Seed dispersal and distributions of woody plants across temperate forest fragments

Seed dispersal has been shown to limit plant distributions and seedling recruitment for some species within local habitats. However, dispersal has rarely been quantified at larger spatial scales and experimental evidence of dispersal limitation across isolated forests is sparse. To test for dispersal limitation within and between forest fragments, I examined species composition of forest fragments across isolation gradients, measured seed and seedling distributions, and added seeds to examine recruitment. As predicted under dispersal limitation, native species richness of 26 forest fragments in southeastern Michigan decreased with increasing forest isolation. In addition, native bird-dispersed shrubs with heavier seeds (poor dispersers) displayed stronger declines across isolation gradients than species with lighter seeds (good dispersers).; To quantify seed dispersal within and between fragments, I collected seeds from traps placed at 40m intervals throughout five fragments (4–8 ha). Within fragments, seed dispersal was limited for most species. Animal-dispersed seed was poorly dispersed, hitting fewer traps than wind-dispersed seed. Heavy-seeded wind-dispersed species required high source densities to saturate fragments with seed. To quantify between-fragment dispersal, I took advantage of species absences from individual forest fragments. Dispersal into fragments was detected for only four of seventeen species, all four of which were disturbance-adapted. Although three species (Acer rubrum, Betula alleghaniensis, Tsuga canadensis) had widely-dispersed seed within fragments containing adults; I detected no between-fragment dispersal. Such restricted seed distributions highlight the importance of incorporating seed dispersal across scale in forest modeling and management.; Seed and seedling distributions were correlated for four species. However, conditions could be inappropriate for recruitment in locations without seed. To test this, I added seed of Lindera benzoin to unoccupied microsites within fragments with and without natural populations. After three years, seed additions significantly increased recruitment in three of four fragments, with no difference in final recruitment densities among fragments. However, insect herbivory was higher on Lindera seedlings in fragments without natural populations, suggesting multiple factors may limit recruitment. Results demonstrate that dispersal can limit species distributions in fragmented landscapes, and that additional factors may act to constrain recruitment and distributions. Such complex responses should make management of fragmented landscapes challenging.