Aim We investigated how current and historical land use and landscape structure affect species richness and the processes of extinction, immigration and species turnover. Location The northern part of the Stockholm archipelago, Baltic Sea, Sweden. We resurveyed 27 islands ranging from 0.3 to 33 ha in area. Methods We compared current plant survey data, cadastral maps and aerial photographs with records obtained from a survey in 1908, using databases and a digital elevation model to examine changes in plant community dynamics in space and time. We examined the effects of local and landscape structure and land use changes on plant species dynamics by using stepwise regression in relation to eight local and three landscape variables. The eight local variables were area, relative age, shape, soil heterogeneity, bedrock ratio, number of houses, forest cover change, and grazing 100 years ago. The three landscape variables were distance to mainland, distance to closest island with a farm 100 years ago, and structural connectivity. Hanskis connectivity measure was modified to incorporate both connectivity and fragmentation. Results The investigated islands have undergone drastic changes, with increasing forest cover, habitation, and abandonment of grassland management. Although the total species richness increased by 31% and mean island area by 23%, we found no significant increase in species richness per unit area. Local variables explain past species richness (100 years ago), whereas both local and landscape variables explain current species richness, extinctions, immigrations and species turnover. Grazing that occurred 100 years ago still influences species richness, even though grazing management was abandoned several decades ago. The evidence clearly shows an increase in nitrophilous plant species, particularly among immigrant species. Main conclusions This study highlights the importance of including land use history when interpreting current patterns of species richness. Furthermore, local environment and landscape patterns affect important ecological processes such as immigration, extinction and species turnover, and hence should be included when assessing the impact of habitat fragmentation and land use change. We suggest that our modified structural connectivity measure can be applied to other types of landscapes to investigate the effects of fragmentation and habitat loss.