Success of restoration projects requires an understanding of the direction and rate of natural and anthropogenic change within a dynamic ecosystem, as well as understanding the best environmental conditions a target species requires for persistence.  Yet predicting future states for optimal restoration is especially challenging for rare species in ecosystems that are neither spatially nor temporally constant and may be particularly vulnerable to fragmentation. Our experimental rare plant reintroductions of U.S. endangered Jacquemontia reclinata and FL endangered Lantana canescens in two South Florida ecosystems revealed both species-specific variables and general patterns impacting restoration in urban ecotones.  Associated with J. reclinata persistence in coastal strand was topographic position from high tide line that was correlated with salt exposure and significantly affected plant survival after hurricanes.  Beyond the species-specific factors impacting plant persistence, both ecotone species had optimal habitats that are spatially wedged between stochastic disturbance and predictable  succession of woody vegetation.  Human activities including road building, beach raking, and fire suppression further constrain the spatial dynamism of ecotones and may eventually eliminate them and their species if trends continue. We present a general conceptual model that incorporates disturbance and succession to predict persistence of populations in ecotones, and suggests management components for restoring and maintaining ecotones to preserve diverse taxa.