Mechanisms of lateral and linear extension of gullies (dongas) in a subhumid grassland of South Africa
Abstract
The formation of deep gullies (called ‘dongas’ locally) in rangeland in KwaZulu‐Natal Province in South Africa is a natural phenomenon. These U‐shaped, very wide gullies have considerable lateral expansion due to the episodic collapse of sidewalls.
The dongas have developed in duplex soils such as Luvisols and Lixisols formed on Permian sedimentary rocks or unconsolidated Quaternary colluvium. This study combined morphological, mineralogical and chemical characterization with measurements of grain‐size content, structural stability and the complete shrinkage curve to detect changes in soil properties of the different horizons located in the gully banks.
The different soil horizons present clear and sharp differences in physical and mineralogical properties. The topsoil with complete grass cover is very resistant to soil detachment. However, the leached E horizon and the BC horizon have low structural stability. The soil profile down to and including the Bt horizon contains exclusively illite in the clay fraction, while the BC colluvial layer and the C horizon (mudstone) contain expandable interstratified illite–smectite. The Bt horizon has a high water content at saturation and high shrinkage, while the BC and C horizons have a high residual shrinkage and a very low water content at saturation.
Because this type of gully expansion is not significantly linked to slope value or the stream power index (SPI) at the gully head, to land‐use change, high rainfall intensities or the threshold of concentrated runoff being exceeded at the gully head, other causes were investigated. It was concluded that the heterogeneity between horizons with different mineralogical properties and structural stabilities, soil types and parent material, anisotropic water‐saturation and shrink‐swell properties are of major importance. This heterogeneity between different soil horizon morphologies and their physical properties can explain why the relationship between the critical slope and the drainage area for gully initiation showed a threshold for gullying much lower than that found elsewhere.