Questions: What are the relative roles of abiotic and grazing management factors on plant community distribution in landscapes? How are livestock type and stocking rate related to changes in vegetation structure and composition? Location: Sub‐alpine grasslands in the central and eastern Pyrenees. Methods: Multivariate analysis and variance partitioning methods were used to evaluate the relative roles of environmental factors in structuring vegetation composition and diversity patterns in three surveys on differently managed grasslands. Results: Vegetation composition within a region was affected by environmental factors hierarchically, changing first according to abiotic factors and then to grazing management. At landscape scales, abiotic factors explained two‐fold more variation in vegetation composition than grazing factors. Within landscape units, cattle grazing increased vegetation heterogeneity at landscape and patch scales, while sheep grazing favoured the presence of a specific set of species with high conservation value. Species composition was highly responsive to management variables compared to diversity components. Conclusions: The combination of sheep and cattle grazing at various stocking rates is an effective tool to preserve the diversity of plant species and communities within a region with a long tradition of livestock management, through the scaling up of effects by local processes occurring in patches at smaller scales.

We studied vegetation responses to disturbances originated by ants and voles in subalpine grasslands in the Eastern Pyrenees. We compared the effects of these small-scale disturbances with those of a large-scale disturbance caused by ploughing. We wanted to know if these soil disturbances promoted species richness through the existence of a specific guild of plants colonizing these areas, and if this guild was the same for all soil disturbances, independently of their extent. In general, grassland vegetation seemed to recover relatively quickly from soil-displacement disturbances, and the effects could be scaled up in time and space in terms of species richness and composition. Vole mound composition was similar to that in the surrounding grassland, suggesting that mounds were rapidly colonized by the neighbouring vegetation. Vegetation composition differed between the grassland and the ant mounds. Grasses and erect dicots coped well with repeated disturbance, while rosette-forming species and sedges were very sensitive to it. Landscape processes could be important to understanding recolonization. Species from xeric grasslands were found in mesic grasslands when disturbed by ploughing and on the tops of active ant mounds. Furrows in mesic grasslands recovered well, but decades after disturbance showed long persistence of some xeric species and increased species richness compared to terraces, while xeric grasslands showed decreased richness. This suggests that, because of those disturbances, within-habitat diversity was increased, although landscape diversity was not. However, specific disturbances showed idiosyncratic effects, which could enhance the species richness globally. In ant-affected areas, the grassland itself showed the highest plant species richness, partially associated to the presence of some species with elaiosomes not, or only rarely, found in adjacent grasslands without ant mounds. Therefore, soil disturbances occurring at different spatial scales contributed to complexity in vegetation patterns in addition to abiotic factors and grazing.