All data were obtained from the Florida DEP STORET database. STORET is a computerized environmental data system containing water quality, biological, and physical data. Total metal concentrations of the LSJR were used in this analysis. EPA methods 200.7, 200.8, and 206.2 were used to measure arsenic; EPA methods 200.7, 200.8, 213.2, and 6010B were used to measure cadmium; EPA methods 200.7, 200.8, 220.2, and 6010B were used to measure copper; EPA methods 200.7, 200.8, 249.2, and 6010B were used to measure nickel; EPA methods 200.7, 200.8, 272.2, and 6010B were used to measure silver; and EPA methods 200.7, 200.8, and 6010B were used to measure zinc.
The LSJR varies in salinity, with the mainstem predominantly freshwater and some of the tributaries ranging from fresh- to full strength seawater. Salinity may affect the toxicity of some metals to aquatic life, therefore the EPA class III Water Quality Criterion (WQC) values may be different for freshwater and marine water. Likewise, for freshwater, hardness, defined as the total concentration of the divalent cations calcium and magnesium, has also been shown to reduce the toxicity of the metals cadmium, copper, lead, nickel, and zinc; therefore, the freshwater criterion is based on an equation which incorporates the hardness of the water body. For the hardness-dependent metals in this analysis, an average hardness value of 100 mg CaCO3/L was used for generating the freshwater criteria.
The WQC for marine (haline; surface chloride concentration ≥ 1,500 mg/L) waters was also used for all of the metals, except for silver, for which no marine water quality criterion has currently been adopted by the U.S. EPA. Therefore, the current proposed WQC value for silver has been used. It must be pointed out that the freshwater and marine WQC are the same for some metals, like arsenic, for example. However, for other metals, like cadmium, the freshwater WQC is substantially different (0.27 mg/L at 100 mg/L hardness) from the marine criterion of 8.8 mg/L. Therefore, for river segments or water bodies that have no saltwater influence, the potential for environmental impacts of certain metals may vary.
Data are presented in box and whisker plots, which consist of a five number summary including: a minimum value; value at the first quartile; the median value; the value at the third quartile; and the maximum value. The size of the box is a measure of the spread of the data with the minimum and maximum values indicated by the whiskers. The median value is the value of the data that splits the data in half and is indicated by the horizontal blue line in the center of the boxes. Data are also presented as yearly mean values and compared to the designated reference values. Graphs are presented for the entire LSJR (including tributaries), the freshwater and saltwater portions of LSJR mainstem, as well as for the tributaries in some cases. Data used from the Florida DEP STORET database are of higher quality but are less abundant than data from the EPA STORET. Only total metal concentrations were used in this report, rather than the preferred dissolved metal concentrations, which are used in calculation of water quality criterion values. Total values were used because the dissolved metal concentrations were not reported to a large extent, and in many cases dissolved values only accounted for less than 5% of the total data reported. Additionally, negative values were removed and values designated as present below the quantitation limit (QL) were replaced with the average of the method detection limit (MDL) and practical quantitation limit (PQL). For “non-detect” values, half the MDL was used; and, for values designated as “zero” the MDL was used. Data were rejected and not used if they had the value qualifier code of K, L, O, or Y. Data designated with a matrix of “ground water”, “surface water sediment,” “stormwater,” or “unknown” were removed. Records with no analytical procedure listed were also removed.
22.214.171.124. Sediment Data Sources
The data used in this report came from several major studies carried out on the Lower St. Johns River from 1983 to 2007. They were conducted by the SJRWMD (Delfino et al. 1992: Delfino et al. 1991a; Durell et al. 2004; Higman et al. 2013) and the Florida Department of Environmental Protection (Delfino et al. 1991a; Pierce et al. 1988), Data were used from the National Oceanographic and Atmospheric Administration’s National Status and Trends Mussel Watch program (NOAA 2007b) and Benthic Surveillance Watch (NOAA 2007a) program. Data from STORET databases managed by the EPA (modern) and DEP were included as well. The STORET data were from studies by the National Park Service Water Resources Division, Florida Department of Environmental Protection, and the Marine Research Institute of the Florida Fish & Wildlife Conservation Commission. Savannah Laboratories (SLES 1988), Cooksey and Hyland 2007, and Dames and Moore 1983 also generated data that were analyzed in this report. The best and most recent data came from an extensive set of studies conducted by the SJRWMD. This study began in 1996 and provides a long-term sediment quality assessment of the LSJR (Durell et al. 2004; Durell et al. 1997; Higman et al. 2013).
A summary of the sources of data is given in Appendix 5.2.A. The database that was generated represents a substantial portion of existing data for LSJR contaminants. It is not exhaustive however, and should be considered a starting point from which omitted past and future studies can be added. In particular, modern pesticides, other important priority pollutants and emerging pollutants, such as endocrine disruptors, should also be included. Future additions of data on concentrations of contaminants in water and organisms will also add to the quality of the assessment.
The contaminants we selected for evaluation had the highest abundance of data available for several years and adequate site information. Sometimes we omitted potentially important contaminants because of analytical differences between studies. The data were first compiled from each source for approximately 200 analytes at nearly 500 sites, over a span of 20 years, and then were culled for location and analytical comparability. We omitted data from some years when the numbers of samples were too few, or when extreme values distorted the analysis. For example, Deer Creek samples in 1991 that consisted of nearly pure creosote (Delfino et al. 1991b) were omitted.
Sediment contamination was assessed by calculating average concentrations, percent exceedances of sediment quality guidelines, and average toxicity quotients, or toxicity pressure. These parameters were compared between years and regions of the river. Data below the detection limit were evaluated as zeroes in these calculations. The numbers of samples for each contaminant, year, and area are given in Appendix 5.2.B.
Trends were assessed by plotting median annual concentrations against time and determining the significance of an upward or downward slope of any line (Spearman Rank correlation coefficients p < 0.05). Because of the limitations of the data, all trends were confirmed by graphical analysis and Pearson Product coefficient > 0.5. Trend statistics are given in Appendix 5.2.C.
Advances in analytical technology during the last 20 years have dramatically reduced the concentration at which some chemicals can be detected. This can skew interpretations of temporal trends, which we attempted to avoid by transforming the zero values in the data to minimum detectable levels. Where possible, the reported minimum detection limits were substituted for zero values. In some cases, we estimated a minimum level of detection by finding the lowest nonzero value in a given year and halving it. Using minimum detection limits reduces the possibility of erroneously concluding there is an increasing trend because of differences in analytical detection limits.
There are numerous sources of variability in reported sediment concentrations, including analytical differences, sampling variations, physical and chemical characteristics of the sediment, and even differences in definitions of reporting parameters such as minimum detectable limits. Furthermore, there are large differences in the numbers of samples in different regions, all taken at irregular intervals. These data gaps limit the applicability of many different standard statistical tests. Thus, major harmful contaminants and their spatial and temporal trends can be difficult to positively identify and requires judicious use of statistics and careful review of all data. Box and whisker plots of the data are given in Appendix 5.2.D, which illustrate the distribution of the values for each contaminant in each region for each year.
126.96.36.199. Sediment Quality Guidelines
Environmental toxicology is the study of the effects of contaminants on ecosystem inhabitants, from individual species to whole communities. While toxicity is often viewed in terms of human health risk, human risk is one of the most difficult toxicity “endpoints,” or measures, to accurately quantify. The effects on ecosystems and aquatic organisms are the focus of our assessment of contaminants in the LSJR although human health effects from mercury in fish are discussed.
The environmental impact of a toxic compound can be evaluated several ways. One way is by comparing the concentrations in the LSJR to various toxicity measures. When the concentration of a contaminant in sediment is greater than the toxicity measure, it is an exceedance. Most sediment quality guidelines for contaminants are based on the impact of contaminants on sediment-dwelling benthic macroinvertebrates, assessing both the individual species’ health and the community structure. Since these organisms are at the beginning of the fisheries food chain, their health is a good indicator of general river health. One toxicity measure that is quite protective of the health of aquatic organisms is a Threshold Effects Level (TEL). This is the concentration at which a contaminant begins to affect some sensitive species. When the number of sites that have concentrations greater than the TEL is high, there is a higher possibility that some sensitive organisms are affected. A second, less protective guideline is the Probable Effects Level (PEL). This is the concentration above which many aquatic species are likely to be affected. The TEL and PEL sediment quality guidelines for marine systems are used in this assessment, with emphasis on the latter. These were the guidelines that were most widely available for the compounds of interest, plus much of the heavily impacted areas are in the marine section of the LSJR. Some alternative guidelines are used and identified for some compounds for which there were no marine TEL or PEL guidelines (MacDonald 1994; NOAA 2008). Specific values are listed in Appendix 5.1.A.
In an approach similar to Long et al. 1995 and Hyland et al. 1999, we evaluated overall toxicity of nearly 40 chemicals on the river ecosystem by calculating a PEL quotient, or toxicity pressure, for each sample. The quotient is the concentration of a contaminant in the sediment divided by the PEL value. If the quotient, or toxicity pressure, is greater than one, adverse impacts on benthic organisms are probable. As the quotient increases, we can assume that the probability of toxic effects increases. The quotients are used to compare the effects of different chemicals and to understand their relative importance in the impairment of the river health.
While sediment quality guidelines are useful tools, it is important to appreciate the limitations of simple comparisons in the extremely complex LSJR. A major difficulty in assessing toxic impacts is that the accessibility, or bioavailability, of a contaminant to organisms may vary with sediment type. Two sediments with similar contaminant concentrations but different physical and chemical features can produce very different environmental impacts, and we know that LSJR sediments are highly variable. Furthermore, each sediment quality guideline can be specific to certain organisms and endpoints (e.g., death of fish, reproductive effects of sea urchin, sea worm community structure, etc.) and cannot easily be extrapolated to other organisms or endpoints. As a consequence, guidelines from different organizations are sometimes different. Finally, separate guidelines are often established for marine and freshwater environments, though few estuarine guidelines exist that apply to the LSJR. These challenges limit our assessment of the impacts of various contaminants on the LSJR to one that is general and relative in scope.
188.8.131.52. Regions of the LSJR
Within the LSJR basin, there is a large variation in the types of ecosystems, land uses, and hydrology. As a consequence, the distribution and potential impacts of contaminants will vary widely within the basin at any given time. To analyze sediment contaminants in the LSJR, we divided it into four regions (Figure 5.2) with roughly similar hydrologic and land use characteristics. Where possible, trends were tracked within each region, and comparisons were made between the regions.
One region, Area 1, is a composite of the basins of three tributaries on the western side of the LSJR. The western tributaries area is composed of the Trout River (including Moncrief Creek and Ribault River tributaries), Long Branch Creek, the Cedar-Ortega system, Big Fishweir Creek, and Rice Creek. Despite their distance from one another, they were combined because they share the unfortunate characteristic of having such high levels of contamination for some chemicals that they mathematically obscure trends in the rest of the lower basin. The northernmost region, Area 2, the north arm, stretches from the coast at Mayport to Talleyrand, and has an extensive maritime industry. It is strongly tidal with a range of salinity from marine to estuarine. Moving south, the next region is Area 3, or the north mainstem, which includes urban Jacksonville and extends down to Julington Creek.