No abstract

Summary1 Disturbance may cause community composition across sites to become more or less homogenous, depending on the importance of different processes involved in community assembly. In north-eastern North America and Europe local (alpha) diversity of forest plants is lower in forests growing on former agricultural fields (recent forests) than in older (ancient) forests, but little is known about the influence of land-use history on the degree of compositional differentiation among sites (beta diversity). 2 Here we analyse data from 1446 sites in ancient and recent forests across 11 different landscapes in north-eastern North America and Europe to demonstrate decreases in beta diversity and in the strength of species-environment relationships in recent vs. ancient forests. 3 The magnitude of environmental variability among sites did not differ between the two forest types. This suggests the difference in beta diversity between ancient and recent forests was not due to different degrees of environmental heterogeneity, but rather to dispersal filters that constrain the pool of species initially colonizing recent forests. 4 The observed effects of community homogenization and weakened relationships between species distributions and environmental gradients appear to persist for decades or longer. The legacy of human land-use history in spatial patterns of biodiversity may endure, both within individual sites and across sites, for decades if not centuries.

In fragmented landscapes, habitat patches are often destroyed and created through time, though most metapopulation models treat patch networks as static. Here we present a generally applicable, modified version of Hanski's Incidence Function Model (IFM) that incorporates landscape dynamics (i.e., habitat patch turnover), and we parameterize the model with data on patch occupancy patterns for forest plants in central Lincolnshire, UK. The modified IFM provided a better, or equally good, fit to species' patch occupancy patterns than logistic regression. Estimated colonization and extinction rates, and the results of logistic regression analyses, varied significantly among species with different life history traits. For example, species with low seed production and predominantly short-distance seed dispersal showed lower rates of colonization and extinction and were more likely to show effects of patch age and connectivity on patch-level presence than species with the opposite set of traits. Model simulations demonstrate a profound negative influence of habitat turnover rate on metapopulation dynamics and persistence, particularly for slow-colonizing species. The incorporation of temporal habitat dynamics into the metapopulation paradigm will permit its application to organisms in a much wider range of real landscapes. Key words: dynamic landscapes; forest plants; Incidence Function Model; life history traits; metapopulation dynamics and persistence. APPENDIX A Results of logistic regression analyses and fitting the incidence function model for all species are available in ESA's Electronic Data Archive: Ecological Archives E085-115-A1. APPENDIX BSimulation results to test the parameter-fitting procedure for the modified incidence function model are available in ESA's Electronic Data Archive: Ecological Archives E085-115-A2.