Abstract:We examined the composition of a soil seed bank in a desert steppe ecosystem under different modes of grassland management. The study was conducted on the desert steppe in the middle of the Ningxia Hui Autonomous Region (NHAR), in northern China. We collected soil seed bank samples in late March 2012, along a gradient of grazing intensities that encompassed 8 distinct treatments: no grazing in enclosures (after a duration of 1, 3, 5 and 7 years); grazing (free range at a medium intensity, and four areas of rotational grazing at a medium intensity), replanting (with Astragalus adsurgen and Alfalfa stiva, in 2006) and open grazing. Following the method of greenhouse germination, soil samples were collected, planted, and left to sprout. The species composition and abundance of the resulting plants were determined, and characteristics of the soil seed bank such as the similarity of soil seed bank and aboveground vegetation, diversity of soil seed bank were also studied. The results were as follows. 1) In total, we recorded 26 plant species belonging to 6 families and 14 genera. Lespedeza potanimii was the only subshrub, while all other species observed were herbaceous plants. The main species in the seed bank of the desert steppe soil were perennial plants, accounting for 65.3%, while 35.7% of plants were annuals. The number of seeds decreased as enclosure time increased, and was highest in the soil from the one-year enclosure and grasslands with no enclosure. 2) We found 2-6 common species in the seed bank compared to what was growing what above ground in all treatments. Long-term enclosure of the steppe grassland may lead to a decline in similarity between the soil seed bank and above-ground vegetation, as the number of species shared between the seed bank and above-ground plant community decreased with as enclosure time increased. In the free range treatment, there was a lower number of shared species. The species which had been replanted to restore the desert steppe did not appear in the seed bank. 3) There was no significant relationship between soil seed bank diversity and management modes. The highest species diversity was measured in the four-area rotation grassland, while the lowest was in the grassland enclosed for 7 years. The highest species richness was found in the one-year enclosure. The highest evenness index was in the free grazing grassland, and the lowest was in the replanted grassland. Seed banks play a central role in the regeneration of vegetation, and managers can rely on soil-stored seed banks for restoration of desert steppe with appropriate grazing disturbance, more so than long-time enclosures. Alien species that were replanted did not benefit seed banks, and the establishment of introduced species may have restricted the germination of certain indigenous pioneer species.