Environmental influences on potential recruitment of pink shrimp, Fatlantopenaeus duorarum , from Florida Bay nursery grounds

Abstract

Two modeling approaches were used to explore the basis for variation in recruitment of pink shrimp, Farfantepenaeus duorarum, to the Tortugas fishing grounds. Emphasis was on development and juvenile densities on the nursery grounds. An exploratory simulation modeling exercise demonstrated large year-to-year variations in recruitment contributions to the Tortugas rink shrimp fishery may occur on some nursery grounds, and production may differ considerably among nursery grounds within the same year, simply on the basis of differences in temperature and salinity. We used a growth and survival model to simulate cumulative harvests from a July-centered cohort of early-settlement-stage postlarvae from two parts of Florida Bay (western Florida Bay and northcentral Florida Bay), using historic temperature and salinity data from these areas. Very large year-to-year differences in simulated cumulative harvests were found for recruits from Whipray Basin. Year-to-year differences in simulated harvests of recruits from Johnson Key Basin were much smaller. In a complementary activity, generalized linear and additive models and intermittent, historic density records were used to develop an uninterrupted multi-year time series of monthly density estimates for juvenile rink shrimp in the Johnson Key Basin. The developed data series was based on relationships of density with environmental variables. The strongest relationship was with sea-surface temperature. Three other environmental variables (rainfall, water level at Everglades National Park Well P35, and mean wind speed) also contributed significantly to explaining variation in juvenile densities. Results of the simulation model and two of the three statistical models yielded similar interannual patterns for Johnson Key Basin. While it is not possible to say that one result validates the other, the concordance of the annual patterns from the two models is supportive of both approaches.