| Abstract |
Patchy landscapes are characterized by abrupt transitions between habitats, forcing species to cross habitat boundaries in order to spread. Although individuals that disperse into unsuitable habitat are often presumed dead in population models, reality may be more nuanced. If survival within an unsuitable habitat is possible (however transiently), then through time, individuals may be able to reach distant, suitable habitat. In this study, conducted with annual plant species in a fragmented Californian grassland, we examined the fate of habitat specialists moving between serpentine patches and an invaded nonserpentine matrix. First, we tested whether habitat specialists use seed dormancy to transiently survive beyond the habitat boundary. We found that dormancy maintained populations of patch specialists deep into the matrix—these patterns were not observed for matrix specialists in serpentine patches. Seeds of patch specialists that lacked morphologies for assisted dispersal accumulated downslope of patches, suggesting that even the most dispersal‐limited species could eventually reach suitable patches even if they first land in the matrix. Second, because an invasive matrix specialist ( Avena fatua ) is highly abundant, we were interested in determining whether seeds spilled into serpentine patches were strong enough to maintain sink populations. We found that A. fatua was largely absent deep in patches, where the components of fitness plummeted and there was no evidence of a dormant seedbank (in contrast to a sizable seedbank of patch specialists in the matrix). Our results suggest that the demographic fates of individuals that land in unsuitable habitat depend on the direction by which the boundary is crossed (patch ➝ matrix ≠ matrix ➝ patch), with different ecological consequences for species in each habitat type. Dormancy is often understood as a mechanism for persisting in the face of temporal variability but may serve as a means of traversing unsuitable habitat in patchy systems, warranting its consideration in estimates of habitat connectivity. |
, Rachel M. Germain 
