USING LINKED BAYESIAN NETWORKS TO PREDICT SEA-LEVEL RISE-DRIVEN GEOMORPHOLOGIC CHANGES TO CRITICAL BARRIER-ISLAND HABITAT

Protecting human development and habitat quality on barrier islands is a persistent focus of coastal management efforts. These objectives, which are sometimes in conflict, will become more challenging as sea-level rise accelerates and storm regimes shift. We developed multiple, independent Bayesian networks that address long-term shoreline change, barrier-island morphology, and piping plover (Charadrius melodus) habitat availability that can be linked to predict overall barrier island evolution to inform management decisions. Bayesian networks are trained with metrics derived from remotely sensed spatial datasets and piping plover presence/absence observations to describe barrier island conditions. Here we describe applications for this approach at Fire Island, New York and Assateague Island, Maryland/Virginia, which are two locations where natural resource and habitat management needs must be considered alongside human development and access. Results from habitat modeling emphasize the importance of periodic disturbance from storms to maintain barrier island morphological characteristics that provide habitat for the species of interest. In particular, sites where storm-driven washovers occur demonstrate strong habitat potential for piping plovers. We also describe how we are extending the approach using spatial and habitat-use datasets spanning 22 sites along the northeastern United States coast that represent different barrier-island geomorphologic and habitat states. Our goal is to use this modeling framework to evaluate scenarios related to coastal management plans and/or future scenarios where shoreline-change rates are likely to exceed those that have been observed historically.