Bounding seed loss from isolated habitat patches

Abstract

Dispersal of propagules (seeds, spores) from a geographically isolated population into an uninhabitable matrix can threaten population persistence if it prevents new growth from keeping pace with mortality. Quantifying propagule loss can thus inform restoration and conservation of vulnerable populations in fragmented landscapes. To model propagule loss in detail, one can integrate dispersal kernels (probabilistic descriptions of dispersal) and plant densities. However, one might lack the detailed spatial information and computational tools needed by such integral models. Here we derive two upper bounds on the probability of propagule loss--one assuming rotational symmetry of dispersal and the other not--that require only habitat area, habitat perimeter, and the mean dispersal distance of a propagule. We compare the bounds to simulations of integral models for the population of Asclepias syriaca (common milkweed) at McKnight Prairie--a 13.7 hectare reserve surrounded by agricultural fields in Goodhue County, Minnesota--and identify conditions under which the bounds closely estimate propagule loss.