Stuck Between a Rock and a Hard Place: Boring Sponges Undermine Oyster Restoration Efforts Outside of a Narrow Salinity Regime in Pamlico Sound

Global oyster populations have declined precipitously, and coupled with these declines are the losses of the wealth of ecosystem services oysters provide, from water filtration to shoreline armoring. After commercial landings crashed to <5% of their historical levels, North Carolina’s Division of Marine Fisheries (NCDMF) designated a network of subtidal eastern oyster spawning sanctuaries to enhance Pamlico Sound’s broodstock. Using NCDMF’s oyster sanctuary monitoring dataset, we: 1) describe decadal network- and individual sanctuary-level oyster population trajectories; 2) evaluate the relationship between oysters and a major pest, clionid sponges; and 3) determine the role salinity plays as a mechanistic driver of sponge-oyster population dynamics. Across the network, reproductive oyster densities increased substantially in the first five years (meeting success criteria during 76% of surveys) but declined over the subsequent 10+ years (meeting success criteria during 59% of surveys). Many individual sanctuaries exhibited similar patterns, though notably two sanctuaries collapsed altogether within five years of establishment. Those sanctuaries are closest to the barrier island inlets that exchange water between Pamlico Sound and the Atlantic, and have the highest mean salinities within the network as well as the greatest occurrence of boring sponge. Restored oyster populations performed best within a narrow regime of mean salinity 14-17 psu and 1-14% salinity occurrences <10 psu. Under fresher regimes, sponge was less prevalent, but conditions were not optimal for oyster growth and reproduction. Contrastingly, more saline regimes facilitated rapid sponge recruitment, higher declines of adult oysters between surveys, and ultimately subsequent sanctuary collapse. In a generalized additive model, the predictors of substrate age, sponge occurrence, and salinity explained 47.2% of model deviance. We recommend managers develop sanctuary-specific success criteria and continue monitoring of this network to evaluate sanctuary lifespan.