Increasing spatial dispersion in ecosystem restoration mitigates risk in disturbance-driven environments

Many vegetated ecosystems, including drylands, coastal dunes, salt marshes and seagrass meadows, inhabit environments frequently disturbed by the erosive forces of wind and water. Once degraded, the restoration of these systems entails a high risk of failure due to the uncertainty in timing and intensity of future disturbances. Risk-mitigation strategies like bet-hedging (i.e. spreading risk over diverse options) have been proven in cross-disciplinary contexts to optimize yield when uncertainty is high. Yet, restoration designs commonly homogenize resources by planting vegetation of similar sizes in grid-like patterns. This decision may unwittingly contribute to the high rate of restoration failure in these environments. Using numerical simulations mimicking vegetation patch dynamics, we demonstrate how avoiding uniform planting designs substantially improves the likelihood of restoration success. These simulations also suggest that the intrinsic risk of failure associated with any planting pattern can be identified a priori by calculating the variance-to-mean ratio of vegetation cover. Synthesis and applications. By introducing a level of spatial overdispersion (variance in vegetation clustering) into restoration planting designs, projects will insure themselves against the uncertainty imposed by disturbances, limited by their willingness to accept a lower rate of recolonization.