1.Overgrazing-induced grassland degradation has become a severe ecological problem worldwide. The diversity and composition of soil microbial communities are responsive to grazing disturbance. Yet, our understanding is limited with respect to the effects of grazing intensity on bacterial and fungal communities, especially in plant rhizosphere. 2.Using a long-term (15 years) grazing experiment, we evaluated the richness and composition of microbial communities in both rhizosphere and non-rhizosphere regions, under light, moderate, and heavy intensities of grazing, in a semiarid grassland. We also examined the relative roles of grazing-induced changes in some abiotic and biotic factors in affecting the richness and composition of microbial communities. 3.Our results showed that the responses of soil bacteria to grazing intensity differed greatly between rhizosphere and non-rhizosphere, and so did soil fungi. Specifically, the bacterial richness decreased markedly under moderate and heavy grazing in rhizosphere soil, whereas little impact on the fungal richness was observed. For microbial composition, with the increase in grazing intensity, an increase in dissimilarity among bacterial communities was observed, and this trend also held true for the fungal communities. Hierarchical partitioning analyses indicated that the bacterial composition in rhizosphere was primarily driven by root nitrogen and soil nitrogen concentrations while that in non-rhizosphere by soil available phosphorus. In addition, soil available phosphorus played an important role in affecting the fungal composition in both rhizosphere and non-rhizosphere regions. 4.Synthesis: This study provides direct experimental evidence that the richness and composition of microbial communities were severely altered by heavy grazing in a semiarid grassland. Thus, to restore the grazing-induced, degraded grasslands, we should pay more attention to the conservation of soil microbe in addition to vegetation recovery.

Passive restoration (without any intervention) has been proposed as an effective strategy for degraded cropland restoration. However, whether the vegetation in abandoned cropland can change towards the desired state and the time needed to reach a stable state are still uncertain. We investigated three abandoned croplands with different recovery times (5, 15 and 20 years) and one natural grassland in each of two different types of grassland (desert steppe and typical steppe) in the agro-pastoral ecotone of northern China to assess the restoration potential of abandoned cultivated grassland. Above- and below-ground productivity as well as species biodiversity increased gradually with increasing recovery time. After 20 years of restoration there was no significant difference between abandoned cropland and natural grassland in the typical grassland site, but above- and below-ground productivity and species biodiversity were still lower in abandoned cropland in the desert steppe site. At the beginning of restoration, the communities were dominated mainly by annual species, especially in the desert steppe. As recovery time increased, the biomass and richness of perennial grasses and forbs increased significantly and replaced annual species as the dominant species. In both desert steppe and typical grassland, species similarity between restored and natural grassland increased over time, suggesting that previously cultivated grassland recovered towards the desired state. Our results indicate that 20 years was long enough for the restoration of previously cultivated grassland in the typical steppe, but more time may be needed in the desert steppe.