Home Introduction Study Area Methods & Procedures Estimation of Nutrient Loads & Water-Quality Analyses Summary & Conclusions References PDF Version

Summary And Conclusions

Biscayne Bay, located along the southeastern coast of Florida, is an oligotrophic, shallow, subtropical estuary that provides habitat for a variety of plant and animal life. In recent years, the ecological health of Biscayne Bay has become a matter of concern due to the presence of nutrient-laden discharges from coastal canals that drain into the bay. This concern, as well as planned diversion of discharges for ecosystem restoration from the urban and agricultural corridors of Miami-Dade County to Everglades National Park, served as the impetus for a study to develop a method for estimating nutrient loads discharged from the east coast canals into Biscayne Bay. More than 200 depth-integrated samples were collected for analysis of total organic nitrogen, ammonia, nitrite, nitrate, nitrite plus nitrate nitrogen, total phosphorus, and orthophosphate concentrations during the 1996 and 1997 water years. At 15 east coast canal sites in Miami-Dade County, a comparison of nutrient concentrations by land-use category was made.

Analytical results indicated that the highest concentration of any nutrients sampled for in this study was 4.38 mg/L for nitrate (as nitrogen) at one site, and the lowest concentrations determined were below the detection limits for orthophosphate at six sites and nitrite at four sites. Total organic nitrogen concentrations ranged from 0.20 to 1.7 mg/L, with a median concentration of 0.75 mg/L for all of the east coast canal sites. The 0.75-mg/L value was the highest median concentration of any nutrients sampled for in this study. Ammonia concentrations ranged from 0.01 mg/L to 1.5 mg/L, with a median concentration of 0.10 mg/L for all the sites. Five sites had concentrations that exceeded the Miami-Dade County DERM freshwater standard of 0.5 mg/L for ammonia. Nitrite concentrations ranged from less than 0.001 mg/L to 0.10 mg/L, with a median concentration of 0.02 mg/L for all the sites. Nitrate concentrations ranged from 0.001 to 4.38 mg/L, with a median concentration of 0.18 mg/L for all the sites. Nitrite plus nitrate nitrogen concentrations ranged from 0.002 to 4.4 mg/L, with a median concentration of 0.20 mg/L for all the sites.

Total phosphorus concentrations ranged from 0.004 to 0.31 mg/L, with a median concentration of 0.02 mg/L. The maximum total phosphorus concentration of 0.31 mg/L was the only nutrient concentration to exceed water-quality standards or guidelines of 0.10 mg/L for control of eutrophication. High concentrations of total phosphorus usually reflect contamination as a result of human activities. Orthophosphate concentrations ranged from less than 0.001 mg/L to 0.26 mg/L, with a median concentration of 0.005 mg/L for all the sites. The 0.005 mg/L value was the lowest median concentration of any nutrients sampled for in this study.

Median concentrations of nitrite, nitrate, and nitrite plus nitrate nitrogen tended to be higher in agricultural areas than in forested/wetland and urban areas. Median concentrations of ammonia, total phosphorus, and orthophosphate tended to be higher in urban areas than in forested/wetland and agricultural areas. Median total organic nitrogen concentrations generally were higher in forested/wetland and urban areas than in agricultural areas. These results coincide with expected differences in nutrient concentrations based on knowledge of point and nonpoint source influences and nutrient cycling.

A water-quality cross-section survey conducted at site S-22 along Snapper Creek Canal during an instantaneous discharge of 414 ft³ /s in October 1997 indicated that dissolved-oxygen concentrations decreased with increasing depth. Water samples collected incrementally from the middle of the canal between the surface and the streambed at site S-22, demonstrated an increase in total nitrogen and total phosphorus concentrations with depth. Additionally, point (grab) samples at 1.0 m deep and depth-integrated samples were collected for analysis of suspended-sediment concentration. Significant differences in concentration were detected between point (grab) and depth-integrated samples (1.0 and 3.0 mg/L, respectively). Suspended-sediment concentrations generally were higher near the streambed during flow conditions than near the surface.

The Wilcoxon signed ranks test (WSRT) was used to compare differences between point (grab) and depth-integrated samples for total nitrogen and total phosphorus concentrations from 12 east coast canal sites. There were no statistically significant differences in total nitrogen concentration between point (grab) samples collected at 1.0 m deep and depth-integrated samples, but statistically significant differences in total phosphorus concentration between the samples were apparent at 25 percent of the sites. No sites showed any statistically significant differences in total nitrogen concentration between point (grab) samples collected at 0.5 m deep and depth-integrated samples, and only one site showed statistically significant differences in total phosphorus concentration. Statistically significant differences in total nitrogen and total phosphorus concentrations between point (grab) samples collected at 0.5- and 1.0-m depths were not detected at any sites.

A fitting procedure, referred to as the line of organic correlation (LOC), was used to compare point (grab) and depth-integrated samples where statistically significant differences exist as defined by the WSRT. Results of the LOC indicated that point (grab) samples at 0.5- and 1.0-m depths underestimate total phosphorus concentrations when compared to depth-integrated samples. Because total phosphorus tends to adsorb to particulate matter, this underestimation is attributed to the reduced suspended-sediment concentrations near the surface as compared to those near the streambed during periods of flow. The suspended-sediment concentration determined from a point (grab) sample was only one-third the concentration determined from a depth-integrated sample. Physical factors, such as distance upstream or downstream from the control structures and configuration of the sampling cross section, could influence the degree of mixing and contribute to differences between point (grab) and depth-integrated samples.

Models were developed to estimate nutrient loads in the east coast canals of Miami-Dade County. The measured loads were mainly based on data collected from depth-integrated samples. Discharge was used as the independent or explanatory variable, and total phosphorus load or total nitrogen load represented the dependent or response variable. The coefficients of determination (R² ) were adjusted to conform with the number of degrees of freedom in the model. The adjusted R² for total nitrogen load models ranged from 0.69 to 0.99, indicating that from 69 to 99 percent of the variation in total nitrogen load is explained by discharge. The average adjusted R² for all the total nitrogen load models was 0.87. The adjusted R² for the total phosphorus load models ranged from 0.23 to 0.99 and averaged 0.76, which was lower than that for the total nitrogen load models. All of the models, except for two total phosphorus load models, were statistically significant at an alpha level of 0.05. Some models for total nitrogen and total phosphorus loads required transformation of the independent variable (discharge) to enhance linearity, and a few models required log transformation of the dependent variable (load) due to heteroscedastic residuals. Because long-term water-quality data exist and continuous discharge is computed at site S-26 along Miami Canal, a software program, ESTIMATOR, was used to estimate total nitrogen and total phosphorus loads at that site.

Models that have the greatest predictive power are those with the highest adjusted R² and the lowest root mean square error or standard error of the regression. Because of time constraints, data were collected over a period of 2 consecutive water years (1996-97). Data collected over longer time periods, such as 5 to 10 years, would enhance the predictive power of the models and would more accurately represent long-term hydrologic conditions in southern Florida.