Water and Phosphorus Budgets of Depressional Wetlands in the Okeechobee Basin, FL

Permanent Link: http://ufdc.ufl.edu/UFE0021306/00001

Material Information

Title: Water and Phosphorus Budgets of Depressional Wetlands in the Okeechobee Basin, FL
Physical Description: 1 online resource (108 p.)
Language: english
Creator: Perkins, Daniel Brent
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007


Subjects / Keywords: hydraulic, hydrology, okeechobee, phosphorus, water, wetland
Soil and Water Science -- Dissertations, Academic -- UF
Genre: Soil and Water Science thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation


Abstract: The Lake Okeechobee, FL catchment basin and lake proper have been given considerable attention in water resources research and management arenas because of the potential shift in trophic status resulting from excess nutrient loads. Depressional wetlands are a signature landscape feature in the Lake Okeechobee basin, comprising approximately 12-16% of the watershed. These wetlands are often surrounded by pasture and grazing land and integrate hydrologic and chemical loads. These wetlands play an integral role in the retention of water and nutrients in the basin, as they are one of the primary mechanisms by which water and nutrients might be stored. This study stems from larger efforts to evaluate the suitability of wetlands at sites in the north Lake Okeechobee catchment to increase water retention, phosphorus (P) transformation, and travel time to the lake. This dissertation emphasizes three areas of ongoing research in: 1) groundwater-wetland hydrologic exchange mechanisms and quantities, 2) quantification of hydrologic pathways and system hydraulic residence times, and 3) phosphorus loading rates and dynamics. Results from this research suggest that groundwater-wetland exchange is important in considering the hydrologic impact of these wetlands within the framework of the overall P budget. Also, P loads from the ditch networks were often found to exceed all other outflows. This research provides a conceptual hydrologic and P model for depressional wetlands in the Lake Okeechobee basin. It also has implications related to hydrologic management of these wetlands for landscape-scale chemical treatment potential and nutrient loading to Lake Okeechobee.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Daniel Brent Perkins.
Thesis: Thesis (Ph.D.)--University of Florida, 2007.
Local: Adviser: Jawitz, James W.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021306:00001

Permanent Link: http://ufdc.ufl.edu/UFE0021306/00001

Material Information

Title: Water and Phosphorus Budgets of Depressional Wetlands in the Okeechobee Basin, FL
Physical Description: 1 online resource (108 p.)
Language: english
Creator: Perkins, Daniel Brent
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007


Subjects / Keywords: hydraulic, hydrology, okeechobee, phosphorus, water, wetland
Soil and Water Science -- Dissertations, Academic -- UF
Genre: Soil and Water Science thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation


Abstract: The Lake Okeechobee, FL catchment basin and lake proper have been given considerable attention in water resources research and management arenas because of the potential shift in trophic status resulting from excess nutrient loads. Depressional wetlands are a signature landscape feature in the Lake Okeechobee basin, comprising approximately 12-16% of the watershed. These wetlands are often surrounded by pasture and grazing land and integrate hydrologic and chemical loads. These wetlands play an integral role in the retention of water and nutrients in the basin, as they are one of the primary mechanisms by which water and nutrients might be stored. This study stems from larger efforts to evaluate the suitability of wetlands at sites in the north Lake Okeechobee catchment to increase water retention, phosphorus (P) transformation, and travel time to the lake. This dissertation emphasizes three areas of ongoing research in: 1) groundwater-wetland hydrologic exchange mechanisms and quantities, 2) quantification of hydrologic pathways and system hydraulic residence times, and 3) phosphorus loading rates and dynamics. Results from this research suggest that groundwater-wetland exchange is important in considering the hydrologic impact of these wetlands within the framework of the overall P budget. Also, P loads from the ditch networks were often found to exceed all other outflows. This research provides a conceptual hydrologic and P model for depressional wetlands in the Lake Okeechobee basin. It also has implications related to hydrologic management of these wetlands for landscape-scale chemical treatment potential and nutrient loading to Lake Okeechobee.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Daniel Brent Perkins.
Thesis: Thesis (Ph.D.)--University of Florida, 2007.
Local: Adviser: Jawitz, James W.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021306:00001

This item has the following downloads:

Full Text




2 2007 Daniel B. Perkins


3 To my wife and children Do or do not. There is no try. Yoda, Jedi Master


4 ACKNOWLEDGMENTS I thank a ll involved in this rese arch that made it possible. I thank my advisor, Dr. James Jawitz for his mentoring and commitment to scien ce as well as his friend ship. I thank the land owners who graciously accommodated our frequent field visits. I thank those involved in the field and laboratory work without who there woul d be considerably less data. I would like to thank the Florida Alumni Fellowship for funding my research and facilita ting my sojourn at the University of Florida. Also, I thank the Fl orida Department of Agriculture and Consumer Services, and South Florida Water Management Dist rict, who had the vision to fund this project. Lastly, Id like to thank my family, Dayna, Canyon, Micah, Meadow, and Bryton who have supported me through this season in life.


5 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........7 LIST OF FIGURES.........................................................................................................................8 ABSTRACT...................................................................................................................................10 1 INTRODUCTION..................................................................................................................12 Background.............................................................................................................................12 Hypotheses and Objectives..................................................................................................... 16 Chapter 2...................................................................................................................... ...16 Chapter 3...................................................................................................................... ...17 Chapter 4...................................................................................................................... ...18 Chapter 5...................................................................................................................... ...18 Site Description......................................................................................................................19 2 WETLAND-GROUNDWATER INTERACTI ONS IN MANAGE D SEASONALLYINUNDATED DEPRESSIONAL WETLANDS................................................................... 24 Introduction................................................................................................................... ..........24 Methods and Materials...........................................................................................................27 Water Budget-Based Groundwater Outflow................................................................... 28 Groundwater Outflow Model..........................................................................................29 Meteorological Data........................................................................................................ 30 Slug tests..........................................................................................................................31 Results.....................................................................................................................................31 Groundwater Outflow Rates............................................................................................ 31 Hydraulic Resistivity....................................................................................................... 32 Model Calibration/Validation..........................................................................................32 Discussion and Conclusions...................................................................................................33 Groundwater Outflow......................................................................................................33 Hydraulic Resistivity....................................................................................................... 34 Model Performance.........................................................................................................35 3 QUANTIFYING HYDROLOGIC P ATHWAYS IN DEPRESSIONAL WETLANDS USING A WATER B UDGET APPROACH......................................................................... 40 Introduction................................................................................................................... ..........40 Water Budget Estimation........................................................................................................ 43 Precipitation and Evapotranspiration..............................................................................45 Groundwater Outflow......................................................................................................45 Groundwater Inflow........................................................................................................47


6 Surface Water Outflow.................................................................................................... 47 Overland Flow.................................................................................................................48 Hydraulic Residence Time.....................................................................................................49 Hydroperiod............................................................................................................................50 Results and Discussion......................................................................................................... ..50 Hydroperiod.....................................................................................................................50 Water Budget................................................................................................................... 51 Upland Area Contribution...............................................................................................56 Hydraulic Residence Time.............................................................................................. 57 Conclusions.............................................................................................................................58 4 SPATIALLY DISTRIBUTED ISOLA TED WETLANDS AS A TREATMENT SYSTEM FOR AGRICULTURAL RU NOFF W ITHIN A WATERSHED.......................... 66 Introduction................................................................................................................... ..........66 Model......................................................................................................................................68 Results and Discussion......................................................................................................... ..72 Conclusions.............................................................................................................................74 5 PHOSPHORUS BUDGETS OF DEPRESSIONAL WETLANDS ....................................... 79 Introduction................................................................................................................... ..........79 Hydrologic Inflows/Outflows.......................................................................................... 82 Phosphorus Budget.......................................................................................................... 84 k-C* Modeling.................................................................................................................86 Results and Discussion......................................................................................................... ..88 Phosphorus Budget.......................................................................................................... 88 k-C* Model......................................................................................................................93 Conclusions.............................................................................................................................94 6 CONCLUSIONS.................................................................................................................... 99 Groundwater-Surface Water Exchange.................................................................................. 99 Water Budget................................................................................................................... .......99 Watershed-Scale Treatment..................................................................................................100 Phosphorus Budget.............................................................................................................. .100 LIST OF REFERENCES.............................................................................................................101 BIOGRAPHICAL SKETCH.......................................................................................................108


7 LIST OF TABLES Table page 1-1 Estimated average percent of isol ated wetlands by selected states.................................... 22 2-1 Wetland area, elevation difference .................................................................................... 36 2-2 Wetland hydroperiod (days).............................................................................................. 37 2-3 Monitoring periods of wetland water surface and upland water tab le elevations.............. 37 2-4 Summary of event-based regressi on between observed groundwater outflow. ................. 37 2-5 Best-fit hydraulic resistivity determined the to tal monitoring record of each wetlandwell pair.............................................................................................................................38 3-1 Geometric mean flow rate of each water budget com ponent and their associated geometric standard deviations............................................................................................ 60 3-2 Percentage of total number of days that each co mponent occurredt................................. 60 3-3 Percentage of the total volume of in flow or outflow of each water budget com ponent.... 60 5-1 Water quality measurements of total P in wetland surface water, ditch water, and groundwater.......................................................................................................................96 5-2 Percentage of total P contri bution from each chemical component.................................. 96 5-3 Input and optimized parameters in k-C* m odel and associated error................................ 96 5-4 Range of percent treatment of P inflow............................................................................. 97


8 LIST OF FIGURES Figure page 1-1 Site map of study wetlands and monito ring well locations. BW1 and BW2 wetlands are lo cated on the Pete Beaty ranch and LW1 and LW2 wetlands are located on the Dixie-Larson ranch. Elevation co ntours are 0.1-m intervals.............................................23 2-1 Example wetland surface water and upland water table elevation behavior at LW 1. Shaded areas indicate drawdown periods used to quantify wetland surface water and groundwater exchange....................................................................................................... 38 2-2 Typical drawdown event used to evaluate correlation between GW and hydraulic gradient between groundwater and wetland (d H ). For these data from BW2-MW2, GW and dH are highly correlated ( r = 0.98)...................................................................... 39 2-3 Cumulative water-budget based groundwater recharge, GW com pared to modeled values using Rtotal...............................................................................................................39 3-1 Cumulative frequency distribution of wetland areas......................................................... 60 3-2 Comparison of daily wetland stage (sto rage) and water budget-estim ated storage for Beaty wetland 2..................................................................................................................61 3-3 Cumulative distribution functions of scaled volum es for the study wetlands................... 61 3-4 Stage-area relationship for study wetla nds (a) as well as their corresponding generalized bathym etric cross-sections (b). The solid vertical lines represent the locations of the critic al ditch elevation.............................................................................. 62 3-5 Percent cumulative frequency plots of s caled ditch flow and wa ter surface elevation for the study wetlands ........................................................................................................ 63 3-6 Cumulative inflow and outflow vol um es for BW1 wetland and LW1 wetland................ 64 3-7 Cumulative density function for the fract ion of contributing upland area to wetland area for the study wetlands. ...............................................................................................64 3-8 Wetland LW1 contributing area and co rresponding rainfall for six days. .........................65 3-9 Cumulative distribution f unction of hydraulic residence tim es for the study wetland...... 65 4-1 Example of wetland treatment analysis A) Geographic placem ent of the single isolated wetland. B) Treatment efficiency as a function of the fractional wetland aerial coverage and decay coefficient................................................................................ 75


9 4-2 Wetland geographic positioning for five si m ulations of increasing wetland overlap in the flow direction (from top to bottom). A) Zero wetland overlap. B)-E) increasing degree of wetland overlap..................................................................................................76 4-3 Random wetland field and corresponding treatm ent. A) Treatment efficiency of the random wetland field as a function of the fractional wetland aerial coverage and decay coefficient (kv). B) Plan view of random wetland field........................................... 76 4-4 Treatment efficiency of a two-wetland sy stem with increasing overlap. A) treatment efficiency as a function of fractional wetla nd aerial coverage and decay coefficient for no wetland overlap (kv). B)-E) increasing ove rlap of wetlands................................... 77 4-5 Fraction of wetland overlap for a two we tland system for relatively high and low kv of 10 and 0.1 respectively..................................................................................................78 5-1 Cumulative mass of P with time for the Larson wetlands................................................. 97 5-2 Cumulative mass of P with time for the Beaty wetlands................................................... 98 5-3 Results from k-C* model compared to m easured P concentrations. A) Larson wetland-1. B) Larson wetland-2. C) B eaty wetland-1. D) Beaty wetland-2...................... 98


10 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy WATER AND PHOSPHORUS BUDGETS OF DEPRESSIONAL WETLANDS IN THE OKEECHOBEE BASIN, FL By Daniel Brent Perkins August 2007 Chair: James W. Jawitz Major: Soil and Water Science The Lake Okeechobee, FL catchment basin and la ke proper have been given considerable attention in water resources res earch and management arenas because of the potential shift in trophic status resulting from excess nutrient load s. Depressional wetlands are a signature landscape feature in the Lake Okeechobee basi n, comprising approximately 12-16% of the watershed. These wetlands are often surrounded by pasture and grazing land and integrate hydrologic and chemical loads. These wetlands play an integral role in the retention of water and nutrients in the basin, as they are one of the primary mechanisms by which water and nutrients might be stored. This study stems from larger efforts to evaluate the suitability of wetlands at sites in the north Lake Okeechob ee catchment to increase water retention, phosphorus (P) transformation, and travel time to the lake. This dissertation emphasizes three areas of ongoing research in: 1) groundwater-wetland hydrologic exchange mechanisms and quantities, 2) quantification of hyd rologic pathways and system hydraulic residence times, and 3) phosphorus loading rates and dynamics. Results from this research suggest that groundwater-wetland exchange is important in considering the hydrologic impact of these wetlands within the fr amework of the overall P budget. Also, P loads from the ditch networks were often found to exceed all other outflows.


11 This research provides a conceptual hydrologi c and P model for depressional wetlands in the Lake Okeechobee basin. It also has implications related to hydrologic management of these wetlands for landscape-scale chemical treatment potential and nutrient loading to Lake Okeechobee.


12 CHAPTER 1 INTRODUCTION Background Lake Okeechobee is the second largest fres hwater lake entirely bounded within the continental United States (approximately 1890 km2). This shallow sub-tropical lake has exhibited signs of hyper-eutrophication since the 1980s, manifested by in-lake phosphorus (P) concentrations and shifts in predominant species of annual algal blooms (Reddy et al., 1995; Goldstein and Berman, 1995; and Haan, 1995, Bottc her et al., 1995). Beef operations in the LOB have come under regulations through th e Okeechobee SWIM Plan (SFWMD, 1993). In 1995, a series of studies conclude d that dairies and beef operations north of Lake Okeechobee were the primary sources of P to the la ke (Campbell et al., 1995; Flaig and Reddy, 1995; Haan, 1995; Reddy et al., 1995; and Goldstein and Berm an, 1995). To meet regulations, the Lake Okeechobee Action Plan (LOAP) was developed by the SFWMD as part of the South Florida Ecosystem Restoration (Harvey and Havens, 1999) Lake Okeechobee is the first water body in Florida for which a total maximum daily load of phosphorus has been established (140 metric tons/year). From 1998 to 2003, annual P loads from the lake watershed averaged 584 metric tons (SFWMD annual project report, 2003, unpublished). As part of the LOAP, the SFWMD included the idea from Bottcher et al. (1995) to use runoff-retention systems, such as naturally-occu rring isolated wetlands (IWs) in the landscape, for P reduction from cow-calf operations. It has been suggested that th ese depressional wetlands may have the potential to filter or retain runo ff P (Bottcher et al., 1995). Under the LOAP, one of the primary mechanisms for reducing P loading to the lake is suggested to be the restoration of isolated wetlands within the basi n. The restoration of these wetla nds may result in the reduction of P loads to the lake via two mechanisms. The first is a reduction of total water reaching the


13 lake. Water stored within the isolated wetla nds and surrounding watershed will be lost to ET thereby not entering the lake and not contributing to the P load. Secondly, isolated wetlands are thought to act as P sinks. The wetlands are hypothes ized to act as P filters within the watershed (Bottcher et al., 1995; Harvey and Havens, 1999). By reducing P loads to Lake Okeechobee it is thought that the original ecosy stem function may be reestablished. This work focuses on depressional wetlands within a managed agroecosystem north of Lake Okeechobee, Florida (USA). Characterizing water quality in isolated wetlands may present challenges because: 1) seasonal and hydrologic extremes alter the wetla nds partitioning mechanism by either accreting or decomposing organic matter that in turn modify chemical release rates from the soil to the water column and 2) chemical inputs are diffuse non-point sources that may be spatially and temporally variable (Parsons et al., 2003). While isolated wetlands are generally considered to be nutrient sinks (Hemond, 1980; Davis et al., 1981; Ewel and Odum, 1984; Neely and Baker, 1989; and Pezzolesi et al., 1998), it is difficult to determine whether or not a pattern exists for chemical transport to downstream water bodies. Th is is due to the paucity of investigations that adequately assess the impact s of hydrologic conditions. The term isolated wetland has been used to describe wetland systems that do not exhibit surface water connectivity with rivers, lakes, oceans, or other wate r bodies (Leibowitz and Nadeau, 2003; Tiner, 2003; Whigham and Jordan, 2003; and Winter and LaBaugh, 2003). Soils, hydrology, and vegetation have been used to classi fy wetlands that intera ct directly with open water. The classification of isol ated wetlands is not as obvious. The ambiguity associated with the classification of isolated wetlands is the resu lt of an inconsistent definition that has been approached from a hydrologic, ecologic, geogra phic, or other perspective. For example,


14 Whigham and Jordan (2003) define an isolated wetland as strictly prohibiting any surface water connectivity with receiving water bodies at any time while allowing for groundwater connections. Other researchers allow for infre quent surface water connect ivity to nearby water bodies (Leibowitz and Nadeau, 2003; Tiner, 2003a; Winter and LaBaugh, 2003). Groundwater interaction with isolated wetlands is explicitly included in the definitions given by Snodgrass et al. (1996) and is recognized to play an important role in cont aminant contribution to receiving water bodies. Ecologists tend to define isolated wetlands by the occurren ce of rare and highly dispersed habitats that are not adjacent to an other body of water, which includes some notion of hydrology (NRC, 1995). Tiner (2003a) defines isol ated wetlands in terms of geography, where the wetland must be completely surrounded by uplands consisting of non-hydric soils and nonaquatic vegetation. Since some definitions in the literature ma y permit periodic surface water connectivity as well as considerable groundwater connectivity to r eceiving water bodies, the use of isolated is perhaps misleading from a strictly hydrologic persp ective. In reality, diffe rent degrees of surface and groundwater connectivity occur and characte rizing the degree of connectivity could be considered in evaluation of wetland interactio ns with distant receiving water bodies and ecological function. Winter and LaBaugh (2003) suggested the use of groundwater travel time as a parameter to establish: 1) the degree of hydrologic isol ation and 2) a standard to characterize a wetland as isolated. Determining groundwater travel times for this purpose often requires rigorous and costly site characterization that is impractic al for large-scale classification of wetland connectivity to receiving water bodies. To avoid the need to characterize highly vari able hydrologic or ecolog ic processes or to assess complex situations where some degree of connectivity exists, perhaps the simplest and


15 most consistent definition of an isolated wetla nd is the geographically-bas ed definition proposed by Tiner (2003a). Leibowitz and Nadeau (2003) al so recommended this definition of isolation. It provides consistency for researchers while still using classical parame ters with which most wetland scientists and regulators are familiar. While the study wetlands are geographically isolated, they are referred to as historically-i solated or depressional we tlands to emphasize the significant degree of surface water connectivity from extensive hydraulic management. Tiner (2003b) reported a comprehensive list of categories of isolated wetlands and their distribution across the contiguous United States of America and Alaska. To illustrate the possible range of isolated wetlands relative to the total amount of wetlands, a subset of data from a study completed by Tiner et al. (2002) that estim ates the occurrence of geographically isolated wetlands is also reported here (Tab le 1-1). This subset of data demonstrates that, regardless of climatologic regime and aerial coverage of wetla nds in a landscape, isolated wetlands are of importance because of their potentially significant proportion of incidence. Despite the seeming commonness of isolated wetlands across the United States of America, th e hydrologic processes that drive the system are often very different. The motivation for this study stems from larger efforts to evaluate the suitability of managing historically-isolated, seasonally-inunda ted wetlands at site s in the north Lake Okeechobee catchment to increase water retention, nutrient transformation, a nd travel time to the lake. The term isolated wetland has been used to describe wetland systems that do not have surface water connectivity with rivers, lakes, oceans, or other wate r bodies (Leibowitz and Nadeau, 2003; Tiner, 2003a; Whigham and Jo rdan, 2003; and Winter and LaBaugh, 2003). However, in the 1950s networks of ditches we re constructed to inte ntionally drain these


16 wetlands, creating semi-permanent surface water connectivity between cow-calf operations and Lake Okeechobee (Flaig and Reddy, 1995). Leaching of nutrients and pesticides for ag ricultural land, generation of non-point source pollution, is one of the critical environmental problems facing hu man society (Carpenter et al., 1998; Reddy et al., 2002; Tilman et al., 2002). N on-point sources of agricultural pollution cause a change in the trophic level of aquatic systems. This elevati on of trophic level may result in loss of habitat, reduction of water quality, and destruction of ecosystems. Eutrophication is a concern throughout the developed wo rld, and specifically in the stat e of Florida where intensive agricultural practices generate la rge quantities of nutrient laden wa ter outflows. In particular, Lake Okeechobee receives P loads that are considerably higher than historical averages (Bottcher et al., 1995; Reddy et al., 2002). Additionally, Lake Okeechobee is part of the larger Everglades ecosystem and associated restoration efforts. Hypotheses and Objectives The principal hypothesis linked to the overall project is that these wetlands currently function, or m ay function (paired with hydrau lic and land use management) as a passive sustainable means of P reduction to Lake Okeec hobee. Many different investigators with a variety of expertise have been i nvestigating this hypothesis. In this work, testing this hypothesis was not feasible because of the s cale and work load that must be integrated from many sources of research (i.e. hydrologic, ve getative, biologic, and ecologic) However, this work does address questions that contribu te to the missing knowledge of sy stem behavior and response. Chapter 2 In Chapter two, groundwater-surface water interaction was hypothesized to play a significant role in the hydrol ogic dynam ics of the study wetl ands. It was furthermore hypothesized that values of wetland-specific hydraulic resistivity (a measure of the conductance


17 of water flow between wetland surface water and groundwater) might be estimated to describe the degree of interaction between groundwater and wetland surface water. Finally, it was hypothesized that groundwater-surface water in teraction might be predicted from linear regression modeling using measured head gr adients between the groundwater and wetland surface water. The first objective of Chapter 2 was to quan tify groundwater flow fr om the study wetlands by using a constrained water budget approach. Th e second objective was to passively determine values of hydraulic resistivity for the study wetl ands by means of a constrained water budget. Lastly, linear regression modeli ng of groundwater outflow from the wetlands (based on a constrained water budget approach) and a linear flux law (Darcys law under saturated flow) was done to predict long-term behavi or and magnitude of groundwate r-surface water exchange. Chapter 3 In Chapter 3, behavior of the overall wetla nd hydrologic budget is quantified to identify the relative contribution of each water budget co mponent. It was originally hypothesized that these wetlands behaved as flow -through wetlands, with groundwat er inflow primarily governing the flow into the system. However, based on measured hydraulic water surface elevations between the upland and wetland surface water, the infrequent occurrence of groundwater inflow and its corresponding low flow rate contributed to the pursuit of a new conceptual hydrologic model. Ditch flow was hypothesi zed to comprise the largest fraction water volume leaving the wetlands. It was also hypothesized that the groundwater outflow from the wetland played a larger role in wetlands with less -intense ditching (lower ditch fl ow). Also, it was thought that a variable-source contributing upl and area was involved in determining overland flow to the wetland. Hydraulic residence in these wetlands wa s also evaluated in this chapter. Through estimates of wetland volume and total hydrologic out flows, frequency dist ributions of hydraulic


18 residence times were developed to better understa nd the time-scales involved in water retention in these types of wetlands. The objectives of Chapter 3 were to first de velop a new conceptual hydrologic model of the study wetlands through quantifying the components in the water budget. It was important to compare differences between hydrologic componen ts at each wetland and develop relationships between the hydrologic differences and site char acteristics (i.e. bathymetry, relative ditch elevation, etc.). Next, the id ea of a variable source contributing upland area was evaluated by correlation of rainfall, overland fl ow, and site topography. The last objective of Chapter 3 was to characterize the hydraulic reside nce times of the study wetlands to better understand how they currently function in te rms of chemical treatment and water retention. Chapter 4 Chapter 4 describes a pseudo-2d first-order upt ake m odel that was used to identify the spatial variability of wetlands in a landscape for P treatment. Assumptions of hydrology are greatly over-simplified, however, estimates of landscape-scale P reduction are described for different spatial arrangements of wetlands in a watershed, uptake rates, and wetland depth. It was hypothesized that some reduction of P would occur at the landscape scale, using one or multiple wetlands in varying spatial arrang ements. The objective of this study was to demonstrate the treatment efficacy of different spatial arrangements of wetlands in a landscape. Chapter 5 In Chapter 5, the P budget for the study wetlands is developed using water budget inform ation from Chapter 3 and measured water quality data. In addition to the P budget estimates, a first-order uptake model (k-C model from Kadlec and Knight (1994)) was calibrated to measured P concentrations in the wetland surface water. The hypotheses related to the P budget were that 1) P associated with ove rland flow would dominate the P inflow to the


19 system, 2) ditch flow from the wetlands would be the primary pathway for P export, and 3) P associated with groundwater would also contribute significantly to the overall P export. It was also hypothesized that the P rem oval or reduction from the wetlands could be estimated with both the P budget and the k-C* model, yielding sim ilar results. It was ex pected that there be some variability in P reduction (treatment performance) from wetland to wetland because of the differences in the relative water budget componen ts. However, it was important to quantify the P load associated with each flow path. The objectives of Chapter 5 were to estim ate the P budget using hydrologic data (from Chapter 3) and measured water quality in the wetland surface water, ditch outflow, and groundwater. These estimates were generated to quantify the relative importance of the components in the P budget as well as determin e a wetland-specific P tr eatment potential. The principal objective of the k-C* modeling was to evaluate the treatment effectiveness (P reduction) of the wetlands. Also, the k-C* mode ling was done to determine the best fit of the uptake coefficient that might enable future prediction of P dynamics. Site Description Agriculture is the m ajor industry in Okeechobe e County (one of the principal counties in the Lake Okeechobee Basin (LOB)), with beef cattle and dairies producing the largest revenue. In 1997, the county had 68,234 beef cattle and 35,707 milk cows (Florida Crop and Livestock Reporting Services, 1997), outnumbering coun ty residents by a ratio of 3.3:1. Depressional wetlands and other water features comprise approximately 21% of the total catchment (Hiscock et al., 2003), with wetland s specifically comprising 15% (Boggess et al., 1995). The spatial extent and inventory of depressi onal wetlands in selected priority basins was more recently given by McKee (2005). These wetlands are located in the lowest physiographic landscape position and may represen t a considerable hydrologic and nutrient store. However,


20 approximately 45% of all depressional wetlands in the basin are hydrologically managed by a network of ditches that serve to drain water from the landscape to rivers that discharge to Lake Okeechobee (McKee, 2005). Depressional wetlands in the Okeechobee basin are typically less than one m deep and exhibit a somewhat bowlor elongated bowl-shape, o ccurring at a relatively small scale compared to the resolution of so il mapping units defined by the USDA Soil Survey for Okeechobee County (Lewis et al., 1997). From onsite characterization of soils immediately below the we tlands (from 0 to 2 m), we observed a higher organic matter content (Munsell co lor of 10 YR 2/1) as well as an order of magnitude lower saturated hydraulic conductivity (compared to upland soils) as determined from onsite slug tests (results in Chapter 2). In a ddition, three 1.35-m intact soil cores, collected within the boundary of LW2, showed a transiti on from a predominantly sandy texture in the upper profile to a clayey texture at approximate ly 90 cm below ground surface (bgs). At these same sites, Dunne et al. (2006) also reported relatively low bul k density values between 0.41 and 0.80 g cm-3 in the top 5 cm of soil. A thorough descri ption of soils and ve getation at the study locations can be found elsewhere (Dunne et al., in press); a brief summary is provided here. Following the USDA soil taxonomic system, upla nd soils at the Dixi e-Larson ranch are classified as Siliceous, hyperthermic spodic, psammaquents (Basinger series), while upland soils at the Pete Beaty ranch are classified as Sandy, siliceous, hyperthermic typic humaquepts (Placid series). Both soils are deep, poorly drained, ra pidly permeable and are formed in beds of sandy marine sediments (Lewis et al., 2001). Typi cal vegetation at these wetlands includes: Juncus effuses, Panicum sp., Pontedaria cordata var. lancifolia Ludwigia repens, Hydrocotyle ranunculoides and Andropogon glomeratus. Wetlands in this region are thought to have been formed from sink-hole karst geology.


21 In the mid 1950s, an extensive network of d itches and canals were created throughout the LOB to lower the regional water table and partiall y drain depressional wetla nds in an effort to use more land for grazing pasture (Flaig and Reddy, 1995). These ditches do not completely drain the depressional wetlands, as they were typically excavated to depths more shallow than the bottom of the wetlands themselves. This hydra ulic management creates a scenario where the water in the wetland must reach a critical depth before water exits the ditch. However, it is not uncommon, with such low topographi c gradients in this region (<1%), for ditch flow rates to drop to undetectable levels usi ng traditional water control structures. Under such management, not only is more grazing land creat ed, but the remaining water in th e wetland serves as a cooling pond for cattle (which are usually ex cavated onsite at dairy operati ons). While this strategy is advantageous for cattle productivity, the ditches hydrologically connect th e pastures with the lake, creating a rapid conduit for water containing P. This P has been integrated from a contributing upland/pasture area that may be deri ved from fertilizers, vegetative senescence, and/or cattle manure. Surface water discharg e from the wetlands also contains direct P contribution from cattle that make use of the cooling pond. The four study wetlands are located in the Lake Okeechobee watershed, Florida (USA) and are part of the larger Everglades ecosystem. Two of the study wetlands (LW1 and LW2) are located at the Dixie-Larson Ranch (27.243 N, -80.83W) and the other two (BW1 and BW2) are located on the Pete Beaty Ranch (27.408 N, 80.945W) (Figure 1-1). The two ranches are separated by approximately 6.6 km. These sites ar e in priority sub-basins, as determined by the Florida Department of Envir onmental Protection (Flaig a nd Reddy, 1995), that have been identified as contributing relatively higher loads of nutrients to Lake Okeechobee, compared to


22 the surrounding lake sub-basins. Both ranches ar e active cow-calf operations with similar forage ( Bahia sp.), pasture slope (<1%), and climate (125 cm average yearly rainfall). More historical information regarding the Dixie-Larson Ranch has been relayed to the project investigators than for the Beaty Ranc h. The two depressional wetlands on the DixieLarson ranch are within 113 ha of pasture land (f enced field area), which was purchased in 1974 and maintained 250 heifers for approximately 20 years. Approximately 1120 kg/ha lime was applied each year during this time period. Fr om 1992 to present, the dairy operation was changed to a cow-calf operation, with about 140 head were kept in the pasture (yearly average). In 1992, a one-time 2242 kg/ha lime treatment was applied, while 224 kg/ha/yr Ammonium Sulfate (25-0-0) has been applied since 1992. Starting in 2005, cattle also received 7711 kg/week hay for 6-8 weeks. Table 1-1 Estimated average percent of isolated wetlands by selected states. State Size of study area (ha) % of study area in wetlands Average % of wetland area predicted as isolated FL 134,486 19.0 43.1 IL 119,776 5.4 59.5 KY 61,520 0.8 46.7 NJ 236,292 24.6 7.3 NE 173,460 2.7 51.0 NM 128,008 0.2 18.7 WA 104,427 1.6 78.3


23 Figure 1-1. Site map of study wetlands and monitoring well locations. BW1 and BW2 wetlands are located on the Pete Beaty ranch and LW1 and LW2 wetlands are located on the Dixie-Larson ranch. Elevation co ntours are 0.1-m intervals.


24 CHAPTER 2 WETLAND-GROUNDWATER INTERACTI ONS IN MANAGE D SEASONALLYINUNDATED DEPRESSIONAL WETLANDS Introduction There are several im portant reasons to unders tand and quantify the frequency, magnitude, and duration of wetland surface water and local groundwater interaction. The first is to determine how wetland ecosystems may respond to managed groundwater drawdown. Currently, no protection policy ex ists for wetlands whose ecosyst em functions are unfavorably affected by local or regional groundwater management. Second, we tland wetting and drying periods are related to exchange of atmospheric gases, some of which have implications for global climate change (Romanowicz et al., 1994). Fi nally, wetland-groundwater interaction is an important component of wetland water retention and chemical treatment potential in managed (e.g. treatment wetlands) and natural systems. It is generally expected that wetlands act as a chemical sink in a landscape because 1) they are often located in re latively low topographic positions where landscape water may collect and 2) they often develop physical, chemical, and biological mechanisms for storing and transfor ming nutrients and other chemicals (Dortch, 1996; Reddy et al., 1999; Pric e and Waddington, 2000). The occurrence and magnitude of surface wate r-groundwater exchange in these wetlands has not been reported in the lite rature. Understanding the rela tionship between wetland surface water and groundwater is paramount in defining the role and function of th ese wetlands from an ecosystem and water quality perspective. This research will focus on characterization, quantification, and prediction of we tland-groundwater exchange in th is region. Results from this study have implications for nutrient load s to Lake Okeechobee via groundwater. Several authors have investigated charac terization and quantif ication of wetlandgroundwater exchange in other types of ecosystems (Devito et al., 1997; Hayashi et al., 1998a,b;


25 Sun et al., 1998; Wise et al., 2000; and Parsons et al. 2003). Devito et al (1997) devised several field experiments to determine the absence or presence of groundwater exchange with prairie pothole wetlands or depression-focu sed recharge. From a water ba lance perspective, Hayashi et al. (1998a) quantified wetland-g roundwater exchange of northern prairies to understand ecosystem-scale implications of salinization in pothole wetlands. The authors concluded that groundwater outflow was significa nt, accounting for 75 % of water leaving the study system. Furthermore, through chloride and bromide fiel d tracer experiments (Hayashi et al. 1998b; Parsons et al. 2004), this groundwater was found to flow laterally into the upland where it was assumed to be transpired by plants. The geolog y of the prairie pothole region (stratified silty sediments) is quite dissimilar from the fractur ed limestone overlain by course sandy sediments found throughout the Florida study s ites in this research. Th ese investigators assessed groundwater exchange with surfac e water by interpreting averaged hydraulic head gradients and soil saturated hydraulic conductiv ity estimates from slug tests in a Darcys Law format. However, these authors focused on quantifying th e total groundwater exchange relative to the other water budget components. Motz et al. (1998) estimated vertical leakage from 11 karst lakes to a deep semi-confined aquifer system (Floridan aquifer) in north-central Florida. In a ddition to deep aquifer recharge estimates, these investigators also used a water budget appr oach to estimate a vertical conductance, Kv/ b [T-1], which is a measure of media conduc tance (vertical saturated hydraulic conductivity, Kv [LT-1]) and depth to aquifer, b [L] (approximately 40 m), for each lake. These vertical conductance values are the inverse of the hydraulic resistivity, R [T], as determined from the residual groundwater term in the water budget. Estimates of R from these 11 lakes ranged from 365 to 9264 days. Reported estimates of hydraulic resistivity from two additional lakes in


26 north-central Florida also fall in this range, with a mean R of 3710 days (Motz et al., 2001; Watson et al., 2001). This current research also used a water budget approach to estimate R and groundwater recharge. However, the Lake Okeechobee isolated wetlands are quite different from the lakes studied by Motz (1998), Motz et al. (2001) and Watson et al. (2001), as they are seasonally inundated and primarily interact w ith local groundwater, rather th an a deep confined aquifer. Additionally, the scale of the wetlands in this research (approximately 100-m diameter) was much different than the lakes described above (1 to 3 km diameter). Hydraulic resistivity was determined from a best fit to the observe d groundwater outflow, as calculated from a constrained wa ter budget. Literature values of R for wetlands of comparable scale to those of this study have only been reported by Wise et al. (2000). These investigators developed a Wetland-Aquifer Inte raction Test (WAIT) method to investigate hydrologic exchange between wetland surface water and groundwater and reported on a 14-day field trial in a 70-m diameter wetland. The WAIT method i nvolves actively pumping surface water out of a wetland and then fitting a model to the observed water level in th e recovery phase. The authors reported good correlation between measur ed hydraulic soil properties and wetland R In the present study, water was not act ively pumped out of the wetlands, rather natural drawdown periods (typical durations of 14 to 22 days ) were observed where the wetland water level naturally declined under the combined infl uence of evapotranspiration and groundwater recharge. These drawdown periods were consider ed passive WAITs. Compared to an active pump test, the methodology in this study may be advantageous because it represents a more passive, and perhaps more natural means of quantifying wetland surf ace water export through groundwater.


27 Based on the findings of these previous studies it was expected that flow between wetland surface water and local groundwater in historic ally-isolated wetlands in the Okeechobee basin could be described in terms of hydraulic resis tivity. In this study, approximately 1000 days of daily water budget data were used to estimate groundwater-wetland exchange with the following specific objectives: 1) evaluate wetland-groundwater exchange usi ng field data in a water budget framework, 2) passively measure R values for comparison to those measured both actively and passively in other systems, and 3) evaluate the predictive capability of Darcys Law using daily site-specific data. The results from this study ha ve implications for water quality monitoring and management strategies aimed to reduce solute export (especially P) from the Lake Okeechobee, FL catchment area to Lake Okeechobee. Methods and Materials Wetland surface water elevation was monitored fo r three years at th e four study wetlands. Internally-logging pressu re transducers (Mini-troll STP, In-situ, Inc.) were deployed in 0.032-m diameter fully-screened PVC wells in the deepes t location of each wetl and. These pressure transducers exhibited 0.01% accuracy over a 0-34 kPa range. Wetland monitoring well installation depth was between 2.5 and 3.0 m bgs with an additio nal 1.5 m of well screen above the wetland ground surface. The above-ground po rtions of the monitoring wells were housed inside a larger diameter PVC casing (0.051-m diameter) to provide structural protection against interactions with cattle. Wetland water surface elevations were reco rded in half-hourly intervals and were subsequently averaged on daily time intervals. The monitoring records for wetland water surface elevations were approximat ely 1000 days for each wetland (Table 2-3). Upland water level monitoring wells were sim ilarly constructed, with screened intervals ranging between 1.5 and 2 m bgs. Seven uplan d wells were equipped with data-logging transducers such that at least one upland well wa s paired with each wetland well. Data logging


28 in upland water level monitoring wells was s ynchronized with wetland surface water monitoring data logging and subsequently averaged over the same 24-hour periods. The monitoring record duration for the seven upland wells ranged from 77 to 433 days (Table 2-3). Isolated wetlands in this re gion are generally less than 150 m in diameter and exhibit an oval to circular geometry. It has been thought that these IWs have been formed from sink-hole geologic processes, with a typi cal depth of one m. The topogr aphy of the four study wetlands was mapped using a line-of-sight laser level in combination with a hand-held Garmin GPS. Surface maps were created from these data usi ng Ordinary Kriging interpolation. Stage-volume and stage-area relationships were determined us ing these topographic maps. In addition, the elevation difference between lowest point in the wetland and the maximum ditch bottom elevation was determined. These parameters and relationships we re important for estimating the water budget at the study sites. Water Budget-Based Groundwater Outflow Elevations o f wetland surface water, Hwet [L], and upland water level, Hup [L], were translated to the same datum using the bathymetric survey to determine hydraulic gradients. These data indicated that gradients were nega tive (groundwater flow away from the wetland) 53% of the monitoring duration, and positive (flow towards the wetland) only 8% of the time. Because of the relatively high percentage of da ys corresponding to outflow from wetlands to groundwater, the emphasis here is on groundw ater outflow from the wetlands. Observed groundwater flow was calculated on a daily time step using a water budget approach, bounded by the wetland surface water. To focus on only groundwater outflow, the daily water budget was calculated to only include da ys when the following three conditions were met: 1) the wetland-groundwater hydraulic gradient diverged from the wetland, 2) the wetland surface water elevation was betw een the ditch outflow elevati on and the wetland bottom (i.e.,


29 days with surface water outflow from the wetland were excluded), and 3) no rainfall occurred. Thus, the wetland water budget for these days included only flow out to groundwater and evapotranspiration, ET [LT-1]. In this study, the Penman-Monteith methodology was used for calculation of ET (Allen et al., 1998). The groundwater outflow, GW [LT-1], was thus determined from the water budget as: GW = Hwet ET (2-1) where Hwet [LT-1] is the daily change in wetland water surface elevation. This constrained water budget provides a record of passively monito red interaction of wetla nd aquifer dynamics. The number of days used to estimate surf ace water outflow to groundwater is shown in Table 2-5 for each paired upland-wetland monitori ng record. An example set of paired wetlandupland well hydrographs is shown in Figure 2-2. Groundwater Outflow Model The wetland-groundwater exchange can also be expressed using a for m of Darcys law: (2-2) where d H is the wetland-groundwater hydraulic head difference. The estimated GW from Equation 2-1 and d H were compared on a daily time step for each wetland-well pair to determine R by linear regression. Hydraulic resistivity does not have a true physical meaning, but might be related to a dynamic zone or annulus in th e ground around the wetland through which water is moving. Two temporal scales were used to determine hydraulic resistivity: an event based approach ( Revent), and an approach using the total monito ring record of each wetland-well pair ( Rtotal). Drawdown events were defined as periods when the following conditions were met: dH R GW 1


30 Water in the wetland Absence of surface water outflow Insignificant rainfall. During these passive drawdown events the groundwater and surface water elevations manifested decreasing trends over the entire length of the subset record (between 14 and 22 days, Table 2-4). The beginning of a dr awdown event was defined when d H was near zero (usually within 2 days of significant rainfall), while the end was defined when a significant uniform drop in surface water occurred (more than 5 cm) across all wetland-well pairs. Seven to nine events were recorded for each wetland during the observation period (Table 2-4). The Revent values were used to identify spatial and/ or temporal trends (i.e. season and initial wetland water stage). However, Rtotal values were considered as the best estimate of hydraulic resistivity for each wetland-well pair. For each wetland-well pair, data were sectioned into multiple calibration and validation sets to determine model prediction capability as a function of calibration record length. Root mean square error and bias we re the objective functi ons used to evaluate the prediction performance of groundwater outf low during the validation period. Meteorological Data Average daily wind speed, daily air tem peratur e limits, net solar radiation, and average relative humidity were recorded from con tinuously logging weather station instruments (Campbell Scientific, Inc.) and served as input parameters in the calculation of evapotranspiration using the Penman-Monteith methodology (Allen et al., 1998). It is possible that this methodology of estimating ET might intr oduce a bias into the water budget, however the magnitude of ET values fell in a range typical of other studies in the sout h Florida area. Because of periodic instrumentation malfunc tions, approximately 50% of the input values originated from


31 onsite instrumentation, while the remaining input data were reco rded from the South Florida Water Management District (SFWMD) weather st ation L001 located approximately 14 km south of the study sites (SFWMD, 2007). Data betw een on-site and SFWMD inputs were highly correlated, with correlation coefficients for aver age air temperature, wi nd speed, solar radiation, and relative humidity of 0.85, 0.70, 0.72, and 0.81 respectively (N=202 days). A global sensitivity analysis was conducted to determine the error between on-site and SFWMD measured meteorological input parameters for estimation of daily ET values. Using the one-at-a-time method (Monod et al., 2006), less than 1% error in ET was observed for any given meteorological input parameter. Slug tests Slug-out tests were conducted in 7 m onitori ng wells at the Dixie-Larson wetlands: 6 in upland monitoring wells and 1 in the monitoring we ll at the lowest bathym etric point of LW2. Slug test data were analyzed to fi nd saturated hydraul ic conductivity (Kslug) using the Hvorslev method (Fetter, 1994). The value of Kslug in the wetland center was si gnificantly different from those in upland soils (0.67 m d-1 in the wetland and a mean value of 1.36 m d-1 in the upland soils). Results Groundwater Outflow Rates Groundwater outflow from each wetland was de term ined using Equation 2-1. Mean and standard deviation for all observed drawdown even ts for each wetland are reported in Table 2-2. Mean GW values were not considerably different between wetlands or ranches. The mean GW value for all wetlands of 0.68 0.25 cm d-1 was slightly higher than the recharge rates of 0.44 0.41 cm d-1 at 11 karst lakes in central Florid a reported by Motz (1998). The daily GW values


32 were multiplied by hydroperiod to determin e annual recharge rates of 1.2 to 2.2 m yr-1 from these isolated wetlands. Hydraulic Resistivity Using Equation 2-2, the drawdown events showed linear dependence between d H and GW with good correlation (T able 2-4). The average Revent values from each wetland-well pair were between 42 and 72 days (Table 2-4) Some variability was found in Revent (CV between 0.32 and 0.49), but this parameter did not exhibit trends wi th initial wetland water elevation (i.e. changing spatial scale of wetland hydraulic resistivity), or season (hydraulic change s due to plants, cattle traffic, etc.). This variability might be related to non-equilibrium of water flow, where R is sampled more frequently than it may be observed as steady-state. Values for Rtotal for each wetland-well pair and the corresponding RMSE are reported in Table 2-5. Observed GW values are compared in Figure 23 to calculated values determined using the best fit value for Rtotal in Equation 2-2 Based on observations from intact soil cores, a value of 0.9 m was used to represent a depth to soil confining unit ( h ). This estimate of h and the mean Rtotal (54 days) from all wetlands were used to back-calculate a saturated hydraulic conductivity ( Kbc) value of 0.02 m d1. This value was compared to Ks slug, which ranged from 0.6-2.4 m d-1 with a mean and standard deviation of 1.3 0.6 m d-1. The Kbc estimate is approximately 64 times less than the mean Ks slug value and 33 times less than the Ks slug value from in the wetland proper, which is likely most representative of the conductivity measured by the passive drawdown. These ratios are similar to those found by Wise et al. (2000), who reported Kbc between 12 and 70 times less than Kslug. Model Calibration/Validation The predictive capability of the pa ssive-drawdown m et hod for determining R to approximate actual GW values on a daily time step was evaluated by defining calibration and


33 validation periods for each upland-wetland monito ring well pair. Cali bration periods were selected to calculate R from Equation 2-2 using data subsets ranging from 10% to nearly all of the data (90%). The wetland-averaged GW RMSE values for these validation periods were all in the range between 0.57 and 0.90 cm d-1, which is comparable to the range of RMSE values based on best-fits to the complete data sets (0.61 to 0.88 cm -1, Table 2-5). Thus, the RMSE for the predicted GW did not significantly improve with longer calibration period, with a mean reduction of 16% when using 90% rather than 10% of the total data for each wetland-well pair. Discussion and Conclusions Groundwater Outflow The isolated wetlands in this study discharg e to surficial groundw ater, where pasture vegetation-d riven ET cycles water to the atmos phere, lowering upland water tables and thus enhancing recharge from surface water. Thes e processes (controlling hydraulic gradients between wetland and upland) and the soil hydraulic propert ies (represented by R in this study) resulted in higher groundwater recharge rates compared to those reported by Motz (1998) for 11 karst lakes in central Florida. However, cons idering the shorter hydroper iod at the wetlands in this study (less frequent occurr ence of standing water in the we tland), the annual recharge (1.2 to 2.2 m yr-1) was found to be similar to values from the 11 karst lakes (Motz, 1998) (average 1.6 m 1.5 m yr-1). Even though the wetlands in this study exhibited lower R values (suggesting greater recharge rates) compared to values from those lakes (55 14 days versus 2380 2683 days respectively), annual cumulative groundwater recharge depth was comparable because of the hydrologic differences between systems The wetlands studied here were all head-of-di tch, with no ditch inflows, thus inflows to these wetlands are driven only by rainfall and ru noff mechanisms. Average yearly rainfall in Okeechobee County is between 1.3 and 1.6 m (Lewis et al., 2001). These inflows are distributed


34 to surface water storage, ET, ditch flow out of the wetland, and groundwater recharge. While a complete water budget was not within the scope of this study, the average annual GW recharge values found here for these isolated wetlands is significant compared to the average annual rainfall total. Thus, groundwater outflow from these wetlands is an important water flow path, with important implications for water quality management practices aimed to reduce solute export from these catchments (particularly nut rients that are relate d to eutrophication of receiving water bodies). Hydraulic Resistivity Di fferences in R between wetlands at the Okeechobee research sites might be explained by differences in the depth to a confining unit, si te-specific saturated hydr aulic conductivity, or land management practices. Processes that c ould lead to variability in estimated Revent values for a given wetland include errors in water budget components, spatia l variability of soil hydraulic properties, cycles of vegetative growth and di e-back, and cattle stocking rates and associated hydraulic impacts (e.g., compaction, physical mixing). Motz et al (1998) found some variability in R (CV between 0.09 and 0.56), but attributed it to errors in water budget components. Identifying and quantifying the mechanisms responsible for temporally variable R values was outside the scope of this study, but further research could quan tify impacts of cattle and plant growth cycles on wetland hydraulic propert ies related to groundwater recharge. The hydraulic resistivity at th e study sites reflects groundwater recharge to a relatively shallow surficial aquifer through primarily sandy soil, and was thus found to be appropriately different than the R values reported by Motz (1998), Motz et al. (2001), and Watson et al. (2001), where groundwater recharge traversed an average of 40 m of clayey confining material. Also, a considerable difference in R was found between the study wetlands ( R between 28 and 73 d) and that was reported by Wise et al. (2000) (R = 8 d). The larger R values at the wetlands evaluated


35 here can be attributed to the lowe r saturated hydrau lic conductivity ( Kbc = 0.02 m d-1) of the wetland sub-soils, compared to the peat soil ( K = 0.083 m d-1) studied by Wise et al. (2000). Comprehensive site characterizati on that included measurement of R which may be estimated from a monitoring record during condi tions describing Equation 2-1, could relate land management practices to long term prediction of groundwater recharge. In addition, a database of R values could be developed to identify sens itive areas of local groundwater recharge. Wetlands might also be ranked by R in terms of potential for hydr aulic and chemical treatment potential, where higher R values might suggest longer residence time and perhaps finer underlying soil materials. Model Performance Prediction o f wetland surface water outflow through groundwater on a daily time-step was not particularly notable, as the RMSE re sulting from the linear regression of Rtotal (which represents the best model fit for the entire data set) for any well pair approached the value of the mean groundwater recharge. Howe ver, it is important to note that the cumulative comparison of modeled and estimated groundwater outflow from all four wetlands was quite close, which illustrates that the model may be used to descri be longer-term time period s (i.e. weeks, months, and years) (Figure 2-3). The bias for the enti re record of each wetland well pair was small (between -0.02 and -0.15 cm d-1), but suggests that the linear regression model slightly underpredicted observed groundwater out flow. Furthermore, when eval uating the distribution of daily observed and modeled groundwater recharge rates for each well pair, one-sided t-test analysis indicated that the hypothesis that the two distributions were not sim ilar could not be rejected (P < 0.01). Predicted groundwater recharge, as also noted by Wise et al. (2000), was quite sensitive to the R parameter. For example, comparison of modeled and observed groundwater recharge from


36 well pair BW1-MW5 using Rtotal = 55 d resulted in a RMSE of 4 cm. Considering that CV = 0.32 was found for Revent for this wetland, estimates one stan dard deviation greater and less than the mean (R = 68 and 40 d, respectively) were also used to predict groundwater recharge over the same period, with corresponding RMSE values of 13 and 18 cm respectively. Thus, because of the variability of GW estimates resulting from Revent, a period of at least several weeks may be required (under conditions describing Equati on 2-1) to develop a predictive model of GW The approach described here was used to m odel groundwater outflow from the wetlands at times when no surface water outflow occurs. This model could be extended to all times when water is in the wetland, which would improve c onfidence in estimates of groundwater outflow from wetland surface water that are currently el usive, leaving surface water outflow through ditches as the only undetermined outflow. Such a water budget would address the relative importance of these different system outflows. Further characterizati on of other regional managed depressional wetlands using the methodology included in this research may ultimately be used to evaluate important solute treatment performance scenarios. Table 2-1 Wetland area, elevation difference betw een the lowest point in the wetland to the highest point in the ditch ch annel (relative ditch elevati on), and the topographic range of the four study wetlands Wetland ID Footprint area (ha) Relative ditch elevation (m) Topographic wetland range (m) LW1 2.6 0.30 0.75 LW2 2.1 0.30 0.80 BW1 1.8 0.50 0.76 BW2 1.6 0.40 0.74


37 Table 2-2 Wetland hydroperiod (days), mean standa rd deviation of groundw ater recharge rates (cm d-1), and mean standard deviation of annual groundwater recharge rates (m yr-1) based on hydroperiod Wetland ID Hydroperiod (days) Groundwater outflow (cm d-1) Groundwater outflow per hydroperiod (m yr-1) LW1 256 0.7 0.2 1.8 0.4 LW2 219 0.5 0.2 1.2 0.4 BW1 302 0.7 0.4 2.2 0.8 BW2 233 0.8 0.2 1.9 0.4 Table 2-3 Monitoring periods of wetland water surface and upland water table elevations for Dixie-Larson wetlands (LW1 and LW2), and Pete Beaty wetlands (BW1 and BW2) Location Monitoring Period Net Number days -----Wetland surface water wells----LW1 7/2/2003 to 3/10/2006 962 LW2 12/18/2003 to 3/10/2006 814 BW1 7/2/2003 to 04/29/2006 1011 BW2 7/2/2003 to 04/29/2006 1033 -----Upland water level monitoring wells----LW1 MW7 3/09/2004 to 3/10/2006 234 LW2 MW3 08/07/2004 to 3/10/2006 206 LW2 MW4 3/09/2004 to 3/10/2006 222 BW1 MW2 7/03/2003 to 3/10/2006 433 BW1 MW5 4/27/2005 to 11/12/2005 77 BW2 MW2 3/09/2004 to 4/04/2005 145 BW2 MW3 4/05/2005 to 3/10/2006 110 Table 2-4 Summary of event-based regression between observed groundwater outflow from the four study wetlands and the wetland-groundwat er hydraulic gradient. Mean values for each wetland include hydraulic resistivity, Revent correlation coefficient, r and coefficient of variation, CV. Wetland Mean Revent (days) CV Mean r Number events Mean days/event LW1 42 0.49 0.71 7 14 LW2 72 0.33 0.73 9 22 BW1 54 0.32 0.74 9 20 BW2 40 0.45 0.83 7 17


38 Table 2-5 Best-fit hydraulic resistivity determin ed the total monitoring record of each wetlandwell pair, Rtotal, and corresponding RMSE. Wetland-upland well pair Rtotal (days) RMSE (cm d-1) Bias (cm d-1) Days LW1 LW1MW7 62.8 0.75 -0.15 181 LW2 LW2MW4 72.6 0.77 -0.04 176 LW2 LW2MW3 61.2 0.73 -0.10 94 BW1 BW1MW2 58.9 0.61 -0.02 349 BW1 BW1MW5 54.5 0.70 -0.14 54 BW2 BW2MW2 48.5 0.78 -0.12 111 BW2 BW2MW3 28.9 0.88 -0.10 84 Figure 2-1. Example wetland surfac e water and upland water table elevation behavior at LW1. Shaded areas indicate drawdown periods used to quantify wetland surface water and groundwater exchange.


39 Figure 2-2. Typical drawdown event used to evaluate co rrelation between GW and hydraulic gradient between groundwater and wetland (d H ). For these data from BW2-MW2, GW and dH are highly correlated ( r = 0.98). Figure 2-3. Cumulative water-budget based groundwater recharge, GW compared to modeled values using Rtotal.


40 CHAPTER 3 QUANTIFYING HYDROLOGIC PA THWAYS IN DEPRESSIONAL WETLANDS USING A WATER BUDGET APPROACH Introduction Naturally occurring depressional m ay play a significant role in landscape-scale water quality issues that impact Lake Okeechobee. During times of high regional water table and rainfall, surface water outflow from depressiona l wetlands through the ditches and canals may carry a significant load of P (R eddy et al., 1995). However, wetla nds in general are considered to have the ability to provide a chemical buffering and contaminant assimilation capacity. The question of whether or not these wetlands function as a source or sink is largely dependent on the hydrology of the system. The mechanisms of flow in and out of these wetlands must first be understood in order to determine their function in landscape-scale discharge of P to the lake. Partial or total elimination of ditch outflow (by damming the ditch outflows of the wetlands) is hypothesized to enhance nutrient treatment poten tial of wetlands through increased hydraulic retention time and wetland wa ter-soil interface contact area. Constructed wetlands implement regimented in flow and outflow, to quantify the amount of water passing through the system and create optimal treatment performance conditions (i.e. hydraulic retention time allows water to interact with the uptake mechanisms in the wetland long enough to achieve an acceptable level of treatme nt). Performance evaluation for any treatment wetlands is predicated on contaminant removal efficiency usually determined by comparing import and exported contaminant loads at well-defin ed surface inflow and outflows. Constructed wetlands may be designed or developed to take a variety of geometries that influence the treatment potential; however, a well-defined inflow and outflow is a common theme (Kadlec and Knight, 1994).


41 Guardo (1999) determined nominal residen ce times (based on wetland volume and average water flow through the system) from a constructed treatment wetland called water conservation area 1, in the Everglades from a water budget approach. This re search involved point inflows and outflows under nearly steady-state hydraulic conditions. Similarl y, Wang et al. (2006) evaluated the Orlando, FL Easterly Wetlands (a lso a treatment wetland), using conservative chemical tracers and hydraulic flow rates to determine wetland treatment efficiency. Wetland hydraulic efficiency was based on the ratio of me an residence time (from tracer data) to nominal residence time. This study was also conducted under nearly steady-state hydraulic conditions. These studies determined hydraulic characteristic s of constructed treatment wetlands that are essential in determining the overall chemical treatment efficiency. Evaluation treatment potential by analysis of hydrau lic efficiency or retention times in naturally-occurring depressional wetlands in the LOB is not simple to calculate because of the transient nature of most water inflows and outflows. Seasonally hi gh surface-water levels, paired with enhanced surface water outflow to groundwater due to evapotranspiration ( ET) in the uplands (creating a larger hydraulic head difference between upland and wetland) from around the perimeter of surface water bodies are a major cause of the co mplex and seasonally dynamic groundwater flow fields associated with surface water (Winter 1999). In addition, performance evaluation of wetlands in the basin is further confounded by the lack of a well-defined inflow, outflow, or both. In this research, a water budget wi ll be estimated for four isol ated wetlands in the LOB to determine the quantities of infl ow and outflow from pertinent processes. The conceptual hydrologic model for these wetlands will be better understood through the water budget estimation. Water budgets have been used in hydrologic assessments of wetlands for several


42 reasons: 1) Assessment of water resources (Mak hlouf and Michel, 1994), 2) Identification of water quality and quantity impacts to wetland ecosystems (Price and Waddington, 2000), and determining links between wetlands, uplands, an d receiving water bodies (Drexler et al., 1999 and Hayashi et al, 1998a). Using these estimate s of inflows and outflows, an analysis of hydraulic residence time will be done to begin to ev aluate the overall potential. This component will involve inflow estimates resulting from su rface runoff (overland flow to the wetland), precipitation on the wetland water surface, and groundwater inflow. The outflows of these wetlands are from ET, groundwater outflow, and surface water outflow through the ditch network. The dynamics and quantities of inflows and outflows of managed wetlands in the LOB have not been reported in the literature, and will be presented in this research. Results from the water budget hold informati on describing the temporallyand spatiallyvariable extent of the surroundi ng uplands that are involved in the treatment process, or hydrologic zone of influence. The idea of a te mporal and spatial component non-point source inflow has been described in river systems and termed variable source zones (Ward, 1984). These zones are controlled by groundwater prox imity, topography, and rainfall rate (Dunne and Black, 1970; Montgomery and Dietrich, 1995; Fra nkenberger et al., 1999; and Lyon et al., 2004). The objectives of this study are to: 1) estimat e the inflows and outflows of each wetland to build a conceptual hydrologic mode l, 2) identify the wetland hydrologic zone of influence within the landscape, and 3) determine the distributi on of nominal residence times to interpret the hydrologic component associated with contaminan t treatment potential. This research will provide a conceptual hydrologic model for depre ssional wetlands in the LOB. It also has implications related to hydrologic management of wetlands for landscape-scale chemical treatment potential and nutrient loading to Lake Okeechobee.


43 Water Budget Estimation Detailed topographic surveys were conducted to determ ine the bathymetry of the four study wetlands. The bounds for these surveys we re typically 10-20 m into the upland (as delineated by a vegetative border of Serenoa repens ). A line-of-sight laser level was used in combination with a hand-held Garmin GPS to collect point elevation data. Ordinary Kriging interpolation of those elevation points, using ARCGIS (ESRI ArcMap 9.1), yielded a continuous surface (average root mean square error, RM SE = 0.175m). Wetland foot print area and relative ditch elevation (elevation difference between lowe st point in the wetland and the maximum ditch bottom elevation) were determined from this surv ey. For this study, it is important to note that the critical flow depth ( hd) (elevation difference between the bottom of the wetlands and the highest elevation of the ditch) for LW 1, LW2, BW1, and BW2 was 0.3, 0.3, 0.5, and 0.4 m respectively. The relative areas of these wetlands were quite di fferent and resulted in volume estimates between wetlands (while changes in depth were similar) (Figure 3-1). The differences in area are critical in understanding the differe nces in flow rates of each component between wetlands. Water budget data for this study was calculate d over approximately 640 days between May 2004 and March 2006. The water budget for th ese isolated wetlands was defined as: dHwet = P ET D + S GW (3-1) where S is overland flow, D is surface water outf low (through ditches), d Hwet is the change in wetland surface water storage, ET is evapotranspiration, and GW is groundwater inflow or outflow. Rainfall and d Hwet were calculated from direct measurement, while ET was estimated using an empirical function using meteorological data. A combination of direct measurement and linear regression modeling was done to determ ine the remaining residual term(s) in the water budget: groundwater flow, ditch flow, and overland fl ow. It was necessary to incorporate linear


44 regression modeling of wetland-groundwater flow because not all processes were directly measured, leaving sometimes two or three unknown water budget components, depending on the hydrologic conditions. Equation 3-1 may be reduced to fewer terms when the wetland stage is lower than hd for ditch flow and/or when rainfall and overland flow do not occur. For the simplest case, no rainfall, runoff, or ditch flow, Equation 3-1 reduced to: dHwet = ET GWout (3-2) In this case, the GWout term in the only unknown, as ET was assumed to be a known component. When rainfall occurred, but no di tch flow occurred, Equation 3-1 reduced to: d Hwet = P ET + S GWmod (3-3) Here, the rainfall, ET, and GWmod (modeled groundwater inflow/outflow from linear regression) were known. This left only the overland flow ( S) as the residual term. When ditch flow occurred without rainfall and runoff, Equation 3-1 reduced to: dHwet = ET -D GWmod (3-4) Ditch flow ( D ) is left as the residual term. During these periods, the ditch flow was linear regressed to Mannings equation and will be expl ained further below in the groundwater section of this chapter. When all water budget com ponents occurred, such as in Equation 3-1, the overland flow was left as the residual term a nd modeled groundwater in flow/outflow and ditch flow ( Dmod) were used as known components: dHwet = P ET +S GWmod-Dmod (3-5) To evaluate the degree error in the estimated water budget, the m easured daily wetland stage was compared to the wetland stage as determined by the sum of the estimated water budget components. Figure 3-2 shows an example of the daily measured wetland stage (storage) and the water budget-estimated storage. Detailed methodol ogies for estimating and/or measuring each


45 water budget component will be desc ribed later in this section. Calculating the volume-based water budget was done by first calculating th e depth-based water budget for each day and comparing estimated wetland water stage against the measured daily values. Then, the depth values for each water budget component were multiplied by the day-specific wetland area using information from the stage-area relationships to generate volume-based estimates. These volume-based estimates assumed that the wetland porosity was equal to one, meaning that the volume of wetland surface water oc cupied by vegetation was neglig ible. In reality, this is probably not the case, as Kadlec and Knight (1994) report porosity values of treatment wetlands between 0.7 and 0.9. Precipitation and Evapotranspiration Precip itation ( P ) was measured onsite at two locations using data logging tipping buckets (Onset Communications Corp. model RG3-M) and the Bassett monitoring station maintained by the South Florida Water Management District (SFWMD) (located on the Beaty Ranch). Approximately half of the rainfall data came fr om onsite measurements, with the remaining half from the Bassett rain gage (SFWMD). The Penman-Monteith method was used to estimate evapotranspiration ( ET) from the wetlands. Approximately half of the meteorological data necessary for this estimation technique was collected onsite with Campbell Scientific instrumentation, while the remaining half was collected at the SFOO1 station, also maintain ed by the SFWMD (located at a distance of approximately 14 km). See Chapter 1 for further details. Groundwater Outflow Groundwater inflow and outflow in water budget analyses have been the most difficult to quantify because of its com plex behavior (Drexl er et al., 1999; hunt et al., 1996). Groundwater outflow ( GWout) from between wetland surface water a nd upland groundwater was determined


46 from a constrained water budget that excludes di tch flow, overland flow, and rainfall. Data logging pressure transducers (Minitroll 500, In-S itu Inc.) were deployed in fully-screened 3.18cm diameter well casings to measure water levels in the upland and wetland surface water. At each wetland, one well was located at the lowest topographic position in the wetland and at least one other was located at the wetland fringe where wetland vegetation transitions to Serenoa Repens (at approximately 50 m from the wetland cente r). These measurements of hydraulic head in the wetland surface water and upland groundwa ter will serve to estimate the groundwater inflow and outflow between wetland and upland. For more specific information regarding well locations and sampling protocol, see Chapter 2. Linear regression modeling was used to approximate groundwater inflow and outflow when overland flow and/ or ditch flow occurred. It was important to develop a means to directly de termine as well as estimate groundwater flow so that overland flow and ditch flow would be left as the residual terms in the water budget (for which no measurements were collected). This study incorporates the data analysis done in Chapter 2 to develop observed outflow estimates as well as linear regression mode led approximations of groundwater outflow. Observed groundwater outflow was derived from a constrained daily water budget that excluded rainfall, runoff, and ditch flow (wetland water stage was below hd). Of the total number of days when groundwater outflow occurred, a wetland -averaged percentage of 34 was directly determined from the constrained water budget. Linear regression of Darcys law and these observed outflow data was used to determin e a wetland-specific hydr aulic resistivity and approximate the observed groundwater outflows. A reasonably good f it (r between 0.71 and 0.85) resulted from the linear regression of gr oundwater outflow (from the water budget) and Darcys law (Chapter 2). For times when gr oundwater outflow conditions did not meet the


47 criteria for the constrained water budget, Da rcys law was used to approximate groundwater outflow and included the fit hydraulic resistivity and upland-wetland pressure head difference. Groundwater Inflow The occurrence of groundwater inflow, as dete rm ined by the days for which upland water elevation was greater than wetland surface water elevation, wa s only 8% of the time that standing water was in the wetland. The c onstrained water budget described in the Groundwater outflow section was not usually viable for direct observation of groundwater inflow, because hydraulic gradients that resulted in groundwater flow toward th e wetland were controlled by rainfall, which in turn added overland flow to the water budget equation as a second unknown. With relatively fewer periods of time when groundw ater inflow occurred, it was assumed that the inflow dynamics could be approximated by Da rcys law and the fit hydraulic resistivity determined from the groundwater outflow data. Surface Water Outflow Estim ation of the water budget for these depressi onal wetlands is not straightforward. Low topographic gradients and vegetation in the ditc hes often contribute to submergence conditions, which limit the design of surface water outflow c ontrol structures. The occurrence of surface water outflow through these ditches is principa lly controlled by hydrologic conditions that nudge the wetland stage above the lowe st elevation of the ditch bo ttom (between 0.2 and 0.5 m above the lowest elevation in the wetland bathymetry ). Once surface water in the wetland has risen above hd, the magnitude of flow is conceptualiz ed using elements of open-channel flow described by Mannings equation: roughness, sl ope, and the geometry of the ditch. Long-throated weirs were installed in the outfl ow ditches of the four wetlands, using flow specifications best suited to low-flow, subm ergence conditions. Water surface elevations measured in the stilling well and behind the weir (approximately 2 m) did not exhibit adequate


48 head-loss to qualify a flow rate. However, observations of surface outflow events through ditches (unidirectional grass ma tting and physical observations) we re noted for all wetlands. Because of the inadequate head-loss across the flumes, another methodology was developed to estimate ditch flow. This met hodology involved another form of the daily water budget determined by the following constraint s 1) water surface elevation was above hd, 2) rainfall/runoff did not occur, 3) groundwater inflow/outflow, derived from linear regression modeling, was considered a known variable. Tabl e 3-3 reports the percent of the ditch flow record for which ditch flow was the residual term in the constrained water budget. The wetlandaveraged percent of days for which ditch flow occu rred (ditch flow was the residual term in the water budget) was 45. The other 55 % of the ditch flow record that included rainfall/runoff (a second unknown) was estimated using Mannings equation. Mannings r oughness coefficient of 0.03 (typical value for lightly vegetated, un-maintain ed channel) was used for all wetlands to reflect the grassy conditions of the ditches. Linear regression of the daily ditch outflow rates (resulting from the constrained water budget as described above) and Mannings equation was done to fit the slope of the ditch to best approx imate ditch flow. Once direct measurements and linearly regressed estimates of groundwater and ditch flow we re established, the water budget had only overland flow as the remaining term. Overland Flow One of the most common m ethods for determini ng runoff is using water control structures (Branfireun and Roulet, 1998; Metc alfe and Buttle, 1999). In these examples, runoff would be comparable to overland flow at the study sites in this paper. When water control structures are not feasible to record overland flow, empirical functions that hi nge on simple rainfall-dependent relationships have been used (Prescott and Tsan is, 1997). Some investigations of water budgets exclude the overland flow component because it is difficult to measure or infrequent enough to


49 be considered insignificant (Drexler et al., 1998 ). However, in this study, overland flow was considered to be important because of shallo w water table conditions that reduce soil-water storage. In addition, while soils in this re gion are primarily sand, drainage is poor (Soil Conservation Service classifi cation A/D or B/D). Capece et al. (1988) reported saturated hydraulic conductivities be tween 1.5 and 15 cm h-1 in the basin, while value between 6 and 24 cm d-1 were reported at the field sites (Chapter 2) Because the overland flow at these wetlands entered the wetlands from around the perimeter, infl ow water control structur es were not viable. Overland flow was considered to be the only residual term, once all other components were either measured or estimated through regr ession modeling (dependi ng on the hydrologic conditions). Hydraulic Residence Time An i mportant hydraulic parameter related to treatment wetland performance (usually in constructed wetlands) is the hydraulic or nominal residence time ( HRT ): (3-2) where V is the wetland volume [L3] and Q is usually defined as the average inflow and outflow rate [L3T-1]. Paired with reaction and uptake coeffi cients, the HRT of a treatment wetland may be incorporated into a continuously-stirred tank r eactor or tanks-in-series modeling approach to evaluate treatment performance. This type of an alysis is typically applied to steady-state flow conditions with controlled inflow and outflow to evaluate wetla nd hydraulic efficiency. These study wetlands do not exhibit thes e characteristics, as the wetland volume is transient and inflows and outflows are not only transient, but result from different hydrologic processes, which exhibit unique rates and dynamics. Q V HRT


50 In this research, a methodology was implemented to compare HRT of traditional treatment wetlands to transient, dynamic wetlands in general by assuming a pseudo-steady-state flow condition. Still applying Equation 3-2, the rates of water budget components were calculated on a daily time step and assumed at steady-state for each day. The Q term in Equation 3-2 was defined as the sum of all outflows (groundwater outflow, ET, and Ditch flow out when it occurred). Following this methodology, a dist ribution of flow rate s was generated and calculation of HRT was done using the resulting distribution. Wetland volume was determined from the daily stage-volume relationship and comb ined with daily total Q to ultimately create a distribution of HRTs based on data from the entire monitoring record of each wetland. Hydroperiod Hydroperiod m ay be expressed on either a depth or volume basis. Dunne et al. (In press) reported hydroperiods in terms of depth for thes e study wetlands over the same time period. In this work, it was important to describe the hydroperiod on a volume basis, because of the implications for wetland treatment efficiency and HRTs. Using the time series wetland water stage data from each wetland (described in the Water Budget Estimation section), the hydroperiod was calculated for each wetland. To better compare hydroperiods between the study wetlands, results from the volume-based analysis of hydroperiod distribution were scaled to their corresponding average wetland volume ( AWV ). Results and Discussion Hydroperiod The depth-based hydroperiod for BW1, BW 2, LW 1, and LW2 was reported as 302, 233, 256, and 219 respectively (Dunne et al., in press). Results from the volume-based hydroperiod analysis showed that there was a difference in AWV distribution between more-intensely ditched wetlands (LW1, LW2, and BW2) and the less-intens ely ditched BW1 wetland. The majority of


51 the wetland hydrologic dyna mics occur within two AWV s, consisting of approximately 85% of the variability of surface water occurrence. At LW1, LW2, and BW2, 73% of the variability was described at or below one AWV whereas 60 % of the variability was described at or below one AWV for BW1. Similar volume-based hydroperiod di stributions were observed between the study wetlands except BW1, which exhibited a cons istently wetter condition (Figure 3-3). The most obvious explanation for differences in AWV distributions is because each wetland had a specific hd that defined the dynamics between more rapid surface water outflow and longer groundwater and ET outflows. While the hd of the wetlands affected the distribution of AWV s, wetland bathymetry was also an important aspect of th e behavior of wetland hydroperiod. For example, at a value of one AWV LW1, LW2, and BW2 exhibit simila r distributions. However, the hd at BW2 was 10 cm higher than that at LW1 and LW2, suggesting that reduced ditch flow out of BW2 should skew the distribution of AWV s toward relatively wetter conditions. This is not the case, suggesting that wetland bathymetry may also be important in describing the variability in AWV distribution. Water Budget In this study, one objective wa s to ev aluate the potential treatment of these wetlands based on hydrology. As presented previously in this paper, a depth-based water budget was completed to develop a close fit between th e observed daily change in wetla nd surface water and the sum of inflow and outflow components in Equation 3-1. Results from this type of analysis were scaled by the corresponding daily values of wetland surface water area to yield a volume balance. The volume-based water budget was used exclusively in this paper to allow more perspective into the potential management of these wetlands, especially for their use as natural treatment wetlands. The absolute values of flow rates among wetlands were quite different. For example, LW1 wetland exhibited consistently large values for each water budget component, relative to the


52 other study wetlands. The differences in the absolu te values between wetland-specific flow rates (Table 3-1) are due to the di fferences in the stage-area rela tionship and corresponding volumeweighted hydrograph (Figur e 3-4) as well as the hd, which constrains the occurrence of ditch flow. The distributions of inflow and outflow rate s for each wetland were log-normal, and thus the geometric mean and corresponding geometric st andard deviation were used (Table 3-1). Trends in relative inflow and outflow rates were similar for all wetlands, with runoff (overland flow) and ditch flow exhibiting the highest values. Overland flow and ditch flow are important components in the conceptual model of these wetlands as a means to improve water quality discharged to Lake Okeechobee. Overland flow is particularly important because it has the potential to transport nutrients from minera lized, decomposing plants and surface-applied fertilizers. Relatively higher ditch flow ra tes were expected because a critical volume (corresponding to hd) of water must be present before water will discharge from the wetlands. Conceptually, these large outflow rates from the ditch are not only important in terms of quantity, but also because of the connectivity between wetland surface water (and associated water quality) and receiving water bodies. Cumulative frequency plots in Figure 3-5 show the distributions of normalized ditch flow (m d-1), which was scaled to the maximum ditch flow rate so that all wetlands could be displayed reasonab ly well on the same plot. Also shown in Figure 3-5 are the cumulative frequency distributions of water surface el evation in the ditches of the study wetlands. These were also included to id entify the occurrence of backwater influence in the ditches. All wetlands exhibit some frequency of standing water in the ditch in the absence of ditch flow, ranging from 2 to 15% of the time that ditch flow occurs.


53 Resulting groundwater inflow and outflow rates were higher than expected, being equal to or greater than ET rates. This contributes significant information to the overall conceptual model of wetland hydrology, as wetland surface water in teracts with groundwater in the surrounding upland landscape at significan t rates. Intrinsic to this interac tion is the water quality aspect and wetland treatment efficiency that is linked to these outflow pathways. In this study, it was important to characteri ze the wetland inflow and outflow rates of each study wetland to identify the role of each hydrolog ic component related to water and chemical transport. The total volume of water associat ed with each water budget component is also important to quantify because it offers information toward development of a conceptual hydrologic model. An important discovery relate d to the hydrologic behavior of these wetlands is the equally large sources of overland flow and rainfall. Rainfall would be expected to dilute chemicals that enter the wetland from outside sources, while overland flow (in these managed agro-ecosystems) may transport chemicals from the surrounding upl and areas. Therefore, the conceptual model of a flow-through wetland does not necessarily apply to these wetlands, as so little groundwater enters the wetlands. For ease of inter-wetland comparison, the per centage of total wetland inflow or outflow was reported, taken from cumulative flow rates at the end of each wetlands monitoring record (Table 3-3). Wetlands LW1 and LW2 (less than 1 km apart on the same ranch) exhibited similar contributions of inflows and outflows, even thou gh the inflow and outflow rates were higher at LW1, with ditch flow comprising approximately 50% of the total outflow. In addition, approximately 30% of the total outflow was attributed to groundwater outflow, with ET making up the remaining 20%. It is important to note that the groundwater component was significant, considering the hydraulic management to drain th e surface water via ditches. Differences in


54 bathymetry between LW1 and LW2 were not si gnificant enough to indepe ndently govern the relative contributions of inflow s and outflows. Corresponding hd for these two wetlands was also similar, which probably contributed to their similar hydrologic dynamics by constraining the occurrence of inflows and outflows to a similar relative degree. The fraction of rainfall, runoff, and groundwater to the total inflow at the Beaty ranch wetlands (BW1 and BW2) were similar to those at LW1 and LW2 (Table 3-3). This implies that the physical properties, such as soil hydrauli c properties related to groundwater inflow and infiltration mechanics related to overland flow, th at govern each inflow do not vary significantly enough to create large differences in relative fracti ons of inflow to the wetlands. Any subtle differences in inflow fractions between wetlands may be attributed to differences in wetland bathymetry as well as errors in the estimates themselves. The more pronounced differences in fractions of outflow from the BW1 and BW 2 wetlands indicated that probably both hd and wetland bathymetry had a significant role in determining their cumulative volume. Ditch outflow from BW1 only made up 17% of the total outflow, compared to 70% at BW2. BW1 had the highest hd, which resulted in relatively low ditch flow and exaggerated ET and groundwater outflow compared to the other study wetlands. One possible explanation for the relatively large volume of d itch outflow is that the wetland bathymetry exhibits relatively small surface area below hd and very large surface area above it (Figure 3-4). It was important to observe that in BW1 (the wetland least impacted by ditching) groundwater outflow made up 53% of the total outflow. Nutrient s transported through groundwater will require a longer time to reach a receiving water body and may be sequest ered by plants or immobilized by sorption to soil and organic matter, or precipitated through metal complexation processes.


55 Wetland hydrologic dynamics in a time-series format offer information that is also important for creating a conceptu al hydrologic model as well as determining consequences of hydraulic management related to nutrient treatment efficiency. To identify the dynamic behavior of the wetlands in the basin, two examples of hydraulic regimes, BW1 representing relatively low impact from ditching compared to LW1 which was more extensively ditched, were evaluated using cumulative AWV s (Figure 3-6). Overland flow to the wetland demonstrated a high intensity low duration behavior, correspond ing with large increas es in wetland surface water storage. Rainfall on the wetland water su rface was more frequent than overland flow, but exhibits lower inflow rates. Evapotranspirati on exhibited significantly sm aller flow rates than the other hydrologic components, partially because the wetland area scaling wetland volume was at or below hd 65% of the monitoring record (with stan ding water in the wetland) and partially because the ET rate itself was relatively lower than the other outflow rates (Table 3-1). Groundwater inflow only occurred under very constrained conditions, where the upland water table elevation was greater than the wetland surf ace water elevation (a we tland-averaged 24% of the time standing water occurs in the wetland). It only appeared to o ccur under relatively wet conditions and was limited by the rate by which water was evapotranspired from the surrounding upland plants that induced divergent outflow from the wetland surface water via groundwater. Unlike overland flow, high ditch flow rates from th e wetlands did not necessarily coincide with large increases in wetland volume. This is pr obably because rainfall/overland inflow events often occurred during times when the wetland contained li ttle or no water, resulting in a large pulse of overland flow to the wetland that, in tu rn, filled the wetland instead of flowing out of the ditch. Once hd was exceeded, the duration of sign ificant ditch outfl ow occurred over considerably long periods of time (in the order of weeks).


56 In addition to hydrologic dynamics in the time-series data, the frequency of each inflow and outflow was also reported in Table 3-2. The frequency of ditch flow at BW1 was less often compared to LW1, occurring only 9% of the hydr operiod versus 20% at LW1 (Table 3-2). Nonetheless, ditch flow in general only occurs over very short time intervals, but the amount of water generated during these outflow events is incredibly significant (Table 3-3). This short duration, high energy outflow may be closely a ssociated with distressing climatic conditions, such as high intensity and dur ation rainfall (e.g. tropical storms). Similarly, runoff to the wetlands was also less-frequent and much more s poradic than rainfall, occurring 15 and 27% of the hydroperiod at BW1 and LW1 wetlands respectiv ely. Parallel with ditch flow, the total volume of runoff was very important in the ove rall water budget, even though the duration was short. It is worth mentioning that the durati on of groundwater outflow was considerably longer than the ditch flow (87% of the hydroperiod at BW1 and 69% of the hydroperiod at LW1), but accounted for less volume (Table 3-3). Upland Area Contribution As overland flow was found to pl ay a critical role in the wetland water budget, it was im portant to identify the extent to which the wetland and upland are conne cted. The contributing upland area associated with overland flow was es timated for the study wetlands. In the water budget estimation, the volume of overland flow (seen in Equation 3-1) was sc aled to the wetland area, so it was necessary to re-scale those values to an analogous upland area. Once those values were re-scaled, a daily value for contributing area was calculated. To re -scale the overland flow volumes to the upland, a correlation between pr ecipitation and observed overland flow (water budget based) was developed to independently calculate a depth of overland flow (dover): ),()1( itit overPPaverage d (3-3)


57 where average( Pt=i,Pt=(i-1)) is the average value for rainfall for the ith time step and the previous time step and is a unit-less runoff coefficient. This approximation is related to the rational method, but it incorporates the idea of routed overland flow from an earlier rainfall events. The modeled overland flow captured between 50 and 60% of the variability of the observed overland flow. The population of fraction of contributing area to wetland area ( Fcont) exhibited a lognormal distribution. Geometric mean and standard deviation values Fcont and for each wetland are reported in Table 3-3. From this table, values of Fcont appear to be, on average, similar to the wetland area, however, the geometric nature of the distribution is worth analyzing in terms of daily values of Fcont, as seen in the cumulative density f unction(cdf) plot (F igure 3-7). From these data, it can be seen that between 45 and 60% of the time, the value of Fcont is less than one, indicating that the upland contributing area is less than the wetland area. It is worth noting that the frequency of overland flow for the entire monitoring record was between 22 and 33% of the time. This fact alone is an important finding, as this flow path contributes at least 50% of the inflow to these wetlands. Another im portant point is that the value of Fcont is highly variable for the remaining days when overland flow occurred, reaching above a value of 10 (Figure 3-7). Six snapshots in time of LW1 bathymetry and cont ributing upland area is shown in Figure 3-8 to better visualize the wetland area overland flow relati onship. In this figure, it is illustrated that larger rainfall amounts correspond to the relative contributing ar ea size, where larger rainfall depth corresponds to larger upland contributing area. Hydraulic Residence Time The hydraulic residence tim e ( HRT) or nominal residence time is defined for treatment wetlands as the system volume divided by the average of the inflow and outflow rates. In this study, the average of the inflows a nd outflows were not used to re present the flow rate in the


58 HRT equation. The study wetlands did not exhibi t steady-state flow rates through the wetland and were constrained by the sum of the outflows. Therefore, the residence time was determined by the sum of the outflows (ditch flow, evapotranspiration, and groundwater outflow). The volume term in the hydraulic residence time c oncept integrated the inflows from multiple sources (rain, overland flow, and groundw ater inflow). A distribution of HRTs was calculated based on the daily wetland volume and sum of da ily outflows. The distribution of hydraulic residence time for LW1 and LW2 we tlands were similar, with less than 30-day residence times occurring 55% of the time (Figure 3-9). The cumulative HRT distribution for BW1 was shifted, denoting generally longer residence times. At BW1, HRT s of less than 30 days occurred only 40% of the time. A skewed HRT distribution was noted for BW2, resulting in shorter overall residence times. This relatively dryer behavior at BW 2 was also seen in trends in hydroperiod, but the HRT analysis offers a quantitative compar ison of hydraulic performance between wetlands. Conclusions Overland flow and ditch flow are two of the m ost important components in the hydrologic regime of the study wetlands for several reason s. While their rela tive frequency was low compared to the total inflow and outflow, the flow rates and total volumes associated with these components are consistently greater than the othe r flow rates. Also, nut rient loads associated with overland flow may be inflated from the part iculate and labile forms of P that collect on the surface of upland soils from decomposing plant ma terial and cattle manure. These loads may be transported from considerable distances in the cat chment. Finally, the ditc h flow integrates the mixing processes from inflows and outflows and creates a rapid conduit of flow to receiving water bodies over considerable periods of time. The climatic conditions associated with the monitoring record in this study we re atypical, as four named hurricanes passed within kilometers


59 of the field sites. The resulting water budget was clearly not representative of long-term meteorological averages, but may instead represent the worst-case scenario of nutrient loading to Lake Okeechobee. Regardless of the hydraulic mana gement of these wetlands, it is likely that surface water connectivity between these pastures and the lake via ditches resulting from extreme, tropical storm caliber intensity/quantity rainfall events would create conditions for considerable nutrient loading. The HRT distributions for these wetlands appear to be quite long, since the hydroperiod of these wetlands was between 219 and 305 days. More work is needed to evaluate the overall treatment efficiency of these wetlands. A firs t-order uptake model paired with this residence time information may shed further light on the treatment efficiency of these wetlands. Furthermore, a larger system domain might be more appropriate for estimation or modeling nutrient treatment efficiency, because wet and dry cycles influence the availability and production of labile P within as well as on top of wetland and upland soil. The hydrologic dynamics of the study wetlands appear to offer some encouragement toward their use as treatment wetlands. Howeve r, further hydrologic management of the ditch outflow may increase the HRT and thereby promote nutrient up take and water retention. The BW1 wetland exhibited hydrologic dynamics th at may most closely reflect those of a hydrologically restored wetland, because its hd was the highest of all study wetlands. However, during extreme rainfall events, which occur relatively often in Florida in the form of hurricanes, surface water floods the entire pastures, crea ting a highly connected and hydrologically unmanageable conditions. Secondary treatment mech anisms, such as retention basins down-stream of these wetlands, may be essential to redu ce nutrient export to La ke Okeechobee for all scenarios of hydrologic conditions.


60 Table 3-1 Geometric mean flow rate of each water budget component and their associated geometric standard deviations Site Rain ET Runoff GWout GWin Ditch flow LW1 [m3 d-1] 39 10 26 4 164 7 88 3 69 4 226 6 LW2 [m3 d-1] 16 13 12 4 42 11 30 4 5 3 120 5 BW1 [m3 d-1] 18 8 15 2 48 22 22 2 11 4 52 7 BW2 [m3 d-1] 1 18 1 5 3 15 1 4 1 6 19 4 Table 3-2 Percentage of total number of days that each component occurred over the monitoring record. Also, the Predicted Ditch column represents the percentage of time that predicted ditch flow was calculated versus observed from the water budget Rain Runoff GWin ET GWout Ditch Predicted Ditch LW1 37 27 32 100 69 20 56 LW2 39 22 28 100 62 21 51 BW1 34 27 8 100 87 9 59 BW2 37 15 27 100 65 15 61 Table 3-3 Percentage of the total volume of infl ow or outflow of each water budget component with estimates of fraction of cont ributing upland area to wetland area ( Fcont) Rain Runoff GWin ET GWout Ditch Fcont LW1 36 59 4 19 27 54 6.0 0.92 1.3 LW2 42 57 1 17 26 57 3.7 1.17 1.5 BW1 50 48 2 30 53 17 3.7 1.03 1.4 BW2 53 45 2 19 11 70 3.6 0.77 1.2 Figure 3-1. Cumulative frequency distribution of wetland areas.


61 Figure 3-2. Comparison of daily wetland stage (storage) and water budget-estimated storage for Beaty wetland 2. Figure 3-3. Cumulative distribut ion functions of scaled vo lumes for the study wetlands


62 Figure 3-4. Stage-area relations hip for study wetlands (a) as well as their corresponding generalized bathymetric cross-sections (b). The solid vertical lines represent the locations of the critical ditch elevation. a b


63 Figure 3-5. Percent cumulative fr equency plots of scaled ditch fl ow and water surface elevation for the study wetlands


64 Figure 3-6. Cumulative infl ow and outflow volumes for BW1 wetland and LW1 wetland Figure 3-7. Cumulative density function for the fraction of contributing upland area to wetland area for the study wetlands.


65 Figure 3-8. Wetland LW1 contributing area an d corresponding rainfall for six days. Figure 3-9. Cumulative distribut ion function of hydraulic reside nce times for the study wetland


66 CHAPTER 4 SPATIALLY DISTRIBUTED ISOLATED WE TLANDS AS A TREATMENT SYSTEM FOR AGRIC ULTURAL RUNOFF WITHIN A WATERSHED Introduction In a rev iew of water quality in isolated wetlands, Whigham and Jordan (2003) report that the chemistry of surface waters in isolated wetlands is highly va riable, being commonly linked to the input water source. The input water source may be heavily influenced by the watershed area and position of wetlands within a watershed. Smaller watersheds ar e conceptualized to contribute relatively lower chemical loads to the wetland from their uplands as compared to larger watersheds. For example, in an agricultu ral setting, nutrient and pesticide loading to an isolated wetland is directly propor tional to the area of la nd that feeds it (Parsons et al., 2003). According to a study conducted in prairie po thole wetlands, lower su rface water salinity correlated to wetlands in higher topographic locations of a waters hed (Driver and Penden, 1977). Surface water quality in isolated wetlands has al so been correlated to the differences in underlying substrate thickness in the wetland and its chemical composition (Driver and Penden, 1977; Schalles, 1989; and Newman and Schalles, 1990). Natural wetland systems have long been considered to be a chemical buffer between uplands and receiving water bodies (Dortch, 1996; Reddy et al., 1999; Price and Waddington, 2000). It was the recognition of su ch a function that spawned the idea of municipal storm-water and point-source, agricultural, e ffluent water treatment using c onstructed wetlands (Gale et al., 1994; Raisin et al., 1997; Guardo, 1999; and Kov acic et al., 2000). In the United States, treatment of such point sources is widespread, but implementing this concept for applications to non-point diffuse sources is less straightforward. In the case of an isolated wetland used fo r treatment of contaminants, twoor threedimensional hydrologic and chemical monitoring may be necessary to estimate the treatment


67 efficiency and down-gradient chemical transpor t in surface and groundwater. A study by Shan et al. (2001) focused on an agricultura l watershed in China that incor porates a network of spatially distributed mature constructed wetlands and d itches through which overland flow is routed before exiting the watershed. To handle diffuse contaminant loading, a calibrated curve number method was used to estimate runoff contributio n to individual wetlands and the groundwater component was assumed to be negligible. Re tention of total phosphorus dissolved phosphorus and suspended solids in this multipond system was 93.9, 90.0, and 94.9%, respectively (Shan et al., 2001). This study suggest s that retention of contaminan ts from a non-point source can greatly increase with a network of isolated wetlands; however, accurate estimation of diffuse contaminant loading to isolated wetlands is sti ll challenging. Groundwater flows will provide an additional complicating factor fo r both water quality/quantity monitoring and system modeling. Additionally, seasonal variabili ty in hydrologic and clima tic conditions may complicate assessment of isolate wetla nd treatment efficiency (P arsons et al., 2003). Using a chloride tracer experi ment in a naturally occurri ng isolated wetland in Canada, Hayashi et al. (1998b) and Parsons et al. (2003) identified chemical cycling between the isolated wetland and the surrounding upland areas. In th ese studies, a barrage of water level and meteorological instrumentation was used to char acterize groundwater flow pathways and water balance inputs. Lateral groundwat er flow, evapotranspiration, a nd deep aquifer recharge were the only hydrologic components that occurred in this isolated wetland. It is completely severed from any surface water interactions with adja cent water bodies and, while not specifically commented on in their discussion, the down-gradient export of contaminants would likely change if the terrain were modified by d itching. Compared to contaminant loading in groundwater outflow, ditching may allow greater export of contaminants from the isolated


68 wetland due to relatively fast er hydrologic response time and reduced mean residence time within the wetland. The findings from these investig ators lend confidence to the idea that isolated wetlands are effective for contaminant removal of diffuse gr oundwater and surface water sources. Also, the degree of surface water connectivity and wate r movement to a nearby water body plays an important role in the removal rates of cont aminant inputs because mean residence time ( ) generally decreases as water velocity increases, as show n by (Mitch and Gosselink, 2000): Aq V (4-1) where V is the volume of water involved in the wetland (L3), is the water fraction in the flow media, and q is the specific discharge (L T-1), and A is the cross-sectional area of flow (L2). The objective of this study was to demonstrate the efficacy of using spatially distributed isolated wetlands for the treatment of P contaminated water within an agricultural watershed. A first order model was developed that simulated water treatment within a watershed similar to those located in the Okeechobee Basin. Simulations were preformed at va rious levels of wetland percent coverage, spatia l distribution of wetlands, and cont aminant degradation rate. These simulations show that spatially distributed wetl and may provide an effective means of treating contaminated water within a watershed. Model A first order contam inant degradation, steady state water flow model was developed in order to evaluate the efficacy of using spatial di stributed isolated wetlands for the treatment of non-point sources of water within a water shed. A general wate r balance for a watershed was written as:


69 QSPAETAGG td Vdout in (4-2) where V was the volume of water stor ed within the watershed (L3), t was time (T), Gin and Gout were the groundwater inflow a nd outflow rates, respectively (L3 T-1), A was the wetland area (L2), ET was the rate of evapotranspiration (L T-1), P was the rate of precipitation, S was the rate of surface run-on or runoff (L3, T-1), and Q was the rate of pipe flow into or out of the wetland (L3 T-1). The steady state water flow assumption require s that all inflow equal outflows. For this model it was assumed that there was no surface flow and no pipe flow into or out of the system. Additionally, ET and P were assumed to be equal so that ov er the long term they were balanced. Therefore, the groundwater flow into and out of the system was equal and the left hand side of Eq. 3 was 0. The chemical model was based on a chemical balance and was kCVACVGCVGCV td CdVwawoutoutwininw w (4-3) where C was the concentration of a particular contaminant within the watershed (M L-3), Vw was the volume of the watershed (L3), Cin was the concentration of the water entering the watershed via groundwater inflow (M L-3), Cout was the concentration of wa ter leaving the watershed via groundwater outflow (M L-3), Ca was the areal contaminant addition rate (M L-3 L-2 T-1), and k was the volumetric decay coefficient (T-1). Assuming that the watershed acts as a completely stirred tank reactor (CSTR) the c oncentration of contaminant as a function of time maybe written as kt a kt ine k AC eCtC 1. (4-4)


70 Obviously the assumption of the entire watershed acting as a CSTR is somewhat farfetched. In order to assure a degree of reality the watershed wa s divided into a regular grid. Each cell within the grid was treated as a CSTR. This breaking up of the watershed requires that the water flow through a single co lumn and not mix between columns. This assumption will only hold in areas with homogenous water potential energy field. This assumption results in a series of tanks that where an exact solution can be dete rmined by recursion. Because the degradation properties of th e wetland and non-wetland portions of the watershed were different it was necessary to determine what proportion of each cell was composed of wetland. This requires that the location and size of each wetland was known. To simplify the model it was assumed that each wetland was circular in plain view and semi-circular in cross section. This resulted in wetlands that were shaped like the bottom portion of a sphere. This allowed for the development of a stage surf ace area relationship so that the area of the wetland within the watershed could be determin ed for any wetland depth. The semi-circular assumption also allowed for the development of a stage volume relationship for each cell. This relationship used the average depth of the wetla nd and the surface area of the wetland within each cell to calculate the volume of wetland within that cell. When the total wetland volume was determined using the model it agreed very well with theoretical wetland volumes. Three scenarios were considered in a constant 1,000,000 m2 (1000 by 1000 meters) landscape area: 1) a single wetland within the la ndscape, 2) two wetlands within the landscape, and 3) a random distribution of six isolated wetla nds. The total wetland area was the same for all scenarios and was constrained to a maximum of 16% based on the average isolated wetland aerial coverage in the Lake Okeechobee Basin, FL (Flaig and Reddy, 1995). The single wetland simulation scheme was implemented to identify the degree of groundwater treatment in the


71 simplest case. The center of the isolated we tland was placed in the geographic center of the landscape (Figure 4-1). It also served as a benchmark for comparing multiple-wetland scenarios of varying spatial arrangements. The two-wetland arrangement was implemen ted to explore groundwater treatment of various spatial arrangements of isolated wetlands in the landscape. In the first of five simulations of two-wetland arrangem ents, the isolated wetlands were offset in the y-direction to avoid overlapping of individual we tland areas and were just far enough apart in the x-direction so that no treated water from the first sequen tial wetland would enter the second. Preserving the offset in the y-direction in the following four simulations, the wetlands were systematically moved closer together in the x-di rection to allow more treatment overlap in the flow-field with each simulation until the second we tlands water flow pathway eclipsed the first (Figure 4-2). One simulation of six randomly distributed isolated wetlands was performed, maintaining the same maximum aerial coverage of 16 % (Figur e 4-3). Results from this simulation were used to compare the relative tr eatment efficiency to previously discussed arrangements. Treatment efficiency (Tr) was used to evaluate watershed-scale P retention: % C C Tro100 1 (4-5) where C is the total P concentration along the watershed outflow boundary and Co is the total concentration added to the watershed for th e duration of the simulation. The term C/Co is the residual normalized concentration at the watershed outflow boundary. Two model parameters were varied w ith each simulation: 1) wetland area ( Aw), calculated as a function of depth of water in the wetland(s) and 2) decay rate ( kv). The wetland area was incrementally increased from 0.01 % of the tota l watershed area to 16 % by increasing the level of water in the wetland from 0 to 1 meter depth. This depth is consistent with typical depths of


72 isolated wetlands in the Lake Okeechobee Basin. The kv was varied from 0.1 to 10 yr-1 for each incremental increase in Aw. Gale et al. (1994) reported kv values for the constructed wetland treatment system in Orlando, FL between 12.8 and 26.3 yr-1. Sompongse (1982) found kv values ranging between 18 and 29 yr-1 for Florida soils receiving agricultural drainage waters. Because low groundwater flow rates were used in the simulations, the treatment of P was fairly insensitive to the kv. A somewhat lower range of kv was used relative to the values reported in some literature in Florida systems in an effort to offset the slow groundwater component and simulate a system in relative non-equilibrium. The Damkohler number ( Da) of a single wetland covering 16 % of the total watershed area was used to identify the relative importance of variables that control treatment: wvakD, (4-6) where kv,w is the decay coefficient for a wetland and is the mean residence time. Results and Discussion The high end of the range of kv values used in the simulations in this study is similar to those reported by other inves tigators (Sompongse, 1982; Gale et al., 1994). These relatively higher kv values resulted in equilibrium decay cond itions that may not reflect rate-limited P dynamics that are usually manifest in natura l systems. Therefore, the low range of kv values were included in the simulations to create more rate-limited P decay, which placed more emphasis on the hydraulic contact area of the wetlands. Although this emphasis is more apparent at low kv values in the assigned range, increased wetland hydraulic contac t area was generally correlated to higher P treatment. The percent treatment for the single isolated wetland simulation (first scenario) ranged from 0 to 20 of the annual P addition, which was 0.841 g L-1 m-2 yr-1. While the percent


73 treatment, kv, and fractional wetland area (Aw) appear exponentia l related, the Aw and kv are of near-equal influence for watershed-scale groundwate r P treatment (Figure 4-1). This type of relationship between percent treatment, Aw, and kv persists for all simulations of different numbers and arrangements of isolated wetlands, but the relative range of treatment varies slightly. For the second scenario of five simulations of two-wetland arrangements, the first arrangement (Figure 4-2A) yielde d the highest maximum P treatment of 28 % (Figure 4-1A). The following four wetland arrangements (Figure 4-2B, 4-2C, 4-2D, and 4-2E) resulted in increasingly smaller maximum P treatments of 27, 25, 23, and 20 % respectively (Figure 4-4BE) and it appears that the kv plays a slightly more significant role (especially closer to 16 % wetland coverage) as wetlands are arranged so th at individual wetland treatment is distributed more evenly across the outflow boundary. Analysis of the two-wetland arrange ments (Figure 4-5) shows that at kv equals 0.1, the length term embedded in w (Equation 1), representing the wate r flow path through wetland grids dominated the Da. This results in an increase in treatment efficiency with increasing fraction of overlap of the two wetlands. The Da generated from a kv of 0.1 with the maximum and minimum wetland aerial coverage was 23 a nd 13 respectively and represent equilibrium P decay processes. Extremely high equilibrium decay conditions were observed (Da>>100) when kv approached 10. In this case maximum treatment is observed at zero over lap. Evaluation of wetland overlap a nd P treatment at high and low kv values adds insight to the importance of the relative geograp hic position of isolated wetla nds in the landscape. Wetland overlap was defined as the radial overlap of the two wetlands in the flow direction divided by the wetland diameter (as both wetlands were the same size). Simulations using relatively lower kv


74 values show that, while the treatment difference between scenarios of complete or no wetland overlap is small (~4%), more wetland overlap results in better treatment because the water treatment flow path controls the Da. Simulations involvi ng relatively higher kv values revealed that higher P treatment was observed when the tw o wetlands were not over lapping in the flow direction. In these scenarios where instantaneous P decay existed, the change in length of the wetland flow path was not significant enough to produce differences in treatment from a single wetland scenario. It was more important to ha ve more of the total volume of water passed through the landscape treated, in the scen ario of wetlands th at did not overlap. The third and final scenario of six rando mly distributed isolated wetland simulation estimated maximum P treatment across the outflow boundary to be 28 % (Figure 4-3B). Spatial distribution of P concentration for each grid cell throughout the watershed was also evaluated (Figure 4-3B). When water enters a wetland, immediate and almost complete treatment is achieved, but after exit ing the wetland, the concentration of input P from cows significantly influences the groundwater P concen tration. This amount of watershe d treatment is similar to the scenario made up of two wetlands, offset in th e yand x-directions, such that no overlap occurred in the flow field direction (Figure 42A). Higher P treatment efficiency corresponding to isolated wetland arrangements that overlap less in the x-direction. Conclusions No regulatory agency or action have listed criteria that define a wetland as isolated, therefore, various definitions emerge in the liter ature that depend on the discipline defining it and its scientific application. For all intensive pur poses, we conclude that the geographic definition offered by Tiner (2003a) suffices because it provi des consistency for regu lators and scientists while applying commonly used soil and vegetation parameters wit hout the need to intensively characterize hydraulic connectivity to receiving water bodies.


75 Modeling P reduction in an isolat ed wetland landscape provided insight into their utility to be used as treatment cells for agriculturally impacted groundwater. In a single isolated wetland system, at the maximum aerial coverage, and combined with a relatively high kv reduction of P in groundwater discharge along the outflow boundary was approximately 20%. Reduction of P in the landscape increased to 28% for simulation of a system of two isolated wetlands with the same fractional area of isolated wetlands that di d not overlap in the dire ction of flow. But, increased overlap in the flow di rection corresponded to decrease d P reduction. Modeling efforts to estimate P treatment using spatially distributed isolated wetlands in a landscape show that typical conditions of the Lake Okeechobee Basin are favorable for an effective means of reducing P from agriculturally impacted waters. Figure 4-1. Example of wetland treatment anal ysis. A) Geographic pl acement of the single isolated wetland. B) Treatment efficiency as a function of the fractional wetland aerial coverage and decay coefficient. B


76 Figure 4-2. Wetland geographic pos itioning for five simulations of increasing wetland overlap in the flow direction (from top to bottom). A) Zero wetland overlap. B)-E) increasing degree of wetland overlap. Figure 4-3. Random wetland field and corresponding treatment. A) Treatment efficiency of the random wetland field as a function of the fractional wetland aerial coverage and decay coefficient (kv). B) Plan view of random wetland field. A B E D C A B


77 Figure 4-4. Treatment efficiency of a two-wetland system with increasing overlap. A) treatment efficiency as a function of fractional wetland aerial coverage and decay coefficient for no wetland overlap (kv). B)-E) increasing overlap of wetlands. A B E C D


78 Figure 4-5. Fraction of wetland overlap for a two wetland system for relatively high and low kv of 10 and 0.1 respectively.


79 CHAPTER 5 PHOSPHORUS BUDGETS OF DEPRESSIONAL WETLANDS Introduction Estimating P dynamics in wetlands can be quite difficult because of the complex nature of processes involved in the P cycle and the number of parameters that must be defined to develop a conceptual or mathematical model. Reddy et al. (2005) gives a comprehensive description of P biogeochemistry in wetland systems. In short, the P cycle in wetlands has some important distinctions. First, the re dox conditions influence the solu bility, and thus, the solution concentration of P (Ann et al., 2000). Dependence of P specia tion becomes more important in wetlands because pH and electri cal conductivity (Eh) may vary dr amatically from hour to hour and season to season. In addition, chemically reducing conditions in wetlands are common and create more ideal conditions for low Eh (increased P solubility), but also mobilizes iron and aluminum which bind P in the form of precipitates. Secondly, total organic carbon (TOC) has also been shown to be a considerable P-binding agent in wetlands (Reddy et al., 2005). This becomes important because of the continuous accretion rate of new organic matter to the wetlan d from dying plants and microbes. In wetlands, microbes largely mediate active cycling of orga nic P by influencing mineralization rates through metabolic processes. Furthermore, wetland microorganisms alter ionic composition, redox, pH, and other conditions that affect P binding efficiency to particulate organic matter, clays, and other soil materials. Finally, typical wetland vegetation interacts with the P cycle in a distinctive manner. They are a significant P sink via root uptake in wetland pore water, but seasonally also contribute to the P load upon death and senescence. This se nescence may release 20 to 50% of the total P content within hours and 65 to 85% over longe r periods (Wetzel, 2001). They may create an


80 oxygenated layer at the soil-surface water interface, which affects the use mineralization rate of P. The net long-term effect of wetland plants is sequestration, once P is incorporated into the vegetation/microbe cycle (Reddy et al., 2005). Early studies of P budgets input and output indicated that net build-up of P in for the northern Lake Okeechobee watershed indicated that net build-up of P threatened to impede or halt lake restoration (Fonyo et al ., 1991; Boggess et al., 1995). Boggess et al. (1995) estimated TP runoff from agricultural land in the basin from major land use types and mean annual concentrations that were calculated from measured water quality data collected by the SFWMD. Using a GIS framework, these estimates were applie d to the entire basin to estimate loading rates to the lake from the basin. In this early study, nearly 20 % of the total P mass input to the basin was estimated to be specifically retained in wetlands. Between 1985 and 1989, the average annual P load to the lake was approximately 300 metric tonnes per year. By 2001, the P loading to the lake had increased to 582 metric tonne s per year (SFWMD, 2001). The South Florida Water Management District (SFWMD) implemented the Surface Water Improvement and Management plan and Lake Okeechobee Protecti on Program to develop water quality objectives (SFWMD, 1997, 2002). It is important to note that the agricultural land us es itself does not solely contribute to the total P load to the la ke. For example, surface water discharge from tributary sources, such as the Upper Kissimmee Chain-of-Lakes and Istokpoga, were reported to be increasing in the last decade to over 140 me tric tonnes per year (Harvey and Havens, 1999). Combined, a significant P load originates from the upper and lower watershed, where estimates of P loading rate are approximately 198 metr ic tonnes per year above the target (SFWMD, 2001).


81 Zhang et al. (2002) evaluated the potential P load reductions under the Lake Okeechobee regulatory Program using a distributed mass ba lance approach coupled with GIS and soils information. These authors reported P load re ductions for two scenarios: 1) P discharge was reduced to limits within the regulatory program (0.35 mg/L in runoff water), resulting in a 68 % reduction of P load to the lake and 2) an areal P loading rate of 0.114 kg/ ha (typical of natural rangeland land use), which yielded a 95 % reduction in P load. These results do not specifically address the means by which the load reduction will occur, but only that they may occur. Since these modeling efforts do not account for the proce sses that control the re ductions in P export, the time-scale and feasibility as sociated with the proposed P e xport equilibriums are vague at best. Hiscock et al. (2003) evaluate d the P budget in several Lake Okeechobee priority basins to characterize differences in P loading from ch anges in best management practices (BMPs) between 1991 and 2003. Using a simplistic black -box approach, the role of wetlands in the watershed for reduction of P was considerable, with wetlands stor ing approximately 32 % of the runoff P from the watershed. This study did not explicitly evaluate the role of wetlands and water features within the catchment, but simply took the difference of ins and outs and attributed any P loss to wetland uptake. Under ne w BMPs (primarily related to changes in land uses), net P imports to the watershed were re duced by 28 %. However, improved pasture land and dairies remained considerable contributors of net P input to the watersheds (reduced from 91% to 60% of the total). The authors of this st udy noted that a more efficient means of P load reduction to the lake might be achieved by decreasing P runoff from the watershed. The objectives of this study were to devel op a P budget for these depressional wetlands, determine potential export to receiving water bodi es, and identify the treatment potential of these


82 types of wetlands using a first-order uptake ap proach (e.g., Kadlec and Knight, 1994). This study will describe the extent to which these wetlands may currently function for retaining P. The current retention dynamics of these wetlands may help to determine their utility for landscape-scale P reduction as well as their role among all efforts at the basin-scale to reduce P load to the lake. Finally, these results may offe r insights into the future role of these wetlands under hydraulically and chemically managed scenarios. Treatment potential determined from the P budge t will also be compared to the P transport results from the two-dimensional landscape-scal e distributed hydrologic m odel of Perkins et al. (2005) which also incorporated first-order uptake processes. That study indicated that wetlands may reduce landscape-scale P loading between 19% and 28%, assuming steady-state flowthrough the wetlands. However, this assumpti on likely does not adequately represent the hydrologic behavior of these systems. In th is study, hydrologic infl ows and outflows were derived from a water budget approach (Chapter 2). Hydrologic Inflows/Outflows During the two years of productive data colle ction, four significant tropical depressions affected the site hydrology. While tropical dist urbances are a reasonable phenomenon to include in hydrologic studies of any nature (in this part of Florida), the number of large storms that occurred during the data collecti on for this study was atypical to sa y the least. However, this data may represent the worst-case scenario fo r nutrient runoff, which may be desirable in design parameters of water and P control strategies such as retention basins. Hydrologic data collection and water budget component calculation is described in detail in Chapters 2 and 3. A brief summary of that research is presented here. The depth-based water budget for these de pressional wetlands was defined as: dHwet = P ET D + S GW (5-1)


83 where dHwet is the change in wetland surface water storage, P is rainfall, ET is evapotranspiration, D is surface water outflow (through ditches), S is overland flow, and GW is groundwater inflow or outflow. Rainfall and dHwet were calculated from direct measurement, while ET was estimated using an empirical function using meteorological data (Allen et al., 1998). Before the 1950s, wetland surface water was hydraulically disconnected from receiving water bodies (Flaig and Reddy, 1995). Once cow-cal f and dairy operations started, the forests and cypress domes were converted to pasture la nd. Along with this land use conversion, the hydraulic management of pastures emphasized drie r pasture conditions. Ne tworks of ditches (no more than one meter deep) were created to drai n surface water from the pastures and maintain lower groundwater levels. These ditches are not completely effective, as the degree to which they were excavated was often more shallow than wetland. This inefficient ditching was intentionally created to yield a semi-drained hydraulic management regime, where some surface water could be used for cattle cooling ponds a nd drinking water, while increasing the land area available for grass. Because of the low gradients in the watershed (<1% typical slope), water control structures were difficult to implement, due to backwater effects and inadequate head-loss. The ditch flow out of the wetlands was left as a residual term in the water budget equation. During times when ditch flow occurred and rain fall (and by definition over land flow as well) did not, equation 1 was used to determine the ditch flow (as ditch flow was the residual term). Mannings surface water flow equation was calibrated using these data to fit a ditch slope that best described the ditch flow estimates from eq uation 5-1. Thus, ditch outflow became a known parameter. Ditch flow was a significant portion of the total outflows (approximately 50%).


84 Daily groundwater outflow from the wetland surface water to the uplands was determined as the residual term in a constrai ned form of equation 5-1, where only ET and dHwet occurred. These groundwater outflow values were used in a linear regression model to determine the best fit of a hydraulic resistivity coefficient as part of a linear flux law (a form of Darcys law). The calibrated linear flux equation was then applied to all other time s in the monitoring record and assumed to be a known component. The same resistivity was also applied to groundwater inflows, assuming isotropy. Groundwater flow between the wetland and upland was assumed to occur radially, because differences in upland water table elevation around the perimeter of the wetland were not considerably different. Gr oundwater outflow from the wetland represented approximately 30% of the total outflow from the wetland. Surface runoff to the wetland, or overland flow, wa s not explicitly measured in the field. This term was the unconditional residual term in equation 1. It was also one of the largest hydrologic inputs to the wetland, which was important in the overall conceptual model of these wetlands (approximately 50% of the total inflow to the wetland). The initial assumption was that the wetland was a flow-through wetland, receiv ing principally regional groundwater flow. Phosphorus Budget Each water budget component had an associated chemical counterpart, as described by the P budget: GW ET D S P WWMMMMM CV d t d (5-2) where VW is the wetland water volume (L3), CW is the P concentration in the wetland surface water (M L-3), MS is the P flux in overland flow (M T-1), MGW is P flux in groundwater to and from the wetland (M T-1), MP is the P flux in precipitation (M T-1), and MD is the P flux out of


85 the wetland via the ditch (M T-1). Assuming the firstorder uptake kinetics, (M T-1) was defined as: 2 exp*)(C H CCwet init (5-3) where Cinit is the initial concentration in the wetland surface water (M L-3), C* is the equilibrium surface water concentration (M L-3), is the first-order uptake rate coefficient (L T-1), is the mean hydraulic residence time (T), and Hwet is the wetland surface water elevation (L). In most treatment wetlands, the wetland depth is a constant value, as inflows and outflows are regulated. In this more natural setting, it was an essential co mponent in the treatment model because it varied significantly as a function of time. The P concentration in the wetland surface water was measured from grab samples collected approximately monthly over the study period and more intensely for the last few weeks of the study. An automated water sampling device was deployed for the last several weeks of the study period and samples were collected every three days at midnight Groundwater samples were drawn from 2-m deep, fully-screened groundwater monitoring wells along the perimeter of each wetland at a similar time interval as wetland surface water sampling. Total P values for ditch outflow were averaged from water qualit y samples from the ditch during and after runoff events. Total P was measured in the laborat ory according to IEPA method 365.1 (USEPA, 1992). Concentrations of TP in rainfall P and runoff were based on those reported by Hiscock et al. (2003) in the same basin. Of course some va riability in overland flow concentration would be expected, depending on land use and intensity of chemical management. Hiscock et al. (2003) reported that P concentration in overland flow from high intensity dairy farms may exceed 10 mg L-1, however, for improved pastures, such as the ones in this study, a more appropriate overland


86 flow P concentration wa s reported as 1.32 mg L-1. This value was used exclusively for calculations in the current work, as no values were measured onsite. One of the earliest reported wetland models wa s presented by Kadlec and Hammer (1998). This self-proclaimed simple model was too extens ive for application to this study, implementing just less than 50 input parameters. Thus an even simpler approach was taken to characterize the P dynamics. First-order uptake kinetics were used as it is very simple to apply and does not require in-depth knowledge of th e wetland P cycle processes and dyna mics. First order kinetics, as represented by the k-C* model have been used extensively to model treatment wetland dynamics (Kadlec and Knight, 1994). For the purposes of this research, and the amount of data available for comparison of model accuracy, this first-order approach was satisfactory. This first-order approach is a lumped parameter approach and may be quite variable, with fit values of varying 100-fold. An assumption in this model is that there is a constant P equilibrium in the surface water to which, depending on the rate cons tant, the net system behavior will converge. In reality, an equilibrium phosphorus concentratio n in the surface water is probably not constant in time, because of P mineralization and organic matter accumulation that are dependent on the site hydrology. However, this first-order ap proximation of P uptake has been applied with success in modeling chemical uptake in treatment wetlands (Dortch, 1996; Kadlec, 1999; Srodes and Normand, 1999; and Wang et al., 2006). k-C* Modeling Using equation 5-3, a generalized Reduced Gradient nonlinear optimization code (Microsoft Excel Solver) was used to determine the best fit between and wetland TP concentrations while varying the uptake coefficient, over the entire time period. A value of 0.05 mg L-1 was initially used to represent C*, which was the mean P equilibrium concentration


87 measured by Plant et al. (2002) in a newly constr ucted treatment wetlands in a wetland in the okeechobee basin that was proposed to be converted into a stormwater treatment area. However, onsite wetland water-soil P information were also in corporated into this study to evaluate a range of resulting treatment efficiencies. Onsite soil column studies, it was reported that values for equilibrium P concentration (EPCw)were between 0.12 and 0.57 mg L-1 at the Larson wetlands and 0.12 and 0.28 mg L-1 at the Beaty wetland (Dunne et al. 2006). Equilibrium P concentration is analogous to equilibrium P con centration at zero sorption (EPCo), where adsorption equals desorption. These values were determined fr om a 28-day period as Pspiked water ponded on the soil core (0.1 mg L-1) was allowed to equilibrate with the underlying soil. In a second trial in the same study, those investigators used a hi gher P-spiked water con centration (1.0 mg L-1) over the same time interval to evaluate differences in resulting EPCw values. Significant differences were observed in the resulting EPCw values, where the second trial yielded values between 0.79 and 1.30 mg L-1 at the Larson wetlands and between 0.84 and 1.30 mg L-1 at the Beaty wetlands. This first trial yielded values that were rela tively higher than those reported by Plant et al. (2002), but were on the lower end of the spectrum in the pertinent literature. The second trial was of particular interest in the current resear ch because it represented a 60-day wet antecedent soil-water condition, which is co mmon for the study wetlands and al so resulted in significantly higher EPCw values. The onsite values from the sec ond trial were also incorporated into the analysis of treatment effectiveness using the k-C* modeling approach. A value of 1.3 mg L-1 was used for Cinit, which is a value of TP concentrations in runoff from improved pastures in the Okeechobee basin re ported by Hiscock et al (2003). In reality, the concentrations associated with runoff from each wetland was probably different, however, based on the information available, the value reported by Hiscock et al. (2003) was used. The Larson


88 ranch was been more intensely managed compared to the Beaty ranch, and would therefore be expected to exhibit higher concentrati ons of P in overland flow. Because EPCw values observed at the sites for saturated initial conditions were relatively close to the value assigned to P concentration in runoff (Dunne et al., 2006), the treatment effectiveness of these wetlands would likely be reduced significantly from what is reported in Table 5-3. From the P budget dynamics in Figure 5-1 and 5-2, these wetlands may behave as sources of P, rather than sinks. The k-C* modeling effort also reflects this, as the EPCw values of BW1, BW2, and LW2 are all greater than the average wetlands P concentration. Because of the non-linear and transient behavi or of the wetland residence time behavior, it was important to incorporate the specific wetland surface water residence times () for each day that P treatment effectiveness was estimated. The values for were the nominal residence times (volume of wetland divided by the sum of th e hydrologic outflows) derived from daily calculations using the wate r budget (Chapter 3). Results and Discussion Phosphorus Budget Average concentrations of TP measured in surface water, gr oundwater, and ditch flow at each study wetland are reported in Tabl e 5-1. It is important to note from these data that the surface water and ditch flow concentrations were not significantly different. The groundwater concentrations were however significantly diffe rent from both surface water and ditch water TP concentrations. Results from th e P budget are reported in several ways (Table 5-2): percent of the total TP mass input/output, to compare the relative P fluxes over the study period, mean and coefficient of variation of daily flux for each co mponent, to identify important loading rates and their duration, and TP fluxes are scaled to th e Lake Okeechobee basin to compare loading rates from the sites to observed loads to the lake. While scaling up P loading results was not


89 completely within the scope of this study, it wa s useful to identify comp are linearly scaled P loading estimates from this study to larger scale TP loading estimates to the lake (derived from multiple sources within the basin (Boggess et al., 1995)). Cumulative mass of P (g) from the wetland P budgets at Larson a nd Beaty wetlands are shown in figures 5-1 and 5-2 respectively. Al so shown in these plots is the uptake/release variable ( ) from Eqn. 5-2, but was calculated as th e residual term in the daily P budget and not by the k-C* model given in Eqn 5-3. The dynamics of each component are important to identify, as well as any differences between wetlands. The temporal dynamics of the P budget were driven by the hydrology, as mean P concentrations were used to represent all times of each component. For example, P mass a ssociated with overland flow (Mover) exhibited periodic and relatively infrequent behavior, compared to mass of P corresponding to groundwater flow out (Mgwout). Mass of P associated with rainfall (Mp) was slightly more frequent, compared to Mover, but was also event-driven. Perhaps a more interesting observation is that is temporally variable, where a wetland might behave as a P sink and then a source, depending on the dynamics at the time. Phosphorus temporal dynamics at LW2 appear to be quite different from the other wetlands, as the mass P associated with ditch flow (Mditch) nearly parallels the Mover over the entire record. It is easy to see that, with the other components of the P budget do not significantly contributing to the overall P dynamics, P that enters the wetland via overland flow exits the ditch almost immediately. This infl ow/outflow dynamic appears to be damped in the other wetlands. The amount of P input to the wetland was overwhelmingly dominated by overland flow (Table 5-2). While water volumes of overland flow and rainfall to the wetlands were similar


90 (approximately 50% each) the higher P concentra tion associated with overland flow was the driving force of P loading to the wetland. It is now clear that better estimates of P loading to the wetland via overland flow are necessary to bette r determine the P budget, however, the results from this study will build the conceptual framework for such an analysis. While groundwater outflow comprised approxima tely 35% of the total hydraulic outflow from the study wetlands, the principal component responsible for the majority of P export was ditch flow, which yielded between 7.6 and 43.7 % (wetland-averaged) of the total TP outflow. This finding is critical in the development of a conceptual model of the P dynamics in these wetlands because the treatment effectiveness is a function of not only what comes out the ditch as surface water export, but also th e load to groundwater that will ultimately arrive at the lake. The wetland BW1, which acted most like a rest ored wetland, was highly influenced by the amount of P load coming to the wetland associ ated with overland flow, but also had a considerably larger uptake component compared to the other wetlands (83% of the total outflows). Total P loads associ ated with rainfall and groundwa ter inflow (which represented considerable volumes of wate r in the water budget) were not significant pathways, only accounting for between zero and 2.6 % of the tota l TP mass inflow to the wetland. The opposite was true for groundwater inflow, concentrations of P in the pore water were considerable, but the occurrence of groundwater inflow was not high enough to transport comparable amounts of P to the wetlands. Daily loading rates of P for each wetland flow component (Equation 5-2) were important to characterize, because the variability of the system dynamics may be inferred. Also, when paired with the occurrence and timing of each component, these data may be useful for determining appropriate management strategies aime d at reducing P export from these fields. It


91 is obvious that special attention ne eds to be given to the relatively rapid nature of ditch flow and overland flow when implementing management strate gies to reduce P loading to the lake. While these events are not as frequent as the other components in the ch emical budget, their associated P flux rates are at nearly an order of magnitude larger. Extrapolating these data to the basin scal e is a daunting task, but some idea of the estimated P loads from the basin may be used to compare with estimates of P export from this research. Using the product of the wetland ditch ouflow rates from Table 5-2 and the area of the Okeechobee basin corresponding to isolated wetlands (16%), an estimate of basin-scale P load was calculated. Values between 0.2 and 9.6 metr ic tonnes per year were estimated represented a significantly smaller estimate compared to basin-scale estimates of 415 tonnes yr-1 (Boggess et al., 1995) and 582 tonnes yr-1 (SFWMD, 2001). The estimate in this work assumes that all historically isolated wetlands function hydrologically similarly within the landscape and have had similar historic and current la nd use practices (fertilizer, cattle st ocking, etc.). This estimate of basin-scale load is only pertinent to the fraction of runoff passing through depressional wetlands and out ditches. It does not incorporate P load related to regional drainage from ditch networks, land use contribution of dissimilar nature (dairies and other re latively high-intensity operations), and P load from tribut ary and in-stream internal loads. The TP treatment effectiveness listed in Tabl e 5-3 is derived from the sum of chemical outflows divided by the sum of the chemical inflows (similar form as in equation 3-5). When the term treatment is used in this document, it is somewhat different from the strict definition of the word as commonly used in analysis of cons tructed wetlands. Constructed wetlands do not usually consider the export of P via groundwat er discharge from the wetland, as they are designed to allow minimal groundwat er export. However, in the depressional wetlands in this


92 study, groundwater out flow may be considerable and carry a signi ficant P load. When the term treatment is used, is it the sum of the outflow s (via ditch and groundwater ) divided by the sum of the outflows (rainfall, overland fl ow, and groundwater inflow). Therefore, the treatment is a measure of not only the amount of P exiting the ditch, but also wh at is exiting with groundwater. The estimates of treatment woul d increase if groundwat er outflow was not considered in the term, but would not accurately or strictly represent the hydrology of the systems. These data suggest that thes e wetlands may sequester P, even though several extreme hydrologic events occurred during the data collection that may ha ve skewed the results. An average of 8 % of the total P inflow was assimila ted in the Larson wetlands, while an average of 62.5 % of P entering the Beaty wetlands was retain ed (and assumed to be sequestered) (Table 53). For BW1, this might be due to the fact th at ditch flow occurs less frequently, allowing P chemical transformation and sequestration more productive. Also, wetland vegetation might be more productive under the more predictable (i.e slightly wetter hydrol ogic conditions) at the Beaty wetlands. The sensitivity of the P budge t to input concentrations for ditch flow, groundwater inflow/outflow, and overland flow was evaluated to identify the degree of variability associated with the results of this methodology. Wetland 1 at the Larson ranch was used as an example of the sensitivity of the wetlands. Each P concen tration of the aforementioned flows was varied individually (one-at-a-time method, Monod et al., 2006) bounded by the limits of their corresponding standard deviation. Percent treatment effectiveness (Cin/Cout) was re-calculated to determine the degree of sensitivity of each P conc entration in the budget. Using mean values of measured TP concentrations for ditch flow and gr oundwater as well as the va lues for TP in rain and overland flow reported by Hiscock et al. (2003), the Cin/Cout for LW1 was 55.9%. When


93 CGW was varied between its coefficient of variability (0.5 .9 mg L-1), the percent treatment efficiency was -4.2 to 24.5% (where the negative value may be interpreted as a net load to the wetland surface water). As the concentrati on of surface water runoff to the wetland (Cs) was varied between a standard deviati on similar to that of measured TP variability in ditch flow (1.3 0.5 mg L-1), Cin/Cout was 39.7 to -75.9%. Varying ditch fl ow concentration using the range of the standard deviation (1.2 0.5 mg L-1), Cin/Cout was -11.6 to 32.0%. From this simple analysis, it may be seen that the TP concentration associ ated with the overland flow component is the most sensitive parameter. This analysis also sh ows that the net response of these wetlands might be to sequester P or to expor t it, depending on the hydrologic conditions. The other wetlands exhibit similar behavior in sensitivity (Table 5-4). A more time-inte nsive monitoring program needs to be done to better determine the times of sequestration and export of P. k-C* Model The optimized values of C*, Cinit, and for each wetland, as determined from equations 52 and 5-3, are shown in Table 5-3. The model fits were evaluated based on comparisons to the time series of measured TP concentrations in the wetland surface water (Figure 5-3). Root mean square error (RMSE) of the modeled and measur ed surface water concentrations for LW1, LW2, BW1 and BW2 was 0.35, 0.32, 0.05, and 0.58 mg L-1 respectively. This level of error in prediction was acceptable for the objectives of th is study, given that mean measured surface water concentrations were generally larger th an their corresponding erro r (Table 5-3).. Treatment performance was significantly diffe rent for each wetland, ranging from 0.22 to 0.97 (Table 5-3). Wetland 1 at the Beaty ra nch (BW1) exhibited the highest degree of P treatment. This was due to the limited ditch ou tflow of P, which provided longer residence times that likely aided P uptake by vegetation and microbe s. Assuming that the P load to the wetlands were similar, BW1 also exhibited relatively higher intrinsic uptake rates, compared to the other


94 study wetlands. Uptake rates might be attributed to differences in soil physical and chemical properties (i.e. organic matter, ir on, aluminum clay content) a nd vegetative species distribution and corresponding species uptake ef ficiencies. The values of upt ake coefficients in this study seem relatively low compared to those reported to be typical for Flor ida (Reddy et al., 1995). The treatment performance of the study wetlands were more va riable than those estimated from chapter 4 (spatially-distributed flow-thr ough wetlands). The idealistic nature of the modeling in chapter 4 may mask the actual we tland treatment performance and associated variability. However, while estimates of treatment potential in this chapter indicate higher variability between wetlands, these estimates may also fall somewhat short of the actual treatment dynamics. While RMSE was reasonable for the model estimates of observed TP, the observable variability in the concentration of wetland surface water P was perhaps not adequately frequent to determine strong mode l correlation. Also, the lack of good strong correlation of modeled and observed TP values might be due to the assumption that the equilibrium P concentration (C*) is constant over time. The measured values of TP concentration in the wetland surface water was not pa rticularly variable in time, which resulted in a rather small uptake coefficien t (limiting the variability of the modeled TP). This may not be the case for these wetlands, especially give n their highly dynamic hydrology, vegetation, and agricultural management. Conclusions A P budget was estimated for four depressiona l wetlands in the Lake Okeechobee basin to identify important P flow pathways and quantif y their effect on downstream water quality. Phosphorus associated with overl and flow was the dominant mass input to these systems, where P from groundwater inflow and rainfall combined accounted for less than te n percent of the total P influx. Groundwater outflux of P was proportional to the P outflux associated with ditch


95 outflow in two of these wetlands, suggesting that groundwater outflow is an important part of the conceptual P budget model. Phosphorus export associated with ditch outflow was either comparable to or more significant th an groundwater outflow for each wetland. The effectiveness of the st udy wetlands to reduce P load into the system was only significant for BW1 wetland, where ditch flow P was constrained by the infrequency of ditch water flow. All other wetlands e xhibited insignificant intrinsic uptake of P. A non-trivial error may be associated with these estimates due to the lack of time-intensive P sampling and further work should focus on concentrations and fl ow of overland flow to the wetland. The k-C* modeling effort produced acceptable levels of error considering the somewhat infrequent measured TP concentrations in the we tland surface water. The fit parameters may be further validated with future measurements of mo re frequent water quality data. In this study, commencement of higher frequency water sample collection coincided with dry hydrologic site conditions. Phosphorus treatment in these wetlands was nearly what was estimated in Chapter 4, using the landscape-scale pseudo-2d modeling appro ach, except for BW1, which showed more indication of being capable of significantly treating future P loads to the wetlands before export. Under current managed wetland hydrol ogy, the utility of th ese wetlands to retain P does not fully result in P load reduction to the lake, but w ith measures to increase retention time in the wetlands, higher treatment potentia l might be expected. The duration of P treatment under a modified hydraulic regime, such as blocking the export of P load ing via ditches, is not known and would require a more in-depth look at the mechanisms and limits of the P cycle.


96 Table 5-1 Water quality measurements of tota l P in wetland surface water, ditch water, and groundwater Wetland Surface Water Ground water Ditch water Mean [mg/L] CV Mean [mg/L] CV Mean [mg/L] CV LW1 1.2 (N=27) 0.4 0.6 (N=35) 0.9 0.9 (N=109) 0.5 LW2 1.0 (N=16) 0.4 0.5 (N=31) 0.7 0.8 (N=170) 0.6 BW1 0.1 (N=27) 0.6 0.4 (N=29) 1.0 --BW2 0.6 (N=16) 0.8 0.7 (N=15) 1.0 1.0 (N=7) 0.9 Table 5-2 Percentage of total P contribution from each chemical component, mean TP daily flux, and TP loads scaled to the Lake Okeechobee ba sin. Percent efficiency is also shown as a fraction of total P outflows to the total P inflows %MP %MGWin %MS %MD %MGWout %Uptake LW1 0.8 1.3 97.8 43.7 46.1 10.2 LW2 1.7 0.5 97.8 76.2 18.3 5.1 BW1 2.5 2.6 94.9 9.8 7.6 82.6 BW2 0.0 0.1 99.9 43.3 14.8 42.0 Mean flux MP MGWin MS MD MGWout Uptake Release LW1 [mg d-1] 3.0 4.0 310 138 146 284 251 CV 2.9 0.6 2.7 1.1 0.6 2.3 1.0 LW2 [mg d-1] 1.0 <1 84 65 16 266 99 CV 4.1 1.1 2.7 2.0 1.3 0.9 <1 BW1 [mg d-1] 1.0 1.0 31 3 3 63 7 CV 2.6 1.4 1.9 1.5 1.8 0.2 <0.1 BW2 [mg d-1] <1 <1 15 7 2 34 11 CV 1.8 1.3 2.7 1.5 1.0 0.5 <0.1 Table 5-3 Input and optimized parameters in k-C* model and associated error Site Cinit C* k (m d-1) k-C* based % treatment P budget based % treatment LW1 1.3 0.05 0.003 10 10 0.25 LW2 1.3 0.05 0.001 5 6 0.34 BW1 1.3 0.05 0.035 92 83 0.05 BW2 1.3 0.05 0.005 42 42 0.30


97 Table 5-4 Range of percent treatment of P inflow based on the high and low value of each concentration as determined by the coefficient of variability (low value -high value). The Cmean is the percent treatment based on the mean values of each component. Site C mean (%) CGWout (%)C GWin(%)C overland(%)C ditch(%) LW1 10.2 -4.2 -24.5 11.2 -9.1 39.7 --75.9 -11.6 -32.0 LW2 5.1 -2.7 -13.0 5.6 -4.7 36.3 --85.7 -32.9 -43.2 BW1 91.7 88.8 -94.6 91.9 --91.5 94.4 -84.2 90.0 -93.5 BW2 42.0 30.4 -53.5 42.0 --41.9 61.3 --16.0 20.3 -63.6 Figure 5-1. Cumulative mass of P with time for the Larson wetlands.


98 Figure 5-2. Cumulative mass of P with time for the Beaty wetlands. Figure 5-3. Results from k-C* model compared to measured P concentrations. A) Larson wetland-1. B) Larson wetland-2. C) B eaty wetland-1. D) Beaty wetland-2


99 CHAPTER 6 CONCLUSIONS Groundwater-Surface Water Exchange Hydrologically, groundwater outflow plays a ma jor role in the water budget. Hydraulic resistivity was determined for the four study wetla nds and was compared to values measured in several relatively large lakes in Florida and a wetland in south Florida of comparable size to the study wetlands. The hydraulic resistivity values determined for the study wetlands may have been most influenced by the soil properties underlying the wetlands. The sandy, poor-drainage soils under the wetlands were likely re sponsible for the relatively lower R values, compared to the wetland in south Florida (Wise et al., 2000). However, when compared to the R values at the lakes, the clayey subsurface geologic layering between the lake and the aquifer resulted in relatively higher R values, compared to those determined at the study wetlands. Linear regression of Darcys law and observed groundwater outflow data was adequate to determine long-term estimates of groundwater-surface water exchange. Water Budget Surface water outflow via the ditch dominate d the water budget outflows. However, groundwater outflow was a significa nt outflow pathway. While ET occurred more frequently than any other water budget component, it did no t contribute as significan tly to the water budget, except in BW1 which was not well ditched and reta ined surface water for longer periods of time. While ditch flow was infrequent, it was a crit ical component of the water budget, being responsible for approximately half of the total outflows. Ditch bottom elevation, relative to the wetland bathymetry (surface water drainage effec tiveness) was important in the amount of water that left via the ditch. The wetland bathymetry was also found to be an important control in the quantity of surface water export. Overland flow to the wetland was equally as important as

PAGE 100

100 direct rainfall on the wetland water surface. Groundw ater inflow was less frequent than the other wetland water budget components and also had relatively low corresponding inflow rates. Contributing upland area around th e wetland was significant, being as great as or up to three times greater than the wetland area. The hydrol ogic connection between the wetland and upland was considerable, but was manifest through two different hydrologic pa thways: overland flow and groundwater outflow. Watershed-Scale Treatment The spatial arrangement of wetlands in the watershed, based on the hydrologic assumptions in this modeling effort, was important in determining the degree of P treatment. Increased groundwater interception by wetlands re sulted in increased P treatment at the watershed boundary. According to this simulation, expected P treatment at the landscape scale might be up to 28%. Phosphorus Budget Estimation of the P budget indicated that overland flow dominated the inflows, representing approximately 95% of the total P mass to the wetland. Export of P via the ditches was also considerable, representing approximately 60% of the total mass of P out of the wetland system. The effectiveness of these wetlands to sequester P was also determined from the P budget. A net P uptake was determined from the P budget, ranging from 18 to 89% effectiveness, with BW1 having th e greatest effectiveness. The k-C* model was calibrated to measured surface water P values to develop the best fit using the first-order uptake coefficient as the optimization parameter. The results from that analysis also yielded a net loss of P, ranging from 20 to 90 % efficiency.

PAGE 101

101 LIST OF REFERENCES Allen, G.A., L.S. Pereira, D. Raes, and M. Sm ith, 1998. Crop evapotranspiration Guidelines for com puting crop water requirements FAO 399. Irrigation and Drainage Paper 56, FAO, Rome Italy, 78-86. Ann, Y., K.R. Reddy, and J.J. Delfino, 2000. In fluences of redox potential on P solubility in chemically amended wetland soils. Ecol. Eng. 14, 169-180. Boggess, W., E.G. Flaid, and R.C. Fluck, 1995. P budget-basin relationships for Lake Okeechobee tributary basins Ecol. Eng. 5, 143-162 Bottcher, A.B., T.K. Tremwel, K.L. Campbell, 1995. Best management practices for water quality improvement in the Lake Okeechobee Watershed. Ecol. Eng. 5, 341-356. Branfireun, B.A.and N.T. Roulet, 1998. The baseflow and storm flow hydrology of a precambrian shield headwater peatland. Hydrol. Proc. 12: 57. Campbell, K.L., J.C. Capece, and T.K. Tr emwel, 1995. Surface/subsurface hydrology and phosphorus transport in the Kissimmee River Basin, Florida. Ecol. Eng. 5, 301-330. Capece, J.C., K.L. Campbell, and L.B. Baldwin, 1988. Estimating runoff peak rates from flat, high-water-table watersheds. Tr ans. A.S.A.E. 0001-2351/88/3101-74. Carpenter, S.R., N.F. Caraco, D. L. Correll, R.W. Howarth, A.N. Sharpley, V.H. Smith, 1998. Nonpoint pollution of surface water with phosphorus and nitrogen. Ecol. App. 8, 559568. Davis, C.B., J.L. Baker, A.G. van der Valk, and C.E. Beer, 1981. Prairie pothole marshes as traps for nitrogen and phosphorus in agricultural runoff. pp. 153-63. Devito, K.J., Waddington J.M., and Brandireun B.A, 1997. Flow Reversals in Peatlands Influenced by Local Groundwater Sy stems. Hydrol. Proc. 11, 103-110. Dortch, M.S, 1996. Removal of solids, nitr ogen, and phosphorus in the Cache River Wetland. Wetlands. 16, 358-365. Drexler, J. Z., B. L. Bedford, R. Scognamiglio and D. I. Siegel, 1998. Quantification of the water budget and nutrient loading in a small peatland. Journal of the American Water Resources Association 35, 753 769. Drexler, J.Z. B.L. Bedfore, R. Scognamiglio, and D.I. Siegel, 1999. Fine-Scale Characteristics of groundwater flow in a peatla nd. Hydrological Processes 13: 1341. Driver, E.A. and D.G. Peden, 1977. The chemistry of surface water in Prairie ponds. Hydrobiologia 53: 33-48.

PAGE 102

102 Dunne, E.J., J. Smith, D.B. Perkins, M.W. Clark, J.W. Jawitz, K.R. Reddy, 2007. Phosphorus Storages in Ecosystem Compartments of Emergent Marsh Wetlands and Surrounding Uplands, Ecol. Eng., In press. Dunne, E.J., K.R. Reddy, M.W. Clark, 2006. Phosphor us release and retention by soils of natural isolated wetlands. Int. J. of Env. and Poll. 28, 496-512. Dunne, T. and R.D. Black, 1970. Partial area contributions to storm runoff in a small NewEngland watershed. Water Resour. Res. 6, 1296-1308 Ewel, K.C. and T.T. Odum (Eds.), 1984. Cypre ss Swamps. University Presses of Florida, Gainesville, FL, USA. Fetter, C.W, 1994. Applied Hydrogeology, Macmillan, New York. Flaig, E.C. and K.R. Reddy, 1995. Fate of phosphorus in the Lake Okeechobee watershed, Florida, USA. Ecological Engineering 5, 163-181. Florida Crop and Livestock Reporting Services,1997. Florida agricultural statistics 1997 summary. Florida Department of Environmental Protectio n (FDEP), 2001. Total maximum daily load for total phosphorus Lake Okeechobee, Florida. Tallahassee, FL. Fonyo, R.C., C., R. Fluck, W. Boggess, H. Dink ler, and L. Stanislawski, 1991. Biogeochemical behavior and transport of P in the Lake Okeechobee basin. Area 3 Final Report. Departments of Agricultural Engineeri ng and Food and Resources Economy, IFAS, University of Florida. Frankenberger, J.R., E.S. Brooks, M.T. Walter, M.F. Walter, T.S. Steenhuis, 1999. A GIS-based variable source area model. Hydrol. Proc. 13, 804-822. Gale, P.M., K.R. Reddy, D.A. Graetz, 1994. Phosphorus retention by wetland soils used for treated wastewater disposal. J. of Env. Qual. 23, 370-377. Goldstein, A.L. and W. Berman, 1995. Phos phorus management on confinement dairies in southern Florida. Ecol. Eng. 5, 357-370. Guardo, M. 1999, Hydrologic balance for a s ubtropical treatment wetland constructed for nutrient removal. Ecologi cal Engineering 12, 315-337. Haan, C.T., 1995 Fate and transport of phosphorus in the Lake Okeechobee Basin, Florida. Ecol. Eng. 5, 331-339. Harvey, R. and K. Havens, 1999. Lake Okeechobee Action Plan. South Florida Water Management District. West Palm Beach, FL. pp. 1-43.

PAGE 103

103 Hayashi, M. van der Kamp, G. Rudolph D.L. 1998a, Water and Solute Transfer between a Prairie Wetland and Adjacent Upland, 1. Water balance. J. Hydrol. 207, 42-55. Hayashi, M. van der Kamp, G. Rudolph D.L. 1998b, Water and Solute Transfer between a Prairie Wetland and Adjacent Upland, 2. Chloride Cycle. J. Hydrol. 207, 56-67. Hermond, H.F., 1980. Biogeochemistry of T horeaus Bog, Concor d, MA, Ecological Monographs 50, 507-526. Hiscock, J.G., C. Scott Thourot, J. Zhang, 2003. Phosphorus budget land use relationships for the northern Lake Okeechobee watershed, Florida. Ecol. Eng. 21, 63-74. Hunt, R.J., D.P. Krabbenhoft, and M.P. Anderson, 1996. Groundwater inflow measurements in wetland systems. Water Resour. Res. 32, 495-507 Kadlec, R.H., 1999. The limits of phosphorus re moval in wetlands. Wetland Ecol. and Manag. 7, 165-175. Kadlec, R.H. and R. L. Knight, 1996. Treatment Wetlands. Lewis Publishers, Boca Raton, LA. Kovacic., D. A., M.B. David, L.E. Gentry, K.M. Starks, and R.A. Cooke 2000. Effectiveness of constructed wetlands in re ducing nitrogen and phosphorus e xport from agricultural tile drainage. J. Env. Qual. 29, 1262-1274. Leibowitz, S.G. and T. Nadeau, 2003. Isolated wetlands: state-of-the-science and future direction. Wetlands 23, 663-684. Lewis, D.L. K.J. Liudahl, C.V. Noble, and L. J.Carter, 2001. Soil Survey of Okeechobee County, Florida. USDA/NRCS in cooperation with the University of Florida, Institute of Food and Agricultural Sciences, Agricultural Experimental Stations and Soil Science Department; and Florida Department of Agriculture and Consumer Services. Lyon, S.W., M.T. Walter, P. Gerard-Marchant, and T.S. Steenhuis, 2004. Using a topographic index to distribute variable source area runoff predicted with the SCS curve-number equation. Hydrological Processes 18, 2757-2771. Makhlouf, Z. and C. Michel, 199 4. A two-parameter monthly water balance model for French watersheds. Journal of Hydrology 162, 299-318. Mckee, K.A., 2005 Predicting soil phosphorus storage in historically isolated wetlands within the Lake Okeechobee priority basins. M.S. Thesis University of Florida online reserve. Marston Science Library. Gainesville, FL 32611. Metcalfe, R.A. and J.M. Buttle, 1999. Semi-distributed water balance dynamics in a small boreal forest basin. J. Hydrol. 226, 66.

PAGE 104

104 MicGill, G., R.A. Gatewood, C. Hutchinson, and D.D. Walker, 1976. Final report on the special project to prevent the eutrophication of Lake Okeechobee. Re port DSP-BCP-36-76, Division of State Planning, Tallahassee, FL. Mitch and Gosselink, 2000. Wetlands (Third Edition). John Wiley, New York, NY, USA. Monod, H., Naud, C., Makowski, D., 2006. Uncertai nty and sensitivity analysis for crop model. In Wallach, D., Makowski, D., Jones, J.W., (Eds.), 2006. Working with dynamic crop models. Evaluation, analysis, parameterizatio n and applications. Elsevier, Amsterdam, pp. 447. Montgomery, D.R. and W.E. Dietrich, 2004. Hydrologic processes in a lowgradient source area. Water Resour. Res. 31, 1-10. Motz, L.H., 1998. Vertical leakage and vertically averaged vertical conductance for karst lakes in Florida. Water Resour. Res.. 34, 159-167. Motz, L.H., G.D. Sousa, M.D. Annable. 2001. Water budget and vertical conductance for Lawry (Sand Hill) Lake in north-central Fl orida, USA. J. Hydrol. 250:134-148. National Research Council (NRC), 1995. Wetlands: Characteristics and boundaries. National Research Council Committee on Characterizati on of Wetlands, National Academy Press, Washington DC, USA. Neely, R.K. and J.L. Baker, 1989. Nitrogen and phosphorus dynamics and the fate off agricultural runoff. P. 92-131. In A.G. va n der Valk (Ed.) Northern Prairie Wetlands. Iowa State University Press, Ames, IA, USA. Newman, M.C. and J.F. Schalles, 1990. The water chemistry of Carolina bays: A regional survey. Archiv Fr Hydrobiologie 118: 147-168. Parsons, D.F., M. Hayashi, and G. van der Kamp, 2004. Infiltration and solute transport under a seasonal wetland: bromide trace r experiments in Saskatoon, Canada. Hydrol. Proc. 18, 2001-2027 Perkins, D.B. and J.W. Jawitz, Wetland-groundw ater interactions in managed seasonallyinundated depressional wetlands. Jour nal of Hydrology. In review. Perkins, D.B., J.W. Jawitz, and A.E. Olsen, 2005. Spatially distributed isolated wetlands for watershed-scale treatment of agricultural r unoff. In: Dunne, E.J., K.R. Reddy, and O.T. Carton (Eds.), Nutrient management in agricultural watersheds: A wetlands solution Wageningen Academic Publishers, The Netherlands. pp. 80-91. Perkins, D.B., J.W. Jawitz, and M.D. Annable, Quantifying hydrologic pathways in depressional wetlands using a water budget approach In preparation for J. Hydrol.

PAGE 105

105 Pezzolesi, T.P., R.E. Zartman, E.B. Fish, and M.G. Hickey, 1998. Nutrients in Playa wetland receiving wastewater. J. Env. Qual. 27, 67-74. Plant, H.K., K.R. Reddy, and R.M. Spechler. 2002. Phosphorus retention in soils from a prospective constructed wetland site: environmental implicat ions. Soil Sci. 176:607-615. Prescott, K.L., and I.K. Tsanis,1997. Mass balan ce modeling and wetland restoration. Ecol. Eng. 9, 1. Price, J.S. and J.M. Waddington, 2000. A dvances in Canadian wetland hydrology and biogeochemistry. Hydrol. Proc. 14, 1579-1589. Raisin, G.W., D.S. Mitchell, R.L. Croome, 1997. The effectiveness of a small constructed wetland in ameliorating diffuse nutrient loading from an Australian rural catchment. Ecol. Eng. 9, 19-35. Reddy, G.A. OConnor, C.L. Schelske, 2000. Phosphorus Biogeochemistry of Sub-Tropical Ecosystems. Lewis Publishers Inc. Boca Raton, Florida. Reddy, K.R., O.A. Diaz, L.J. Scinto, M. Ag ami, 1995. Phosphorus dynamics in selected wetlands and streams of the Lake Okeechobee basin. Ecol.Eng. 5, 183-207. Reddy, K.R., R. H. Kadlec, and E. Flaig, 1999. Phosphorus retention in streams and wetlands: A review. Critical Reviews in Enviro nmental Science and Technology 29, 83-146 Reddy, K.R., E. Lowe, T. Fontaine, 2002. Phosphor ous in Floridas ecosystems: Analysis of current issues. In: Reddy K.R., G.A. OC onnor, and C.L. Schelske (Eds.), Phosphorous biogeochemistry in subtropical ecosystems. Reddy, K.R., R.G. Wetzel, and R.H. Kadlec, 200 5. Biogeochemistry of phosphorus in wetlands. In: Sims, J.T. (Ed.), Phosphorus: Agricu lture and the Environment, Agronomy Monograph no. 46. Richardson (Ed.) Riparian Management in Forest s of the Continental Eastern United States. Lewis Publishers, Boca Raton, FL, USA. Romanowicz, E.A., Siegel D.I., and Glaser P. H., 1994. Hydraulic Reversals and Episodic Methane Emissions during Drought Cycles in Mires. Geology. 21, 231-234. Schalles, J.F., 1989. Comparative chemical limnology of Carolina ba y wetlands on the Upper Coastal Plain of South Carolina. In: Sharitz, R.R. and J.W. Gibbons (Eds.), Freshwater Wetlands and Wildlife. pp. 89-111. US DOE Office of Science and Technical Information, Oak Ridge, TN, USA.

PAGE 106

106 SFWMD (South Florida Water Management Dist rict), 1993. Update of Okeechobee SWIM Plan. South Florida Water Management Distri ct, West Palm Beach, Florida. January, 1993 SFWMD (South Florida Water Ma nagement District), 1997. Surface Water Improvement and management (SWIM) Plan update for Lake Ok eechobee. Volume I: Planning Document. West Palm Beach, FL. SFWMD (South Florida Water Ma nagement District), 2001. Surface Water Improvement and management (SWIM) Plan update for Lake Okeechobee. West Palm Beach, FL. SFWMD (South Florida Water Ma nagement District), 2002. Surface Water Improvement and management (SWIM) Plan update for Lake Okeechobee. West Palm Beach, FL. SFWMD (South Florida Water Management Dist rict), 2007. DBHYDR O Browser database, http://glades.sfwmd.gov/pls/dbhydro _pro_plsql/show_dbkey_info.main_page Shan, B., C. Yin, and G. Li, 2002. Transpor t and retention of phosphorus pollutants in the landscape with a traditional, multipond system. Wa ter, Air, and Soil Pollution 139, 15-34. Snodgrass, J.W., A.L. Bryan, Jr., R.F. Lide, and G.M. Smith, 1996. Factors affecting the occurrence and structure of fi sh assemblages in isolated wetlands of the upper coastal plain, U.S.A. Canadian Journal of Fish eries and Aquatic Sciences. 53, 443-454. Sompongse, D., 1982. The role of wetland soils in nitrogen and phosphorus removal from agricultural drainage water. Ph.D. Dissertation, University of Florida, Gainesville (Diss. Abstr. DA 8226434). Tilman, D., K.G. Cassman, P.A. Matson, R. Mayl or, S Polasky, 2002. Agri cultural sustainability and intensive production pract ices. Nature. 418: 671-677. Tiner, R.W., H.C. Bergquist, G.P. DeAlessio, and M.J. Starr, 2002. Geographically isolated wetlands: a preliminary assessment of their char acteristics and status in selected areas of the Unites States. U.S. Department of the Interior, Fish and Wildlife Service, Northeast Region, Hadley, MA, USA (web-based report at: wetlands.fws.gov ). Tiner, R.W., 2003a. Geographically isolated wetl ands of the United States. Wetlands. 23, 494516. Tiner, R.W., 2003b. Estimated extent of geographically isolated wetlands in selected areas of the United States. Wetlands 23, 636-652. USDA (United States Department of Agricult ure), 2001. Soil Survey of Okeechobee County, FL. Natural Resource Conservation Service

PAGE 107

107 USEPA (United States Environmental Protecti on Agency), 1992. Standard methods for the examination of water and wastewater (SM). 18th edition, American Public Health Association, Washington D.C. Waddington, J.M. and N.T. Roulet, 2000. Car bon balance of a boreal patterned peatland. Global Change Biology 6, 87-97 Wang, H., Jawitz, J.W., White, J.R. and Martin ez, C.J. and Sees, M.D, 2006. Rejuvenating the largest municipal treatment wetland in Florida. Ecol. Eng. 26, 132-146. Ward, R.C., 1984. On the response to precipita tion of headwater streams in humid areas. J. Hydrol. 74:171-189. Watson, B. J., L.H. Motz, and M.D. Annable, 2001. Water budget and vertical conductance for Magnolia Lake. Journal of Hydr ologic Engineering 2, 208-216. Wetzel, R.G., 2001. Limnology: Lake and river ecosystems. 3rd ed. Academic Press, San Diego. Whigham, D.F. and T.E. Jordan, 2003. Isolated wetlands and water quality. Wetlands 23, 541549. Winter, T.C., 1990. Relation of streams, lakes and wetlands to groundwater flow systems. Hydrogeology Journal 7:28-45. Winter, T.C. and J.W. LaBaugh, 2003. Hydrologi c considerations in defi ning isolated wetlands. Wetlands. 23, 532-540. Wise, B.W., M.D. Annable, J.A.E. Walser, R.S. Switt, and D.T. Shaw, 2000. A wetland-aquifer interaction test. J. Hydrol. 227, 257-272. Zhang, J., S.A.F. Ray, and A Steinman, 2002. Potential phosphorus load reduction under the Lake Okeechobee regulatory program. J. Am. Water Res. Assoc. 38, 1613-1624.

PAGE 108

108 BIOGRAPHICAL SKETCH I was raised in rural Caliente, NV, where I m ade use of my time in school, sporting, theatre, and outdoor life. I served a full-time mi ssion for the Church of Jesus Christ of LatterDay Saints in Paris, France for 2 years where I learned another language and culture. I then attended Brigham Young University where I received a B.S. in agronomy. My wife and I had our first child, Canyon just before graduating. Next, I received an M.S. from Purdue University in agronomy, where I split my coursework a nd research with Agronomy and Environmental Engineering. Our second son, Micah, was born in Lafayette, IN. Afterward, I attended the University of Florida for work toward a Ph.D We had our first (and only) girl, Meadow. Toward the end of my time as a Ph.D. student, we had Bryton join our family.