional erosion models fail to adequately capture subsurface controls. This study evaluates the influence of lithology and morphotectonics on soil erosion risk within the Central Eastern Cape (CEC), South Africa, to improve land degradation neutrality planning and erosion prioritisation. Soil erosion risk was estimated using the Revised Universal Soil Loss Equation (RUSLE), supported by satellite-derived rainfall and soil datasets, field sampling and laboratory analyses. The study’s novelty lies in the explicit incorporation of geological controls through lithostratigraphic mapping and morphotectonic zonation derived from geomagnetic data. Results indicate a mean soil loss of 0.49 t ha −1 yr −1 , totalling about 1.81 million tons, with moderate-to-very high erosion concentrated in structurally dissected terrains. High morphotectonic zones dominated by dolerite intrusions account for 18.8% of moderate-to-very high erosion classes, reflecting tectonically induced relief, drainage modification, and soil instability. Lithological controls are pronounced, with soils derived from sandstone (1340 km 2 ), dolerite (1277 km 2 ), and mudstone (1275 km 2 ) exhibiting the highest susceptibility to erosion (classified as moderate-to-very-high soil erosion). Wavelet coherence analysis confirms strong seasonal agreement between satellite-derived and observed rainfall, validating the use of remote-sensing data. Satellite-derived soil texture and structure reveal overall accuracy of 65.7% and 64.5% (K = 0.37, p < 0.001) with field observations, and particularly in perfect agreement for permeability. To validate predictive fidelity, an XGBoost regression model was implemented and demonstrated moderate predictive accuracy (MSE = 0.0705; R 2 = 0.578), capturing regional erosion trends while reflecting the stochasticity of localised land-cover dynamics and rainfall variability. The study demonstrates that integrating geological controls substantially enhances the discrimination of erosion risk beyond what RUSLE alone provides. The proposed framework supports evidence-based implementation of land management policies and provides a transferable approach for advancing land degradation neutrality objectives in erosion-prone landscapes.