The Use of GAM Modeling Techniques to Evaluate the Effects of Fresh-water
Flow into Florida Bay-Part 2- Application of Models to Predict Community
Composition in Extreme Wet and Extreme Dry Years
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 and GAM models for 10 species and pink shrimp were developed from it. These 11 species models were used collectively as a community index to predict the densities of the forage community for the wet
(August) and dry (April) seasons of an extreme wet (1995) and extreme dry year
(1990) for each habitat of each basin of the bay. Areal coverage of each habitat of each basin was used to calculate abundance and biomass for the entire bay in each of time period. The approach allows a bay-wide assessment to be made based on changes in freshwater input and any resulting changes in salinity or seagrass. The project objective was to determine whether certain species densities or the combined 11-species density might serve as a performance measure to evaluate water management alternatives. We hind-casted densities for wet and dry seasons of a previous wet and dry year to evaluate the potential usefulness of the models as predictors (hydrologic and hydrodynamic model output was unavailable).
Models predicted that bay-wide abundance was seasonal and higher during wet seasons (August) than dry seasons (April). Predicted abundance of the 11 species was slightly higher in the wet year (1995) than the dry year (1990). The highest predicted abundance was during the wet season of the wet year (August 1995).
The total predicted abundance in April 1995 (dry season, wet year) was 1.05 times higher than in April 1990 (dry season, dry year), and the total predicted abundance in August 1995 (wet season, wet year) was 1.07 times higher than in
August 1990 (wet season, dry year).
Predicted response differed by species. The predicted bay abundance of Lucania,
Eucinostomus, Floridichthys, and Farfantepenaeus was seasonal and was highest in August. Highest predicted numbers of Lucania were in August of the wet year, and highest predicted numbers of the other three species were in August of the dry year. Highest predicted numbers of Microgobius, Gobiosoma, Anchoa,
Syngnathus , and Hippocampus occurred during the wettest period, August 1995.
The predicted abundance of Lagodon was highest during the driest period, April
1990. The predicted abundance of Opsanus was highest during times of intermediate conditions: wet season/dry year followed by dry season/wet year.
On a Bay-wide basis, there were no major changes in predicted species
1

composition among the four time periods. Four taxa were consistently dominant in terms of percent composition by number during all time periods: Lucania,
Farfantepenaeus, Anchoa , and Eucinostomus . The relative importance of the individual species varied seasonally. Eucinostomus were more important in April than August. Opsanus and Floridichthys were most important in the driest period.
Percent composition of Syngnathus was higher in both seasons of the wet year.
Patterns of density for the combined 11 species were similar across all habitats.
When bay-wide total abundance was greatest (August 1995), basin habitat, followed by the mainland shoreline habitat, had the greatest predicted density, and bank habitat the lowest. Predicted densities were higher in August than April and in 1995 than in 1990 for both seasons. Individual species responses were habitat dependent. Predicted density was highest in basin and mainland shoreline habitats and lowest on banks. Highest predicted density of Anchoa was in channel habitat followed by basin habitats and lowest predicted density was on banks. Eucinostomus predicted densities were highest in island and mainland shoreline habitats and lowest in channel habitat. Predicted densities of
Farfantepenaeus were highest in island shoreline and basin habitat and lowest in channels. Floridichthys predicted densities were highest on banks, mainland shorelines, and island shorelines and lowest in basins and channels. Predicted densities of Gobiosoma were highest in mainland shoreline habitat and lowest in island shoreline habitat. H ippocampus predicted densities were highest around island shorelines and lowest along mainland shorelines. Predicted densities of
Lagodon were similar in all habitats with highest in basin and channel habitats.
Lucania predicted densities were highest in basin and channel habitats and lowest on banks and island shorelines. Predicted densities of Microgobius were highest in channels and basins and lowest on banks and island shorelines. Opsanus predicted densities were highest in basin and channel habitats and lowest on banks. Predicted densities of Syngnathus were highest on bank and mainland shoreline habitats and lowest in channels and basins.
Predicted percent composition differed more on a habitat basis. In bank habitat
(21% of bay area), the highest ranking dominants during the driest period were, in order, Floridichthys, Eucinostomus , Lagodon , and Anchoa , while the wettest period was characterized by Eucinostomus , Floridichthys, Anchoa, Farfantepenaeus, and Syngnathus . Lucania, Anchoa, and Eucinostomus were the predicted dominants in basin habitat (78% of bay area) in all periods. The importance of Farfantepenaeus, Opsanus, and Floridichthys varied among time periods. The first two species were more important in the dry season/wet year when Floridichthys predicted numbers were reduced.
A spatially averaged species evenness index, limited to our 11 species, was
2

calculated from model predictions to compare dry (April) and wet (August) seasons of an extremely dry (1990) and extremely wet (1995) year. Predicted evenness for the wet year (0.75-0.77) was higher than predicted evenness for the dry year (0.62-0.71). Wet-season evenness was higher than dry-season evenness.
Evenness was greater for wet-year predictions had a more even composition than dry-year predictions. Habitat was a major factor affecting species diversity. The basin habitat had the highest evenness and the bank habitat the lowest, which may indicate that the environment on banks is more stressful than that in basins.
Salinity also affected evenness. A least square line of predictions suggested a negative correlation of evenness with salinity for the 11-species assemblage.
Highest evenness was in the salinity range from 21-30 ppt followed by salinities less than 20 ppt. Lowest evenness was at the upper range, 51-60 ppt. These relationships are consistent with what may be expected for an estuarine fauna.
Predicted biomass was highest during the wet season of both years, but higher in the wet season of the dry year than that of the wet year. Biomass was dominated by three species ( Opsanus, Eucinostomus , and Lagodon ). Opsanus dominated biomass in all periods, except the driest when Lagodon dominated biomass.
Predicted biomass of both Opsanus and Eucinostomus were higher in the wet season but higher in the dry than the wet year. Lagodon predicted biomass was higher in both seasons of the dry year. Predicted bay-wide percent composition of
Opsanus was greater in the dry season in the wet than the dry year.
Predicted densities were highest during wet years in basins with the lowest salinities. Salinities ranged from 8 to 35 ppt in the wet year and from 28 to 60 ppt in the dry year. Predicted responses of species to salinity followed four general trends. Three species ( Farfantepenaeus, Gobiosoma, and Lagodon ) had a positive relationship to salinity at low salinities, optima approaching 30 ppt, and a negative response at high salinities. The predicted densities of three species
( Anchoa, Hippocampus, and Syngnathus ), decreased in relation to salinity in both years. Three species ( Floridichthys, Lucania, and Microgobius ) had a negative linear relationship in the wet year (at low salinities) and a near flat relationship in the dry year (high salinities). Two species ( Opsanus and Eucinostomus ) had a negative relationship at high salinities (dry year) and a flat relationship at lower salinities. High variability in predicted response suggests strong habitat influence.
Models suggest wet conditions produce a diverse and slightly more abundant estuarine forage community in Florida Bay but not greater biomass.
Johnson, Darlene, NOAA Fisheries, 75 Virginia Beach Drive, Miami, FL 33149
Phone 305 361-4490, Fax 305 361-4478, darlene.johnson@noaa.gov
, Question 5
3