Afforestation is the most effective method of revegetation and choosing suitable afforestation tree species is regarded as a key performance criterion. In this study, 4 shrub species (Nitraria tangutorum, Caragana korshinskii, Berberis vernae and Caragana brevifolia, planted in Xining Nanshan, Prefecture, China in 2009) were studied to measure microbial abundance, enzyme activities and nutrient characteristics of rhizosphere and non-rhizosphere soils. The variables measured included abundance of soil bacteria, fungi and actinomycetes, catalase activity, acid phosphatase activity, urease activity, pH, organic matter (OM), total nitrogen (N), total phosphorus (P), total potassium (K), hydrolysis N, available P and available K. The effects of shrubs on soil properties were evaluated by making a comprehensive comparison of the soil fertility of rhizosphere and non-rhizosphere soils. The results showed: 1) Microbial abundance and enzyme activities of rhizosphere soil were generally higher than non-rhizosphere soil except for the abundance of actinomycetes and urease activity. Specifically, only the total microbial abundance of rhizosphere soil under N. tangutorum was less than that of non-rhizosphere soil, and urease activity in rhizosphere soils under all 4 shrubs were lower than their non-rhizosphere soils; 2) Soil pH, N, P and total K contents of rhizosphere and non-rhizosphere soils were not significantly different. OM, P and available K of rhizosphere soils were significantly higher than non-rhizosphere soils, whereas available N was higher in non-rhizosphere than rhizosphere soils. These results showed that the presence of the 4 shrubs could improve soil OM, available P and available K contents through root aggregation; 3) There was no significant correlation between soil enzyme activities and soil microbial abundance, OM was significantly correlated with soil bacterial and fungal abundance, and soil available P content was significantly correlated with abundance of soil bacteria, fungi and actinomycetes. The soil available P content was significantly correlated with soil catalase and acid phosphatase activities and the soil total N and available N were significantly positively correlated with urease activity. From these results, the correlation between soil nutrients and enzyme activities differed from the correlation between soil nutrients and microbial abundance, thus soil enzyme activities and microbial abundance were proven to have a significant influence on soil nutrients; 4) The results of principal component analysis showed that soil bacterial abundance, fungal abundance, catalase activity, acid phosphatase activity, pH, OM, total N, total K and available P contents showed high factor loadings and these indexes could provide important information on soil fertility. In addition, principal component factor scores and synthetic scores of soil fertility levels showed that soil fertility of rhizosphere soils was significantly higher than non-rhizosphere soils. It showed that the soil fertility of rhizosphere soils ranked as follows: B. vernae > C. korshinskii > C. brevifolia > N. tangutorum. These results indicated that the soil improvement effect of B. vernae and C. korshinskii were better than the other 2 species, indicating that they may be suitable for afforestation in Xining Nanshan. B. vernae and C. korshinskii could be selected as afforestation tree species to help improve soil fertility and make contributions to revegetation. N. tangutorum was not suitable for afforestation in Xining Nanshan. Our research could provide important basic data and theoretical support for choosing suitable afforestation tree species for revegetation by artificial afforestation in Xining Nanshan.