xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 2200385Ka 4500
controlfield tag 001 002069340
007 cr mnu|||uuuuu
008 100421s2009 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0003211
Malone, Sarah J.
Agricultural nonpoint source pollution management :
b water quality impacts of Balm Road Treatment Marsh, Hillsborough County, Florida
h [electronic resource] /
by Sarah J. Malone.
[Tampa, Fla] :
University of South Florida,
Title from PDF of title page.
Document formatted into pages; contains 108 pages.
Thesis (M.S.)--University of South Florida, 2009.
Includes bibliographical references.
Text (Electronic thesis) in PDF format.
ABSTRACT: Balm Road Treatment Marsh is a 12 ha constructed wetland treatment system in south-central Hillsborough County, Florida created to improve water quality in Bullfrog Creek and ultimately Tampa Bay. The treatment system was designed to treat runoff from approximately 741 ha of upstream agricultural land prior to discharging into the creek, with the primary goals of reducing sediment and nutrient loads. Water quality data from four sites on Bullfrog Creek were analyzed to determine impacts to ambient water quality and pollutant load reductions downstream. Results were compared to the performance of other wetlands to treat both nonpoint and point source pollution. Impacts to ambient water quality in the creek were found to be minimal, if any, and although significant load reductions were found downstream, they could not be attributed to wetland treatment affects with confidence. In general, nonpoint source pollution, particularly from agriculture, was found to be treated less effectively than point sources. The importance of monitoring the performance of stormwater projects while employing a strategic sample design and including receiving water impacts is highlighted.
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
Advisor: Philip Reeder, Ph.D.
Total suspended solids
t USF Electronic Theses and Dissertations.
Agricultura l Nonpoint Source Pollution Management : Water Quality Impacts of Balm Road Treatment Marsh, Hillsborough County, Florida b y Sarah J. Malone A thesis submitted in partial fulfillment of the requirements for the degree of Master of Sc ience Department of Geography College of Arts and Sciences University of South Florida Major Professor: Philip Reeder, Ph.D. Kamal Alsharif, Ph.D Graham A. Tobin, Ph.D. Date of Approval November 16 2009 Ke ywords: constructed wetland, agricultu ral runoff sediment, total suspended solids nutrients Copyright 2009, Sarah J. Malone
i Table of Contents List of Tables i v List of Figures v Abstract vi ii Chapter One Introduction 1 Chapter Two Background 5 Water Quality in the United Stat es and Florida 5 Water Quality in Tampa Bay 5 Agricultural Nonpoint Source Pollu tion 7 Agri cultural Pollution Legislation 10 Treatment Wetlands for Agricultural Pollution Management 11 Treatment Wetland Processes 12 Balm Road Treatment Marsh 14 Chap ter Three Review of the Literature 19 Wetland Treatment Performance 19 Factors Affecting Performance 21 Process es and Design 26 Data Analysis 27 Chapter Four Research Design 30 Problem Statement 30
ii Research Questions 30 Study Significance 30 C hapter Five Study Area 32 Location 32 Climate 32 Soil 33 Land Use 3 5 Hydrology 3 6 Chapter Six Methods 3 7 Sample Collection and Laboratory Analysis 3 7 Data Organization 4 3 Statistical Analysis 4 5 Chapter Seven Results and Discussion 4 8 Wate r Quality Descriptive Statistics 48 Discharge and Precipitation 57 Ambient Water Quality Impacts 58 Pre Post Comparisons 5 8 Up stream Downstream Comparisons 63 Loading Impacts to Tampa Bay 69 Comparison to other Wetlands 7 5 Chapter Eight Conclus ions 7 9 Summary 7 9
iii Data Limitations and Future Needs 81 Recomme ndations for Project Managers 8 3 List of References 8 6 Appendices 9 4 Appendix A : Pictures of Balm Road Treatment Marsh 9 5 Appendix B: Histograms 10 2
iv List of Tables Table 1 Detecti on limits, units, and metho ds for parameters of interest 40 Table 2 Station ID numbers and dates for re trieval from online databases 4 1 Table 3 Overall descriptive statistics for entire dataset at each site 48 Table 4 Typical statewide percentile values f or Florida streams 49 Table 5 Wet and dry season descriptive statistics for the pre phase 52 Table 6 Wet and dry season descriptive statistics for the post phase 53 Table 7 Ma nn Whitney results using PASW 5 9 Table 8 Wilcoxon matched pairs signed rank test results using PASW 6 4 Table 9 Average annual load reduction at the Downstream 3 site 67 Table 10 Pollutant reductions for Balm Road Treatment Marsh 7 6 Table 11 Wetland performance by source water using values from the literature 7 7
v List of Figures Figure 1 Location of Balm Road Marsh 2 Figure 2 Bullfrog Creek Watershed 3 Figure 3 EPA nat ional water quality assessment 6 Figure 4 Fl orida water quality assessment 6 Figure 5 Average monthly precipitation in Parish, Florida 33 Figure 6 Soil map for Balm Road Treatment Marsh site 3 5 Figure 7 Bullfrog Cr eek water quality sample sites 3 9 Figure 8 P icture of USGS Gaging Station 4 2 Figure 9 Diagra m describing grouping of data 4 3 Figure 10 Diagram depicting data comparis ons using nonp arametric tests 4 6 Figure 11 TSS dataset boxplot 49 Figure 12 TN dataset boxplot 50 Figure 13 TP dataset boxplot 50 Figure 14 TSS boxplot by phase with seasons combined 54 Figure 15 TSS boxplot by phase and season 54 Figure 16 TN boxplot by phase with seasons combined 55 Figure 17 TN box plot by phase and season 55 Figure 18 TP boxplot by phase with seasons combined 56 Figure 19 TP boxplot by phase and season 56
vi Figure 20 Annual average discharge and precipitation 57 Figure 21 Pre/Post pollutant load reduct ions at the Downstream 3 site 71 Figure 22 Pre/Post mea n discharge and precipitation 72 Figure 2 3 Pre/Post mean pollutant concentration at the Downstream 3 site 73 Figure 24 Pre/Post mean pollutant con centratio n at the Inflow site 7 4 Figure A 1 Balm R oad Marsh Property aerial, 2004 94 Fig ure A 2 Balm Road Marsh Property aerial, 2005 94 Figure A 3 Balm Road Treatment Marsh se dimentation basin, 9/26/2009 95 Figure A 4 B alm Road Treatment Marsh cell #1 9/29/2009 95 Figure A 5 Balm Road Treatm ent Marsh cell #2, 9/29/2009 96 Figure A 6 Balm Ro ad Treatm ent Marsh cell #3, 9/29/2009 96 Figure A 7 Balm Road Treatm ent Marsh cell #4, 9/29/2009 97 Figure A 8 Upstr eam sampling site, 9/29/2009 97 Figure A 9 Diversion structure (left) and canal t o wetland (right), 9/26/2009 98 Figure A 10 Diversion stru cture allowing base flow to Bullfrog Creek, 9/26/2009 98 Figure A 11 Treatment system outfall str ucture in cell #4, 9/26/2009 99 Figure A 12 Treatment system outfall, 9/26/2009 99 Figure A 13 Looking upstream on Bullfrog Creek from the Downstream 1 sampl e site, 9/26/2009 100 Figure B 1 Upstream TSS histograms 101 Figure B 2 Downstream 1 TSS histograms 102 Figure B 3 Downstream 2 TSS histograms 102
vii Figure B 4 Downstream 3 TSS histograms 103 Figure B 5 Downstream TN histograms 103 Figure B 6 Downstream 1 TN histograms 104 Figure B 7 Downstream 2 TN histograms 104 Figure B 8 Downstream 3 TN histograms 105 Figure B 9 Upstream TP histograms 105 Figure B 10 Downstream 1 TP histograms 10 6 Figure B 1 1 Downstream 2 TP histograms 106 Figure B 1 2 Downstream 3 TP hist ograms 107
viii Agricultural Nonpoint Source Pollution Management : Water Quality Impacts of Balm Road Treatment Marsh, Hillsborough County, Florida Sarah J. Malone ABSTRACT Balm Road Treatment Mar sh is a 12 ha constructed wetland treatment system in s outh central Hillsborough County, Florida created to improve water quality in Bullfrog Creek and ultimately Tampa Bay. The treatment system was designed to treat runoff from approximately 741 ha of upstream agricultural land prio r to discharging into the c reek with the primary goals of reducing sediment a nd nutrient loads Water quality data from four sites on Bullfrog Creek were analyzed to determine impacts to ambient water quality and pollutant load reductions downstream. Results were compared to the performance of other wetlands to treat both nonpoint and point source pollution. Impacts to ambient water quality in the creek were found to be minimal, if any, and although significant load reductions were found downstream, they could not be attributed to wetland treatment affects with confidence. In general, nonpoint source pollution, particularly from agriculture, was found to be treated less effectively than point sources. The importance of monitoring the performance of stormwater projects while emp loying a strategic sample design and including receiving water impacts is highlighted.
1 Chapter 1 Introduction Balm Road Treatment Mar sh is 12 ha constructed surface flow wetland system in south central Hillsborough County, Florida created to improve wate r quality in Bullfrog Creek and ul timately Tampa Bay ( Figure 1 ). The treatment system is located near the headwaters of Bullfrog Creek, which has been partially diverted to flow through the wetland along with any overland runoff from the upper parts of th e watershed. Bullfrog Creek then empties into Tampa Bay approximately 32 km downstream. The treatment system was designed to treat runoff from approximately 741 ha of upstream agricultural land prio r to discharging into the c reek with the primary goal o f reducing sediment and nutrient loading to Tampa Bay while improving water quality in Bullfrog Creek (Figure 2) The system was constructed in 2004 through a joint effort between Hillsborough County and the Southwest Florida Water Management District (SW FWMD). This research use s water quality data from Bullfrog Creek upstream and downstream from the treatment system to examine its affects on the water quality in Bullfrog Creek and loadings to Tampa Bay. The treatment performance of this treatment wetlan d system is compared to other performance data available in the literature to determine whether constructed wetland treatment systems are a useful tool in managing agricultural nonpoint source pollution.
2 Figure 1. Location of Balm Road Marsh. This document outlines the research in its relevant context. Background information is presented including the current status of water quality in the United States and Florida. The role of agricultural nonpoint source pollution is discussed along with detail ed impacts of nutrients and sediments on water resources. A brief history of related policy, both at national and state levels, is then outlined. Wetlands as pollution treatment systems are discussed including history and processes. To conclude the back ground section, the design of Balm Road Treatment Marsh is described.
3 Figure 2. Bullfrog Creek watershed. Subbasins that drain to Balm Road Marsh are highl ighted in yellow A dapted from Dames & Moore, 2000 A review of the literature as related to constructed treatment wetlands follows. Litera ture reviewed includes treatment wetland performance investigations studies Upstream subbasins Balm Road Treatment Marsh
4 determining factors affecting performance, literature on processes and design, and sources for data analysis reference. This, a long with the background section, se t the framework for the research. The study purpose is to determine the water quality impacts of Balm Road Treatment Marsh in order to gain a better understanding of the performance of constructed treatment wetlands for agricultural pollution. The specific research questions regardin g the treatment system are presented as follows: What are the resulting ambient water quality impacts on Bullfrog Creek? Was there a subsequent pollutant load reduction to Tampa Bay? How does the performance of constructed wetlands used to treat agricultural pollution compare to wetlands used to treat other pollution? The comparisons and questions are intended to help solve the overarching problem of whether or not constructed wetlands ar e appropriate for agricultural pollution management. The study area is described including climate, soil, land use, and hydrology. Next, the specific research methods are outlined. This includes sections on sample design and data collection and data ana lysis. Finally, the results and conclusions are discussed which include the determination of impacts to water quality in Bullfrog Creek, load reductions to Tampa Bay, and the discussion of treatment wetlands as potential management strateg ies for agric ultural nonpoint source pollution.
5 Chapter 2 Background Water Quality in the United States and Florida The United States Environmental Protection Agency (EPA) has reported that approximately 44% of river reaches, 64% of lake area, and 30% of estu arine area assessed do not fully meet their water quality standards (EPA, 2009). The state of Florida reports similar results with 32% of stream reaches, 64% of lake area, and 98% of estuarine area not meeting water quality standards (FDEP, 2008). These numbers can be seen in Figures 2 and 3 Improving surface water quality has been a national goal in the United States since the passage of the Clean Water Act (CWA) in 1972. Although the CWA was largely successful in reducing point source pollution, nonp oint source pollution remains the major cause of water body degradation. Nutrients, sediment, bacteria, metals and oxygen depleting substances have been found to be the most common causes of water body impairment. The leading source of these pollutants i s from ur ban and agricultural runoff, known as nonpoint source pollution (EPA, 2002). In fact, agricultural nonpoint pollution has been identified as the number one source of water quality impairments to streams and lakes in the United States (Parry, 1998 ; EPA, 2009).
6 Figure 3 EPA national water quality assessment. Waters that do not meet the standards for the ir designated uses shown in red (EPA, 2009). Figure 4 Florida water quality assessment. Waters that do not meet the standards for the ir designated uses shown in red (FDEP, 2008). Water Quality in T ampa Bay urban and agricultural runoff. In fact, nonpoint source pollution accounted for 63% of nitrogen loading to t he bay from 1999 2003, nearly half of which is from agricultural lands (TBEP, 2006). Total nitrogen loading to the bay from nonpoint sources for this time period was approximately 2,321 metric tons per year, total phosphoru s was 747 metric tons per year, and totals suspended solids was 37,068 metric tons per year (TBEP, National Wa ter Quality Assessment Florida Water Quality Assessment
7 2005). The major contributing basin of concern for the proposed research is the Coastal Hillsborough Bay basin, which includes the Bullfrog Creek basin. Th e Coastal Hillsborough Bay basin represents only 7.5% of the Tampa Bay watershed area (FDEP, 2001). Estimated loading to this basin for the same five year period was 465 metric tons per year of total nitrogen, which represents approximately 20% of loading s to th e Tampa Bay watershed. It was estimated that 50% of the load for this basin was from nonpoint sources (TBEP, 2005). Agricultural Nonpoint Source Pollution Approximately 50 to 70% of water bodies assessed have been found to be adversely affected by agricultural nonpoint source pollution (Ritter & Shirmohammadi, 2001). Agricultural runoff carries sediments from erosion resulting from row crops and overgrazing as well as nutrients, primarily nitrogen and phosphorus, originating from fertilizer applic ation. Approximately 30% of phosphorus and 18% of nitrogen applied to agricultural land in the form of fertilizers is utilized in plant production (Isermann, 1991; Carpenter, 1998). The remaining nutrients either runoff to surface water or accumulate in agri cultural soils whi ch may eventually erode and also runoff to surface water. Nitrogen export from agricultural land also occurs through leaching and infiltration which eventually deposits nitrogen to ground and surface waters. Nutrients also accumula te in a similar manner from animal waste and manure (Carpenter et al 1998). Soil erosion is the source of 99% of the total suspended solids (TSS) loads found in water bodies (Ritter & Shirmohammadi, 2001) and sediments and the pollutants attached to sedi ments are the most widespread source of pollutants in surface waters of the United States (Gianessi & Peskin, 1989). Sediments affect water bodies by degrading
8 wildlife habitat, decreasing water storage capabilities, and may result in the need for costly dredging activities (Ritter & Shirmohammadi, 2001). Increased sediment loads also interfere with recreational use and cause water clarity problems, decreasing the aesthetic value of water bodies (USDA, 1997). Sediments are also harmful to aquatic organi sms, result in temperature changes, and cause oxygen depletion. Effects on benthic invertebrates and algae populations vary from reduced growth rates to mortality (Hynes, 1970; Newcombe & MacDonald, 1991). Increased suspended sediment loads can cause a re duction in fish growth rate and disease resistance, modify migration patterns, reduce the number of organisms available for fish to feed on, interfere with fishing activities, and can be lethal at higher concentrations (Newcombe & MacDonald, 1991). In addition to the direct effects of suspended sediments, soil particles also degrade water quality by transporting other pollutants to surface waters. Phosphorous, nitrogen, and pesticides bind to soil particles on agricultural land and are washed into waterways after irrigation or rain events (Ritter & Shirmohammadi, 2001). Soil erosion accounts for 80% of the total phosphorous and 73% of the total Kjeldahl nitrogen found in waterways of the United States (USDA, as cited by Ritter & Shirmohammadi, 200 1). Nutrients transported to surface waters either attached to soil particles or dissolv ed in runoff have been identified as the number one cause of impairment by nonpoint source pollution in lakes and estuaries (Baker, 1992). The primary nutrients of concern are nitrogen and phosphorus (Carpenter et al 1998). While nitrogen can be toxic to humans at certain concentrations, phosphorus is not considered to be directly toxic to humans or animals. Rather than toxicity concerns, water bodies are listed a s impaired for excessive nutrients because they lead to accelerated eutrophication, or excessive plant
9 growth (EPA, 1999). Phosphorus is the limiting nutrient in the majority of freshwater lakes, and nitrogen is generally the limiting nutrient for estuarie s (Baker, 1992). In Florida, some regions are composed of soils with large deposits of phosphorus, and nitrogen becomes the limiting factor for lakes in these regions (Florida Lakewatch, 2000). Eutrophication of water bodies in the United States is a gro wing problem that accounts for about 50% of impaired lake area and 60% of impaired river reaches (Carpenter et al 1998). According to the University of Florida, 57% of Florida lakes are considered either eutrophic or hypereutrophic (UF IFAS, 2009). Eut rophication is a process caused by increased nutrient loads in a water body that results in excessive algae and plan t growth (Correll, 1998). P lants and animals require nutrients for growth, and nitrogen and phosphorous occur naturally in aquatic environm ents at levels below 0.3 and .01 mg/L, respectively. When nitrogen and phosphorus are introduced into aquatic ecosystems above these natural levels, plant production increases which can lead to eutrophication (EPA, 1999). Eutrophication can severely impa ability to attain its designated use standards The most obvious impact is that the overgrowth of algae and aquatic weeds impairs the fisheries, aquatic life, recreation, and drinking water supply uses. In addition, increased decomposit ion of dead plant matter results in oxygen shortages which can cause fish kills (Carpenter et al, 1998). Eventually oxygen in the bottom of lakes can become depleted which leads to toxic releases from sediments affecting the fisheries and aquatic life use s (EPA, 1999). Drinking water supplies are impaired by cyanobacteria blooms that result from eutrophication. Excessive algae cause foul tastes and smells in drinking water, can clog water treatment plant filters, and form potentially carcinogenic trihalo methane
10 during the chlorination process. Excessive plant growth and odors also interfere with recreational uses such as swimming, fishing, and boating (EPA, 1999). Agricultural Pollution Legislation Although the Clean Water Act (CWA) was mainly targe ted at point source pollution, nonpoint source pollution was addressed as well. Section 208 called for the development of watershed management plans and all sources, including nonpoint sources, were to be included in the plans (Malik, 1994). States were directed to identify and control nonpoint source problems and to implement appropriate controls; however, due to the prevalence and severity of point source pollution problems, nonpoint sources were routinely overlooked (Adler et al 1993). Section 303 of the CWA outlined the Total Maximum Daily Load (TMDL) program. The program called for states to identify waters that do not meet water quality standards, determine the maximum pollutant loads that would bring water quality to standards, and to develop b asin management action plans to implement the TMDL. The TMDL was to be split between all sources, both point and nonpoint (Houck, 2002). The program moved very slowly until more recent years, but implementation plans are currently being developed that wi ll push pollution reduction strategies. In 1987, Congress passed amendments to the CWA including section 319 which set up state programs to address nonpoint source pollution problems. States were directed to identify sources of nonpoint source pollution a nd implement management programs to control the sources that included best management practices (BMPs), or land use controls and land management practices (Malik, 1994). Management practices can be either structural or managerial in nature. Examples of m anagerial BMPs for
11 agricultural pollution control include rotational grazing, nutrient management, pesticide manag ement, and conservation tillage. Structural BMPs include the use of treatment lagoon s or ponds, terraces, and sediment basins (EPA, 2003). T he most efficient and accepted approach by land owners to control agricultural pollution is a combination of these BMPs along with offsite natural or constructed wetlands located in various areas throughout the watershed designed to receive nonpoint source pollution from larger areas (Hammer, 1992). In 1999, many years following the passage of the CWA, Florida Legislature enacted the Florida Watershed Restoration Act (FWRA) in order to establish the TMDL program in accordance with the federal requirement s ( Section 403.067, Florida Statutes ) The Florida Department of Environmental Protection (FDEP) was authorized as the lead agency in determining impaired waters and TMDL development. The Department of Agricultu re and Consumer Services (DACS) was establi shed as the lead agency responsible for FWRA enforcement involving agricultural nonpoint source pollution. Under the FWRA, DACS may develop and adopt BMPs to meet the load allocations for agricultural nonpoint source pollution resulting from TMDLs (UF IFA S, 2005). Treatment Wetlands for Agricultural Pollution Management Natural wetlands have been used for wastewater discharge sites for at least one hundred years in some locations around the world. However, their water quality benefits were not recogniz ed until monitoring of some of these natural wetlands began in the 1960s (Kadlec & Knight, 1996). The first constructed wetland was designed to receive wastewater and underwent extensive scientific investigations beginning in 1952 (Kadlec & Knight, 1996; Campbell & Ogden 1999). Widespread use of constructed wetlands for
12 wastewater treatment began in the United States in the 1970s. Industrial stormwater and process water began to be treated by constructed wetlands in 1975 and in the 1980s constructed wetl ands were beginning to be designed for urban stormwater treatment (Kadlec & Knight, 1996). The use of t reatment wetlands for nonpoint sources can be more complex than their use for point source pollution. For example, storms can have a large effect on th eir pollutant removal efficiency. High flows into the wetland can severely impair pollutant retention and can even cause release of nutrients (Mitsch & Gosselink, 2000). As the construction of treatment wetlands increased, so did the research and underst anding of their processes and functions in regard to water treatment (Campbell & Ogden, 1999). Constructed wetlands have the benefits of being self sustaining and having relatively low maintenance requirements (Kadlec, 2001). However, the increasing popul arity of using constructed wetlands for water quality treatment can be primarily attributed to their efficiency in pollutant reduction and relatively low cost (Hammer, 1992). The success of using treatment wetlands to treat point sources and later nonpoin t sources, has led to interest in their use to treat agricultural runoff (Kovacic et al 2000). In fact, wetlands have been recognized as potentially the most cost effective pollutant sinks in many agricultural landscapes (van der Valk & Jolly, 1992). De spite the increased use and recognized importance of treatment wetlands in agricultural pollution control, few studies have been published on wetland effectiveness in reducing agricultural runoff pollution in the United States (Mitsch & Gosselink, 2000). Treatment wetland processes Pollutant removal in treatment wetlands occurs by a variety of physical, chemical and biological processes. Wetlands have important characteristics that influence their
13 pollutant reduction capabilities. The gas exchange rate s between wetland soils and the atmosphere are very low due to the fact that they are usually inundated or at least saturated, which causes wetland sediments to be mostly anaerobic (Mitsch & Gosselink, 2000; Bix, 1993). This causes organic material to ac cumulate on top of the bottom sediments because decomposition is significantly slowed in anaerobic conditions. In, addition, because wetlands are generally fairly heavily vegetated, there is an overabundance of organic material within wetland systems. Th e layer of organic matter on the wetland bottom combined with the vegetation provides a large surface for microbial growth (Bix, 1993). Although sediments are highly anaerobic, a very thin oxidized layer is usually present on the surface of the soil. Thi s layer contributes to sediments having a high oxidation reduction potential which is important in the chemical transformations that occur in wetlands (Mitsch & Gosselink, 2000; Bix, 1993). This combination of characteristics gives wetlands their high cap ability of transforming nutrients (Bix, 1993). Suspended solids are removed by the purely physical processes of sedimentation and filtration (Bix, 1993). Although resuspension may be common in some shallow lakes and floodpla in wetlands, sedimentation is generally an irreversible process in most wetlands, including constructed wetlands (Johnston, 1991). In addition to the natural process that occurs to remove sediments in wetlands, many constructed wetlands are designed with some type of sediment basin or mechanical pretreatment unit to remove sediments before they even enter the wetland (Bix, 1993, Higgens et al 1993). The processes involved in nitrogen removal in wetlands include ammonification or mineralization, nitrification, and denitrification. A mmonification refers to the series of
14 biological transformations that convert organic nitrogen to ammonia which occurs when organic matter is decomposed by microorganisms (Kadlec & Knight, 1996; Ritter & Shirmohammadi, 2001). Nitrification then takes plac e as ammonia is oxidized to nitrate by microbes in the aerobic zone. Nitrates can either be immediately assimilated by plants or microbes, or are converted into nitrogen gas by microbes in the anaerobic zone through a process called denitrification (Bix, 1993; Mitsch & Gosselink, 2000). it releases the gas into the atmosphere (Mitsch & Gosselink, 2000). Phosphorus retention in wetlands can occur as either sho rt term or long term storage. Although a large number of temporary phosphorus storage processes and transfers occur within a wetland, the primary process involved in permanent phosphorus removal is soil sorption (Kadlec & Knight, 1996). This occurs throu gh adsorption, complexation, and precipitation with aluminum, iron, calcium and clay minerals present in wetland sediments (Bix, 1993). However, the capacity of wetland soils to sorb phosphorus is highly variable and may only last a short period of time. Phosphorus that is attached to sediment particles is lost through the physical process of sedimentation. Phosphorus removal also occurs through plant uptake, however, it has been suggested that this should not be considered a long term retention process (Kadlec & Knight, 1996). Balm Road Treatment Marsh Pictures of Balm Road Treatment Marsh can be found in Appendix A and show many of the features described in this section. The t reatmen t system was designed as a series of shallow vegetated cells located in the floodplain on the north west side of Bullfrog Creek. It has a wetland to watershed ratio of 2 % which is the recommended
15 minimum for successful treatment of pollutants (Carleton et al 2001). The system receives flow diverted from the creek near t he end of McGrady Road. A diversion ditch and two structures were constructed to route the water from the creek into the system. Two existing channels in the creek diverge near the south end of McGrady Road. A diversion structure was placed in each of t he two channels, so that water enters the system from each channel of Bullfrog Creek. The structures are constructed of sheet pile and slotted to provide base flow to the historic creek channel. The constructed di version ditch is approximately 2.7 m deep and begins at the previously existing western channel and flows approximately 400 m west to the sedimen tation basin The sedimentation basin is approximately 4.6 m deep at its deepest point, 91.4 m wide at the bank and sloping to 7.6 m wide at the bottom of the po nd. A series of four cells are separated by berms. The system was designed to avoid "dead zones", or areas of no flow. Water flow between the four cells is maintained by 1.2 m diameter pipes. The system was designed so that the majority of dry season flow and at least the first flush of runoff from the upstream watershed resulting from storms are diverted into the wetland. This was accomplished by placing structures in the existing channels that would be overtopped during the 100 year flood ev ent, thus do not increase the 100 year peak water elevations. Another important design element was ensuring embankments were protected from erosion and overtopping during the 100 year flood, while providing adequate treatment time during periods of low fl ow. At the time of design, the normal pool elevation was expected to be approximately 25 m NGVD during the dry season with small fluctuations following minor rainfall events. During the wet season, elevations were expected to fluctuate somewhat above the dry season elevation. At 25 m NGVD,
16 the maximum water depth in the ponds would be approximately 0.5 m deep in the few deep water areas. In May 2005 staff gauges were installed in cell number three and four and water levels were recorded during monthly s ampling events. The mean water level from the period of observations from May 2005 to September 2007 was 25.7 m, and the lowest observation which occurred during the dry season of 2006 was 25.3 m. At 25.7 m, the maximum depths were approximately 1.2 m w ith average depths at approximately 0.5 m. Depths are an approximation based on design plans; actual depths may very due to possible soil swelling and lift after saturation (Kadlec & Wallace, 2009) Water levels in the wetland remained higher than anticip ated. The original planting plan called for low elevations of the four cells (24.5 24.7 m NGVD) to be planted with groupings of spatterdock ( Nuphar luteum ). These elevations would be approximately 0.5 m deep under normal conditions, but may be submersed in up to 2.0 m of water during seasonal high stages. The upper portions of the anticipated dry season seasonal high water elevations (25.0 25.3 m) were planted with pickerelweed ( Pontederia cordata ) and arrowhead ( Sagittaria latifolia ) with smaller amo unts of sawgrass ( Cladium jamaicense ) and fireflag ( Thalia geniculata ). Areas above the dry season high water elevation, but within the anticipated wet season normal pool elevations (25.3 25.9 m) were dominated by pickerelweed and maidencane ( Panicum h emitomon ) along with several other herbaceous species and some trees and shrubs. Upper elevations of the wetland area (25.9 26.8 m) were planted to resemble a pine flatwoods community with species such as slash pine ( Pinus elliottii ), wax myrtle ( Myrica cerifera ), and saw palmetto ( Serenoa repens ). These elevations were expected to only be inundated for very short periods of time following large storm events.
17 By August, 2005, many of the plants had been destroyed by nutria and apple snails. Replanting of the site was completed by December, 2006. Areas throughout the four cells with no coverage remaining were planted with spikerush ( Eleocharis intersticta ) in elevations of 24.7 25.3 m and spikerush ( Eleocharis intersticta ), bulrush ( Scirpus validus ), and maidencane ( Panicum hemitomon ) in elevati ons of 25.3 25.8 These plants were chosen based on their ability to withstand apple snail infestations. Nutria were trapped and removed from the site. A site visit on September 26, 2009 revealed there had been a major shift in ve getation. As seen in Appendix A water paspalum ( Paspalum repens ) dominated every pond along with the submerged invasive species hydrilla ( Hydrilla verticillata ). There were still small amounts of pickerelweed, maidencane, duck po tato, spike rush, and arrowhead remaining and t he non native wild taro ( Colocasia esculenta ) and torpedo grass ( Panicum repens ) w ere becoming established Typically, displacement of planted species by other species will not alter treatment efficiency (Kad lec & Wallace, 2009). However, the lack of established vegetation during much of the period of study could be a factor in performance. Annual l oad reduction estimates for the treatment system were made prior to construction using a model developed for Hil lsborough County. The model calculations were based on EPA approved runoff calculations, which were developed using land use and soils data for the project area. Event mean concentrations of parameters were developed from NPDES p ermit sampling performed by the C ounty. For the purposes of the model run, it was assumed that all of the pollutan ts from the modeled drainage basin enter the creek and are routed through the wetland system. The wetland system was identified in the model as a wet detention best management practice (BMP) with removal
18 efficiencies estimated using previous field and literature research collected by SWFWMD and Environmental Research and Design (ERD), Inc. (M. Moore personal communication, September 12, 2000). L oad reductions were es timated at 85% TSS, 30% TN, and 65% TP which equaled 125 ,060 kg TSS, 8 ,700 kg TN, and 13,690 kg TP per year.
19 Chapter 3 Review of the Literature Wetland Treatment Performance A review of the literature revealed that the performance of many types of treatment wetlands have been assessed in a variety of studies from around the world. For example, Yang et al (1995) studied the removal efficiency of a vegetated subsurface flow bed used to treat municipal wastewater in Shenzhen, China. Monthly samples were taken at the inflow and outflow of the wetland for a period of three years and the data were analyzed to determine the percent reduction for a suite of parameters. The removal efficiencies found were 92.6% total suspended solids, 23.2% total nitrogen and 30.6% total phosphorus. The results were used for a comparative st udy with other similar wetlands. The studied wetland was highest in total suspended solid removal but much lower than the highest performing wetland in nitrogen and phosphorus remov al. In Estonia, three different types of treatment wetlands were studied. A vertical flow sand/plant filter, a semi natural wet meadow, and a drainage channel planted with macrophytes which were all designed to treat wastewater were sampled on a monthly basis for several parameters (Mander & Mauring, 1997). Nitrogen removal efficiencies ranged from 36 67% and phosphorus removal ranged from 69 74%. Statistical analysis t test, Kruskal n technique to compare results for the different types of wetlands. The method of data
20 anal ysis used in this study was examined for it s applicability to the present study on Balm Road Treatment Ma rsh as described by Zar (1984). A surface flow wetland, which is the same type of wetland as Balm Marsh was studied in Italy (Borin et al 2001). Although this wetland was designed to treat agricultural waste water, it is much smaller than Balm Road Treatment Marsh, and receives less water from a smaller agr icultural area. Nitrogen was the only water quality parameter analyzed and it was sampled on a daily basis. Reductions were found to be almost 90%. In Thailand, a constructed wetland was studied to determine its efficiency for removing pollutants from s eafood industry wastewater (Yirong & Puetpaliboon, 2004). The wetland consisted of a series of ponds with differing process designs with the final pond in the series design ed as a free water surface wetland. Samples were collected once per week for a per iod of only four months after approximately one year of the wetland becomin g operational. N itrate concentrations were found to be higher at the w etland outflow than the inflow, however total K jeldahl nitrogen removal was 56%. Suspended solids removal was 95%. In Polk County, Florida, a natural cypress dome was studied that has been used to treat municipal wastewater since 1985 (Martin et al 2001). Water quality was monitored on a monthly basis at the inflow, center, and outflow of the wetland for a p eriod of eight years, which allowed for the evaluation of long term performance. Average r emoval efficiencies for the eight years were 38% total suspended solids, 90% total nitrogen, and 48% total phosphorus based on mass.
21 Althoug h treatment wetland perf ormance has been studied around the world, pe rformance varies due to several site specific factors includi ng wetland design soil, plant species and number, fauna, hydrology, climate, receiving water and source water This creates difficult ly in using th e results from a particular study to assess another wetland (Kadlec & Knight, 1996; Borin et al 2001; Carleton et al 2001). Even though the performance of a variety of treatment wetlands is well represented in the literature, there are fewer studies tha t describe the performance of constructed wetlands to treat nonpoint source pollution and fewer that focus specifically on agricultural runoff. Even results from the few existing studies cannot be used to accurately characterize the performance of a diff erent constructed treatment wetland for agricultural runoff, bec ause the available data contain no clear performance trends based on characteristics (Kadlec & Wallace, 2009). Studies that were found that address agricultural runoff focus on pollutant remo val efficiencies and not the overall a ffects to downstream ambient water quality (Koskiaho et al 2003; Kovacic et al 2000; Tanner et al 2005). Receiving water impacts appear to be lacking for all wetland types and pollution sources. Factors Affectin g Performance The factors causing variability in the performance of treatment wetlands have been studied. Kuehn and Moore (1995) compare data from constructed wetlands treating pulp mill effluent for reduction in biochemical oxygen demand and total suspe nded solids. Ponds were constructed with varying retention times and vegetation and a replicate pond was constructed for each, so that there were pairs of nearly identical ponds for comparison. Samples were taken from the inflow and outflow of each of th e ponds and the resulting data compared. The results showed that similar pairs of ponds had very
22 low performance variability. Significant variation occurred between all other ponds. The factors leading to variation included vegetation type and retention time as well as variation over time according to the season. Other studies of comparable ponds have shown similar results (Gearh eart, 1992). These studies demonstrate some of the important factors affecting variability in the performance of treatment we tlands and support the fact that wetland performance results cannot be extended across wetlands. Carleton et al (2001) compared pollutant reduction efficiencies from forty nine wetland systems used to treat direct stormwater runoff flows or runoff impac ted surface water. When the results from all forty nine wetlands were combined and compared to values reported for wastewater treatment wetlands, nitrogen removal efficiencies were very similar. Stormwater treatment wetlands, however, showed much higher variability than wastewater treatment wetlands, which is generally expected due to the nature of stormwater and variable flows. Removal rate constants for several parameters were calculated and compared to those constants reported in the literature for wa stewater treatment wetlands and found to be similar. This study suggests that it is reasonable to expect stormwater treatment wetlands to have removal rate constants similar to wastewater removal rate constants, which have been extensively studied and pub lished in the literature compared to stormwater removal rate constants. The rate constants can be used in determining the pond area needed to achieve a specific reduction of pollutants by a stormwater wetland. A number of studies have explored the phos phorus retention capacity and removal efficiency of treatment wetlands (Liikanen et al, 2004; Moustafa, 1999; Nova k et al, 2004; Casey & Klaine, 2001; Richardson, 1985; Dierberg, 2001). Liikanen et al
23 demonstrates the importance of soil characteristics in phosphorus removal efficiency. Soil properties were studied before the construction of the treatment wetland and used to determine its ability to retain phosphorus. Soil samples were used in laboratory studies to determine their ability to remove phosph orus and water samples were taken at the inflow and outflow of the wetland once it was operational to d etermine its efficiency. The study found that if soils on the wetland project site contain high amounts of phosphorus, it is essential to remove the soi ls prior to construction because they can lower phosphorus removal of the wetland. This research is significant in that it demonstrates the importance of soil characteristics in phosphorus removal. Another factor involved in phosphorous retention capaci ty of wetlands is the extractable aluminum content of the soil (Richardson, 1984). Soils from a wide range of natural wetlands were sampled to determine their phosphorus sorption capacity. Actual measurements of phosphorus exports from the same wetlands correlated to soil sorption capacities. The sorption capacity was then compared to other soil characteristics such as percent organic matter, pH, and extractable aluminum, iron, and calcium. Statistical analysis showed a direct correlation between the am ount of extractable aluminum present and soil sorption capacity. This study reconfirms the importance of soil characteristics in phosphorus removal efficiencies. The data also indicated that initial phosphorus removal rates of a wetland may be followed b y large exports of phosphorus within a few years. There are other factors influencing phosphorus retention in treatment wetlands as demonstrated by Moustafa (1999). Moustafa examined data from approximately one hundred wetlands to determine their phosp horus loading rates, morphology, and hydrological characteristics. The research found that water depth plays a key role in
24 phosphorus retention and showed that shallow water depths within a wetland increase the amount phosphorus removal. Phosphorus remov al efficiencies were also demonstrated to be a function of water and phosphorus loading rates into the wetland. The relationship can be used to predict phosphorus removal efficiencies An in stream wetland that receives water from an agriculturally inten sive subwatershed in North Carolina was examined for phosphorus retention (Novak et al 2004). Weekly samples for dissolved phosphorus were taken along with flow data to determine inflow and outflow dissolved phosphorus load estimates and retention and rel ease rates. Water column dissolved phosphorus samples were also collected at two points within the wetland along with soil samples that were analyz ed for phosphorus. The data were then used to determine the sorption or desorption tendency of the wetland sediments by comparing the water and soil samples. Water column sediments were also sampled and analyzed for dis solved phosphorus. The data were used to produce dissolved phosphorus concentration profiles under varying management conditions, including fl ooding, draining and shifts in dissolved phosphorus concentrations. These results can be used to determine optimal ranges for variables that affect phosphorus retention including residence time and sediment surface area. An important conclusion drawn fro m this research is the fact that this particular wetland did not provide effective long term dissolved phosphorus retention. The results here indicate that long term detention in phosphorus laden wetlands may be unlikely. If inflow phosphorus concentrati ons are reduced resulting in less phosphorus present in the water column than the underlying sediments, the sediments may release phosphorus resulting in higher phosphorus discharges than inflows, creating a negative phosphorus retention rate.
25 Phosphorus release from underlying sediments can negatively impact treatment wetland removal rates. Nitrogen retention by wetlands has been studied as well. A study by Felberova et al reported seasonal variations in nitrogen retention (1993). Removal efficiencies were determined to be greater in the summer months. Nitrogen retention was also shown to be affected by plant species. A constructed wetland that received wastewater treatment flow was designed with four subsurface horizontal flow treatment beds. Pairs of beds were planted with a different wetland species. The treatment beds with different species showed varying removal efficiencies, while similar beds displayed similar results. This study was important in describing factors that affect nitrogen reten tion in wetlands. A more thorough investigation of vegetation and temperature effects on nitrogen removal efficiency was performed by Bachand and Horne (2000). The study was intended to determine the design features of a constructed treatment wetland that may contribute to increased nitrogen removal performance. Species were planted in six treatment cells; two cells contained only bulrush ( Scirpus spp .), two cells only cattail ( Typha spp .) and the last two cells contained a combination of the two. Th e cells received water with nitrogen concentration similar to that from a wastewater treatment plant. Water samples were collected at the inflow and outflow of the cells on varying frequency, at times as often as every day. Plant and soil samples were al so taken and all three sample types were analyzed for nitrogen concentration. Nitrogen removal rates between cells with different plant composition showed significant differences. The mixed vegetation displayed the greatest efficiency followed by the cat tail and the bulrush species. The study was combined with a thorough review of the literature to make
26 detailed suggestions of vegetation composition for the most efficient nitrogen removal. By comparing nitrogen concentrations, it was found that sediment and plant uptake only accounted for a fraction of the nitrogen removed from the water column, concluding that denitrification was the primary responsible process. It was further concluded that dissolved oxygen concentrations and nitrate availability did not affect denitrification, but that water temperature was likely the most influencing factor. This research suggests that vegetation effects, water temperature and seasonal variations should all be taken into consideration when examining nitrogen remova l efficiency. There appears to be a general consensus in the published literature that pollutant removal efficiencies show seasonal variatio n. This suggests that data should be examined on a seasonal basis in addition to long term comparisons Processes and Design C hemical, physical and b iological cycles and processes in treatm ent wetlands are important factors in pollutant removal. Kadlec (1999) presents some of these cycles and describes their effects on pollutant removal. For example, solar radiati on drives photosynthesis influencing plant processes on an annual cycle. Pollutant uptake and burial is regulated by the biogeochemical cycle and rain and evapotranspiration influence the wetland water budget which in turn affects pollutant removal. Due to many of the cycles involved, nitrogen and phosphorus removal may vary seasonally due to temperature dependent processes. Hammer has published a substantial amount of work on treatment wetlands in peer reviewed journals as well as written and edited b ooks on the topic, especially concerning treatment wetland design (Hammer, 1989a; Hammer 1989b; Hammer, 1992;
27 Hammer 1994; Hammer 1997). His work was reviewed extensively an d incorporated into the present research, an example of which is presented here Hammer (1992) provides good background information in the historical use of wetlands, both natural and constructed, for their water treatment capabilities. He also discusses the four principle components in the pollution reduction functions of wetlands vegetation, water column, substrates, and microbial populations. Hammer then presents a detailed discussion of designing treatment wetlands for livestock wastewater treatment. This includes the use of an optional settling basin just upstream of the wetla nd to remove solids, site selection criteria, the required treatment area, suggested number of treatment cells, cell shape, water control structures, pond bottom and liners, and vegetation. The above criteria are then adjusted and presented along with add itional recommendations for adapting the design for pasture or crop field runoff. T he design details of the Balm Road Tre atment Mars were evaluated and compared against design criteria found in the literature. Kadlec and Knight (1996), a chemical engin eer and a wetland ecologist, have both been studying treatment wetlands since 1970. They combined their efforts in 1996 to produce the first engineering design manual for treatment wetlands. Most of the literature published since this book, reference the manua l at least once, and it was re ferred to often for this research. Although the work is primarily focused on treatment of wa stewater, rather than nonpoint source pollution the underlying concepts are generally the same. Topics included in this work range from wetland structure and function, soils, hydrology, microbial communities, plants, wildlife, effects on water quality with detailed processes, modeling efforts and values for rate constants and regression parameters, wetland design, operation and maintenance, and case studies.
28 Although the text presents only a limited amount of information on monitoring and performance determination, which i s the main focus of the present research, the information presented in the text was necessary to present the research in its relevant context. Data Analysis Data analysis for the determination of treatment wetland performance found in the existing literature has relied primarily on the comparison of inflow and outflow constituent concentration averages som etimes combined with discharge data to find a concentration reduction or mass removal (Kuehn & Moore, 1995; Yang et al 1995; Mander & Mauring, 1997; Borin et al 2001; Martin et al 2001; Yirong & Puetpaiboon, 2004). How ever, outflow pollutant concentrat ion and discharge data for Balm Road Treatment Marsh are not available and the current research focus is the affect on receiving water quality There has been a vast array of literature published on water quality data analysis which was exami ned in relati on to the present research (Hirsch et al 1982; van Belle & Hughes, 1984; Helsel, 1987; Lettenmaier, 1998; Berryman et al 1998; Loftis et al 1991 ; Hirsch et al 1991; Harcum et al 1992). An important consideration in this research was the use of parame tric verses nonparametric statistical analysis which is discussed at length. Nonparametric methods have distinct advantages when analyzing data without normal distributions and many outliers. This literature was the basis for choosing statistical method s for data analysis to determine impacts to ambient water quality data Methods of calculating pollutant loads under typical conditions where discharge data are available at near continuous intervals, but water quality data are collected less
29 frequently have been studied extensively (e.g., Dolan et al, 1981; Walling & Webb, 1981; Ferguson, 1987; Richards & Holloway, 1987; Cohn et al, 1989; Preston et al, 1989; Kronvang & Bruhn 1996). The methods used to produce load estimates using limited water quality data can be split into three general categories: averaging approaches, regression models, and ratio estimators. Averaging is considered to be the simplest approach and is based on some form of average used in calculations with available discharge and wa ter quality data (Preston et al, Richards 1996). There have been a number of different averaging approaches suggested with varying degrees of accuracy and precision (Dolan et al, 1981; Walling & Webb, 1981; Preston et al, 1989). Although it has been foun d that regression and ratio methods are often more accurate than averaging methods, they frequently lack precision and produce inconsistent results. Some averaging methods, although they often greatly underestimate loads, tend to be fairly precise among e stimates and may be the more appropriate choice in certain situations (Walling & Webb, 1981; Richards, 1996). These studies were used in determining the most appropriate method for estimating pollutant loads
30 Chapter 4 Research Design Problem Statement The purpose of this study was to determine the water quality impacts of Balm Road Treatment Marsh in order to gain better understanding of the performance of constructed treatment wetlands for agricultural pollution management Research Questions T hree research questions were answered in order to address the problem statement What were the resulting ambient water quality impacts of Balm Road Treatment Marsh on Bullfrog Creek ? W as there a subsequent pollutant load reduction to Tampa Bay ? How does the performance of constructed wetlands used to treat agricultural pollution compare to wetlands used to treat other pollution? The se answer s aided the determination of whether or not constructed treatment wetlands are appropriate for agricultura l pollution management, which in turn will help water resource managers design effective pollution reduction strategies for agricultural nonpoint source pollution. Study Significance As previously noted, agricultural nonpoint source pollution is the number one source of water quality impairments to most surface water in the United States (Parry, 1998). It is therefore imperative to find effective tools and management practices to reduce pollution from this source in order to ensure water bodies meet their designated
31 standards. Surface water is important for use as a source for drinking water, navigation, recreation, and habitat for wil dlife and fish among others. M eet ing quality standards for these uses is dependent on effective management practices that lead to m aintaining and improving water quality. The proposed research will address a specific management practice that is being used with increasing frequency but for which there is little information concerning its effectiveness (Mitsch & Gosselink, 2000). Van der Valk and Jolly (1992) found that studies which address the effectiveness of constructed wetlands as nutrient sinks are one of the most impo rtant research needs regarding the use of wetlands to treat agricultural p ollution. This research is an importa nt step in filling the information gap that exists on the effectiveness of constructed wetlands to reduce agricultural nonpoint source pollution. In addition, the information on overall a ffect on receiving water bodies is limited. This is of particular i mportance when the treatment objective is to improve water quality in receiving water s, for example to meet water quality standards. The information on pollutant removal efficiency of wetlands available in the literature rarely includes overall affects on downstream water quality.
32 Chapter 5 Study Area Location The Bullfrog Creek basin is 100 square kilometers located between the Alafia and Little Manatee Rivers in southern Hillsborough County. It drains to the Hillsborough Bay segment of Tampa Bay j ust south of the Alafia River (Dames and Moore, 2000). The basin has been grouped with the Coastal Hillsborough Bay major basin for loading the east with rapid declines to sea level moving west to the bay (Dames and Moore, 2000). Balm Road Marsh is located near Bullfrog portions of th e Bullfrog Creek basin (Figure s 1 and 2 ). The 12 ha treatment system was built on the southeast c orner of a 121.4 ha portion of c ounty land. The Balm Road upland areas to less than 19.8 m NGVD in the stream channel located in the west end of the site (Ayres, 2000). Climate The area climate is subtropical, with long humid summers and mild short winters. The majority of rainfall occurs between the summer months of June a nd September as seen in Figure 5 Rainfall is highly variable both spatially and temporally with the
33 maj ority of rain resulting from isolated summer thunderstorms. Intense rainfall may result from hurricanes, tropical storms, or tropical depressions. Winter rainfall is light (Dames & Moore, 2000). Historical data retrieved from the nearest Southeast Regio nal Climate Center weather station loc ated in Parish, Florida, reveal that the average maximum summer temperature is approximately 33 C with an average minimum of 22 C. Winter average maximum is 23 C and average minimum is 11C. Figure 5. Average Monthly Precipitation in Parish, Florida (SRCC, 2007). Soil The dominant soil type in the Bullfrog Creek basin is Myakka, which is a fine, poorly drained sand with no or extremely low slopes. Bullfrog Creek and its tributaries are dominated by Winder fine sands, which is frequently flooded and either flat or nearly flat. The dominant hydrological soil group is D in the naturally undrained condition and B where the soils have been artificially drained. Group D soils are described as having
34 high runoff potential and low infiltration rate. They are mostly shallow clays with a high water table. Group B soils have a moderate infiltration rate. They are moderate to deep, with a moderately fine to moderately course texture, and are moderately well drained (Dames and Moore, 2000). The soil survey for the Balm Ro ad property is shown in Figure 6 (USDA, 2006). The highest elevations on the site consist of mostly Archbold fine sand, labeled 3 on the map, and some Pomello fine sand (41). Ayres studied historic al aerial photographs and adjacent undisturbed habitat to determine that this area formerly supported a scrub habitat (2000) Myakka fine sand (29), which is generally associated with pine flatwoods, covers almost half of the property area and the majorit y of the actual wetland site. Other soils found on the property include Basinger, Holopaw, and Samsula soils (5) located in the natural flatwoods pond on site, St. Johns fine sand (46) which is typically found in areas of natural overland flow, and Winder fine sand (60) found in the Bullfrog Creek floodplain (Ayres 2000; USDA 2006).
35 Figure 6. Soil Map for Balm Road Treatment Marsh site (USDA, 2006). Land Use The Bullfrog Creek basin consists of 65% agricultural lands including field and row crops, citr us, and pasture. Residential is the second highest land use which comprises 8% of the total area. Other minor land uses include natural lands and industrial. However, future land uses are projected to be primarily residential with agricultural lands bei ng quickly developed into residential areas. (Dames and Moore, 2000). The Balm Road property is mostly uplands with some natural wetlands. The land was previously converted to row crops, which involved the removal of native vegetation, grading, and the c onstruction of an extensive network of drainage ditches throughout the
36 surrounding uplands. One of the larger ditches, just upstream of the site and parallel to McGrady Ro ad, receives runoff from a few hundred hecta res of pasture and citrus groves. Prior to construction of the treatment marsh, the entire property was used for cattle grazing (Aryes 2000). The area upstream to the inflow of the marsh site consists of approximately 741 ha of land used primarily for pasture, citrus groves, and tropical fish farms, and a few single family residential areas. Hydrology The major conveyance in the Bullfrog Creek basin is Bullfrog Creek. The creek has several tributaries from the east, with the largest b eing Little Bullfrog Creek ( Figure 3). The creek flows fr om the southeast to the northwest, with the longest segment flowing directly to the north. The flow is relatively quick in the lower reaches and slow in the wetland sections in the upper reaches and near the headwaters (Dames and Moore, 2000). Detailed hy drologic studies and modeling have been performed for the Bullfrog Creek/Wolf Creek Watershed and were later modified by Ayres for use specific to the Balm Road property (Dames & Moore, 2000; Ayres, 2000). Ayres found that the 2.33 year storm event has a peak flow rate of 13 m 3 /s with a 26.93 m stage and the 100 year storm event has a peak flow rate of 45 m 3 /s with a 27.57 m stage at the marsh site.
37 Chapter 6 Methods Sample Collection and Laboratory Analysis Water quality data from four loc ations on Bu llfrog Creek were an alyzed to answer the research questions In order to establish base line conditions, ambient water quality monitoring on Bullfrog Creek began six years prior to the construction of Balm Road Treatment Marsh in 1998. The first water qu ality sample collection site on Bullfrog Creek was located just upstream of the proposed inflow to the marsh system at the end of McGrady Road For the purpose of this research, this site is called Upstream This site continued to be monitored throughout the construction phase and post construction until the end of the study The Upstream site was located downstream from a culvert on Bullfrog Creek after merging with a drainage ditch. The area was wide and water flow slowed and cre ated a small pool betw een the upstream and a second downstream culvert. The creek split here and water either flowed through the first diversion structure continuing down the first branch of Bullfrog Creek or down a canal which led to the second d iversion structure. The seco nd structure diverted baselin e flow to a second branch of Bullfrog Creek. All other flows went though the treatment system. The second sample collection site of in terest was monitored beginning in 2001, over two years before construction of the treatment system was completed The site was located on Bullfrog Creek just downstream from the planned wetlan d discharge. The site
38 was approximately 1.3 km downstream from the Upstream site and after the treatment system was complete, it included both the untreat ed baseline flow though Bullfrog Creek and the treated wetland discharge. The data collected here represents the overall impacts of the treatment system to Bullfrog Creek. Th is site is named Downstream 1 Additional water quality sample sites were loca ted further downstream from the ambient water quality. The second downstream site was located in Bullfrog Creek Scrub, a 650 ha nature preserve approximately 9 km downstream from the treatme nt system. This site is named Downstream 2. Water quality monitoring began at this site in August of 2002 and continued through the end of the study The final monitoring site, Downstream 3, was located at a United States Geological Survey (USGS) flow gauge, and is the only site with f low rate data for the creek. The site is located approximately 12.5 km downstream from the treatment system at Big Bend Road Monitoring here began in 1998 and continued until study completion The site locations can b e found in Fi gure 7 with the Upstream and Downstream 1 sites just abov e and below the area labeled Balm Road Marsh and the other downstream sites locat ed further downstream from the m arsh Sampling for all four sites was conducted on a monthly basis unti l the project concluded in September of 2007. Samples and field measurements were collected by the Southwest Florida Water Management District (SWFWMD). Samples for the final four months of monitoring were collected by the researcher, and previous data w ere collected by other SWFWMD staff. Monthly sampling was schedu led at the convenience of SWFWMD staff, so it usually occurred on a different day every month without regard to previous rainfall. Therefore, some
39 sampling events may have occurred immediate ly following storm event s while others occurred during extended dry periods and sampling intervals vary month to month. Each site was sampled within a few hours on the same day. F igure 7 Bullfrog Creek water quality sample sites. Sites are sym bolized as red dots along Bullfrog Creek. The sites of interest for the proposed research are labeled as Upstream, Downstream 1, Downstream 2, and D ownstream 3 Monthly water quality measurements included a suite of parameters. Field measurements were taken using a YSI 6 Series Sonde and included temperature, specific conductance, dissolved oxygen, pH, total (stream) depth, and sample depth. Samples for laboratory analysis were then collected following the Florida Department of Environmental Protection Samples were Upstream Downstream 1 Downstream 3 Downstream 2
40 collected at half of the total depth at the sample site. Nutrient samples were immediately preserved with sulfuric acid to a pH of less than 2 All samples were immediately put on ice for prese rvation. Samples were transported to the SWFWMD laboratory in Brooksville, Florida for analysis of total suspended solids, nitrogen, ammonia, nitrate, nitrate/nitrite, phosphorus, orthophosphate, chlorophyll a, chlorophyll b, chlorophyll c, phaeophytin, t urbidity, total coliform, and fecal coliform. Sampling, analyses, and associated tasks were performed in accordance with federal (USEPA), state (FDEP), and regional (SWFWMD) quality assurance requirements. The SWFWMD laboratory is certified by the Nation al Environmental Laboratory Accreditation Program (NELAP) under the Florida Department of Health for all parameters analyzed. The list of SWFWMD NELAP c ertified methods can be found in the FDEP NELAP Certified Laboratories Database available online (FDEP, 2009 b ). The paramete rs of intere st for the present research are total suspended solids (TSS), total nitrogen (TN), and total phosphorus (TP). TSS is the measure of suspended material present in a sample and includes sediments and other particulates Ni trogen and phosphorus are present in surface waters in a variety of forms and TN and TP includes each of these forms (Florida Lakewatch, 2000). Table 1 lists the detection limits, units of measure, and methods used by SWFWMD for each analysis which can be found at the original sources (EPA, 1983; Greenburg et al 1992 ). Parameter Detection Limit Units Method Total Suspended Solids 0.01 mg/L S.M. 18th ED. 2540 D Total Nitrogen 0.16 mg/L E.P.A. 353.2 Total Phosphorus 0.03 mg/L E.P.A. 365.1 Table 1. De tection limits, units, and methods for parameters of interest.
41 Analytical results were received from the laboratory in the form of a hardcopy report and entered along w ith field measurements into Excel worksheet s by the researcher or other SWFWMD staff. These Excel files were used as the source for all analysis for this research. The field data were also entered into a separate spreadsheet and sent via email to the SWFWMD laboratory staff to be combined with laboratory data and uploaded to the state and federal storage and retrieval databases called STORET. Raw data can be retrieved from either FDEP STORET using organization identification code 21FLSWFD (FDEP, 2009c; SWFWMD, 200 9 ). The s tation names in the databases will not match those used here, so identificati on numbers are listed in Table 2 Station Name Station ID Dates Available Upstream 17927 4/1998 9/2007 Downstream 1 17982 4/1998 9/2007 Downstream 2 17737 8/2002 9/2007 Downstream 3 17925 12/2002 9/2007 Table 2. Station ID number s and available dates for retrieval from online databases D ischarge, or flow rate, is measured by a United States Geological Society (USGS) gaging station on Bullfrog Creek at the Downstream 3 sample site (Figure 8 ) The gaging station on Bullfrog Creek is a real time system that sends instantaneous discharge data to USGS via satellite. Discharge is monitored indirectly and calculated using stage height and the predetermined rating curve for this lo cation. Stage height is recorded by a stilling well which consists of a float inside a vertical pipe attached to a
42 bridge on Old Big Bend Road where it crosses Bullfro g Creek. The float is attached by a pulley to a data logger and satellite (USGS, 2009 b ) Data was downloaded from the USGS Instantaneous Data Archive site number 02300700 / Bullfrog Creek near Wimauma FL (USGS, 2009a). Figure 8 Picture of USGS Gaging Station This site is number 02300700 Bullfrog Creek Near Wimauma, FL 9/26/2009.
43 Data Organization Total nitrogen, total phosphorus, and total suspended solids data for the Upstream and D ownstream 1 3 sites were extracted from the exist ing data set. Data were grouped into three time per iods: baseline, pre, and post (Figure 9 ) Th e baseline data were not used in the research due to the lack of data at two of the four stations. The pre phase represent ed the time period prior to the treatment system becoming fully operational The post phase represented the time period after the tr eatment system was fully operational Pre and post data were then further split into wet season and dry season s Wet and dry season determinations were based on historical rainfall data from the data collection site at nearby Parish, Florida (Figure 5) For the purposes of t his research the wet season was from June to Septe mber, and the dry season was from October to May. F igure 9. Data grouping diagram. The dataset was split into three subsets: baseline, pre, and p ost. Dataset May 1998 Sept 2007 n = 113 Baseline May 1998 Nov 2001 n = 44 1. Upstream 2. Downstream 3 Pre Dec 2002 Dec 2004 n = 36 1. Upstream 2. Downstream 1 3. Downs tream 2 4. Downstream 3 Post Jan 2005 Sept 2007 n = 33 1. Upstream 2. Downstream 1 3. Downstream 2 4. Downstream 3
44 The baseline dataset consist ed of data from the Inflow and Downstream 3 sites collected from May 1998 to November 2001. During this time period a total of 44 water quality samples were collected from the both the Inflo w site and Downstream sites. No sampl es were collected at the other two site s duri ng this time. These data were not used due to lack of available data at the Downstream 1 and 2 site s The pre dataset consist ed of data collected in December 2001 through December 2004. During this period 36 sample events occurred at all sites except Downstream 2 where 25 samples were collected The treatment system became fully operational in late December 2004, so samples collected prior represent untreated conditions in the creek. This dataset was analyzed to find overall median TN, TP, and TSS as well as wet and d ry season median s These data were used in co mpar isons with data from the post phase from each site to determine water quality impacts of the treatment sys tem to Bullfrog Creek. The post dataset include d data collected since Ja nuary 2005, after the treatment system was fully operational Thirty three samples were collected for each of the four sites. Data from the Downstream 1 3 sites during this time peri od reflect the impacts of the trea tment system. D ata were to ambient water quality in Bullfrog Creek and load reductions to Tampa Bay Both overall and wet and dry season medians were found for both the Upstream and Downstream 1 3 sites. The post Upstream and Downstream 1 3 datasets were compared to both t he corresponding pre datasets as well as the post Upstream in order to determine the water quality impacts of Balm Road Treatment Marsh on Bullfro g Creek.
45 Downstream 3 site data were also combined with discharge data and pre/post comparisons were made to determine load reductions to Tampa Bay. Statistical Analysis The software package PASW 18 .0 (formerly SPSS) was used for all statistical analyses. Seasonal means and other descriptive stat istics for each dataset were determined. The comparisons between datasets to determine treatmen t impacts on water quality and pollutant loads were then made. Like most w ater quality data the datasets were not n ormally distributed and log t ransformations were not appropriate due to the presence of heavy tails. Histogram s for each da taset can be found in Appendix B Due to the lack of a normal distribution and presence of outliers, nonparametric methods were ch osen for statistical analysis (Hirsch et al 1982 ). The Mann Whitney test was used to compare the pre/p ost medians for each sample site The test determine s whether or not the datasets come from different populations by comparing medians and determine d whether or not the values were larger in the p re Downstream 1 3 datasets when compared to the p ost Downstream 1 3 dataset s. If the p re Downstream concentrations are found to be significantly larger than the p ost Downstream concentrations a reduction in p ollutant conc entration may be attributed to Balm Road Treatment Marsh. Wet season, dry season and overall TP, TN, and TSS for each dataset were compared following this example as depicted in Figure 10 Data from the Upstream si te were compared to data fr om th e Downstream sites 1 3 for the p ost treatment system time period. The Wilcoxon matched pairs sign ed rank test was used for these comparisons (Figure 10 ) This test is similar to the Mann Whitney test, except that it compares the differences in the p aired medians to determine
46 which come from a larg er population. If the Upstream site has significantly higher pollutant concentrations than the Downstream site s a reductio n in pollutant concentration may be attributed to Balm Road Treatment Marsh. Post Upstream Mann Whitney Pre Upstream Mann Whitney Pre Downstream 1 Post Downstream 1 Wilcoxon Post Upstream Mann Whitney Pre Down stream 2 Post Downstream 2 Wilcoxon Post Upstream Mann Whitney Pre Downstream 3 Post Downstream 3 Wilcoxon Post Upstream Figure 10 Diagram depicting sample co mparisons using nonparametric test s Each datase t on the left was compared to the datasets on the right using the method listed red uction estim ates to Tampa Bay were determined by comparing annual pollutant loads at the Downstream 3 sit e both p re and post treatment system Instantaneous d ischarge data from the USGS stre am ga ging station at the Downstream 3 site are av ailable online i n fifteen minute increments. Sampling time s were recorded to the nearest five minutes, so water quality data were pair ed to a discharge rate within five minutes of the sampling time The aim of this research is to accurately detect the change in load ra ther than to quantify the actual load, so precision is more important than accuracy, and it was determined that averaging was the most appropriate technique for load estimation
47 Walling and Webb (1981) analyzed six averaging techniques and found two that provide th e most consistent results Both methods were used to estimate pre and p ost average annual loads at the Downstream 3 site and produced similar results. The results from tions to Tampa Bay (1981) : annual load n i=1 C i n i=1 Q i /n) where: annual load = estimated annual load (kg/year) K = conversion factor to take account period of record and weight units (60*60*24*365*0.000001) C i = instantaneous concentration associated with individual samples (mg /L) Q i = instantaneous discha rge at time of sampling (L/sec) n = number of samples Wet and dry season loads were calculated for each year during both the pre and p ost time periods using the formula above. The seasonal mean concentration for the correspondin g phase was used when monthly water quality data were missing. An overall annual load was found by adding the time weighted wet and dry season loads for each year. This method assumes that the values of concentration and discharge associated with the indiv idual monthly samples may be averaged to provide representative mean values for the associated time of record.
4 8 Chapter 7 Results and Discussion Water Quality Descriptive Statistics Overall pollutant concentration descriptive statistics including the mi nimum, maximum, median, mean and standard devia tion are found in Table 3 f or each sample site with pre and p ost phases combined N Minimum Maximum Median Mean Standard Deviation Total Suspended Solids (mg/L) Upstream 112 0.20 32.72 2.44 4.20 5.43 Do wnstream 1 68 0.87 22.10 3.25 4.81 4.21 Downstream 2 59 0.16 9.95 0.54 1.22 1.91 Downstream 3 111 0.50 45.25 3.20 4.73 6.01 Total Nitrogen (mg/L) Upstream 112 0.25 3.94 0.90 0.93 0.60 Downstream 1 68 0.22 3.05 0.87 0.96 0.50 Downstream 2 59 0.30 1.2 2 0.55 0.58 0.20 Downstream 3 110 0.11 3.33 0.72 0.83 0.42 Total Phosphorus (mg/L) Upstream 113 0.04 0.58 0.11 0.15 0.11 Downstream 1 68 0.05 0.64 0.12 0.16 0.10 Downstream 2 59 0.12 0.60 0.26 0.29 0.13 Downstream 3 111 0.08 0.60 0.24 0.26 0.11 Tab le 3 Overall descriptive statistics for entire d ataset available at each site. TSS means ranged from 1.22 mg/L at Downstream 2 to 4.80 mg/L at Downstream 1. TN means ranged from 0.58 mg/L at Downstream 2 to 0.96 mg/L at Downstream 1. TP means ranged f rom 0.15 mg/L at Upstream to 0.30 at Downstream 3. For comparison, typical statewide values are provided in Table 4. Boxplots are shown in Figures 11 13.
49 Parameter (mg/L) 10 th Median 9 0th TSS 2 7 26 TN 0.5 1.2 2.7 TP .02 .09 .89 Table 4 Typical s tatewide percentile values for Florida streams Median TSS values fall below the 50 th percentile in statewide comparisons, but maximums at two of the four site fall above the 90 th percentile. Median TN values fall below the 50 th percentile in statewide comparisons, howe ver maximum values at three of the four sites fall above the 90 th per centile. Total phosphorus medians and maximum s fall above the 50 th percentile. Figure 11 TSS dataset boxplot.
50 Figure 12 TN dataset boxplot. Figure 1 3 TP dataset boxplot.
51 According to the box plots there are many outl iers displayed as circles, and ext reme values, displayed as stars, for most of the datasets Outliers are more than 1.5 times the interquartile range and extreme values are greater th an 3 times the interquartile range. The data are bound at the minimum detection limit for the given parameter and contain occasional high values, which makes the datasets highly skewed with a non normal distribution. These are common characteristics of w ater quality data (Helsel, 1987). For more information on the distributions, see the histograms in Appendix B Descriptive statistic s were also found after splitting data into p re and post phases for both combined seasons and wet and dry seasons Ta ble 5 and Table 6 contain wet and dry season descriptive statistics. B ox plots for combined, wet and dry seas ons are displayed in Figures 14 19
52 Site Season N Minimum Maximum Median Mean Standard Deviation Pre TSS (mg/L) Upstream Wet 12 0.6 4 17.92 4.09 5.09 4.98 Upstream Dry 24 0.80 17.73 2.50 3.30 3.41 Downstream 1 Wet 11 2.42 14.44 4.90 5.98 3.33 Downstream 1 Dry 24 0.87 12.25 2.20 2.97 2.56 Downstream 2 Wet 8 0.50 2.89 0.66 1.04 0.85 Downstream 2 Dry 18 0.50 7.36 0.60 1.24 1.67 Dow nstream 3 Wet 11 1.42 12.52 4.60 5.27 3.06 Downstream 3 Dry 23 1.10 29.63 2.88 5.04 6.84 Pre TN (mg/L) Upstream Wet 12 0.52 1.85 1.26 1.28 0.43 Upstream Dry 24 0.25 3.35 0.91 1.02 0.72 Downstream 1 Wet 11 0.58 1.54 1.19 1.11 0.35 Downstream 1 Dry 24 0.22 3.05 0.57 0.86 0.65 Downstream 2 Wet 8 0.55 0.83 0.74 0.73 0.09 Downstream 2 Dry 18 0.35 1.22 0.54 0.63 0.26 Downstream 3 Wet 11 0.59 1.12 0.90 0.92 0.15 Downstream 3 Dry 23 0.11 3.33 0.69 0.86 0.61 Pre TP (mg/L) Upstream Wet 12 0.11 0.53 0 .18 0.25 0.15 Upstream Dry 24 0.05 0.30 0.11 0.12 0.07 Downstream 1 Wet 11 0.14 0.64 0.17 0.23 0.14 Downstream 1 Dry 24 0.05 0.36 0.11 0.12 0.06 Downstream 2 Wet 8 0.32 0.54 0.46 0.45 0.08 Downstream 2 Dry 18 0.12 0.60 0.20 0.25 0.13 Downstream 3 We t 11 0.16 0.55 0.39 0.37 0.18 Downstream 3 Dry 23 0.11 0.51 0.23 0.24 0.10 Table 5 Wet and dry season descriptive statistics for the pre phase.
53 Site Season N Minimum Maximum Median Mean Standard Deviation Post TSS (mg/L) Upstream Wet 12 0.6 3 13.2 2.40 3.85 3.75 Upstream Dry 20 0.55 6.48 1.64 2.39 1.79 Downstream 1 Wet 12 1.36 17.07 4.52 6.37 4.81 Downstream 1 Dry 21 0.92 22.10 4.26 5.39 5.20 Downstream 2 Wet 12 0.16 7.11 0.75 1.24 1.86 Downstream 2 Dry 21 0.16 9.95 0.50 1.27 2.46 Down stream 3 Wet 12 0.87 10.80 5.42 5.57 3.31 Downstream 3 Dry 21 0.67 7.27 1.65 2.40 1.70 Post TN (mg/L) Upstream Wet 12 0.36 1.57 0.84 0.86 0.42 Upstream Dry 21 0.33 1.98 0.70 0.86 0.47 Downstream 1 Wet 12 0.56 1.99 0.93 1.10 0.50 Downstream 1 Dry 21 0.40 1.88 0.81 0.91 0.34 Downstream 2 Wet 12 0.49 0.85 0.63 0.65 0.11 Downstream 2 Dry 21 0.30 0.70 0.44 0.44 0.11 Downstream 3 Wet 12 0.53 1.32 0.82 0.89 0.24 Downstream 3 Dry 21 0.35 0.98 0.54 0.59 0.19 Post TP (mg/L) Upstream Wet 12 0.11 0.58 0.18 0.23 0.14 Upstream Dry 21 0.40 0.12 0. 8 0 0.79 0.22 Downstream 1 Wet 12 0.12 0.53 0.16 0.22 0.13 Downstream 1 Dry 21 0.05 0.35 0.10 0.12 0.07 Downstream 2 Wet 12 0.29 0.48 0.31 0.38 0.07 Downstream 2 Dry 21 0.13 0.36 0.20 0.21 0.06 Downstream 3 Wet 12 0.24 0.44 0.38 0.32 0.07 Downstream 3 Dry 21 0.08 0.31 0.16 0.17 0.06 Table 6 Wet and dry season statistics for the post phase.
54 Figure 14 TSS box plot by phase with seasons combined. Figure 15 TSS box plot by phase and season.
55 Fi gure 16 TN box plot by phase with seasons combined. Figure 17 TN boxplot by phase and season.
56 Figure 18 TP box plot by phase. Figure 19 TP box plot by phase and season.
57 Pre TSS means ranged from 1.17 mg/L at the Downstream 2 site and 5.1 1 mg/L at the Downstream 3 when data from both season were combined. Post TSS mean range increased to 1.26 mg/L at Downstream 2 and 5.75 mg/L at Downstream 1. Pre TN ranged from 0.66 mg/L at Downstream 2 to 1.10 mg/L at the Upstream site. Post TN mean r ange decreased to 0.52 mg/L at Downstream 2 to 0.99 mg/L at Downstream 1. Pre TP means ranged from 0.16 mg/L at the Downstream 1 site to 0.31 mg/L at Downstream 2. Post TP mean ranges decreased to 0.13 mg/L at the Upstream site and 0.27 mg/L at Downstrea m 2. All wet season means were higher than dry season means except Downstream 2 pre and post TSS and post Upstream TP. Discharge and Precipitation A nnual average discharge at the USGS gaging station located at the Downstream 3 site were found for the peri od of study (Figure 20 ) For thi s purpose the period of study was considered to be the entire years from 2002 to 2007, even though December 2001 water quality data is included in pre phase and the final three months of 2007 are not included the post phase water quality data. Monthly total precipitation data from (SCADA) site was used to find yearly totals (Figure 21 ). The site is located approximately 3.2 km northwest of the Upstr eam site in the Bullfrog Creek watershed. available online (SWFWMD, 2009).
58 Figure 2 0 Annual average discharge and total precipitation. As expected, discharge an d precipitati on appear to be directly related Average discharge and total precipitation were greater during the pre (2002 2004) than the post period (2005 2007) The average discharge for the pre phase was approximately 1,700 L/sec and decreased to 885 L/sec during the post phase. Average precipitation for the pre phase was 167 cm and decreased to 122 cm during the post phase. Annual average precipitation for the area based on historical data is 138 cm. (SRCC, 2007). Precipitation not only a ffects discharge, but i s also important when examining water quality data, particularly for pollutants whose primary source is from non point source pollution. Ambient Water Quality Impacts to Bullfrog Creek Pre Post Comparisons Mann Whitney comparisons of the pre and post ph ase conditions at each sample site on Bullfrog Creek were m ade for each parameter with both the seasons combined and the data split into wet and dry seasons. Statistically significant changes ( p < 0. 0 5) in pollutant concentration were found for several si tes during comparisons of the pre and p ost phases (Table 7 ).
59 Median Concentration (mg/L) Parameter and Season N Pre/Post Pre Post z Significance ( p ) Upstream Site TSS Combined 36/32 2.65 1.99 1.364 .173 TSS Wet 12/12 4.09 2.40 0.246 .356 TSS Dry 24/20 2.50 1.64 1.320 .187 TN Combined 36/33 0.97 0.76 1.562 .118 TN Wet 12/12 1.26 0.84 0.431 .024 /decrease TN Dry 24/21 0.91 0.70 0.466 .641 TP Combined 36/33 0.121 0.099 1.808 .071 TP Wet 12/12 0.180 0.181 0.677 .908 TP Dry 24/21 0.106 0.078 2.663 .008 /decrease Downstream 1 Site TSS Combined 35/33 2.71 4.26 1.319 .187 TSS Wet 11/12 4.90 4.52 .246 .805 TSS Dry 24/21 2.20 4.26 1.433 .152 TN Combined 35/33 0.69 0.88 1.184 .236 TN Wet 11/12 1.26 0.84 .431 .667 TN Dry 24/21 0. 57 0.81 1.718 .086 TP Combined 35/33 0.134 0.122 0.558 .577 TP Wet 11/12 0.169 0.160 .677 .498 TP Dry 24/21 0.112 0.102 1.126 .260 Downstream 2 Site TSS Combined 26/33 0.70 0.54 1.342 .180 TSS Wet 8/12 0.80 0.75 .155 .877 TSS Dry 18/21 0.68 0.50 1.88 0.60 TN Combined 26 /33 0.57 0.50 2.512 .01 2 /decrease TN Wet 8/12 0.72 0.63 1.466 .143 TN Dry 18/21 0.54 0.44 2.832 .005 /decrease TP Combined 26/33 0.298 0.253 .939 .348 TP Wet 8/12 0.416 0.376 1.929 .054 TP Dry 18/21 0.204 0.203 .66 2 .508 Downstream 3 Site TSS Combined 34/33 3.70 2.42 1.467 .142 TSS Wet 11/12 4.60 5.42 0.369 .712 TSS Dry 23/21 2.88 1.65 2.197 0 28 /decrease TN Combined 34/33 0.83 0.61 2.163 .0 31 /decrease TN Wet 11/12 0.90 0.82 0.492 .622 TN Dry 23/21 0. 70 0.54 2.620 .009 /decrease TP Combined 34/33 0.248 0.200 1.913 .056 TP Wet 11/12 0.388 0.313 1.477 .140 TP Dry 23/21 0.225 0.157 2.585 .010 /decrease Table 7. Mann Whitney results using PASW Pre and Post comparisons for each sample site on Bullf rog Creek.
60 Reductions were found at the Upstream site for wet season TN and dry season TP. No statistically significant changes were found at the Downstream 1 site. The Downstream 2 site showed a decrease in overall TN alo ng with reduced TN concentrations during the dry season. The Downstream 3 site showed reductions during the dry season in TSS. TN reductions were seen when the seasons were combined with a significant reduction during the dry season being the major contr ibutor. TP demonstrated reduced concentrations during the dry season. Although some statistically significant reductions were found for TN, TSS, and TP at the Downstream 2 and 3 sites, it is difficult to attribute the reductions to Balm Road Treatment Mar sh with confidence for several reasons. First, although reductions were found at some of the downstream sites, significant reductions were also found at the Upstream site for TN and TP. The upstream site acts as a control site, receiving no influence fro m the treatment system. Reductions found at this site, without treatment system impacts, lend to the possibility that factors other than the treatment system may have impacted reductions at the downstream sites as well. Also, the reductions in input to th e treatment system suggest that results may be due in part to decreased inputs and not treatment of TN and TP. Second, no reductions were found at the Downstream 1 site. Significant reductions at this site would have prov ided evide nce for positive impac ts to ambient water quality The site was located only a few hundred meters downstream from the treatment system, and the contributing drainage basin is only slightly larger than that of the wetland and the Upstream site. Not finding reductions at this s ite gives rise to the
61 possibility that reductions found at the other downstream sites were due to factors in their contributing basins independent of treatment system impacts. D ownstream 2 and 3 sites were located approximately 9 and 12.5 km downstream f rom the treatment system. There are numerous other factors that may affect water quality this far downstream. The are a of the contributing watershed is much greater at these points along the creek and any changes within the watershed could affect wat er q uality downstream. The drainage basin for the Downstream 3 site is 75.4 km 2 compared to the Upstream site basin which is only 7.4 km 2 or approximately 10% of that for the Downstream 3 site (USGS, 2009). An examination of the drainage basin was made usi ng the FDEP Map Direct Conso lidated Application available online (FDEP, 2009a). A pproximately sixty five National Pollutant D ischarge Elimination System (NPDES) stormwater permits were issued within the contributing watershed to the Downstream 3 site. Al l sixty five of these permits were for construction sites greater than 0.4 ha. O nly one of the sixty five site s was The search revealed only two permitted wastewater discharges in the Downstream 3 basin and both were well downstream from the treatment system. Land use in the larger Downstream 3 basin is similar to that of the treatment system basin, mainly agriculture with only slight increases in residential and suburban areas. However, the small differences in land use correlate to the numerous NPDES stormwater permits for construction in the Downstream 3 watershed that could lead to increased pollutant inputs at the d ownstream sites. These differences make it difficult to correlate results from these sites to impacts from the treatment system. If the construction activities or wastewater discharges produce greate r pollutant outputs during the p re treatment system
62 phase, reduced pollutant concentration s would be observed during the p ost phase independent fro m treatment system impacts. These observations were based on visual examination of the permitted facility locations on a map, along with contours, flow lines and land use and not geospatial analysis, so numbers are approximate. Third, reduction in pollut ant concentration at the downstream sites may have b e en due to changes in precipitation rather than the treatment impacts of Balm Road Treatment Marsh. Figure 20 in the previous section showed that both annual average discharge and annual precipitation w ere less during the p o st than during the p re phase As discussed earlier, the primary source of TSS, TP, and TN pollution in Bullfrog Creek is from agricultural nonpoint source pollution. The pollutants are picked up from the surrounding landscape by run off and washed into the cr eek. Less precipitation in the p ost time frame could be the cause of lower pollutant concentrations due to less storm events to carry pollutants to the creek. There is no strong evidence from the Mann Whitney results that water quality in Bullfrog Creek was positively impacted by Balm Road Treatment Marsh. However, no t finding significant impacts to ambient water quality at this site does not necessarily imply the treatment system is unsuccessful in treating the pollutan ts The treatment system was designed to capture flow s resulting from storms while leaving a fairly stable baseline flow through the cree k. The wetland receives the first flush after a storm that would be expected to be high i n se diments and nutrients and also c aptures the le ss pollutant concentrated waters that may be experienced during longer rain events. This water would dilute pollutant concentrations in the creek and since the wetland receives this water rather than the creek, pollutants may be more concent rated at the Downstream 1 site after
63 some storms when compared to pre treatment system conditions. The wetland significantly altered the hydrology in the upper portions of Bullfrog Creek. The hydrological impact may have masked the pollutant reductions. Pollutant l oads, however, may still be reduced, but this does not aid in determining ambient water quality impacts of the treatment system to Bullfrog Creek. In addition, ambient water quality monitoring was not designed to establish the performance of t he treatment wetland. Stormwater monitoring at the inflow and the outflow would have more accurately determined the pollutant reduction of the treatment system. Although it may be reasonable to expect pollutant reductions year round because water from th e creek is always flowing through the system, the system was designed primarily to reduce pollutant loads from agricultural runoff. Therefore, the only way to accurately measure the effectiveness of the system would be through stormwater monitoring, rathe r than ambient water quality monitoring. Upstrea m Downstream Comparisons Wilcoxo n matched pairs signed rank tests were performed to compare ambient water quality bet ween the Upstream site and downstream sites on the creek during the p ost phase Recall that the Upstream site is located near the headwaters of the creek and upstream of the treatment system, the Downstream 1 site is just downstream from the system and the other two sites are further downstream. Both overall and wet and dry season comparis ons were made. Statistically significant differenc es ( p < 0 .05) between the Upstream and other sites were found for the majo rity of the comparisons (Table 8 ).
64 Median Concentration (mg/L) Parameter and Season N Upstream /Test Site Upstream Test Site z Significance ( p ) Downstream 1 Site TSS Combined 32/33 1.99 4.26 2.599 .009 /increase TSS Wet 12/12 2.40 4.52 1.334 .182 TSS Dry 20/21 1.64 4.26 2.203 .028/increase TN Combined 33/33 0.76 0.88 1.832 .067 TN Wet 12/12 0.84 0.84 1.961 .050 TN Dry 21/ 2 1 0.70 0.81 0.852 .394 TP Combined 33/33 0.099 0.122 2.251 .0 24 /increase TP Wet 12/12 0.181 0.160 .471 .638 TP Dry 21/21 0.078 0.102 3.215 .001/increase Downstream 2 Site TSS Combined 32 /33 1.99 0.54 3.889 .000/decrease TSS Wet 12/ 12 2.40 0.75 3.059 .002/decrease TSS Dry 20/21 1.64 0.50 2.576 .010 /decrease TN Combined 33/33 0.76 0.50 3.940 .000/decrease TN Wet 12/12 0.84 0.63 1.883 .060/decrease TN Dry 21/21 0.70 0.44 3.493 .000 /decrease TP Combined 33/33 0.099 0.253 4.7 08 .000/increase TP Wet 12/12 0.181 0.376 2.746 .006/increase TP Dry 21/21 0.078 0.203 4.015 .000 /increase Downstream 3 Site TSS Combined 32/33 1.99 2.42 1.047 .295 TSS Wet 12/12 2.40 5.42 1.490 .136 TSS Dry 20/21 1.64 1.65 0.000 1.000 TN Comb ined 33/33 0.76 0.61 3.788 .000 /decrease TN Wet 12/12 0.84 0.82 1.334 .182 TN Dry 2 1/ 2 1 0.70 0.54 3.667 .000 /decrease TP Combined 33/33 0.099 0.200 4.530 .000 /increase TP Wet 12/12 0.181 0.313 2.353 .019 /increase TP Dry 21/21 0.078 0.157 3.980 000 /increase Table 8. Wilcoxon matched pairs signed rank test results using PASW Post construction comparisons between the Inflow site and other sites on Bullfrog Creek downstream from the treatment system. Note: tests. Pollutant concentrations were found to be greater at the Downstream 1 site for overall TSS and TP as well as dry season TSS and TP. There were differences for all test parameters between the Upstream and Downstream 2 site. TSS and TN for both overall
65 and individual seasons were less, and TP for both overall and individual seasons were greater at the Downstream 2 site. Combined season TN and dry season TN were less at the Downstream 3 site than the Upstream site. Both combined and individual season TP increased at the Downstream 3 site. The increased pollutant concentrations at the Downstream 1 site provide s evidenc e that the treatment wetland had a negative impact on ambient water quality in Bullfrog Creek. There are two possi ble explanations: either the treatment wetland was exporting TSS and TP or something else caused the higher concentrations at the Downstream 1 site. It is high ly possible that the wetland exported TP. This occurrence has been extensively noted in the li terature Soil properties prior to construction of a wetland can influence phosphorus removal. The Balm Road Marsh property was previously used for agriculture, so it is likely that the soils were high in phosphorus. High phosphorus content in soils can (Liikanen et al, 2004). When phosphorus concentrations are low in the wetland inflow, sediments may release phosphorus back into the water column (Novak et al 2004). Phosphorus retention is dependent on water depth and according to staff gauge measurements in the wetland, Balm Road Marsh water levels remained higher than anticipated. Phosphorus retention decrease s as water depth increases, so higher than anticipated water levels could have diminished ph osphorus removal efficiency (Moustafa, 1999). Additionally, research has shown that long term phosphorus retention may not occur in wetland systems and storage may be only temporary (Richardson, 1984 ; Kadlec & Knight, 1996 ).
66 T he wetland may also be expor t ing suspended solids Particulates are removed from the water column by sedimentation that occurs as the water slows down in the sedimentation basin and three wetland cells. Shorter residence times may affect the amount of particu lates that settle. Flo wing waters pick up sedi ments by erosion and resuspension of bottom sediments, which are high energy actions that are unlikely to occu r in the slow moving waters of the treatment wetland. However, wind and wave action have been shown to cause resuspension of sediments in shall ow lakes and could have similar affects in wetlands (Kadlec &Wallace, 2009). The presence of emergent vegetation reduces resuspension by wind and waves (Horpilla & Nurminen, 2001). However, t he establishment of vegetation in the Bal m Marsh Treatment System had early setbacks and replanting was necessary to overcome the effects of nutria and exotic apple snails. The treatment cells still contain large open water areas, as seen in the photographs in Appendix A Wind and wave action i n the wetland could have been a factor in increased TSS concentrations at the Downstream 1 site. Additionally, TSS measurements do not include only sediments Other particulates including suspended algae and other organic material are included in TSS me asurement s If phytoplankton is being exported from the system, it will appear in TSS results downstream (Mays, 2001). The final c ell in the wetland has a large area of open water with vegetat ion only around the perimeter This configuration is suscepti ble to high algae production which may have influenced TSS measurements at the Downstream 1 site (Kadlec & Wallace, 2009) Examining available aerials from the SWFWMD General Map Viewer, the FDEP Map Direct, and Google Earth revealed several algae blooms over the years in various cells. In 2005 there was an algae bloom throughout the
67 entire wetland seen mainly along the shorelines (Appendix A, Figure A 2) 2007 aerials revealed what appear ed to be algae mats in cell one 2008 aerials showed a large bloom in the sedimentation basin, and 2009 aerials showed a bloom in the sedimentation basin and cells one, two a nd possibly three These observations support the possibility that algae may have contributed to TSS downstream from the wetland. In addition, a s trong positive correlation between post phase Downstream 1 chlorophyll a and TSS is evidence that algae exports influence d TSS values (r s (32) = 0.70, p < 0.05). Finding n o reduction in TN is not unexpected based on inflow concentrations to the treatment s ystem Median TN values at the Upstream site were 0.9 mg/L, which is below the 50 th percentile for typical statewide stream concentrations (Table 4) There is strong evidence that wetlands either pass through or produce a background level of approximatel y 1 2 mg/L of organic nitrogen and up to 2.5 mg/L TN. Outflow concentrations will li kely be as high as 2.5 mg/L, therefore inputs of 0.9 mg /L would not be expected to be a ffected by treatment and may actually increase to background levels (Kadlec & Wallac e, 2009). TSS re ductions could be a ffected in a similar manner; the inflow concentrations are so low there is little roo m for improvement. During the p ost phase, median TSS was 1.99 mg/L which is in the 10 th percentile for streams in the state. This low concentration of suspended sediments is difficult to improve upon This condition does not, however, apply to TP. Median TP at the Upstream site during the post phase was 0.11 mg/L which is above the 50 th percentile in statewide comparisons (FDEP, 2000). High removal efficiencies have been found at lower inflow concentrations for other constructed wetlands treating agricultural runoff for example 80% reduction at 0.075 mg/L inflow,
68 76% reduction at 0.075 mg/L, and 58% reduction at 0.067 mg/L (Tanner et al 2003 and SWFWMD unpublished data as cited in Kaldec & Wallace, 2009). Background concentration in the southeast is approximately 0.01 mg/ L; th erefore there was large margin available for TP improvement by wetland treatment. As discussed in the previou s section, examination of ambient water quality data is not the preferred method of determining treatment efficiency. Low median inflow TSS and TN does not represent the entire range of conditions that occur in the stream. As previously stated, monthly s ampling occurred at the convenience of SWFWMD staff, without regard to precipitation patterns. It is likely that peak influx of sediments and nutrients, which would be expecte d to occur with storm events, were not captured in the data set. Non point source pollutants are typically found in highest concentrations during the first flush of a storm event after an extende d antecedent dry period. Storm water sampling would be a more appropriate choice to capture peak performance of a treatment system designed to treat pollution resulting from runoff. Flows enter the wetland year round regardless of prec ipitation, but the highest concentrations and therefore the best opportunity for large reductions, occur with storm events. Low inflow concentrations an d poor pe rformance during ambient monthly sampling event s do not indicate poor performance over t he entire range of conditions. L arge amounts of pollutants may have been retained from storm flows; however the monitoring scheme was not designed to capture performan ce under these conditions. Interestingly, increased TSS and TP at the Downstream 1 site only were only found during the dry season. No statistically significant changes were found during the wet season.
69 Another possibility is that the increased TSS and TP were not exported from the wetland. There is one small tributary to Bullfrog Creek in between the Upstream and Downstream 1 site. The tributary serves a small drainage basin of mainly agricultural land and some upland forest The water quality of the tribut ary is unknown, but based on the small contributing basin and similar land us e to the Upstream site basin the pollut ant loads would be expected to b e much smaller than those from the larger basin that contributes to the wetland. However, with no d ata to confirm this, the possibility remains that the tributary could contribute significantly to pollutant concentrations at the Downstream 1 site. There were both increases and reductions of pollutant concentrations at the Downstream 2 and 3 sites compar ed to the Upstream site. The decreased TSS and TN concentrations are not likely due to the treatment system because there were no reductions found immediately downstream from the system at the Downstream 1 site The reductions must have be en due instead to other factors. Two possibilities are the dilution by downstream tributa ries or attenuation through physical or chemical processes and assimilation as the pollutants travel downstream Increased TP at the Downstream 2 and 3 sites may be due in part to exports from the treatment system; however conc entrations are higher than at the Downstream 1 site, so phosphorus loading from either runoff from the surrounding watershed or the permitted point sources must be involved as well. Load ing Impacts to Tampa Bay Pollutant l oad reductions were found at the Downstre am 3 site as seen in Table 10 and Figure 21
70 Year TSS Load (kg/year) TN Load (kg/year) TP Load (kg/year) Wet Season 2002 29507 5607 2035 2003 81795 14234 6439 2004 340379 54711 21791 2002 2004 AVG (Pre) 150560 24851 10088 2005 87283 11183 4100 2006 110763 17411 5350 2007 11935 2722 1041 2005 2007 AVG (Post) 69994 10439 3497 Reduction (kg/year) 80566 14412 6591 % Reduction 54 58 65 Dry Season 2002 69898 11623 2776 2003 44472 9043 2962 2004 103564 14406 4604 2002 2004 AVG (Pre) 72645 11691 3447 2005 42556 7346 2565 2006 4763 1805 498 2007 26189 7263 1862 2005 2007 AVG (Post) 24502 5471 1642 Reduction (kg/year) 48142 6220 1806 % Reduction 66 53 52 Combined Seasons 2002 99404 1723 0 4811 2003 126266 23277 9401 2004 443943 69117 26395 2002 2004 AVG (Pre) 223205 36541 13536 2005 129839 18529 6665 2006 115525 19216 5848 2007 38124 9985 2902 2005 2007 AVG (Post) 94496 15910 5139 Reduction (kg/year) 128709 20631 8397 % Reduction 58 56 62 Table 1 0. Average annual load reductions at the Downstream 3 site.
71 Figure 21 Pre/Post pollutant load reductions at the Downstream 3 site. When comparing the pre and post average annual loads, reductions were 58% for TSS, 56% for TN and 62% for TP. This translates into reductions of approximately 129,000 kg TSS, 20,600 kg TN and 8,300 kg TP entering Tampa Bay each year. The calculated reduction for TSS is very near the estimated reduction based on modeling performed prior to constructio n of the wetland. The estimated reductions were 125 ,060 kg TSS, 8 ,700 kg TN, and 13,690 kg TP per year. However, loads were calculated based on a method to provide precision and not accuracy. The method chosen has been found to underestimate loads by as much as 80%, so the similarities between estimated and calculated load reductions may be misleading (Walling & Webb, 1981). In addition, estimated load reductions were based on modeled inflow and typical performance data and calculated load reductions we re based on loads pre and post treatment wetland at a site downstream from the treatment system. These differences make comparing
72 calculated and estimated load reductions problematic. The majority of the pollutant loading occurred during the wet season f or each parameter. This is typical for pollutants whose major source is runoff from the surrounding watershed. There were load reductions in the post phase both during the wet and dry season. Figures 2 2 and 23 demonstrate how discharge, precipitation, and pollutant concentration reductions contributed to load reductions. Monthly total precipitation data rby Romp 49 Balm Park SCADA site was used to find wet season and dry season annual averages. Figure 2 2 Pre/Post mean discharge and annual average precipitation Discharge is from the USGS gauging station at the Downstream 3 site and precipitation is from the SWFWMD site at nearby Balm Park.
73 Figure 23 Pre/Post mean pollutant concentration at the Downstream 3 site. Discha rge at the Downstream 3 site decreased during the post phase. This is likely due to a decrease in precipitation during the post phase as depicted in Figure 23. Precipitation is the annual average amount for the three years during the post and pre phase, while discharge is the mean instantaneous flow at the time each sample was taken, so it is not expected that the differences in pre and post discharge and precipitation would be proportional. Both reductions in pollutant concentration and discharge during the post phase contribute d to load reduction as shown in Figures 22 and 23. As shown in both Figure 23 and the Mann Whitney results in Table 7, wet season concentrations for all three constituents remained largely unchanged, therefore wet season load red uctions can be attributed mainly to decreased flow through the Downstream 3 site during the post p hase, and not treatment system a ffects. However, dr y season pollutant reductions did contribute to overall load reductions for all three parameters.
74 Figure 2 4 is provided to compare mean pollutant concentration reductions at the Downstream 3 site to reductions at the Upstream site. Figure 24 Pre/Post mean pollutan t concentration at the Upstream site. From the graphs in Figures 23 and 24 it appears tha t there were pollutant reductions at both the Inflow and Downstream 3 site when comparing pre and post phases, indicating that Downstream reductions may have be en at least partially due to reductions in pollutant inputs to the treatment system, rather than affects of the treatment system. However, there were reductions during the post phase for more parameters at the Downstream 3 site than the Upstream site. The Mann Whitney test results in Table 7 indicate d that only TN during the dry season and TP durin g the wet season had statistically significant reductions at the Upstream site when comparing pre and post phases, whereas TSS during the dry season, both overall and dry season TN, and dry
75 season TP demonstrated significant reductions at the Downstream 3 site. Th erefore, input reductions did not contribute to all reductions at the Downstream site. The results do not implicitly indicate that the treatment wetland was responsible for the load reductions found. Decreased discharge during the post phase rat her than a reduction in pollutant concentration may be the primary cause In ad dition, decreased concentration during the dry season at the Downstream 3 site which contributed to decreased loads was not necessarily due to treatment by the wetland as dis cussed in the previous sections. Comparison to other Wetlands Kadlec and Wallace (2009) compiled wetland treatment performance data for constructed systems designed to treat agricultural runoff. They found pollutant reductions by comparing mean pollutant concentration entering the wetland with mean pollutant co ncentr ation leaving the wetland following storm events The m ean pollutant reductions from the compiled data were 52 % TSS, 30 % TN and 22 % TP TSS reductions were present for all fourteen wetlands examined while two out of nineteen experienced TN increases and four out of twent y four had TP increases The largest TN increase reported was 11 % an d the largest TP increase reported was 76 %. Assuming the increase in pollutant concentrations were du e to wetland impacts and not the small tributary or other factors, the performance of Balm Road Treatment Marsh can roughly be compared to other treatment wetlands found in the literature. TSS and TP medians at the Downstream 1 site were found to be highe r than at the Upstream site during the post phase according to Wicoxon tests (Table 8). Although not statistically significant, both median and mean TN were also higher at the Downstream 1
76 site than the Upstream site. D irect comparison s to the reductions found by Kadlec and Wallace are difficult because the Downstream 1 site is composed of both wetland discharge and base flow through Bullfrog Creek and sampling was conducted without regard to storm events but are provided to give a general idea of where Balm Road Treatment Marsh falls among other wetlands. Balm Road Treatment Marsh reductions are in Table 10 Parameter Mean Upstream (mg/L) Mean Downstream 1 (mg/L) Reduction (%) TSS 2.94 5.75 96 TN 0. 86 0.98 14 TP 0. 133 0.154 16 Table 10 P ollutant reductions for Balm Road Treatment Marsh. Negativ e reductions represent increases *TN changes were found to be statistically insignificant using the Wilcoxon matched pairs test. All reductions in Table 10 are negative indicating that there w ere actually increases for each parameter. TP increase is well within range of that reported by Kadlec and Wallace and TN is nearly within range (2009). TSS concentration almost doubles and there were no reported increases for other wetlands. Reductions from wetlands designed to treat agricultural nonpoint source pollution can be compared to reductions reported for wetlands tre ating other source water to aid in determining whether constructed wetlands are a good option for treating agricultural runoff. W hen Balm Road Treatment Marsh reductions are combined with those reported by Kadlec and Wallace, mean reductions are 42 % TSS, 28 % TN, and 20 % TP with ranges from 96 to 97% TSS, 14 to 67% TN, and 76 to 60% TP (2009) However, since
77 Balm Road Treatment M arsh reductions were not based on stormwater sampling, they were not included in the comparisons in Table 11 Mean % Reduction in Constituent Concentration Ag Runoff Urban Runoff Wastewater TSS 52 64 72 TN 30 35 53 TP 22 44 56 Table 11 Wetland performance by source water using values from the literature. Urban stormwater treatment by constructed wetlands was reported as having mean reductions of 64% TSS, 35% TN and 44% TP (Kadlec & Wallace, 2009). Wastewater treatment by both constructed and n atural wetlands, including municipal and industrial wastes, has been reported as having average concentration reductions of 72% TSS, 53% TN, and 56% TP (Kadlec & Knight, 1996). Agricultural runoff treatment by constructed wetlands appears to be less eff ective t han treatment of pollutants in urban stormwater or wastewater Nonpoint source pol lution, whether from agricultural or urban sources, is not treated as effectively as wastewater. The decreased reductions for nonpoint source pollution are likely d ue to the fact that the amount of water and pollutant concentration entering the system is highly variable over time due to the dependence on precipitation. Municipal and industrial wastewater typically has a fairly constant flow rate and pollutant concen tration The reasons for differences between agricultural and urban runoff treatment efficiency are unknown. The research on agricultural runoff treatment by constructed wetlands is limited and the reductions wer e calculated based on the performance of o nly 12 to 24 wetlands varying based on parameter Performance of
78 urban runoff treatment was based on only 19 wetlands, while wastewater treatment performance was based on 48 to 71 wetlands. More research is needed to more accurately characterize the per formance of wetlands treating nonpoint source pollution and determine factors affecting performance.
79 Chapter 8 Conclusions Summary The goal of this research was to determine the water quality impacts of Balm Road Treatment Marsh in order to gain b etter understanding of the performance of constructed treatment wetlands for agricultural pollution management In order to a ccomplish the research goal, three questions were posed: What were the resulting water quality impacts of Balm Road Treatment Mar sh to ambient conditions in Bullfrog Creek? Was there a subsequent pollutant load reduction to Tampa B ay? How does the performance of constructed wetlands used to treat agricultural pollution compare to wetlands used to treat other pollution? It was pro posed that a nswer ing these questions would help determine whether or not constructed treatment wetlands are appropriate for agricul tural pollution management, which would assist water resource managers in design ing effective pollution reduction strategies for agricultural nonpoint source pollution. Beneficial ambient w ater quality impacts of Balm Road Treatment Marsh to Bullfrog Cr eek appear to be minimal if any. No significant changes in pollutant concentration could be found immediately downstream from the treatment wetland when comparing pre and post treatment wetland pollutant concentrations When comparing data fr om upstream and downstream of the treatment wetland, some of the pollutants
80 were actually more concentr ated downstream Pollutant reducti ons were found at sites several kilometers downstream from the treatment system, however due to the distance from the treatment system and large increase in contributing drainage basin to these sample sites, there are too many uncertainties to attribute th e reductions to the treatment system with any confidence. Results show large reductions in loads to Tampa Bay, but again there is not enough evidence to attribute the reductions to Balm Road Treatment Marsh. The load reductions may be due in part to decr eased pollutant inputs at the headwaters of the creek and therefore fewer pollutants entering the treatment system. Load reductions were a function of both decreased pollutant concentrations and discharge but only discharge impacted wet season load reduc tions. Decreased concentrations may have been due to factors in the contributing basin rather than treatment system affects. Pollutant reduction percentages of Balm Road Treatment Marsh were all negative, indicating there was actually an increase in poll utant concentrations downstream from the system. When comparing reported treatment wetland pollutant reductions for agricultural runoff to those of urban runoff and wastewater agricu ltural runoff treatment is l e ss effective than treatment of other pollut ion sources. The decreased treatment efficiency, along with the increased possibility of pollutant exports could lead to the conclusion that constructed wetlands may not be the best optio n for treating agricultural non point source pollution. However, av ailable data for removal efficiencies of agricultural runoff treatment are limited and more research should be conducted before drawing conclusions. In addition, Balm Road Treatment Marsh data were not optimal for making the comparisons, since outflow dat a was partially composed of flows that were not treated by
81 the wetland, and only ambien t data, rather than stormwater data were available. In addition, more research needs to be done to determine why pollutant removal is le ss efficient and whether new tec hnology or improved design can improve treatment. Data Limitations and Future N eeds Unfortunately, the sample design was not optimal for determining the efficiency of the treatment system and additional sampling needs to be performed to successfully ans wer the research questions. Some of the proposed changed to the sampling design can be accomplished in future studies; however some of the elements recommended should have been included in the original design prior to constr uction of the treatment system. Although water quality at the inflow to the treatment system was known due the close proximity of the Upstream sample site, the treatment system discharge water quality was unknown. An additional sampling site at the outflow of the wetland, prior to merg ing with Bullfrog Creek would have provided important information. In addition to being able to determine overall impacts to Bullfrog Creek through evaluation of the Downstream 1 site data, the treatment system impact on water flowing through the wetland could have been determined. In addition flow rate data at the treatment system inflow and outfall would have allowed for calculation of pollutant mass reduction. This could have been expressed as a percentage w hich would have allowed for additional com parison to values found in the literature for a variety of treatment system types and pollutant sources. This information would aid in determining the effectiveness of wetland treatment of agricultural nonpoint sourc e pollution compared to other sources. A future study can be designed incorporating this site and flow rate information.
82 Water quality data from the small tributary to Bullfrog Creek located between the Upstream and Downstream 1 site would have provided essential information regarding the wate r quality impacts to Bullfrog Creek. In order to make conclusions about the impacts of the treatment system to Bullfrog Creek, it must be assumed that the influence of the tributary was minimal. More data is needed to support or contradict this assumptio n Support of the assumption would have increased confidence in the conclusion that the treatment system did not positively impact ambient water quality in Bullfrog Creek immediately downstream If data contradicted the assumption, overall impacts to the creek still could not have been determined with confidence. A future study could be designed incorporating data from the tributary. However, the main goal of the treatment system project was to decrease loading to Tampa Bay, which could have been accomp lished even though significant improvements to Bullfrog Creek were not found. Since the treatment system was designed to achieve the greatest pollutant reductions following storm events, stormwater sampling would have added valuable information. Automati c sampling devices installed at the Upstream site and Down stream 1 sites both during the pre and p ost phases would have allowed for a more complete analysis of impacts to the creek and loading to Tampa Bay The samplers could be programmed to begin sampl ing after a specified amount of precipitation was detected. A preprogrammed volume of water per unit time would be collected, and if flow is measured as well, a flow weighted composite sample would be analyzed for the parameters of interest. These data c ould have been compared both using the Mann Whitne y test to compare Downstream 1 pre and p ost data and the Wilcoxon matched pairs test to compare Upstream and Downstrea m 1 data during the p ost phase. I t is
83 expected that these analyse s would have detected the greatest pollutant reductions. Future studies could incorporate stormwater sampling int o the design; however Wilcoxon pre p ost comparisons would not reflect the updated sample design. Recommendations for Project Managers Although no absolute conclusi ons were made as to the effectiveness of constructed wetlands for agricultural nonpoint source pollution treatment, the research still provides important lessons for nonpoint source pollution managers regarding impact and pollution reduction studies. Care ful planning and sample design is important prior to spending large sums of money and several years collecting data to determine project impacts. In this study, monthly water quality samples were collected and analyzed for a period of nine and a half year s. Although the first few years prior to construction of the treatment system were necessary to collect baseline information to aid in design, the designs were complete mid 2002, and the remaining five years were to monitor changes in ambient water qualit y. With the corre ct sample design, the project time period a nd number of samples could very likely have b een reduced to counter the additional costs associated with additional sample sites and equipment needed to collect stormwater samples and measure flo w. Monthly grab sampling is typical for ambient water quality monitoring, but perhaps not the best choice for impact studies. One aspect of this study that does not often occur in effectiveness studies is the overall ambient water quality impact to recei ving water bodies. Typically the treatment system is studied as separate and complete system and overall impacts to in stream water quality (or other affected water bodies) are overlooked. I f reducing downstream pollutant concentration is a project goal, for example to meet water quality standards ambient
84 water quality im pacts should be studied in addition to pollutant load reduction s If major improvements only occur following storm events or if the treatment system alters flows so that pollutant conce ntrations are affected, improvements may not necessarily be detected in ambient water quality data. Monitoring impacts to downstream conditions is an important aspect of effectiveness studies that is often overlooked. This research demonstrates that it i s an important component to aid in pollution management decisions. Al though this research did not come to definitive conclusions regarding the water quality i mpacts in Bullfrog Creek by the treatment system, it does appear that ambient water quality was not positively impacted This demonstrates the importance of selecting treatment options. It has been shown that constructed wetlands do not always perform as expected, and pollutants, especially phosphorus, have been shown to be exported under some con ditions. Often, stormwater treatment projects are constructed with no subsequent effectiveness studies. This research demo nstrates the importance of such studies in order to fill the existing data gap, especially in treating agricultural and other nonpoi nt source pollution. The information will help managers select appropriate treatment options to successfully reduce pollution and limit the misuse use of resources. This research demonstrates that constructed wetland systems to treat agricultural nonpoin t source pollution may not be as effective as wetlands designed to treat other sou rces of pollution Additionally, pollutant exports from these systems are possible. Although more research is needed, managers may choo se to other options for reducing agri cultural nonpoint source pollution until more research becomes available For example using BMPs on individual farms to reduce the amount of pollutants reaching streams may be a better option than treatment with in the watershed.
85 Balm Road Treatment Mars h was not found to positively impact ambient water quality in Bullfrog Creek and although there was a significant load reduction of nutrients and TSS to Tampa Bay, it could not be attributed to treatment by the wetland with confidence. However, the sampl e design was lacking, and more research is recommended before final conclusions as to the success of treatment and impacts to water quality are drawn. The proposed future research will produce results that can be effectively compared with pollutant remova l efficiencies of wetlands to treat other sources of pollution found in the literature. The comparisons will be useful in the determination of the appropriateness of using constructed wetland s to treat agricultural nonpoint source pollution. This researc h demonstrated the importance of monitoring the performance of pollution management projects, strategic s ample design, and including receiving water impact s in monitoring studies while adding to the limited existing information of the effectiveness of usin g constructed treatment wetlands to manage agricultural nonpoint source pollution.
86 List of References Adler, R.W, Landman, J.C., & Cameron, D.M. 1993. The Clean Water Act 20 Years Later Washington, DC: Island Press. Ayres Associates. 2000. Balm R oad Wetland/Stormwater Treatment Facility. Report for Hillsborough County. Bachard, P.A.M. & Horne, A.J. 2000. Denitrification in constructed free water surface wetlands: II. Effects of vegetation and temperature. Ecological Engineering 14, 17 32. Baker, L.A. 1992. Introduction to nonpoint source pollution in the United States and prospects for wetland use. Ecological Engineering 1, 1 26. Berryman, D, Bobee, B., Cluis, D, & Haemmerli, J. 1988. Nonparametric test for tren d detection in water q uality time series. Water Resources Bulletin 24(3), 545 556. Borin, M., Bonaiti, G.,Santamaria, G., & Giardini, L. 2001. A constructed surface flow wetland for treating agricultural waste. Water Science and Technology 44(11 12), 523 530. Brix, H. 1993. Wastewater Treatment in Constructed Wetlands: System Design, Removal Processes, and Treatment Performance. In Moshiri, G.A. (Ed.), Constructed Wetlands for Water Quality Improvement (pp. 9 22). Boca Raton, FL: CRC Press. Campbell, C.S., & Ogden M.H. 1999. Constructed wetlands in the sustainable landscape New York, NY: John Wiley & Sons, Inc. Carleton, J.N., Grizzard, T.J., Godrej, A.N. & Post, H.E. 2001. Factors affecting the performance of stormwater treatment wetlands. Water Research 3(6), 1552 1562. Carpenter, S.R., Caraco, N.F., Correll, D.L., Howarth, R.W, Sharpley, A.N., & Smith, V.H. 1998. Nonpoint pollution of surface waters with phosphorous and n itrogen. Ecological a pplications 8(3), 559 568.
87 Casey, R.E., & Klaine, S.J. 2001. Nutrient attenuation by a riparian wetland during natural and artificial runoff events. Journal of Environmental Quality, 30, 1720 1731. Clean Water Act, 33 U.S.C. Â§Â§1251 1387 (1972) Cohn, T.A., DeLong, L.L., Gilroy, E.J., Hirsch, R.M., & Wells, D .K. 1989. Estimating constituent loads. Water resources research 25(5), 937 942. Correll, David. 1998. The role of phosphorous in the eutrophication of r eceiving waters: a r evi ew. Journal of environmental q uality 27(2), 261 266. Dames & Moore. 2000. Hillsborough County Bullfrog Creek/Wolf Branch Watershed Management Plan. Hillsborough County, Florida. Dierberg, F.E., Debusk, T.A., Jackson, S.D., Chimney, M.J. & Pietro, K. Submerged aquatic vegetation based treatment wetlands for removing pho sphorus form agricultural runoff: Response to hydraulic and nutrient loading. Water Research 36, 1409 1422. Dolan, D.M.; Yui, A.K.; & Geist, R.D. 1981. Evaluation of River Load Estimation Methods for Total Phosporus. Association of great lakes r esea rch 7(3), 207 214. Felberova, L, Raunch, Ota, & Kvet, Jan. 2001. Wastewater treatment, with emphasis on nitrogen removal, in a constructed wetland planted with Phragmites Australis and Glyceria Maxima In J. Vymazal (Ed.), Transformations of Nutrients in Natural and Constructed Wetlands. (pp. 235 241). Leiden, The Netherlands: Backhuys Publishers. Ferguson, R.I. 1987. Accuracy and precision of methods for estimating river loads. Earth surface processes and landforms 12, 95 104. Florida Departm ent of Environmental Protection (FDEP). 2000. 2000 Florida Water Quality Assessment: 305(b) Report. Florida Department of Environmental Protection. Florida Department of Environmental Protection (FDEP). 2001. Tampa Bay Basin Status Report. Florida Department of Environmental Protection Division of Water Resource Management. Florida Department of Environmental Protection (FDEP). 2004. DEP SOP 001/01, FS 2100 Surface Water Sampling.
88 Florida Department of Environmental Protection (FDEP). 2008. Integrated Water Quality Assessment for Florida: 2008 305(b) Report and 303(d) List Update Florida Department of Environmental Protection Divisi on of Environmental Assessment a nd Restoration Bureau of Watershed Management Florida Department of Environm ental Protection (FDEP). 2009 a Map di rect c onso lidated a pplication Retrieved October 6, 2009 from http://www.dep.state.fl.us/gis/portal.asp Florida Department of Environmental Protection (FDEP ). 2009 b NELA P certified laboratories d atabase Available at http://www.dep.state.fl.us/labs/cgi bin /aams/org_results.asp?choice=1&lab_id_o=6002&lab_id_a=6214&pt_name= Drinking+Water&sort=3&B1=Submit Florida Department of Environmental Protection (FDEP). 2009c. STORET Public Access Website Available at http://storet.dep.state.fl.us/WrmSpa/ Florida Lakewatch. 2000. nutrients. Information Circular 102 Gainsville, FL: Department of Fisheries and Aquatic Sciences, Institute of Food and Agricultural Sc iences, University of Florida. Florida Watershed Restoration Act, Fla. Stat. Â§403.067 (1999). Gearheart, R.A. 1992. Use of constructed wetlands to treat domestic wastewater, City of Arcata, California. Water Science and Technology 26(7 8), 1625 1637. Gianessi, Leonard P., & Peskin, Henry M. 1981. Analysis of national water pollution control policies 2: Agricultural sediment control. Water resources r esearch 17(4), 803 821. Greenburg, A.E., Clesceri, L.S., & Eaton, A.D. (Eds.). 1992. Standard methods for the examination of water and waste water. 18 th edition. Washington, DC: American Public Health Association. Hammer, D.A. 1989a. Constructed wetlands for treatment of agricultural waste and urban stormwater. In: S.K. Majumdar et al. (Eds) Wetlands Ecology and Conservation: Emphasis in Pennsylvania (pp. 333 348). The Pennsylvania Academy of Science. Hammer, D.A. 1989b. Constructed wetlands for wastewater treatment: Municipal, industrial, and agricultural Chelsea, MI: Lewis Publis hers. Hammer, D.A. 1992. Designing constructed wetlands systems to treat agricultural nonpoint source pollution. Ecological Engineering 1, 49 82.
89 Hammer, D.A. 1997. Creating freshwater wetlands Boca Raton, FL: CRC Lewis Publishers. Hammer, D.A., & K night, R.L. 1994. Designing treatment wetlands for nitrogen removal. Water Science and Technology, 29(4), 15 17. Harcum, J.B., Loftis, J.C., & Ward, R.C. 1992. Selecting trend tests for water quality series with serial correlation and missing va lues. Water Resources Bulletin 28(3), 469 478. Helsel, D.R. 1987. Advantages of nonparametric procedures for analysis of water quality data. Hydrological Sciences 32(2), 179 1990. Higgens, M.J., Rock, C.A., Bouchard, R., & Wengrezynek, B. 1993. Co ntrolling agricultural run off by the use of constructed wetlands. In Moshiri, G.A. (Ed.), Constructed Wetlands for Water Quality Improvement (pp. 359 368). Boca Raton, FL: CRC Press. Hirsch, R.M., Slack, J.R., Smith, R.A. 1982. Techniques of trend a nalysis for monthly water quality data. Water Resources Research 18(1), 107 121. Hirsch, R.M., Alexander, R.B., & Smith, R.A. 1991. Selection of methods for the detection and estimation of trends in water quality. Water Resources Research 27(5), 803 813. Houck, O.A. 2002. The Clean Water Act TMDL Program: Law, policy, and implementation. Washington, D.C.: Environmental Law Institute. Hynes, H.B.N. 1970. The ecology of running waters. Toronto, Canada: University of Toronto Press. Isermann K. 1991. Share of agriculture in nitrogen and phosphorus emissions into the surface waters of Western Europe against the background of their eutrophication. Fertilizer R esearch 26, 253 269. Johnson, C.A. 1991. Sediment and nutrient retention by fre shwater wetlands: Effects on surface water quality. Critical Reviews in Environmental Control 21(5 6), 491 565. Kadlec, R.H. 1999. Chemical, physical and biological cycles in treatment wetlands. Water Science and Technology 40(3), 37 44. Kadlec, R.H. 2001. Phosphorus dynamics in event driven wetlands. In: J. Vymazal (Ed), Transformations of Nutrients in Natural and Constructed Wetlands. (pp. 365 391). Leiden, The Netherlands: Backhuys Publishers.
90 Kadlec, R.H. & Knight, R.L. 1996. Treatm ent Wetlands Boca Raton, Florida: CRC Press. Kadlec, R.H. & Wallace, S.C. 2009. Treatment Wetlands 2 nd Ed. Boca Raton, Florida: CRC Press. Koskiaho, J.; Ekholm, P.; Rty, M.; Riihimki, J.; Puustinen, M. 2003. Retaining agricultural nutrients in constructed wetlands experiences under boreal conditions. Ecological Engineering 20, 89 103. Kovacic, D.A., David, M.B., Gentry, L.E., Starks, K.M., & Cooke, R.A. Effectiveness of Constructed Wetlands in Reducing Nitrogen and Phosphorus Export fr om Agricultural Tile Drainage. Journal of Environmental Quality, 29(4), 1262 1274. Kronvag, B., & Bruhn, A.J. 1996. Choice of sampling strategy and estimation method for calculating nitrogen and phosphorus transport in small lowland streams. Hydrologi cal processes 10, 1483 1501. Kuehn, E. & Moore, A. 1995. Variability of treatment performance in constructed wetlands Water Science and Technology 32(3), 241 250. Lettenmaier, D.P. Milivariate nonparametric tests for trend in water quality. Wate r Resources Bulletin 24(3), 505 512. Liikanen, A., Puustinen, M., Koskiaho, J., Vaisanen, T., Martikainen, P., & Hartikainen, J. 2004. Phosphorus removal in a wetland constructed on former arable land. Journal of Environmental Quality 33, 1124 1132. Loftis, J.C., McBride, G.B., & Ellis, J.C. 1991. Considerations of scale in water quality monitoring and data analysis. Water Resources Bulletin 27(2), 255 264. Malik, A.S., Larson, B.A., & Ribaudo, M. 1994. Economic Incentives for Agricultual Nonp oint Source Pollution Control. Water Resources Bulletin 30(3), 471 480. Mander, U, & Mauring, T. 1997. Constructed wetlands for wastewater treatment in Estonia. Water Science and Technology 53(5), 323 330. Martin, J.R., Keller, C.H., Clarke, R.A., & Knight, R.L. 2001. Long term performance summary for the Boot Wetland treatment system. Water Science and Technology 44(11 12), 413 420. Mays, L.W. 2001. Stormwater collection system design handbook New York: McGraw Hill. Mitsch, W.A., & Gosse link, J.G. 2000. Wetlands New York: John Wiley & Sons, Inc.
91 Moustafa, M.Z. 1999. Analysis of phosphorus retention in free water surface treatment wetlands. Hydrobiologia 392, 41 53. Newcomb, C.P., & MacDonald, D.D. 1991. Effects of suspended s ediments on aquatic ecosystems. N orth American journal of f isheries m anagement 11, 72 82. Novak, J.M., Stone, K.C., Szogi, A.A., Watts, D.W., & Johnson, M.H. 2004. Dissolved phosphorus retention and release from a coastal plain in stream wetland. Jou rnal of Environmental Quality 33, 394 401. Parry, Roberta. 1998. Agricultural phosphorus and water q uality: A U.S. Environmental Protection A gency perspective Journal of environmental q uality 27(2), 258 261. Preston, S.D., Bierman, V.J., & Silliman, S.E. 1989. An evaluation of methods for the estimation of tributary loads. Water resources research 26(6). 1379 1389. Richards, P.R. 1996. Estimation of pollutant loads in rivers and streams: A guidance document for NPS programs. Tiffin, OH: U.S Environmental Protection Agency Region VII. Richards, R.P., & Holloway, J. 1989. Monte Carlo studies of sampling strategies for estimating tributary loads. Water resources research 23(10) 1939 1948. Richardson, C.J. 1985. Mechanisms controlling phosphorus retention capacity in freshwater wetlands. Science 228, 1424 1427. Ritter, William F., & Shirmohamma di, Adel. 2001. Agricultural n onp oint source p ollut ion: Watershed management and hydrology. Boca Raton, Florida: CRC Press Southwest F lorida Water Management District (SWFWMD). 2009. Water management information system. Retrieved June 30, 2009 from http://www8.swfwmd.state.fl.us/WMIS/ResourceData/ExtDefaul t.aspx Southeast Regional Climate Center (SRCC) 2007. Historical Climate Summaries for Florida. Parrish, Florida Station Number 086880 Retrieved November 9, 2007 from http://www. sercc.com/cgi bin/sercc/cliMAIN.pl?fl6880 Tampa Bay Estuary Program (TBEP). 2005. Estimates of total nitrogen, total phosphorus, total suspended solids, and biochemical oxygen demand loadings to Tampa Bay, Florida: 1999 2003, Tampa Bay Estuary Program Technical Publication # 02 05. St. Petersburg, FL. Tampa Bay Estuary Program (TBEP). 2006. Charting the course for Tampa Bay: the comprehensive conservation and management plan for Tampa Bay. St. Petersburg, FL.
92 Tanner, C.C.; Nguyen, M.L.; Sukias, J.S. 2005. Constructed wetland attenuation of nitrogen exported in subsurface drainage from irrigated and rain fed dairy pastures. Water Science & Technology 51(9), 55 61. United States Department of Agriculture (USDA). 1997. Water quality and agri culture: Status, conditions, and trends. Working Paper #16. United States Department of Agriculture, Natural Resources Conservation Service. United State Department of Agriculture (USDA). 2006. Soil map, Hillsborough County, Florida. Soil web survey 2.0, national cooperative soil survey USDA, Natural Resource Conservation Service. Retrieved November 9, 2007 from http://websoilsurvey.nrcs.usda.gov/app United States Environmental Protection Ag ency (EPA). 1983. Methods for chemical analysis of water and wastes, EPA 600/ 4 79 020. Cincinnati, OH: United States Environmental Protection Agency. United States Environmental Protection Ag ency (EPA). 1999. Protocol for developing n utrient TMDLs. EPA 841 B 99 007. Washington DC: United States Environmental Protection Agency, Office of Water. United States Environmental Protection Agency (EPA). 2009. National water quality inventory: report to Congress: 2004 reporting cycle. EPA 841 R 08 001. Washington, DC: United States Environmental Protection Agency, Office of Water. United States Environmental Protection Agency (EPA). 2003. National Management Measures for the Control of Nonpoint Pollution from Agriculture. EPA 841 B 03 004. Washingt on, DC: United States Environmental Protection Agency, Office of Water. United States Geological Society (USGS). 2009a. Instantaneous Data Archive. Retrieved September 20, 2009 from http://ida.water.usgs.g ov/ida/ United States Geological Society (USGS). 2009 b National Streamflow Information Program Retrieved September 30, 2009 from http://water.usgs.gov/nsip University of Florida, Institute of Food and Agricultural Sciences Extension (UF IFAS). 2005. Handbook of Florida water regulation. FE579. Gainesville, FL: University of Florida, Food and Resource Economics Department F lorida Cooperative Extension Service, Institute of Food and Agricultural Scie nces
93 University of Florida, Institute of Food and Agricultural Sciences Extension (UF IFAS). 2009. Plant management in Florida waters. University of Florida, Food and Resource Economics Department F lorida Cooperative Extension Service, Institute of Fo od and Agricultural Sciences Retrieved July 19, 2009 from http://aquat1.ifas.ufl.edu/guide/trophstate.html van Belle, G., & Hughes, J.P. 1984. Nonparametric tests for trend in water qu ality. Water Resources Research 20(1), 127 136. van der Valk, A.B., & Jolly, R.W. 1992. Recommendations for research to develop guidelines for the use of wetlands to control rural nonpoint source pollution. Ecological Engineering 1, 155 134. Wallin g, D.E., & Webb, B.W. 1981. The reliability of suspended sediment load data. Erosion and Sediment Transport Measurement. IAHS Publication no.133, 177 194. Yang, Y., Zhencheng, X., Kangping, H., Junsan, W., & Guizhi, W. 1995. Removal Efficiency of th e constructed wetland wastewater treatment system at Bainikeng, Shenzhen. Water Science and Technology 32(3), 31 40. Yirong, C. & Puetpaiboon, U. 2004. Performance of constructed wetland treating wastewater from seafood industry. Water Science and T echnology 49(5 6), 289 294. Zar, J.H. 1984. Biostatistical Analysis New Jersey: Prentice Hall, Inc.
95 Appendix A Pictures of Balm Road Treatment Marsh A 1. Balm Road Marsh Property aerial, 2004. A 2. Balm Road Marsh Property aerial, 2005.
96 Appendi x A (continued) A 3 Balm Road Treatment Marsh Sedimentation Basin 9/26/2009. A 4 Balm Road Treatment Marsh Cell #1 9/26/2009.
97 Appendix A (continued) A 5 Balm Road Treatment Marsh C ell # 2 9/29/2009. A 6 Balm Road Treatment Marsh Cell #3 9/26/2009.
98 Appendix A (continued) A 7 Balm Road Treatment Marsh Cell #4 9/26/2009. A 8 Upstream sampling site 9/26/2009. Maintenance crews had recently removed sediments and hyd rilla from the creek bed.
99 Appendix A (continued) A 9 Diversion structure on the left and canal to Balm Road Treatment Marsh on the right 9/26/2009. A 10 Diversion structure allowing base flow to Bullfrog Creek 9/26/2009. All additional flow s are directed through the canal on the left that flows to Balm Road Marsh.
100 Appendix A (continued) A 11 Treatment system outfall structure in cell #4 9/26/2009. A 12 Treatment system outfall 9/26/2009. Merges with Bullfrog Creek approximate ly 200 m downstream.
101 Appendix A (continued) A 13 Looking upstream on Bullfrog Creek from the Downstream 1 sample site 9/26/2009. Bullfrog Creek on the right merges with the treatment system outflow on the left.
102 Appendix B Histograms B 1. Ups tream TSS Histograms.
103 Appendix B (Continued) B 2. Downstream 1 TSS Histogram. B 3. Downstream 2 TSS Histogram.
104 Appendix B (continued) B 4. Downstream 3 TSS Histogram. B 5. Upstream TN Histogram.
105 Appendix B (continued) B 6. Downstream 1 TN Histogram. B 7. Downstream 2 TN Histogram
106 Appendix B (continued) B 8. Downstream 3 TN Histogram. B 9. Upstream TP Histogram.
107 Appendix B (continued) B 10. Downstream 1 TP Histogram. B 11. Downstream 2 TP Histogra m.
108 Appendix B (continued) B 12. Downstream 3 TP Histogram.