The Use of GAM Modeling Techniques to Evaluate the Effects of Freshwater Flow Into Florida Bay- Part 1- Forage Fish Models
Darlene Johnson and Joan Browder
NOAA Fisheries, Miami, FL
An integrative database was assembled from three decades of forage fish studies conducted in Florida Bay. GAM models for 10 species of fish and pink shrimp were developed from this database. Samples were linked to other information: tidal amplitude, sea level, rainfall, freshwater discharge indices, wind, salinity, and habitat type (physical habitat, seagrass density, and seagrass type). The 11-species community selected for statistical models comprised 87% of the throw-trap samples by number for 1980’s and 1990’s. The group comprised 86% of the 1980’s trawl data, 66% -89% of the 1990’s trawl data, and 74% of the 1990’s seine samples. During the 1970's, the 11-species group comprised 64% of the fauna sampled in the western bay, 92% in the central bay, and 39% in the eastern bay. A second database was prepared that combined ENP creel census data with environmental variables.
Freshwater flow was a significant model variable for all species, while month, seagrass density, seagrass type, and tidal amplitude were significant variables for 10 of the 11 species, followed by depth (9 species), habitat and temperature (8 species), and salinity and rainfall (7 species). Gear and wind (6 species), and sea level (5 species) were less important variables in predicting the abundances of forage species in the Bay.
All the resident species (except goldspotted killifish) and the pelagic species, bay anchovy (which spawns both offshore and inshore), showed a positive relationship with freshwater flow into the Bay. In contrast, the three species that spawned outside the Bay showed either a negative relationship with the previous month’s freshwater flow that flattened out at higher flows (pinfish) or a negative/positive relationship to lagged freshwater flow.
1
Three species (bay anchovy, code goby, and gulf pipefish) were negatively correlated with salinity and two species (dwarf seahorse and mojarras) declined under hypersaline conditions. Pink shrimp and gulf toadfish showed a parabolic relationship to salinity peaking at 29 ppt and 32 ppt respectively. One species (code goby) was positively correlated with salinity. Predicted abundance of the two smallest species (rainwater killifish and goldspotted killifish) was highest at extreme conditions (low and high salinity) which suggests competitive exclusion or increased predation at moderate salinities.
.
The density of two resident species (rainwater killifishand clown goby) was negatively related to rainfall, one resident species (code goby) was positively related to rainfall, and another resident species (goldspotted killifish) had a parabolic relationship with rainfall. Among the outside spawners, pinfish was positively correlated with rainfall (previous month), and mojarras were negatively correlated with lagged rainfall.
Two resident species (rainwater killifish and gulf pipefish) and the three offshore spawning species (pinfish, mojarras, and pink shrimp) were positively correlated with seagrass density. Predicted densities of gulf toadfish, dwarf seahorse, and goldspotted killifish were significantly higher at moderate seagrass densities. Densities of the pelagic bay anchovy were also significantly higher at moderate seagrass densities. Densities of the two gobies were highest in sparse seagrass. Increased densities of five species (gulf toadfish, dwarf seahorse, clown goby, code goby, and bay anchovy) may have resulted from decreases in seagrass density due to seagrass die-offin the late 1980's. The models suggest that declines in seagrasses from moderate to sparse would result in increases in the two goby species, but declines in gulf toadfish and dwarf seahorse. The models predict that declines in dense seagrass would lead to declines in rainwater killifish, mojarras, pinfish, pink shrimp and gulf pipefish.
Seagrass type was important for all species but gulf toadfish, with significant higher densities in Syringodium (pinfish, dwarf seahorse, rainwater killifish), Halodule (pink shrimp, gulf pipefish, goldspotted killifish), mixed seagrass with Thalassia dominant (code goby), mixed seagrass with Syringodium dominant (mojarras, clown goby). The bay anchovy showed the strongest correlation to areas with no seagrass.
Densities of the three offshore spawners--pink shrimp, pinfish, mojarras were positively related to one or more physical parameters that may be indices of transport mechanisms for recruitment: tidal amplitude, sea level, and wind forcing. Densities of bay anchovy, a species that may spawn outside of the Bay as well as within it, also were positively related to these factors. In contrast, those species that spawned exclusively within the Bay and may be less dependent upon recruitment mechanisms showed less correlation with these factors.
1
Johnson, Darlene, NOAA Fisheries, 75 Virginia Beach Drive, Miami, FL 33149
Phone: 305 361-4490, Fax 305 361-4478, , Question 5
1