Abstract
This research investigates the reference water body approach to development of numeric criteria for dissolved oxygen in freshwater water bodies, using Total Maximum Daily Load (TMDL) targets for several impaired waters in the Everglades West Coast (EWC) basin of Florida as examples. The goal of this research is to critically evaluate the reference water body approach as it applies to setting site-specific numeric nutrient targets for use in TMDLs located in the EWC basin. The reference approach assumes that information about nutrients in the reference waters during a period when they are assumed to be unimpaired is sufficient to characterize unimpaired conditions in similar waters for which data are insufficient. The reference site approach is one method specified by U.S. Environmental Protection Agency (USEPA), which is used when biological end points are not sufficiently understood or characterized. The Florida agency responsible for analyses of the target waters, the Florida Department of Environmental Protection (FDEP), chose to specify a target as the 75th percentile of all reference waters' median concentration, a choice this research contests. The research uses existing data from waterbodies judged to be unimpaired, and applies statistical methods to chemical data on total nitrogen (TN) and dissolved oxygen (DO) conditions in those waters over the available period of record. The research considered variation among waterbodies, to investigate whether a single unimpaired archetype characterizes all south Florida flowing waters; variation over time, to investigate whether a single long-term average is adequate to judge the health of a south Florida waterbody; whether those waterbodies conform to two metrics (one each for two separate seasons, wet and dry instead of one; and considers the impact of selecting a period of record) on the computed numeric target. Conclusions note how the nutrient target for the EWC basin can change based on various decisions regarding the processing of the reference water body data. Data are shown to be sensitive to the period of record, as the 75th percentile of the median data collected outside of the verified period yields TN level of 1.0955 mg/L as opposed to 0.74 mg/L. Statistical similarities between waterbodies for TN also change when taking into account data collected before and after the verified period, this additional data also changes the waterbodies overall mean for both TN and DO. Results with DO demonstrate the inadequacy of selecting a single numeric target, based on a central tendency of a constituent in each of multiple water bodies, followed by a central tendency of those waterbodies in aggregate. This is inadequate because the distribution of DO varies substantially and a single central-tendency target would allow for conditions that fail to conform to important upper-tail characteristics of the distribution, thus failing to remove the impairment. Instead the target should include a central tendency (such as median) as well as at least one target elsewhere in the distribution, such as the 75th or 90th percentile of the aggregate data. The results also show DO is the preferred target constituent, as changes in DO, the stated problem, do not mathematically track historical data of changes in TN. Historical data also clearly demonstrate the DO in these south Florida water behaves very differently, such that a target selected as an annual average may not be protective of waterbodies for entire seasons; long enough to affect ecosystems in the long term, thus failing to remove the impairment. Further data suggests DO nutrient fluctuation during wet and dry seasons, with DO being impaired during the wet Florida months (June-Sept) and relatively unimpaired during the dry months (Jan-May, Oct-Dec) which suggests that target levels should recognize these seasonal changes, instead of one nutrient target being set for the entire year.