Even with the in-depth implementation of forestry ecological projects, such as restoring farmland to forest (grass) in the loess area of northern Shaanxi, the characteristics of soil macrofauna communities and their coupling relationship with environmental factors after vegetation restoration in the study area are yet obscure. However, the soil macrofauna community characteristics are of great significance for evaluating the effectiveness of vegetation restoration in the study area. Therefore, the study aims to reveal the characteristics of the soil macrofauna community and their coupling relationships with the environment in the loess area of northern Shaanxi. In this study, all organisms of the five typical vegetation types in the study area were collected by manual sorting (Armeniaca sibirica and Populus simonii mixed forest (M), Robinia pseudoacacia (P), Populus simonii (S), Populus hopeiensis (H) and Hippophae rhamnoides (R)), and the adjacent abandoned grassland (G) was used as a control group. The group number and the individual number of soil macrofauna of different vegetation types in the study area and their coupling relationships with environmental factors are studied, and the following conclusions were drawn. (1) The study shows that there are certain differences in the environmental factors of different vegetation types in the study area, which include the significant differences in the alkaline nitrogen content of various vegetation types (p < 0.05). (2) The effects of different vegetation on soil macrofauna community were different. There were no significant differences in the soil macrofauna community structure between Armeniaca sibirica and Populus simonii mixed forest, Robinia pseudoacacia, Populus simonii and Populus hopeiensis, but there was a large difference from that of the abandoned grasslands. The community density of soil macrofauna in Armeniaca sibirica and Populus simonii mixed forest and Populus simonii were significantly higher than that in the abandoned grassland (p < 0.05), but the other indexes showed no significant differences. The Shannon–Wiener index of Robinia pseudoacacia and Populus hopeiensis were much lower than that of the abandoned grassland (p < 0.05). (3) The diversity of soil macrofauna communities was mainly affected by pH, alkaline nitrogen, potassium available, vegetation coverage and litter production. (4) Different groups of soil macrofauna were closely related and reacted differently to environmental factors, and vegetation coverage, litter production and alkaline nitrogen content were the key factors affecting the composition of soil macrofauna communities.

Vegetation restoration is essential for the stability of the ecological system structure and function in the loess region of North Shaanxi Province. Natural and artificial restoration are the primary modes for vegetation recovery and soil quality improvement in this region. In this study, two adjacent watersheds with similar ecological environment conditions but different restoration modes were selected for research; one watershed is restored naturally (He Gou watershed), and the other is restored artificially (Chai Gou watershed). According to the study of soil stoichiometric characteristics and soil quality after the vegetation restoration in these two watersheds, the results showed: (1) Compared with the natural restoration, artificial restoration was more effective in increasing the content of soil organic carbon and total nitrogen, however, the total phosphorus content of the soil in the natural restoration mode was higher than that in the artificial restoration mode. (2) The ratios of soil C:N, C:P, and N:P showed a decreasing trend with the increase of soil depth at these two restoration modes. (3) In the 0–60 cm soil layer, the soil quality under the artificial restoration mode was better than that of the natural restoration, especially for the soil layer beneath 20 cm. (4) The minimal data set on the soil quality evaluation in the study area included soil organic carbon, capillary water holding quantity, available potassium, soil water content and available phosphorus. It showed a linear relation with the total index data set (y = 0.829x + 0.058, R² = 0.76) and can reflect the soil quality more sensitively than the total indicator data set.