Background/Question/Methods A adult tree in the canopy of an intact forest can produce millions of seeds over its lifetime. On average, only one of those seeds will replace it as a reproductive adult in the canopy. Diverse forests, however, are often assumed to be maintained by broadly dispersed propagules from a regional seed pool that can preserve populations of locally rare species against competitive exclusion, or drive models of how forest stands remain diverse through input from a hypothetical regional community. Theories based on the importance of dispersal mechanisms, therefore, do not align with the reality of extreme mortality of seeds. Direct tests of the importance of seed dispersal in maintaining biodiversity have been inhibited by a lack of processing power needed to run models that can simulate realistic dispersal and recruitment of millions of seeds across landscapes. We use an analytical model and a simulated forest to test the hypothesis that LDD can contribute significantly to stand biodiversity at a range of scales and distances.

Results/Conclusions We demonstrate through both our analytical analysis and simulations that at important ecological scales long-distance dispersal (LDD) has little impact on forest biodiversity. We show that local seed inputs overwhelm migrating seeds to such a degree that LDD cannot importantly affect diversity unless species have strong differences in competitive ability over the course of their life-histories. Because of the high density of seeds derived locally, seed immigration is orders of magnitude lower than assumed in previous models that do not rely on dispersal data and ignore competition. LDD will not allow rapid redistribution of species in response to climate change, except in areas lacking local seed sources or for species having strong competitive later-stage recruitment advantages in local habitats. Anticipating the persistence of forest biodiversity in coming decades will require knowledge of species' capacities to exploit habitat heterogeneity.