Abstract:Bromus inermis Leyss. is a dominant rhizomatous grass in the Othidag Sandland. In late autumn and early spring, strong winds move sand dunes quickly, depositing new sand on the edge of lowland meadows and creating sand patches of different sizes within lowland meadows. Frequently, clonal fragments (i.e. interconnected ramets) of B. inermis inhabiting the lowland meadows are partially buried in sand, i.e. some ramets are buried in sand and some are not. We hypothesized that partially-buried clonal fragments benefit from clonal integration. A pot experiment was conducted in a greenhouse in the Otingdag Sandland. In the experiment, each clonal fragment consisted of four interconnected ramets. The two proximal（older）ramets were not buried, while the two distal (younger) ramets were either not buried (unburied control) or buried at depths of 10, 20, or 30 cm in sand (burial treatments). Rhizome connections between proximal and distal ramets were severed or left intact. Rhizome severing dramatically decreased survival of distal ramets buried in sand. A cost-benefit analysis based on plant performance in terms of biomass production, number of shoots, number of leaves, number of rhizome nodes and total rhizome length showed that proximal ramets supported connected distal ramets buried in sand by clonal integration but did not incur any cost from resource export. Furthermore, effects of clonal integration on growth and survival of B. inermis were greater at deeper depths of burial in sand. We speculate that clonal integration may be an important adaptation of B. inermis to harsh burial condition in sand that increases its growth and survival, and thus fitness, in semiarid sandy lands.