Question To what extent does the movement of animals between fragmented habitat patches provide functional connectivity via endozoochorous seed dispersal? Location The Stockholm archipelago, Sweden. Methods We followed all movements of livestock between islands during one grazing season. After each movement, manure was collected and its seed content assessed through seedling emergence. Seedling data were then compared to vegetation surveys from the grazed islands with regard to functional traits. Results Light‐ and nitrogen‐demanding locally abundant species, and those with relatively small and persistent seeds were more likely to be moved between islands. For quantitative traits, only a subset of the available trait ranges were dispersed, with extreme values left behind. Species apparently specialized to other means of dispersal emerged from the manure samples. Neither dispersed traits nor seed density changed with timing of movement, but seed richness and diversity both increased throughout the season. The subsets of endozoochorously‐dispersed species in the established vegetation were more similar than non‐dispersed subsets between islands linked by livestock. Conclusions Grazing networks contribute to the connectivity of the core species in the system, and could provide useful tools for grassland management in fragmented landscapes.

Summary1. Theoretical models show that environmental heterogeneity and dispersal are major determinants of species diversity at multiple scales, yet there are few studies from real landscapes that adequately integrate variation in the surrounding matrix. Understanding how landscape context and management influence species composition and diversity patterns across habitats and scales is an important goal in ecology with relevance for both management and conservation. 2. We used a system of 25 landscapes distributed across islands in the Baltic Sea to investigate the effect of current and historical landscape context and management on plant diversity and composition in grassland communities. Plant diversity was measured at three hierarchical scales (1 m 2 , habitat, landscape) in grazed fields and adjacent wood pastures to calculate a-, b-and c-diversity values across habitats and scales. 3. Structural equation modelling was used to model and quantify the effects of landscape context on species diversity and spatial turnover, and constraint analysis of principal coordinates to relate variation in species composition to landscape variables. 4. Proportion of open land, spacing and grazing intensity positively affected species diversity in both habitats, whereas the effect of historical landscape context was only significant in open fields. Plant diversity in field pastures was mainly determined by the number of species found at a small scale, while both local species density and spatial turnover were key determinants of diversity in wood pastures. 5. Habitat proximity influenced species composition as compositional similarity was higher between adjacent field and wood pastures compared to randomly paired habitats. Although increasing flow of propagules from adjacent patches can promote local coexistence, dispersal can result in spatial homogenization. 6. Synthesis. Plant diversity in grassland communities is substantially influenced by species occurring in adjacent habitats. While the effect of landscape context and management on small-scale diversity was consistent across habitats, the effect on spatial turnover was habitat specific. Our study shows that plant diversity is structured through the interplay between local and landscape processes and highlights that plant communities in specific habitat types cannot be considered in isolation from the surrounding landscape matrix.