Abstract
Shallow landslides can pose a major threat to human lives and infrastructure over significant portions of the global land surface and occur primarily from weakened soil shear resistance due to water infiltration. Although there is growing interest in using vegetation to stabilize hillslopes against landslides, we noted the scarcity of studies examining temporal variations in slope stability, particularly with regard to different land uses. In three tropical and temperate landslide-prone regions (Laos, Costa Rica and France), we combined soil moisture monitoring to 1.2-1.8 m depths in the field, soil shear resistance measurements and numerical modeling to compare slope stability under competing land uses for 2-3 years. Slope stability tracked temporal changes in soil moisture, with smaller contributions from root mechanical reinforcement. Land uses with denser vegetation had greater stabilizing impacts than those with sparser vegetation, which lasted for six to twelve months per year and coincided temporally with growing or rainy/dry seasons. Greater stability under denser land use persisted into wet seasons in one of the sites and were minimized or reversed in the other two sites. Site-specific factors such as climate, soil and species may explain these differences in the vegetational control on slope stability. A review of the data in the literature found that woody vegetation increased slope stability and decreased temporal variation in stability compared to herbaceous vegetation. However, while variations in slope stability decreased in increasingly humid climates, indicating that the largest fluctuations in stability, and hence potential to improve slope integrity with land-use changes, will be found in arid to sub-humid regions. Our results show that dense vegetation provides greater stability and protection against landslides from rainfall. Land managers need to take into account this biological control on hydrology when managing vegetated slopes. Incorporating the vegetation-driven deep soil moisture dynamics will also improve predictive utility of models of specific events.