The vertical distribution of soil microorganisms in soil indicates the restoration degree of degraded soil ecosystems. We took the untreated bare land and vegetation restoration sample plot in the red soil erosion area of southern China as the object of study; comparatively analysed the soil bacterial community changes in the 0 to 10, 10 to 20, 20 to 30 and 30 to 40 cm soil layers; and explored the environmental factors driving the change in the soil bacterial community. The poor nutrient conditions created by soil erosion increased the competitiveness of autotrophs and made Chloroflexi the dominant phylum of bacteria. Soil erosion led to the gradual similarity of soil bacterial communities in the 0 to 10, 10 to 20 and 20 to 30 cm soil layers. However, only the relative abundance of Actinobacteria changed in different soil layers in the erosion area, mainly due to the inconsistent distribution of soil organic carbon caused by erosion affecting the change in the Actinobacteria relative abundance in the soil layer. After vegetation restoration, the soil properties of the eroded land were obviously improved, and the dominant bacterial phylum changed from autotrophic bacteria ( Chloroflexi) to heterotrophic bacteria ( Actinobacteria). The change in community structure existed only in the 0 to 30 cm soil layer in the restoration area, while the community structure changed to mainly Proteobacteria in the 30 to 40 cm soil layer. The change in the respective proportions of Chloroflexi, Proteobacteria and Actinobacteria was the main reason for the difference in soil bacterial community structure among soil layers. The change in soil aggregates caused by vegetation restoration was the main environmental factor driving the variation in soil bacterial community structure, and the formation of aggregates was closely related to soil organic carbon. The vertical distribution of Actinobacteria in different soil layers can indicate the degree of soil ecosystem restoration in the red soil erosion area of southern China, and the relationship between Actinobacteria and soil organic carbon was significant.

The water content is a crucial factor in evaluating the causes of Benggang collapse. The soil–water characteristic curve (SWCC) is an important parameter for the quantitative study of soil water content. However, limited research has been carried out on the SWCCs of the Benggang soil profile. We studied two typical collapsing gullies in southeast China and conducted desorption experiments using a pressure plate extractor to analyze the SWCCs of the undisturbed soils of collapsing walls. The results show large variations in the SWCCs for different soil horizons of a collapsing wall that can be accurately fitted by the van Genuchten (VG) model (NSE≥0.90). With increasing soil depth, the a and θs parameters of the VG model first decrease and then increase, red soil layer had the highest a and θs (the average value of 0.046 and 0.369, respectively), whereas the n parameter of the VG model exhibits the opposite trend, sand soil layer had the highest n (the average value of 1.563). The θr parameter of the VG model decreases with increasing soil depth, red soil layer had the highest θr (the average value of 0.194). The red soil layer has the highest water-holding capacity, whereas the sandy soil and detritus layers have lower water-holding capacities. The SWCCs are related to the soil material composition, particle composition and porosity. The gravel content and the particle morphology (the aspect ratio, sphericity, and specific surface area) are also the significant influence factors for the SWCC that cannot be neglected. The difference among the SWCCs for the soil profiles of collapsing walls can be used to explain the mechanism for the collapse of collapsing wall. The results of this study provide a theoretical basis for understanding the process of the collapse of collapsing wall in Benggang in southeast China.