Effect of selected mechanical properties of agroforestry tree roots on shallow-seated landslide prone areas on mt Elgon, Uganda

Abstract

Landslides continue to occur in the Elgon region despite interventions such as tree planting initiatives aimed at restraining them. The current study explored the efficacy of landslide model hybridization, tree-landslide relationship and selected mechanical properties of tree roots on slope stability with a keen focus on root tensile strength, soil shear strength, and index of root binding. A hybrid model comprising of frequency ratio, index of entropy and weighted overlay characterized landslide risk and its performance was evaluated using the Receiver Operator Characteristics curve. A standard deviation ellipse method was applied in the spatial distribution patterns of selected agroforestry trees. Tree-landslide relationship was tested using the Pearson correlation method while root tensile and soil shear strength variations were tested with a one-way (ANOVA). Study results indicated that Tsume was characterized as very high 4.70 km² (5.17%) and high 22.62 km² (24.90%) susceptibility with population density and soil type as the highest contributors (12.05%) and (10.86%) consecutively while slope least contributed with (3.40%). Overall model performance was very good with ROC (AUC = 0.91). Species distribution results indicate high dispersion of Croton macrostachyus and Markhamia lutea across the study area and high concentration of Albizia coriaria downstream. A weak negative correlation (r = -0.20 < 0.01) was observed between DBH and landslide size. A one way ANOVA test of tensile strength revealed significant difference among species with (F(5, 573) = [18.161], p < 0.001), and Grevillea robusta (3.02±1.217kg/mm²), Albizia coriaria (2.53±1.382kg/mm²), and Markhamia lutea (2.28±1.01kg/mm²) as the best performers. Croton macrostachyus (1.78±1.167)kg/mm² and Cordia africana (1.69±1.153)kg/mm². The best shearing species was Albizia coriaria with average shear strength (of 52.46±10.24) kpa followed by Markhamia lutea (50.70±15.47) kpa. Eucalyptus spp. underperformed with average shear strength (46.75±12.92) kpa. In conclusion, hybridization of single landslide susceptibility models significantly improves landslide mapping and prediction accuracy. The model also showed that population density and soil type are the major drivers of landslide in Tsume micro-catchment. Furthermore, presence of trees reduces landslide risk in an area and DBH is a very important guiding factor. Therefore, mitigation measures should target population control and soil conservation practices such as tree planting specifically A. coriaria, G. robusta and M. lutea which have good slope stability characteristics.