Introduction

I. Title, Authors, Organization, etc.

NOAA Project # NA96OP0234

“The Role of Groundwater in the Nutrient Budget of Florida Bay”

PART I

FINAL REPORT

Principal Investigators: William C. Burnett & Jeffrey Chanton

Graduate Research Assistants: D. Reide Corbett and Kevin Dillon

Department of Oceanography

Florida State University

Tallahassee, Florida 32306-4320

Tel: 850-644-6703

Fax: 850-644-2581

email: wburnett@mailer.fsu.edu

July 14, 2000

Table of Contents

Page

I. Title Page 1

II. Abstract 3

III. Executive Summary 4

IV. Purpose

A. Description of Research Background 5

B. Objectives of the Project 7

V. Approach

Radon and Methane Sampling 7

Seepage Measurements 9

Nutrients 9

Modeling Approach 9

Diffusive Benthic Fluxes 11

Benthic Flux Experiments 12

Tracer Mass Balance Calculations 12

Benthic Flux Calculations 13

Fluxes to the Atmosphere 14

VI. Findings

Tracer Concentrations 14

Tracer Distributions in Florida Bay 15

Diffusive Flux Experiments 22

Water Column Tracer Inventory and Mass Balance 25

Benthic Chamber Experiments 29

VII. Evaluation

A. Results for Project Goals and Objectives

Estimates of Groundwater Input 29

Nutrient Implications 33

B. Dissemination of Project Results 36

Acknowledgments 37

References 38

II. Abstract

This report contains results from our efforts to quantify the groundwater inputs into Florida Bay and how these inputs may influence the nutrient balance of the bay. We were able to establish patterns of potential groundwater input into Florida Bay by use of radon and methane as natural tracers of SGD. The results suggest that the interactions between groundwater and surface water are greatest nearshore along the Florida Bay side of the Florida Keys. Groundwater seepage, measured with the Lee-type meters, appears to be significant in certain areas and may be important for the nutrient budgets of Florida Bay.

Measurements of trace gases, radon and methane, in the waters of Florida Bay indicate substantially higher benthic fluxes than can be accounted for by diffusion alone. These higher fluxes can be supported by groundwater velocities ranging from 0.2 to 8.0 cm day^-1. We suggest a best estimate of approximately 1.9 cm day^-1 for all three sites. These estimates are consistent with the seepage meter data. Groundwater may thus provide a significant quantity of water and nutrients to Florida Bay. Our results have shown that groundwater circulation may provide nutrients to the eastern portion of Florida Bay similar in magnitude to surface inputs from the Everglades.

III. Executive Summary

Surveys for natural chemical tracers of groundwater discharge (²²²Rn and CH₄) were conducted in order to evaluate possible patterns of groundwater/surface water interactions in Florida Bay. Radon and methane concentrations in water samples collected from wells, solution holes, canals, and Florida Bay showed a significant correlation, despite the fact that these two trace gases have independent source terms. Direct measurements of groundwater flux via seepage meters were also made in several locations in Florida Bay. Collectively, results suggest a greater groundwater flow along the bay-side of the Florida Keys. Nutrient flux estimates, based on interstitial nutrient concentrations and groundwater flux measurements, suggest that groundwater in the eastern area of Florida Bay may provide as much nitrogen (110 ± 60 mmol N m^-2 y^-1) and phosphate (0.21 ± 0.11 mmol PO₄^3- m^-2 y^-1) as surface freshwater sources from the Everglades (i.e., Taylor Slough and C-111). However, the inputs are clearly not uniform and areas near solution holes/tidal springs may have a substantially greater nutrient flux into surface waters then these estimates.

Groundwater may represent a significant pathway for nutrients into Florida Bay, especially near the keys where wastewater disposal practices add large amounts of nitrogen and phosphorus to the subsurface each year. In this study, we calculate the benthic fluxes required to maintain these tracer water column inventories and evaluate whether independently measured seepage rates are of the correct magnitude to support these fluxes. Mass balance calculations yield estimates of total benthic fluxes required to support the water column inventories and estimated atmospheric losses between 4.2 - 5.6 dpm m^-2 min^-1 and 5.8 - 30.2 nmoles m^-2 min^-1 for ²²²Rn and CH₄, respectively. Independent estimates of the diffusive flux and porewater concentrations allow us to calculate an advective groundwater velocity, assuming that the difference between the total flux and the diffusive flux is due to seepage driven porewater advection. These calculated velocities ranged from 0.2 to 8.0 cm day^-1 for all sites, tracers, and sampling periods, with a best estimate of approximately 1.9 cm day^-1. Measurements of groundwater flux at the same sites via seepage meters compare very well with these results. Thus, the upward advection of saline groundwaters may provide almost as many nutrients to Florida Bay as waters of the open Gulf of Mexico, currently considered the bay's largest nutrient source.

IV. Purpose

A. Description of Research Background

Submarine groundwater discharge (SGD) is an often overlooked yet possibly significant process in the geochemical and nutrient budgets of marine nearshore waters. According to Johannes (1980) “SGD should occur anywhere that an aquifer is hydraulically connected with the sea through permeable rocks or bottom sediments and where the head is above sea level.” Such conditions are met in most coastal areas. A simple model of SGD from a homogeneous unconfined aquifer indicates that freshwater flows out along the coast through a narrow gap between the freshwater - seawater interface where the water table outcrops at the shoreline. Harr (1962) predicted theoretically that the seepage discharge rate should decrease rapidly from shore in such a setting. This phenomenon has been confirmed in several lake studies and at least a few marine areas. For example, Bokuniewicz (1980) showed that seepage rates decreased roughly exponentially with distance from shore in Great South Bay, New York. Although the zone in which significant seepage occurs is usually thought to be narrow (£100 m in Great South Bay), this may not always be the case. Kohout (1960) stated, for example, that in parts of Florida the salt water front within the sediments (zone of mixing or diffusion) may be as much as 14 km seaward of the coast, thus allowing SGD to occur well offshore. Furthermore, seepage may also occur through breaks or permeable portions of an overlying bed of a confined aquifer.

Groundwater discharge has been documented as being highly significant for nutrient supply in some coastal areas (Valiela et al., 1978; Valiela and Teal, 1979; Capone and Bautista, 1985; Lapointe and O'Connell, 1989; Capone and Slater, 1990; Valiela et al., 1990). SGD may be particularly important in these cases because shallow groundwaters are often enriched in nitrogen, possibly by contamination from septic tanks.

However, most prior studies have addressed the case of a hydraulically-driven freshwater aquifer in contact with typical coastal marine or lake environments. The situation in the Florida Keys is much different than most coastal environments because: (1) most subsurface fluids are saline to hyper-saline; and (2) the driving force is thought to be tidal-driven rather than topographic. Groundwater entering Florida Bay along the north coast, however, would be characterized by more typical topographic gradient flow. Subsurface samples collected in this region vary from relatively fresh (<1 psu) to hyper-saline (>36 psu). Waters sampled from wells within Florida Bay, as deep as 10m, tend to have elevated salinities (>35 psu).

We use the term “groundwater” for the Florida Keys in a very general manner, i.e., it includes recirculated seawater from the Atlantic Ocean and/or Florida Bay, meteoric water, and wastewater components. Nearly all “groundwaters” sampled by shallow wells in Florida Bay are saline to hypersaline.

Moore (1999) referred to this subsurface region of mixing between meteoric water and sea water in coastal aquifers as "subterranean estuaries." He makes the point that water entering surface waters from the subsurface may be considered groundwater, regardless of its salinity. Younger (1996) has referred to all water flowing to the sea, irrespective of its source, as "total" groundwater discharge, in an attempt to distinguish between fresh groundwater of a coastal aquifer from recycled seawater. However, the individual sources of groundwater discharge may not be as important as the dissolved constituents that the final fluid brings into surface waters after mixing within the "subterranean estuary." The mixing of these water masses in the subsurface creates an active chemical environment. Chemical reactions between aquifer solids and a mixture of seawater and meteoric water modifies the fluid composition, and eventually this altered mixture returns to surface waters as a consequence of SGD (Runnles, 1969; Back et al., 1979; Moore, 1999). In addition to these natural processes, wastewater disposal in the Florida Keys adds yet another source of water to the subsurface environment.

It is evident from the salinity in groundwaters below the Keys and Florida Bay that there has been much interaction of surface waters with the subsurface fluids. Therefore, the definition of groundwater discharge adopted here includes recirculated seawater from the Atlantic Ocean and/or Florida Bay, meteoric water, and wastewater. The direction of groundwater flow beneath the Keys is thought to oscillate as the fluctuating Atlantic tides create a differential head with respect to sea level in Florida Bay where tides are significantly damped. The Atlantic Ocean can have tides on the order of one meter, while tides in Florida Bay range from approximately 0.1m around Key Largo to 0.25m near Long Key. Halley et al. (1994) showed that there is a positive and negative head differential of the surface of the Atlantic relative to that of the Bay and the difference can be as great as 0.7 m. Another study concerning the dynamics of the subsurface seawater circulation around Key Largo has demonstrated that the Florida Keys are surprisingly open systems (Halley et al., 1995). The response of hydraulic heads and flows in a series of wells located in Florida Bay have suggested considerable water movement through the porous carbonate rock formations characteristic of these islands (Shinn et al., 1994). Previous groundwater studies in the Florida Keys have shown a range of transport rates between 1.4 and 420 m d^-1, partially dependent on the geological terrain (Key Largo Limestone or Miami Limestone) and differential tidal influences (Lapointe et al., 1990; Paul et al., 1995; Paul et al., 1997; Dillon et al., 1999a, b).

Florida Bay is a large (2200 km²) shallow (mean depth < 2 m) coastal lagoon lying between the southern tip of the Florida mainland and the coralline Keys (Fig. 1). The bay has witnessed significant changes in the past decade (Boesch et al., 1993) including seagrass die-offs, greater frequency of planktonic algal blooms (Phlips et al., 1995; Phlips and Badylak, 1996), and water quality problems which could be related to excess nutrient input (LaPointe and Clark, 1992). Algal blooms may also have played a role in the die-off of the bay's sponge population (Hunt and Herrnkind, 1993). A number of hypotheses to explain this environmental deterioration have been offered. Lapointe et al. (1990) suggested that sewage-derived inputs of nitrogen and dissolved organic phosphorus to canals and surface waters may be a key factor for initiating the phytoplankton blooms in the area. Also, reductions in freshwater inflow from the Everglades have contributed to hypersaline conditions which may be responsible for the seagrass declines (Boesch et al., 1993). Freshwater is delivered to Florida Bay mainly through Taylor and Shark River Sloughs, although other sources (e.g., C-111) may be locally important. Much freshwater flow has been diverted to the Atlantic through canals, and it has been estimated that less than 25% of the "natural" surface water flow through Taylor Slough is presently occurring (Light and Dineen, 1994).

Few data are available on groundwater flow into Florida Bay. This is unfortunate as groundwater may be important due to the porous nature of the limestone underlying the regions soil and sediments (Boesch et al., 1993). Groundwater flow rates in the Miami Limestone, considered to be the least permeable of the two prominent geologic formations that occur in the area, were reported to be as high as 7.2 m d^-1 (Lapointe et al., 1990; Dillon et al., 1999a, b). This formation is also reported to underlie Florida Bay (Perkins, 1977).

Groundwater in the shallow subsurface has been shown to contain elevated concentrations of dissolved nitrogen species relative to surface waters due to the decay of natural organic materials disseminated within the matrix (Sansone et al., 1990). Another cause of elevated nutrients in the subsurface may be the waste disposal practices in the Florida Keys (Shinn et al., 1994). Sewage in the Florida Keys is discharged into over 600 disposal wells that penetrate the permeable Key Largo Limestone (Pleistocene) at depths of 10 to 30 meters. In addition to these wells, used by hotels and commercial establishments, there are an estimated 24,000 septic tanks and 5,000 cess pools. Subsurface waters in the Florida Keys receive an estimated 2.3 x 10¹⁰ mmol of nitrogen and 9.3 x 10⁸ mmol of phosphate per year due to wastewater practices. In an environment believed to be extremely permeable, groundwater-derived nutrients from the Key’s subsurface, whether from sewage or natural sources, may be important to the overall nutrient budget of Florida Bay (USEPA, 1996).

B. Objectives of the Project

The purpose of this study was to perform a preliminary evaluation of the significance of groundwater discharge into Florida Bay and evaluate the potential of groundwater as a contributor to the surface water nutrient pool. We first attempted to locate areas in the bay where groundwater seepage is most pronounced by reconnaissance surveys of the concentrations of radon and methane in the bay waters. These trace gases function as natural indicators of submarine groundwater discharge into standing bodies of water due to their significantly higher concentrations in groundwaters (Cable et al., 1996a, b; Bugna et al., 1996). Radon is typically elevated in groundwater because of production and recoil processes originating from radium within the aquifer matrix, while methane is produced from the decay of organic matter disseminated in the rock. While both processes occur within aquifers and result in elevated tracer concentrations within groundwaters, the production of each gas is completely independent.

Direct measurements of groundwater discharge with seepage meters were also conducted at selected sites along the Keys and in the bay. Nutrient samples were collected and analyzed from surface and porewaters within the bay, along the reef tract, and in some solution holes, wells, and canals.

We present here measurements of water column inventories of the natural geochemical tracers, radon and methane, in Florida Bay waters. These tracers exhibit standing stocks which are clearly greater than could be supplied by diffusive fluxes across the sediment-water interface. We conducted detailed experiments at three field sites: Rock Harbor, Flamingo, and Rabbit Key Basin, which are representative of the Florida Bay environments along the Keys on the bay side, the North coast, and mid bay settings, respectively (Corbett et al., 1999). We will show that the benthic input rates necessary to maintain the water column inventories are consistent with the rates at which these tracers appear to be supplied by groundwater advection.

V. Approach

Radon and Methane Sampling

Samples for tracer analysis were collected at over 170 stations in Florida Bay during the study period. Radon samples were collected from about 0.3 meters above the bottom at each station using a peristaltic pump and 4-liter evacuated bottles. Standing water was purged from the hose at each depth prior to filling the sampling bottles, and the bottles were immediately sealed to prevent gas loss. Radon gas was extracted and transferred with a modified emanation technique similar to that described by Mathieu et al. (1988). After radon stripping and transfer into alpha scintillation cells, samples were counted with Ludlum flask counters. After the initial radon analysis, the samples were sealed and stored for at least five days for ²²²Rn ingrowth and then sparged again in order to determine the ²²⁶Ra activity. "Excess" (unsupported) radon was determined as the difference between the "total" ²²²Rn in samples and the supported ²²²Rn , assumed to be equal to the ²²⁶Ra activity. The excess ²²²Rn values were decay-corrected to the time of sampling in order to assess the in situ concentrations. All radon results presented are excess radon values unless otherwise noted.

Figure 1. Florida Bay separates the Florida Keys, located off the southern tip of Florida, from the mainland. Water samples were collected primarily from north of Long Key and East of Flamingo. Groundwater samples from offshore wells were collected where indicated by the circles. Letters refer to locations mentioned in the text: A. Carysfort Reef; B. Algae Reef; C. French Reef; D. Molasses Reef; E. Rock Harbor; F. Porjoe Key; G. Black Betsy Keys; and H. Tavanier Basin.

Methane samples were collected in Wheaton BOD bottles and stored on ice until analysis. Upon return to the laboratory, water samples were transferred to 50-mL disposable syringes which were pre-flushed with nitrogen. An extraction volume of 10 mL of N₂ to 40 mL of water was added to each syringe, and the methane extracted via headspace equilibration. Samples were run on a Shimadzu flame ionization gas chromatograph equipped with a 2-m stainless steel column packed with Poropack Q (McAuliffe, 1971).

Samples for ²²²Rn and CH₄ in groundwater were also obtained from monitor wells at depths ranging from 5 to 60 meters. The locations of these sites were primarily within Florida Bay, onshore and offshore of Key Largo, and at the Keys Marine Laboratory located on Long Key (Fig. 1).

Seepage Measurements

Direct measurements of groundwater seepage were made with an instrument design modified from Lee (1977). The seepage meter is simply an open-bottomed chamber (0.25 m²) implanted in bottom sediments which has an open port where a plastic bag can be attached to collect seepage over measured time intervals. A measurement of the volume in the collection bags can then be converted to a seepage flux as the collection time and area are known. All seepage meters were placed in areas which had sufficient sediment to provide a seal between the meter and surrounding sediment. Four liter plastic bag collectors were used and were prefilled with 1000 mL of bay water to prevent short-term artifacts (Shaw and Prepas, 1989). An initial volume also allows for measurement of negative seepage, i.e., recharge into the underground aquifer. The lower reliable limit of measurement for seepage meters depends upon the length of deployment and the conditions under which the sampling occurs—based on our experience with these meters, we normally expect a lower useful limit of 3-5 mL m^-2 min^-1 (Cable et al., 1997).

Nutrients

Nutrient concentrations (ammonia, nitrate, and phosphate) were determined at sampling sites by standard methods. Water samples taken in the field were kept in the dark on ice until analysis. A procedure involving vanadium reduction followed by chemiluminescence detection of NO_x was used for nitrate plus nitrite analysis and also total dissolved nitrogen after a persulfate digestion (Braman and Hendrix, 1989). Ammonia and ortho-phosphate concentrations were determined with the phenate and with the ascorbic acid method, respectively (Strickland and Parsons, 1972).

Modeling Approach

In order to evaluate the connection between surface water tracer inventories and groundwater inputs, we constructed a balance of all possible inputs and outputs of these natural tracers. While both ²²²Rn and CH₄ are enriched in groundwater relative to surface waters, they are present throughout the environment and therefore other sources must be considered. A simple box model for both radon (Fig. 2A) and methane (Fig. 2B) may be used to describe their source and sink terms. Our approach was to assess all flux terms and to estimate the groundwater contribution by difference. Specifically, we estimated the groundwater tracer contribution by measurements of the tracer water column inventory, calculation of the total benthic flux required to support these inventories (including atmospheric evasion), and an independent assessment of the diffusive component of this flux. The advective component was accessed by application of an advective-diffusion model after measuring groundwater and porewater tracer concentrations.

Figure 2. Box model depicting sources and sinks supporting water column inventories of ²²²Rn (A) and CH₄ (B) in waters of Florida Bay.

Diffusive Benthic Fluxes

Two types of approaches are generally applied to measure fluxes of any component across the sediment-water interface: (1) measurements of gradients in the sediment and/or water column (Broecker, 1965; Hesslein, 1976; Key et al., 1979; Gruebel and Martens, 1984; Berelson et al., 1990); and (2) in situ benthic chambers (Martens et al., 1980; Gruebel and Martens, 1984; Berelson et al., 1990). In addition, Martens et al. (1980) described a depth-independent mathematical approach specifically for the determination of diffusive fluxes of ²²²Rn. This approach allows the radon flux to be calculated by equilibrating sediment grab samples (~80 g wet sediment) collected from the field with surface waters (~100 mL) over at least a 30-day period in sealed containers. After this time period, the equilibration concentration of radon supported by radium in the sediments should be established and the ²²²Rn concentration in the water phase is measured. Assuming steady state conditions with negligible advective transport, the radon flux across the sediment-water interface can then be calculated from the following:

(1)

where J is the flux of radon from the sediments (dpm m^-2 min^-1); l is the decay constant for ²²²Rn (min^-1); D_S is the effective wet bulk sediment diffusion coefficient in the sediments (m² min^-1) corrected for temperature and tortuosity (Peng et al., 1974; Ullman and Aller, 1982); C_eq is the equilibrium concentration of radon measured in the laboratory (dpm m^-3; wet sediment); and C_o is the radon in the overlying water multiplied by the sediment porosity to obtain a value corresponding to the concentration in wet sediment (dpm m^-3). The D_S value was approximated at each site from the free solution diffusion coefficient, D_o, of radon (1.14 x 10^-5 cm² sec^-1 at 18 ^oC; Rona, 1917).

In order to be certain that the sediment equilibration approach would provide accurate estimates of the diffusive flux of radon, we elected to perform an independent assessment of this flux in a manner similar to that performed by Corbett et al. (1998). A 14-cm diameter sediment core was collected off Rock Harbor in a custom-made device and transported back to the laboratory. Florida Bay water (collected at the same site) was then added to this core and the radon concentration in the overlying water was monitored over time to produce a direct measurement of diffusive transport. The radon concentration was monitored for over a 2-week period and a direct estimate of the diffusive flux of ²²²Rn was made by evaluating the increase in concentration while accounting for decay in the overlying water during the experiment. After these initial time-series experiments were complete, a subcore was collected and partitioned into several depth intervals. These sediment samples were sealed in glass bottles with approximately 80 mL of Florida Bay water. The wet sediments were analyzed immediately (by radon emanation) for ²²²Rn to determine radon porewater concentrations before any significant ingrowth of ²²²Rn could occur due to ²²⁶Ra in the sediment. The samples were then re-analyzed after a thirty-day holding period in order to obtain the “sediment equilibrated” ²²²Rn for each sample. Porosities and wet bulk densities of these sediments were calculated at the same depth intervals from measured water contents and grain densities determined on companion sediment samples collected in a second subcore.

Estimates of the diffusive flux of methane were conducted at the Rock Harbor and Flamingo field sites by measuring the porewater concentration gradient in the sediments with a peeper, a close interval porewater sampling device described by Hesslein (1976). A 0.2 mm polycarbonate membrane covered the open 10 mL compartments of the peeper (initially filled with deionized water) and the entire assembly was inserted into the sediments and allowed to equilibrate with in situ porewaters for approximately 6 days. After this equilibration period, samples were collected with a syringe and needle and placed on ice until analysis. Once the analyses are complete and the concentration gradient estimated graphically, Fick’s law can be used to estimate the diffusive flux, using a diffusion coefficient (1.72 X 10^-5 cm² s^-1), estimated by Chanton et al. (1989) for similar types of sediments and conditions.

Benthic Flux Experiments

Clear acrylic chambers, enclosing 59 L of ambient water and covering 0.21 m² of bottom sediment, were used to assess benthic fluxes during August 1998 and January 1999. After inserting the chambers into the sediment, the chambers were stirred for approximately 5 minutes and initial samples were collected for ²²²Rn and CH₄. Intermediate samples were occasionally collected at about 6 hours before final samples were collected after approximately 12 hours. At that point, the chambers were either reset, by opening ports to the ambient surface water and mixing for approximately 30 minutes, or they were removed and re-inserted at a different location. The flux of ²²²Rn and CH₄ across the sediment-water interface could then be calculated by relating the change in concentration inside the chamber to the enclosed water volume, deployment time, and the enclosed sediment area.

Tracer Mass Balance Calculations

In order to assess the groundwater flux entering Florida Bay via radon and methane measurements, all sources and sinks of the two gases must be considered. A simple box model for ²²²Rn (Fig. 2A) and CH₄ (Fig. 2B) may be used to describe the measurements necessary to complete a mass balance at each experimental site. Mathematically, the ²²²Rn balance can be expressed as

(2)

and for CH₄,

(3)

where J_benthic represents the combined advective (J_seep) and diffusive (J_diff) flux of either ²²²Rn or CH₄ to the overlying water column (dpm m^-2 min^-1 or nmoles m^-2 min^-1); l is the decay constant of ²²²Rn (min^-1); lC_Ra and lC_Rn accounts for production and decay of radon in the water column (dpm m^-3 min^-1), respectively; z is the depth of the water column at each site (m); J_atm is the flux lost to the atmosphere (dpm m^-2 min^-1 or nmoles m^-2 min^-1); and J_production and J_oxidation refer to the amount of methane produced and consumed within the water column, respectively (nmoles m^-2 min^-1). Horizontal transport of the trace gases was not considered in the mass balance equations since the horizontal concentration gradient at each study site was observed to be small (Corbett et al., 1999). The most likely additional source for ²²²Rn in the surface waters of Florida Bay is via benthic flux (Eq. 2), since production of ²²²Rn in the water column is fairly small due to the low ²²⁶Ra concentrations (~1500 dpm m^-3) relative to typically observed ²²²Rn concentrations (>5000 dpm m^-3). Sources of radon across the sediment/water interface include both diffusive and advective fluxes. All sediments contain trace amounts of uranium and daughter products including ²²⁶Ra, the progenitor of ²²²Rn. Thus, diffusion of ²²²Rn would add some amount of unsupported radon to the overlying waters due to its relatively high mobility in the environment as a noble gas.

The production and oxidation of CH₄ in coastal and offshore waters of the northeastern Gulf of Mexico has been studied extensively by Bugna et al. (1996). Incubations of unfiltered seawater produced a mean net oxidation rate of 0.8 ± 0.7 nmol m^-2 min^-1 and a mean methane production rate of 0.4 ± 0.5 nmol m^-2 min^-1 in nearshore waters. We have used the mean values of the oxidation and production rates of methane from Bugna et al. (1996) in all calculations.

Benthic Flux Calculations

The benthic fluxes were assessed experimentally in the field via benthic flux chambers, and from water column tracer concentrations by solving the mass balance equation (Eq. 2 and 3) after estimating all other sources and sinks. The total benthic flux (J_benthic) will exceed the estimated diffusive flux (J_diff) if there is also an advective flux (J_seep) present, such as from groundwater flow or a turbulent transfer mechanism like bioirrigation or bioturbation. Bugna et al. (1996) employed a simple equation to estimate the advective flux of groundwater into surface waters with methane as a tracer:

calculation 4 (4)

(5)

the seepage rate is the velocity of groundwater across the sediment-water interface; and [CH₄]_pw is the maximum porewater methane concentration measured at each site. This same approach has been used for the methane results collected in Florida Bay.

We used a one-dimensional, vertical advection-diffusion model to estimate radon exchange from sediments. This model was initially developed by Craig (1969) to describe radiocarbon profiles in the deep sea and later applied by Cable et al. (1996) to model ²²²Rn fluxes from marine sediment. The flux of radon supplied to the overlying water can be estimated with:

(6)

where C is the radon concentration in the sediments; z is depth positive downwards; K_z is the vertical diffusivity; ¶C²/¶z² and ¶C/¶z are the ²²²Rn concentration gradients across the sediment-water interface for diffusion and advection, respectively; w is the vertical advective velocity; P is a zero-order production term; and lC is radioactive decay. To model radon activity in sediments, K_z is set equivalent to D_s, the effective wet sediment diffusion coefficient, after corrections for temperature and sediment tortuosity. For ²²²Rn in sediments, P is the result of radon present in pore fluids due to recoil after production by decaying ²²⁶Ra in mineral grains (P = lC_eq, where C_eq is defined for Eq. 1). Advection, w, and radioactive decay, l, represent losses from the sediments and are defined as negative terms in the model. We used this approach to estimate the advective flux necessary to balance the total radon inventories (including that lost via gas exchange to the atmosphere) for the three study sites in Florida Bay.

Fluxes to the Atmosphere

Gas exchange across the air-water interface is a significant sink for dissolved gases in coastal waters. The total flux to the atmosphere depends on the molecular diffusion produced by the concentration gradient across this interface and turbulent transfer, which is dependent on physical processes. The flux (J_atm) of a soluble gas across the air-water interface can be calculated from the equation from MacIntyre et al. (1995):

(7)

where C_w and C_atm are the concentration of the gas of interest in surface water and air, respectively (dpm m^-3); a is Ostwald’s solubility coefficient (dimensionless); and k is the gas transfer coefficient (m min^-1). The gas transfer coefficient is a function of the kinematic viscosity, molecular diffusion of the gas, and turbulence at the interface, which is dependent on wind speed (Jahne et al., 1987; Wanninkhof, 1992; MacIntyre et al., 1995; Bugna et al., 1996; Corbett et al., 1997). Gesell (1983) showed that the atmospheric ²²²Rn concentration was on the order of 220 to 890 dpm m^-3. An average value of 560 dpm m^-3 was used for the flux calculations presented here.

VI. Findings

Tracer concentrations

Groundwater samples collected on and offshore exhibited elevated tracer concentrations relative to surface waters (Table 1). Both radon and methane display considerable spatial variation in groundwaters (82 - 1,124 dpm L^-1 and 10 - 16,604 nM, respectively). Although we do not have extensive results for either parameter at one location, we did note that radon did not vary significantly in the same well in measurements collected over a year apart (April 1995 = 291 ± 58 dpm L^-1, June 1996 = 342 ± 118 dpm L^-1). Although the two gases are produced by different processes, there is a statistically significant correlation between them in these groundwater samples (r = 0.46, n = 47, p < 0.01). Radon and methane concentrations in groundwater averaged 80 and 50 times greater, respectively, than that of surface waters in Florida Bay.

Surface water radon and methane concentrations in Florida Bay (Table 2) varied from <1 dpm L^-1 to >20 dpm L^-1 and 5 to 100 nM, with an overall average and standard deviation of 4.8 ± 2.7 and 27 ± 26, respectively (note that the average excludes samples collected from canals and solution holes). Radon and methane samples collected from the reef-side of the Keys varied from <1 dpm L^-1 to approximately 20 dpm L^-1 and 4 to 40 nM, with an overall average of 1.5 ± 1.4 and 11 ± 6, respectively (Table 2). Highest concentrations were observed nearshore, not along the reef tract. As with groundwaters, radon and methane were also statistically correlated on both the bay-side (r = 0.51, n = 191, p < 0.01) and the reef-side (r = 0.81, n = 84, p < 0.01) of the Keys. Radon and methane were statistically correlated in all surface waters sampled throughout the Keys, consistent with a common source such as groundwater discharge.

Table 1: Concentrations of ²²²Rn and CH₄ in groundwater wells. Uncertainties represent the standard error of the mean for multiple measurements at several sites (n represents the number of different sites).

Date/Site	Rn-222	Methane
	(dpm/L)	(nM)

February 1995
NURC, Key Largo	537 ± 4 (n = 2)	96 ± 78 (n = 2)

April 1995
Offshore Wells, Atlantic-Side	455 ± 35 (n = 12)	465 ± 150 (n = 11)
Offshore Wells, Bay-Side	641 ± 169 (n = 3)	655 ± 122 (n = 3)
Ranger Station, Key Largo	338 ± 47 (n = 2)	322 ± 172 (n = 2)

May 1996
Keys Marine Lab, Long Key	245 ± 13 (n = 28)	998 ± 503 (n = 2)
Ranger Station, Key Largo	442 ± 100 (n = 2)

December 1996
Offshore Wells, Bay-Side	615 ± 59 (n = 16)	2520 ± 1189 (n = 16)

June 1997
Offshore Wells, Bay-Side	294 ± 21 (n = 8)	545 ± 176 (n = 8)

Total Average =	398± 24 (n = 73)	1241 ± 452 ( n = 44)

Tracer Distributions in Florida Bay

General trends in surface water concentration were established by contouring data from each tracer survey with a kriging method by use of the software package Surfer^® (Golden Software). Although kriging interpolates between data points, creating some artifacts, the general trends described are independent of the contouring method or a reasonable change in contouring concentration. Examination of these contour plots showed very little apparent seasonal variation throughout the study period. The lack of a seasonal trend may have resulted from our sampling periods which favored the summer months or this may be a result of the primary fluid driving force near the Keys, i.e. tides, which do not have a seasonal trend. Later results did suggest seasonal variations throughout the Florida Bay (Burnett et al., 1998; Brand and Top, 1998).

Table 2: Average ²²²Rn and CH₄ concentrations from samples collected in various surface waters.

Site	Rn-222 (dpm/L)	Methane (nM)

Canals	19 ± 11 (n = 10)	830 ± 1140 (n = 10)

Garden Cove Spring, Key Largo	66 ± 19 (n = 4)	141 ± 176 (n = 4)

Garden Cove Surface Water, Key Largo	4.3 ± 1.2 (n = 4)	41 ± 11 (n = 2)

Lois Key Spring, Sugarloaf Key	122 ± 2 (n = 2)	493 ± 41 (n = 3)

Porjoe Key Interstitial Fluid (seepage meter)	67 ± 1 (n = 1)	176 ± 11 (n = 3)

Porjoe Key Surface Water	0.2 ± 0.1 (n = 1)	7.0 ± 0.2 (n = 3)

Bay Surface Water Average	4.8 ± 2.7 (n = 178)	27 ± 26 (n = 173)

Reef Surface Water Average	1.5 ± 1.4 (n = 57)	11 ± 6 (n=57)

During each period we sampled, high concentrations of both tracers were observed near the Keys. Plots for fall 1995 and summer 1997 show the typical trends observed (Fig. 3-4). As an alternative way to present the natural tracer data and evaluate spatial differences, samples were grouped into four different categories according to region. Regions include samples taken near the north coast (within ~3 km of the Everglades coastline), Keys bay-side (within ~3 km of the Keys coast), mid northeast bay (east of Black Betsy Keys, Fig. 1 G), and mid bay (west of Black Betsy Keys). Samples collected from the Keys bay-side were consistently more elevated in radon and methane than were samples from the other regions within the bay throughout the study period (Table 3). In particular, one of the narrowest areas of Key Largo (near Rock Harbor, Fig. 1 E) continually showed some of the highest tracer concentrations in surface waters on both the bay and reef side of the Keys, excluding canals and solution holes. Taken at face value, the tracer results suggest that the greatest degree of groundwater/surface water interaction in Florida Bay occurs along the bay-side of the Keys, and that groundwater input into the mid-bay, mid northeast bay and north bay regions is of lesser importance.

Table 3. Average tracer concentrations by region, and significant difference relative to Keys Bay-side. Keys bay-side was defined as sites located on the Florida Bay side of the upper Keys (Key Largo, Plantation Key and the Matecumbe Keys). Mid bay sites were typically basins within the mud-banked areas of the middle bay. North coast sties were along the Everglades Coast in muddy bottomed areas. Mid NE sites were in the northeastern areas of the bay and typically had very little sediments overlying a rock bay floor.

	Natural Tracers

Florida Bay Region	²²²Rn (dpm^.L^-1)	²²⁶Ra (dpm^.L^-1)	CH₄ (nM)	¹⁵N (‰)

Keys Bay-side (s, n)	4.38 (3.24, 73)	1.44 (0.59, 73)	38.2 (23.3, 73)	7.89 (2.54, 23)

Mid Bay (s, n, p)	2.23 (1.43, 40, <0.01)	1.42 (0.41, 40, 0.86)	16.4 (8.8, 40, <0.01)	3.92 (1.98, 26, <0.01)

N. Coast (s, n, p)	2.89 (2.15, 33, 0.02)	1.49 (0.48, 33, 0.63)	22.0 (19.3, 30, <0.01)	5.83 (2.26, 13, 0.02)

Mid N.E. (s, n, p)	2.40 (1.96, 32, <0.01)	1.95 (0.61, 32, <0.01)	13.2 (10.8, 30, <0.01)	5.57 (2.46, 7, 0.04)

Samples collected along the reef-side of the Keys showed very little variation throughout the study period. Surface water concentrations were relatively low on the reef-side (Table 2), except near Rock Harbor (Atlantic-side). Tracer concentrations near Rock Harbor were typically 2-4 times higher than other sampling stations on the reef-side for both radon and methane. Samples were also collected along the reef tract and from cracks within some of the healthy (e.g., Molasses, French; Fig. 1 C, D) and degraded reefs (e.g. Algae, Carysfort; Fig. 1 A, B). There was not any statistically significant difference between tracer concentrations from samples collected from cracks and surface waters or between degraded and healthy reefs. Concentrations along the reef tract were generally lower than samples collected near shore. These differences in concentration between the reef and near shore surface waters, as well as the lack of differences between surface water along the reef and water within the reef, is probably the result of the highly energetic environment along the reef tract and may indicate seepage in some nearshore areas. Dilution in the high-energy environment along the reef tract is expected as water within the reef is quickly exchanged with ambient surface water. At any rate, with the exception of some Atlantic-side areas near the Keys, especially near Rock Harbor, our data do not provide any evidence for groundwater directly discharging along the reef tract. This does not necessarily mean that the phenomena does not occur. Early studies by Simmons and Love (1987) indicated a mixture of freshwater and recirculated seawater may be discharging out on the reef. Subsequent reports have shown that the subsurface waters within the reefs were saline to hypersaline, demonstrating that there must be some exchange of surface and subsurface waters (Simmons, 1992; Shinn et al., 1994).

figure 3a
figure 3b

Figure 3. Contours of radon (A) in dpm L^-1 and methane (B) in nM for samples collected in October 1995. Solid crosses indicate sampling locations. Note the darker contours, indicating higher concentrations of both parameters, near the upper Keys.

figure 4a figure 4b

Figure 4. Contours of radon (A) in dpm L^-1 and methane (B) in nM for samples collected in June/July 1997. Solid crosses indicate sampling locations. Note the darker contours, indicating higher concentrations of both parameters, near the upper Keys.

Within the Keys, samples collected from artificial canals/trenches solution holes, locally referred to as springs, were extremely elevated in tracer concentrations and generally more saline than surficial waters at the time of sampling (Table 2). Three solution holes, or "submarine springs" were identified and investigated during this study: (1) Garden Cove spring, located on the Atlantic-side of N. Key Largo (25^o 10.22', 80^o 22.02', 1.5 meter ambient depth, Fig. 1); (2) Lois Key spring on the Atlantic-side of Surgarloaf Key (24^o 36.11', 81^o 27.48', 3 meter ambient depth); and (3) a solution hole located on the bay side of Big Pine Key, locally called "Four Corners" spring (24^o 45.00’, 81^o 24.28', 4.5 meter ambient depth, Fig. 1). Elevated tracer concentrations were measured in Lois and Garden Cove springs and in several canals, suggesting that subsurface fluids are actively seeping into and out of these features, and from them may spill into Florida Bay/Atlantic Ocean. The solution holes (Lois and Garden Cove) appear to be heavily influenced by the Atlantic tide. During high tide in the Atlantic, surface waters were pulled into the holes. During periods of low Atlantic tides, waters moved out of the holes at relatively high flow rates (Table 4). This is consistent with other observations of tidal-driven groundwater flow beneath the Keys. Water samples were collected during both high and low tides whenever possible. Not surprisingly, tracer concentrations of solution holes appear to fall on a mixing line between surface waters and groundwaters (Fig. 5). The natural tracer concentrations in groundwaters and samples collected from solution holes (Lois and Garden Cove) have a significant correlation (Fig. 5, r = 0.98, n = 9, p < 0.01). Radon and methane ratios for the two water masses are almost identical (groundwater Rn:CH₄ = 0.32 ± 0.12 dpmL^-1/mM, spring water Rn:CH₄ = 0.30 ± 0.19 dpmL^-1/mM; ratios are based on averages for each water mass). The similarities in the water masses suggest groundwater as a major source for the solution holes rather than rapidly recirculated surface water (reef-side surface water Rn:CH₄ = 0.13 ± 0.14 dpm L^-1/mM, bay-side surface water Rn:CH₄ = 0.12 ± 0.15 dpmL^-1/mM). Flow rates from Garden Cove spring at low tide were strong enough to produce a boil on the surface of the water on an outgoing tide.

Figure 5. Radon and methane concentrations in waters sampled throughout the Keys. The groundwater tracer concentrations are based on the overall average of all samples collected.

Unlike the other solution holes, Four Corners spring did not appear to be moving water in or out of the solution hole, which measured about two feet in diameter, during high or low tides. Samples for tracer analysis, collected in May, 1997 during a relatively dry period for the area, had similar concentrations to that of surface water The low rainfall and possible low water table may explain the lack of flow from the spring. Thus, some solution holes may be more dependent on rainfall than tidal influences.

Table 4. Nutrient concentrations of springs, groundwater, and surface waters.

Site	Flow Rate (mL m^-2 min^-1)	NH₄⁺ (mM)	N0₃^- (mM)	PO₄^3- (mM)	Salinity (ppt)

KML Well (15' and 60')¹ (n = 2)		13.3 ± 0.04	0.62 ± 0.48	0.98 ± 0.18


Canals/Trenches (n = 3)		6.2 ± 4.7	0.90 ± 0.33	0.07 ± 0.03

Garden Cove Spring, Key Largo (n = 3)²	(2.5± 0.3) x 10⁷	0.53 ± 0.15	0.40 ± 0.16	0.08 ±0.04	31

Garden Cove Surface, Key Largo (n = 3)		BD³	1.24 ± 0.09	BD	29

Lois Key Spring, Sugarloaf Key		12.03	0.1	0.94	38

Porjoe Key Interstitial Fluid (seepage meter)⁴	288 ± 48	15.17	0.68	0.03	25.7 ± 1.9 n = 4

Porjoe Key Surface		BD	1.14	BD	28.5 ± 0.5 n = 2

Bay Average (n = 27)		1.2 ± 1.5	1.1 ± 0.96	BD

Reef Average (n = 49)		BD	0.30 ± 0.38	BD

¹KML refers to Key Marine Laboratory located on Long Key, wells were within 10 meters of Class V sewage

injection well.

²Flow rate measured by a General Oceanics flow meter with low flow propeller.

³BD = Below Detection (NH₄⁺ = 0.01 uM; NO₃^- = 0.001 uM; PO₄^3- = 0.001 uM)

⁴Sample taken directly from seepage meter port. Seepage meter covers an area of 0.25 m².

Canals had a low tracer ratio (0.02 ± 0.03 dpmL^-1/mM) due to the higher methane concentrations measured in these features, probably due to the higher organic content in the underlying sediments. Canals are typically a low energy environment and therefore act as a sink for particulate matter. We thus observe low Rn:CH₄ ratios since decaying organic matter is a source for methane without corresponding radon production. The high organic content and low energy of the canals also tends to lead to eutrophic conditions (Lapointe and Clark, 1992; FDPC, 1973). In spite of these differences, the high radon concentrations in the solution holes and canals/trenches are consistent with a significant groundwater influx. It is likely that when these features were dredged, less permeable layers in the rock were removed, which resulted in greater conductivity between subsurface and surface waters.

Diffusive Flux Experiments

Results obtained from the laboratory core experiment provided three independent approaches to estimate the diffusive flux of radon: (1) direct observation of radon build up in the overlying water; (2) porewater profile; and (3) the sediment equilibration technique. Sediment characteristics of the laboratory core were similar to samples collected from other sites, although the ²²⁶Ra concentrations in the sediment were slightly higher (Table 5). Initially, ²²²Rn was monitored in the overlying water after the apparatus was sealed, analogous to a laboratory flux chamber (Fig. 6). The flux was then estimated by relating the inventory (I) of ²²²Rn in the overlying waters of the experiment as a direct function of the flux (J) with consideration for decay over time (t):

calculation 8 (8)

Table 5. Calculated and measured parameters for sediment samples collected from the laboratory core experiment.

Sediment Parameters
Porosity (F, mL cm^-3)^*	0.82 ± 0.06
Wet Bulk Density (g cm^-3)^†	1.2 ± 0.1
²²⁶Ra (dpm g^-1)^‡	1.9 ± 0.7
Effective Diffusion Coefficient (cm² sec^-1)	(7.6 ± 0.6) X 10^-6
²²²Rn Equilibration Concentration (dpm L^-1 porewater)^§	530 ± 90

, where W_D is the fraction of water present in the sediments, r_w is the density of sea water, and r_dry is the measured dry grain density.

^†

^‡Value is the mean and standard deviation of subsamples collected from core.

^§Reported value is the mean and standard deviation of top three sub-sections of core, representing the top 8 cm of the sediment bed.

Fitting Eq. 8 to the observed data over the entire sampling period (Fig. 6, solid line) resulted in an estimated diffusive flux of 1.8 ± 0.3 dpm m^-2 min^-1 (Table 6). The dashed lines represent 95% confidence limits of the fitted data.

After the completion of the experiment, the overlying water was then drained and a sub-core was extracted from the main core. The sub-core was sectioned into small intervals and analyzed immediately for ²²²Rn to obtain the porewater ²²²Rn concentrations. The resultant ²²²Rn profile showed a mid-depth maximum at approximately 5 cm, indicating diffusion both upward and downward (Fig. 7A). The change in ²²²Rn concentration at depth is due to a lower ²²⁶Ra content of the sediments deeper in the core (Fig. 7B, closed squares), which appears to be associated with the high organic matter content of the sediments deeper in the core (Fig. 7B, open circles). Carbonate muds dominated the top 6-8 cm of the core with an average ²²⁶Ra concentration of 2.3 ± 0.5 dpm g^-1, while the bottom section was composed primarily of peat with an average ²²⁶Ra concentration of 1.3 ± 0.3 dpm g^-1. A theoretical profile of ²²²Rn in the pore solutions may be constructed and an estimate of the diffusive flux of ²²²Rn into the overlying water can be obtained using Eq. 6 (Fig 7A, solid line), assuming advection is negligible and using the best fit of the top 8 cm of the core. The estimated diffusive flux obtained from the sediment profile was 0.9 ± 0.1 dpm m^-2 min^-1 (Table 6).

Figure 6. Rn-222 inventory in overlying water of experimental core. Solid line represents best fit to Eq. 8. Dashed lines represent 95% confidence limits of equilibrium concentration.

Table 6. Rn-222 flux estimates from experiments performed in the laboratory from sediments taken from Rock Harbor, Florida Bay.

Measurement Method	Radon Flux (dpm m^-2 min^-1)
Accumulation in Overlying water	1.8 ± 0.3
Porewater Profile	0.9 ± 0.1
Sediment Equilibration	1.1 ± 0.2

Figure 7. Porewater ²²²Rn concentrations in the experimental core (A). Theoretical profile based on best fit of top 8 cm. Dashed lines represent 95% confidence limits. Radium activity and organic matter content (represented by the % lost on ignition) display opposite trends in the experimental core (B).

Finally, surface sediment from the experimental core was run by the same sediment equilibration technique used for all the other samples in this study. Rn-222 equilibration values (C_eq) from samples collected in the top 8 cm averaged 530 ± 90 dpm L^-1 porewater. The diffusive flux estimated from Eq. 1 was 1.1 ± 0.2 dpm m^-2 min^-1 (Table 6). The sediment equilibration technique thus did not produce a statistically different diffusive flux than the other two methods used in the core experiment (t-test, a = 0.05). While the fluxes from the core experiment, at about 1-2 dpm m^-2 min^-1, are somewhat higher than those indicated by the sediment equilibration runs from the other sites (see below), they are consistent with the higher radium content in the experimental core (Fig. 7B) compared to that observed in the other surface sediments. At a ²²⁶Ra concentration of 2-3 dpm g^-1, the surface sediments in the core have a radium content about a factor of two greater than those from the other samples analyzed. In summary, we feel that these experiments validate the sediment equilibration approach and we will use those benthic flux estimates for our balance calculations.

Sediment samples collected from each study site were used to estimate the porosity, wet bulk density, effective diffusion coefficient, and ²²²Rn equilibration concentration (Table 7). Calculated diffusion coefficients from each site are in good agreement with other theoretical and experimental values obtained in comparable studies (Berner, 1980; Broecker and Peng, 1974; Martens et al., 1980; Hartman and Hammond, 1984). Estimated diffusive fluxes measured from the sites ranged from 0.5 to 0.7 dpm m^-2 min^-1 (Table 8), with no statistical difference between the field sites (t-test, a=0.05). Samples collected during the two time periods and duplicates run on splits from the same site were generally within 10-20 %. The average diffusive flux from each site will be used in all subsequent calculations.

The methane diffusive flux was estimated at Rock Harbor and Flamingo from the porewater concentration gradients obtained with peepers in January 1999. Measured porewater methane concentrations from the two sites resulted in similar concentration gradients in the upper portions of the sediment column (Fig. 8). Since the porewater concentrations were measured in situ, the resultant gradient represents a potentially enhanced diffusive flux due to the advection of groundwater increasing the concentration gradient near the sediment/water interface. However, it should be noted that these measurements were only made once during this study and does not account for potential seasonal variation. The estimated methane diffusion from Rock Harbor and Flamingo were 0.3 ± 0.1 and 0.4 ± 0.1 nmoles m^-2 min^-1, respectively (Table 8). Since the diffusive flux at these two sites were so similar and all the field sites are not very different geologically, the average diffusive flux has been applied for subsequent calculations at all three sites.

Water Column Tracer Inventory and Mass Balance

Initially, a mass balance of the tracers was completed utilizing Eq. 2 and 3 and the simple box models (Fig. 2). Assuming that all other sources and sinks can be estimated, the mass balance equation can be solved for the benthic flux necessary to balance the tracer inventory. Rn-222 samples have all been corrected for ²²⁶Ra in the water column (see “Approach”). By initially subtracting the ²²⁶Ra concentration in the water column from the total ²²²Rn, giving excess ²²²Rn, in situ production of ²²²Rn can be neglected in further calculations, i.e., the production term has already been incorporated into the ²²²Rn results. The average excess ²²²Rn concentration in the water column was used to estimate the loss due to decay. Estimates of the production and oxidation of methane in the water column were taken from Bugna et al. (1996). Wind speed data used to estimate the loss of the gases to the atmosphere were obtained from NOAA weather stations located at Key West and Miami, Florida. An average wind speed from these two locations over 30 years was 3.8 ± 0.8 m s^-1., giving a calculated value of the exchange coefficient (k) for ²²²Rn and CH₄ of (5.8 ± 1.8) X 10^-4 and (6.9 ± 2.1) X 10^-4 cm s^-1, respectively. This long term average agrees with the wind speeds observed during our field experiments in August 1998 and January 1999 of 2.5 ± 0.9 and 4.3 ± 1.6 m s^-1, respectively. The calculated benthic fluxes and the values used to obtain them are presented in Tables 9-10. Rn-222 estimated fluxes are presented as an average of the two sampling periods, since the difference in water column concentrations was not significant. The ²²²Rn total benthic flux at all the sites ranged from 2.8 to 8.0 dpm m^-2 min^-1, with Rock Harbor having the highest flux (Table 9). The CH₄ flux has been separated into the two sampling periods, August 1998 and January 1999 (Table 10), since higher water column concentrations in August led to higher atmospheric evasion rates and thus a corresponding higher benthic flux to balance this sink. Methane concentrations in January 1999 were approximately four times lower than in August, owing to the much lower benthic flux. Methane benthic fluxes ranged at all the sites from 18.2 to 42.3 nmoles m^-2 min^-1 and 2.9 to 11.1 nmoles m^-2 min^-1 in August 1998 and January 1999, respectively. The highest estimated methane flux was calculated for the Flamingo site during both sampling periods.

Table 7. Calculated and measured parameters for samples recovered from three sites in Florida Bay, USA.

Site Parameters	Rock Harbor	Flamingo	Rabbit Key Basin
Mean Water Column Radon Concentration (dpm L^-1)^*	7.5 ± 4.5	8.6 ± 2.0	5.2 ± 1.5
Mean Water Column Methane Concentration (nM)^* August, 1998 January, 1999	42.6 ± 4.5 10.2 ± 5.4	44.8 ± 7.5 12.2 ± 3.8	9.9 ± 1.6
Porosity (F, mL cm^-3)^†	0.57 ± 0.04	0.79 ± 0.05	0.77 ± 0.04
Wet Bulk Density (g cm^-3)^‡	1.5 ± 0.1	1.3 ± 0.1	1.3 ± 0.1
²²⁶Ra (dpm g^-1)	0.8 ± 0.2	1.4 ± 0.4	2.1 ± 0.4
Effective Radon Diffusion Coefficient (cm² sec^-1)	(6.3 ± 1.0) X 10^-6	(6.9 ± 1.0) X 10^-6	(6.9 ± 1.0) X 10^-6
²²²Rn Equilibration Concentration (dpm L^-1 porewater)	400 ± 120	350 ± 160	390 ± 50
Methane Porewater Concentration (nmoles L^-1 porewater)	460^§	1400^§	710 ± 350^**

^*Values represent the mean and standard deviation of several water column samples collected during flux experiments.

^†

, where W_D is the fraction of water present in the sediments, r_w is the density of sea water, and r_dry is the measured dry grain density.

^‡

^§Values are based on the highest concentration measured in the in the sediment column.

^**^{Average and standard deviation of samples collected from shallow
groundwater wells in Florida Bay (n=6).}

Table 8: Field and laboratory measurements of ²²²Rn and methane flux from the sediments at three sites in Florida Bay, USA.

Method	Methane Flux (nmoles m^-2 min^-1)	Radon Flux (dpm m^-2 min^-1)
Flux Chambers
Rock Harbor
August, 1998 (n = 9)	6.6 ± 1.2	4.4 ± 1.7
January, 1999 (n = 7)	7.3 ± 0.7	3.8 ± 0.6
Flamingo
August, 1998 (n = 3)	22.0 ± 7.5	2.3 ± 0.4
January, 1999 (n = 3)	7.2 ± 2.0	3.6 ± 0.9
Rabbit Key Basin (n = 3)	2.7 ± 0.8	1.5 ± 0.5
Diffusive Flux Estimates
Sediment Equilibration
Rock Harbor		0.5 ± 0.2
Flamingo		0.6 ± 0.2
Rabbit Key Basin		0.7 ± 0.1
Peeper
Rock Harbor	0.3 ± 0.1
Flamingo	0.4 ± 0.1

Figure 8. Porewater methane concentrations measured at Rock Harbor and Flamingo. Rock Harbor reaches an equilibrium concentration below 14-cm.

Table 9. Measured and calculated values of parameters used to estimate the total benthic flux necessary to balance the ²²²Rn inventory observed in the water column and that estimated to have been lost to the atmosphere.

Site Parameters	Rock Harbor	Flamingo	Rabbit Key Basin
Inventory (dpm m^-2)	10500 ± 6300	6900 ± 1600	9400 ± 2700
Sources (dpm m^-2 min^-1)
Total Benthic Flux^*	5.6 ± 2.2	5.5 ± 1.8	4.2 ± 1.4
Sinks (dpm m^-2 min^-1)
In Situ Decay	1.3 ± 0.8	0.9 ± 0.3	1.2 ± 0.3
Atmospheric Evasion	4.3 ± 1.4	4.6 ± 1.5	3.0 ± 1.0

^*Estimated using Eq. 2, other values in the table, and solving for the total benthic flux.

^†Values estimated by the sediment equilibration method.

Table 10: Measured and calculated values of parameters used to estimate the total benthic flux necessary to balance the CH₄ inventory observed in the water column and that estimated to have been lost to the atmosphere in August 1998 and January 1999.

Site Parameters	Rock Harbor	Flamingo	Rabbit Key Basin
Inventory (nmoles m^-2)
August, 1998	60000 ± 6000	45000 ± 8000
January, 1999	14000 ± 7000	12000 ± 4000	18000 ± 3000
Sources (nmoles m^-2 min^-1)
Total Benthic Flux^*
August, 1998	28.6 ± 10.4	30.2 ± 11.5
January, 1999	5.9 ± 3.1	7.4 ± 3.5	5.8 ± 3.0
Net Methane Production	0.4 ± 0.5	0.4 ± 0.5	0.4 ± 0.1
Sinks (nmoles m^-2 min^-1)
Net Methane Oxidation	0.8 ± 0.7	0.8 ± 0.7	0.8 ± 0.7
Atmospheric Evasion
August, 1998	28.2 ± 9.2	29.8 ± 10.0
January, 1999	5.5 ± 2.0	7.0 ± 2.4	5.4 ± 1.8

^*Estimated using Eq. 3, other values in the table, and solving for the total benthic flux.

The calculated benthic fluxes are as much as 100 times greater than the estimated diffusive fluxes. The magnitude of the calculated benthic fluxes is primarily determined by the atmospheric evasion rates in this shallow water environment. However, even if loss to the atmosphere was assumed to be negligible, there would still be an inventory of excess tracer (as much as 10 times) over that which diffusion alone could supply. Thus, there must be an additional source of these tracers to the water column. As previously noted, turbulent mixing, gas bubble ebullition, and groundwater discharge can all enhance the benthic flux. Gas bubble ebullition may enhance diffusive benthic fluxes by as much as 2.5 times due to an increase in the surface area of the sediment contact associated with invagination of the sediment by bubble tubes (Martens et al., 1980). The benthic fluxes estimated for Florida Bay appear to be greater than the effects of bubble ebullition alone. Turbulent mixing by bioturbation and current/wave mixing have often been invoked to explain near-bottom tracer inventories greater than that which could be supplied by diffusion (Hammond et al., 1977; Emerson et al., 1984; Hartman and Hammond, 1984; Rutgers van der Loeff et al., 1984). For example, Hartman and Hammond (1984) attributed total ²²²Rn benthic fluxes that exceeded those expected from molecular diffusion by as much as a factor of 4, to macrofaunal irrigation. However, many previous studies in coastal systems have not considered advection of groundwater as a potential source of the parameter of interest.

Benthic Chamber Experiments

An independent estimate of benthic flux was obtained experimentally using benthic flux chambers at all three field sites. Three benthic chambers were deployed at each site over a two day period. Multiple measurements were made on each chamber during this time and the water inside these chambers was monitored for radon and methane to establish a flux rate. The fluxes measured via the benthic chamber approach were consistently higher than could be accounted for by diffusion alone (Table 8). One explanation may be seepage of porewater into the chambers. Although the chambers were apparently closed, it may be difficult to halt the slow seepage of water into them. An experiment performed at the Rock Harbor site showed no significant difference in the tracer flux between a chamber that was apparently closed and another chamber allowed to seep through a small hole in the top. Artifacts caused by sediment resuspension and/or stirring may have contributed to the higher than expected values; however, other studies have also shown that the in situ flux of dissolved constituents (Hg) measured by these devices was elevated relative to the measured diffusive flux (Covelli et al., 1999). In any case, this method has provided some conservative estimate of the total benthic flux, which may include advection as well as diffusion.

There was little difference in the ²²²Rn flux at all sites and the CH₄ flux at Rock Harbor measured from the benthic flux chambers between August and January, although there was a significant change in the methane flux at the Flamingo field site between August and January (Table 8). A similar pattern can be seen in the surface water concentrations of the two trace gases. There was no significant difference in the radon concentration, thus all of the data were pooled for this analysis. However, the methane surface water concentrations changed dramatically at both field sites between the two sampling events, indicating a potential link between the surface water concentration and benthic flux (Table 7).

VII. Evaluation

A. Results for Project Goals and Objectives

Estimates of Groundwater Input

Estimates of the advective velocity of groundwater necessary to support the benthic fluxes determined from both in situ experiments and the water column mass balance calculations can be obtained by establishing the tracer concentration in the advecting fluids (Eq. 4 and 5). Since the fluids are seeping through the sediments, one would ideally use an in situ porewater concentration. However, methane porewater concentrations at the Flamingo field site did not show a constant concentration at depth (Fig. 8). We elected to use the highest concentration of methane measured in the porewaters at each site, since these concentrations are also within the range of previously reported methane groundwater concentrations in wells located throughout Florida Bay (Corbett et al., 1999) and groundwater samples collected during this study. This would also produce the most conservative estimate of groundwater inputs. Since an in situ methane concentration was not measured at Rabbit Key Basin, the average groundwater concentration in wells sampled in the bay during the same period was used as the advective fluid concentration for the site. Radon-222 concentrations in porewater were not obtained from the field sites during the sampling trips. However, equilibrium concentrations measured in the laboratory for the sediment equilibration experiments have been used in previous studies as an estimate of the in situ ²²²Rn porewater concentration (Cable et al., 1996; Corbett et al., 1997). In addition, we note that the estimates of the equilibrium ²²²Rn concentration (Table 7) are within one standard deviation of groundwater values reported in Table 1. Furthermore, these equilibrium values are similar to the porewater concentrations measured in the test core (Fig. 7). Using the estimated porewater concentration, the seepage velocity, which is essentially the upwelling rates of subsurface fluids carrying these trace gases to overlying waters, can be calculated for our results by applying Eq. 4. Estimates of seepage velocity by the mass balance calculations and the flux chambers for ²²²Rn and CH₄ range between 0.3 – 2.0 and 0.3 – 8.0 cm day^-1, respectively (Fig. 9). The average measured and calculated seepage velocities are all very close and typically within the estimated uncertainty of each other. All three study sites give an average advective velocity of 1.9 ± 0.45 cm day^-1.

Average seepage rates for the keys bay side, north coast, and in the mid-bay were 21.2 ± 5.2 (n = 17), 7.2 ± 2.5 (n = 6), and 13.4 ± 2.3 (n = 10) mL m^-2min^-1, respectively. Converting these seepage rates to velocity units show that all sites are in the range of 1 - 3 cm day^-1, in very good agreement with the average velocity estimated via tracers (1.9 cm day^-1). While this agreement could be fortuitous, it is encouraging that these two completely independent techniques converge at a common value for the upwelling velocity.

We have examined all the parameters in our tracer calculations to evaluate how sensitive our results are to the various assumptions and estimations. The calculated advective velocities based on the tracer mass balance estimates are most sensitive to the atmospheric evasion rate and the porewater tracer concentration. Wind speed and the water column tracer concentrations are the most sensitive parameters when calculating the loss to the atmosphere. To constrain our results, we solved Eq. 7 for several reasonable values of these two parameters for ²²²Rn (Fig. 10A). The different curves represent the ²²²Rn water column concentration. Values included in the plot represent almost the entire range of concentrations measured at all three sites. Rock Harbor, Flamingo, and Rabbit Key Basin had ²²²Rn water column concentrations averaging 7.5 ± 4.5 , 8.6 ± 2.5, and 5.2 ± 1.5 dpm L^-1, respectively. We assumed an upper limit for surface water ²²²Rn of 9.0 dpm L^-1. Note that any further increase in water column concentration would result in a further increase in the required advective velocity by the mass balance approach. It is our intention in this sensitivity analysis to verify the lower limit of the results, i.e., to evaluate if we made any assumptions that have resulted in too high an advective velocity. The boxed area in Figure 10 includes the 30-year average wind speed and standard error estimate from the NOAA weather stations at Key West and Miami as well as the reasonable range in ²²²Rn concentrations. Based on these assumptions, we estimate a reasonable atmospheric evasion rate of ²²²Rn ranging from 1.0 to 5.0 dpm m^-2 min^-1. Using these estimates together with a reasonable range for in situ decay (which is well constrained), we estimate that the total benthic flux from the sediments must lie between 1.6 and 6.5 dpm m^-2 min^-1 (Eq. 2). This range in total benthic flux is between 2 and 13 times greater than the amount that diffusion alone could supply to the water column. Even if we assume that there is no loss of ²²²Rn to the atmosphere, there is still a 100% excess of ²²²Rn in the water column relative to that supplied by diffusion.

Figure 9. Seepage velocities calculated from measurements made by flux chambers, tracer mass balance, and seepage meters.

Figure 10. Sensitivity analysis performed on key parameters for calculating advective groundwater velocity following the mass balance model. Radon loss to the atmosphere (A) is most sensitive to wind speed and water column concentration. Curves represent varying water column concentrations. The porewater concentration and atmospheric flux are the most sensitive parameter when estimating the advective flux (B). The curves represent the range of the most reasonable values of total benthic fluxes of ²²²Rn. The best estimate for each field site has also been included.

In order to calculate the advective velocity via ²²²Rn, Eq. 6 not only requires an estimate of the total benthic flux (see above), but the porewater ²²²Rn concentration as well (Fig 10B). Fortunately, we are able to constrain this value fairly well in this system. Corbett et al. (1999) measured the ²²²Rn concentrations from wells in several locations of Florida Bay and the observed range is shown in Fig 10B by the two vertical lines. In addition, the ²²²Rn equilibrium concentrations (C_eq) measured from the three sites all fall within this range (Table 7). Based on the estimated total benthic flux and the observed groundwater concentrations, the calculated advective velocities must lie between 0.4 and 3.0 cm day^-1. Advective velocities calculated specifically for the three field sites are well within these reasonable limits (shown as open symbols in Fig.10B). This same “sensitivity” analysis was performed for the CH₄ model and gave similar results.

Nutrient Implications

Samples for nutrient analyses were collected and analyzed from select surface waters, groundwaters, solution holes, and canals (Table 4). It is interesting to note that many phosphate concentrations are below detection, except for samples from areas also characterized by high concentrations of groundwater tracers. Surface waters, relative to groundwater, canals, etc., were typically low in nutrient concentrations. Nitrate was the only parameter present in all waters sampled. The presence of nitrogen and the continued absence of phosphate in the Florida Bay waters reconfirms that the majority of the bay appears to be a phosphate-limited environment (Powell et al., 1989; Fourqurean et al., 1992a; Fourqurean et al., 1992b; Fourqurean et al., 1993). On average, nitrate and ammonia concentrations were equal to each other within Florida Bay.

Although the nutrient content of these various water masses appears low, the groundwater flux may still be important. For instance, Garden Cove spring waters have relatively low nitrogen and phosphate concentrations, yet contribute approximately 1.2 x 10⁷ mmoles N m^-2y^-1 and approximately 1.0 x 10⁶ mmoles PO₄^3-m^-2y^-1, based on the area of the solution hole, the nutrient concentration in the water exiting the solution hole, and the flow out of the orifice measured by a hand-held mechanical flow meter (General Oceanics). We recognize that estimating nutrient fluxes using seepage meters is somewhat tenuous because of possible artifacts due to the spatial variability of groundwater flow in sediments and possible anomalies associated with the water flux measurements due to wave action and currents (Libelo and MacIntyre, 1994). However, reliable results have been produced under similar conditions as those faced here in other environments (Bugna et al., 1996, Cable et al., 1996, Cable et al., 1997). We thus present these nutrient flux estimates as a first approximation of how the magnitude of groundwater-associated influx compares to freshwater surface discharge estimates.

As with tracer data, seepage results collected throughout each region were pooled into groups according to location: (1) Keys bay-side; (2) north coast; and (3) mid-bay. Average seepage rates for these regions are 21.2 ± 5.2 (n = 17), 7.2 ± 2.5 (n = 6), and 13.4 ± 2.3 (n = 10) mL m^-2min^-1, respectively (error presented as standard errors, n represents the number of sites sampled within the area, each site had at least 3 measurements employing multiple meters). The standard errors of these averaged estimates are primarily due to the large areas over which the measurements were obtained and the inherent spatial variability associated with groundwater flow. However, the same general trend was observed with these direct groundwater flux measurements as those inferred from the natural tracer measurements, i.e., the area of most apparent groundwater/surface water interactions was on the inside of the Keys within Florida Bay.

One seepage meter located near Porjoe Key (Fig. 1F), in a circular seagrass bed, had an extremely high flow rate (288 ± 48 mL m^-2min^-1, Fig. 11A, meter 32) relative to the average rate for the area (13.4 ± 2.3mL m^-2min^-1; Fig. 11A). These circular seagrass beds grow in depressions in the hard limestone bay floor. They are sediment filled and are postulated to have formed as dissolution pits during low stands of sea level (Zieman et al., 1972). Similar features have been observed in the Everglades, and there is radioisotopic evidence that is consistent with groundwater flow through these features (C. Holmes, USGS, pers. comm.). At Porjoe Key, we observed a large flow of brackish water from the site, indicative of conduit flow from a deeper less saline aquifer. The salinity of the interstitial water from the fast flowing seepage meter and another meter in close proximity were significantly different (25.7 ± 1.9 psu, n=4, p<0.05) than the ambient seawater (28.5 ± 0.5 psu, n=2) measured by silver nitrate titration (Fig. 11B). In addition, tracer samples collected directly from the seepage meter showed significantly higher concentrations than that of surface waters (Fig. 11C). The radon and methane ratio of the seeping water was almost identical to that of groundwaters (seeping water Rn:CH₄ = 0.38 ± 0.07 dpmL^-1/mM, groundwater Rn:CH₄ = 0.32 ± 0.79 dpmL^-1/mM). Collectively, these data may indicate a fresher groundwater source.

Figure 11. Seepage rates (A), salinity (B), and tracer concentrations (C) measured at Porjoe Key, north of Key Largo (Fig. 1 F) on May 9, 1996. Salinity of seepage meters is the average of four samples taken from two different meters. There is a significant difference (p<0.05) in the salinity between the meter and the overlying water. Note that tracer concentrations are plotted on a log scale.

From the interstitial nitrogen and reactive phosphate concentrations measured at the Porjoe seepage site (Table 4) and the average seepage rate for the entire mid-bay area (13.4 ± 2.3mL m^-2min^-1), nutrient input associated with groundwater flow was estimated. Based on this calculation, nitrogen and reactive phosphate contribution from advective flow could be as great as approximately 110 ± 19 mmol N m^-2y^-1 and 0.21 ± 0.04 mmol PO₄^3- m^-2y^-1. This may be a conservative estimate since pore water samples collected by Fourquean et al. (1992a) at 18 sites across the Bay had considerably higher nitrogen (NH₄) and reactive phosphate concentrations, 78.6 mM and 0.34 mM, respectively. Although this is described as a groundwater flux, we are not ruling out the possibility that a major source of these nutrients may be from organic matter degradation and recycled nutrients in the surficial sediment. Thus, all of the nutrients may not represent a true groundwater source or “new” nutrient input. However, the advective motion of the groundwater may promote the delivery of recycled nutrients to surface waters.

Surface water inputs from the Everglades are currently considered one of the more important nutrient sources to the eastern portion of Florida Bay (area east of the 80^o40', just west of Black Betsy Keys, Fig. 1G). This area of Florida Bay is relatively isolated from the Gulf of Mexico, which has been shown to be the major contributor of total nitrogen to the Bay's nutrient budget (Rudnick et al., 1999). Measurements made by Rudnick et al. (1999) from major surface water contributors to eastern Florida Bay (Taylor Slough and C-111) were used to estimate the mean annual input of total nitrogen and total phosphate at approximately 1.8 X 10¹⁰ mmolyr^-1 and 8.4 X 10⁷ mmol yr^-1, respectively. In order to compare these surface water contributions from the Everglades to the estimated groundwater input calculated above, we normalized both the surface input and the groundwater flux to the estimated area of the eastern Florida Bay (600 km²). The normalized flux of surface water inputs (TN = 30 mmol m^-2y^-1 and PO₄^3- = 0.1 mmol m^-2y^-1) are similar, although lower, to the groundwater estimates for the Porjoe Key area (TN = 110 mmol m^-2y^-1; PO₄^3- = 0.2 mmol m^-2y^-1). We assume that the groundwater seepage rate and nutrient concentrations to be equal over the entire area. Although this assumption is highly unlikely due to the spatial variability in groundwater flow and in nutrient concentrations of interstitial sediments, the contribution of nutrients from groundwater appears to be at least of the same magnitude as the estimated freshwater sources from the Everglades in the eastern area of Florida Bay. We also note that while tidal springs or solution holes may be relatively rare, their potential contributions to the nutrient flux per unit area appears to be large.

Also, consider the possible nitrogen contribution for an area on the bay-side of the Keys, Tavernier Basin (approximate area and depth = 1.9 x 10⁶ m², 2 m), which is also located in the eastern area of Florida Bay. Seepage measurements collected along the bay-side of the Keys, which included this basin, averaged 21.2 ± 5.2 (n = 17) mL m^-2min^-1. Dissolved N-species and phosphate in groundwaters along the Keys ranged from 15-50 µM and 0.03-1.3 µM, respectively (Shinn et al. 1994). However, Kump (1998) showed nitrogen and phosphate concentrations in groundwaters surrounding a sewage injection facility to be as high as 200 µM and 50 µM, respectively. Areas impacted by these sewage injection wells could potentially provide a much greater nitrogen and phosphorous flux to surface waters. From the more conservative data presented by Shinn et al. (1994), we can then estimate the nitrogen and phosphate flux to Tavernier Basin to be between 160 - 560 mmol N m^-2y^-1 and 0.3 - 10 mmol PO₄^3- m^-2y^-1, respectively. This range in nitrogen flux is similar to estimates of inorganic N input of three large river-dominated Atlantic estuaries: Long Island Sound, Chesapeake, and Pamlico River Estuaries have an estimated nitrogen input from rivers of 400, 510, and 860 mmol N m^-2y^-1, respectively (Nixon and Pilson, 1983). Based on these calculations, it appears that groundwater could be a significant nutrient source to surface waters within the eastern area of Florida Bay, especially in certain areas along the inside of the Keys.

If we accept an upwelling rate of 1.9 cm day^-1 with an average depth of Florida Bay of about 2 m, it would only take approximately 105 days to replace the surface waters in Florida Bay with groundwater alone. Of course inputs from the Gulf of Mexico would reduce this residence time of water within the bay significantly. Rudnick et al. (1999) estimated surface water flow and nutrient fluxes (total nitrogen, total phosphate, dissolved inorganic nitrogen) from the Everglades, the interaction of Florida Bay with the Gulf of Mexico, and provided estimates of atmospheric nutrient deposition. Their results indicate that the integrated water inflow to Florida Bay from the Gulf of Mexico (3.6 X 10¹⁰ m³ y^-1) is much greater than that of Shark River Slough (1.0 X 10⁹ m³ y^-1) and Taylor Slough/C-111 (2.9 X 10⁸ m³ y^-1), the only other sources of surface water. The Gulf of Mexico, although contributing a large volume of water to the over all budget of Florida Bay, probably only skirts the western banks of the bay and then flows south through passes in the lower keys and into the Atlantic (Rudnick et al., 1999; Fourqurean and Roblee, 1999). The flux of water and nutrients from the Gulf of Mexico to the inner northeastern sector of the bay may not be as important as the large volume may indicate, i.e., this area of Florida Bay may be in a “shadow” of Gulf of Mexico waters. This could be an area where the relative nutrient contribution of groundwater is most important.

Although groundwater flow and composition is typically heterogeneous, using our best estimate (1.9 cm day^-1) provides a first approximation on the significance of this source in comparison to those presented by Rudnick et al. (1999). This estimated velocity has been shown to occur at least three different locations throughout the Bay. Assuming this advective velocity occurs over the entire bay (2220 km²), groundwater could provide as much as 1.5 X 10¹⁰ m³ y^-1 of water to the bay, much more than the surface freshwater sources and just under half as much as the Gulf of Mexico. Interestingly, Rudnick et al. (1999) suggests that the Gulf of Mexico provides at least twelve times more total nitrogen (TN) and total phosphate (TP) than any other source measured. The average TN and TP concentration of the Gulf waters measured during their observations were 21 ± 2 mM and 0.4 ± 0.04 mM, respectively. Dissolved TN and TP concentrations as high as 50 mM and 1.3 mM, respectively, were observed in groundwaters along the keys by Shinn et al. (1994). Thus, it is conceivable that groundwater could provide a significant amount of nutrients to Florida Bay.

B. Dissemination of Project Results

The results reported here have been disseminated to the scientific community by presentations at conferences and publication in scientific journals. The most significant of these communications are listed below:

Burnett, W.C., J.P. Chanton, D.R. Corbett and K.S. Dillon, 1998. Natural tracers, nutrients, and groundwater in Florida Bay. In: Proceedings of the 1998 Florida Bay Science Conference, Miami, Florida.
Corbett, D.R., J. Chanton, W. Burnett, K. Dillon, C. Rutkowski, and J. Fourqurean. 1999. Patterns of groundwater discharge into Florida Bay. Limnol. Oceanogr., 44, 1045-1055.
Corbett, D.R., K. Dillon, W. Burnett, and J. Chanton, 2000. Estimating the groundwater contribution into Florida Bay via natural tracers ²²²Rn and CH₄. Limnology & Oceanography, submitted.

Acknowledgments

We would like to thank Paul Carlson of Florida Department of Environmental Protection and Bill Kruczynski of the Environmental Protection Agency, and Gene Shinn and Chris Reich of the USGS for their guidance and assistance. The staff at the National Park Service on Key Largo also greatly facilitated our efforts. Stephen Miller and Otto Rutten of NOAA-NURC of Key Largo were of major assistance in most of the offshore work as well, providing boat, lab and housing support on several occasions. We thank Brian Fry for running some of the ¹⁵N samples at FIU. We would also like to thank Behzad Mortazavi and others for their thoughtful comments.

References

Back, W., B.B. Hanshaw, T.E. Pyler, L.N. Plummer and A.E. Weide, 1979. Geochemical significance of groundwater discharge in Caleta Xel Ha, Quintana Roo, Mexico. Water Resources Res. 15: 1521-1535.

Berelson, W.M., D.E. Hammond, D.O. O’Neill, X-M. Xu, C. Chin, and J. Zukin. 1990. Benthic fluxes and porewater studies from sediments of the equatorial north Pacific: nutrient diagenesis. Geochim. Cosmochim. Acta 54, 3001-3012.

Berner, R.A. 1980. Early Diagenesis: A theoretical approach. Princeton University Press, Princeton, N.J. 241 pp.

Boesch, D., N. Armstrong, C. D'Elia, N. Maynard, H. Paerl and S. Williams. 1993. Deterioration of the Florida Bay Ecosystem: An evaluation of the scientific evidence. Interagency Working Group Report to National Fish and Wildlife Foundation.

Bokuniewicz, H., 1980. Groundwater seepage into Great South Bay, New York. Estuarine Coastal Mar. Sci. 10: 437-444.

Braman, R.S. and S.A. Hendrix. 1989. Nanogram nitrite and nitrate determination in environmental and biological materials by vanadium (III) reduction with chemiluminescence detection. Anal. Chem. 61: 2715-2718.

Brand, L. and Z. Top, 1998. The role of groundwater in the Florida Bay ecosystem. In: Proceedings of the 1998 Florida Bay Science Conference, Miami, Florida.

Broecker, W.S. 1965. An application of natural radon to problems in oceanic circulation. In: Proc. Symp. On Diffusion in the Oceans and Freshwaters. Lamont Geological Observatory, New York, pp. 116-145.

Broecker, W.S. and T.H. Peng. 1974. Gas exchange rates between sea and air. Tellus, 26.

Bugna, G.C., J.P. Chanton, J.E. Young, W.C. Burnett and P.H. Cable, 1996. The importance of groundwater discharge to the methane budgets of nearshore and continental shelf waters of the northeastern Gulf of Mexico. Geochim. Cosmochim. Acta. 60(23): 4735-4746.

Burnett, W.C., J.P. Chanton, D.R. Corbett and K.S. Dillon, 1998. Natural tracers, nutrients, and groundwater in Florida Bay. In: Proceedings of the 1998 Florida Bay Science Conference, Miami, Florida.

Cable, J.E., G. Bugna, W.C. Burnett, J. Chanton, 1996a. Application of ²²²Rn and CH₄ for assessment of groundwater discharge to the coastal ocean. Limnol. Oceanogr. 41(6): 1437-1444.

Cable, J.E., W.C. Burnett, J. Chanton and G.L. Weatherly, 1996b. Estimating groundwater discharge into the northeastern Gulf of Mexico using radon-222. Earth Planet. Sci. Lett. 144: 591-604.

Cable J., W. Burnett, J. Chanton, D.R. Corbett, and P. Cable, 1997. Field evaluation of seepage meters for coastal marine work. Est. Coast. Shelf Sci. 45, 367-375.

Capone, D.G. and M.F. Bautista, 1985. A groundwater source of nitrate in nearshore marine sediments. Nature 313: 214-216.

Capone, D.G. and J.M. Slater, 1990. Interannual patterns of water table height and groundwater derived nitrate in nearshore sediments. Biogeochemistry 10, 277-288.

Chanton, J.P., C.S. Martens, and C.A. Kelley. 1989. Gas transport from methane-saturated, tidal freshwater and wetland sediments. Limnol. Oceanog. 34, 807-819.

Cline, J.D. and I.R. Kaplan, 1975. Isotopic fractionation of dissolved nitrate during denitrification in the eastern tropical North Pacific Ocean. Mar. Chem. 3: 271-299.

Corbett, D.R., J. Chanton, W. Burnett, K. Dillon, C. Rutkowski, and J. Fourqurean. 1999. Patterns of groundwater discharge into Florida Bay. Limnol. Oceanogr., 44, 1045-1055.

Corbett, D.R., W.C. Burnett, P.H. Cable, and S.B. Clark. 1997, Radon tracing of groundwater input into Par Pond, Savanah River Site. J. Hydrol. 203, 209-227.

Corbett, D.R., W.C. Burnett, P.H. Cable, and S.B. Clark. 1998. A multiple approach to the determination of radon fluxes from sediments. J. Radioanalyt. Nucl. Chem. 236, 247-252.

Covelli, S., J. Faganeli, M. Horvat, and A. Brambati. 1999. Porewater distribution and benthic flux measurements of mercury and methylmercury in the Gulf of Trieste. Est. Coast. Shelf Sci. 48, 415-428.

Craig, H. (1969). Abyssal carbon and radiocarbon in the Pacific. J. Geophys. Res. 74, 5491-5506.

D'Elia, C.F., K.L. Webb, and J.W. Porter. 1981. Nitrate-rich groundwater inputs to Discovery Bay, Jamaica: a significant source of N to local coral reefs? Bull. Mar. Sci. 31, 903-910.

Dillon, K., D.R. Corbett, J.P. Chanton, W.C. Burnett and D.J. Furbish, 1999. The use of sulfur hexafluoride as a groundwater tracer of septic tank effluent in the Florida Keys. Jour. Hydrology, 220, 129-140.

Dillon, K., D.R. Corbett, J.P. Chanton, W.C. Burnett and L. Kump, 2000. Rapid transport of a wastewater plume injected into saline groundwaters of the Florida Keys, USA. Ground Water, 38, 624-634.

Emerson, S., R. Jahnke and D. Heggie. 1984. Sediment-water exchange in shallow water estuarine sediments. J Mar. Res. 42, 709-739.

Florida Department of Pollution Control, 1973. Survey of water quality in waterways and canals of the Florida Keys with recommendations. Final Report.

Fourqurean, J.C. Zieman, and G.V.N. Powell, 1992a. Relationships between porewater and seagrasses in a subtropical carbonate environment. Mar. Biol. 114: 57-65.

Fourqurean, J.C. Zieman, and G.V.N. Powell, 1992b. Phosphorous limitation of primary production in Florida Bay: Evidence from C:N:P ratios of the dominant seagrass Thalassia testudinum. Limnol. Oceanogr. 37(1): 162-171.

Fourqurean, J.W., R.D. Jones, and J.C. Zieman, 1993. Processes influencing water column nutrient characteristics and phosphorous limitation of phytoplankton biomass in Florida Bay, USA: Inferences from spatial distributions. Estuarine, Coast. Shelf Sci. 36: 295-314.

Fourqurean, J.W., T.O. Moore, B. Fry and J.T. Hollibaugh, 1997. Spatial and temporal variation in C:N:P ratios, d¹⁵N, and d¹³C of eelgrass Zostera marina as indicators of ecosystem processes, Tomales Bay, California, USA. Mar. Ecol. Prog. Ser. 157: 147-157.

Fourqurean, J.W. and M.B. Robblee. 1999. An overview and history of recent ecological changes and scientific research in Florida Bay. Estuaries. 22, 345-357.

Fry, B., 1994. Introductory address at stable isotope session at Second Coastal Wetland Ecology and Management Symposium. Key Largo, Florida.

Gesell, T.F. 1983. Background atmospheric Rn-222 concentrations outdoors and indoors: A review. Health Phys. 45, 289-302.

Gruebel, K.A. and C. S. Martens. 1984. Radon-222 tracing of sediment-water chemical transport in estuarine sediment. Limnol. Oceanogr. 29, 587-597.

Halley, R.B., H.L. Vacher, and E.A. Shinn. 1995, Geology and Hydrogeology of the Florida Keys. U.S. Geological Survey and Dept. of Geology, University of South Florida.

Halley, R.B., H.L. Vacher, E.A. Shinn, J.W. Haines, 1994. Marine geohydrology: dynamics of subsurface sea water around Key Largo, Florida. Abstracts with Programs, GSA Annual Meeting, Seattle, Washington: A-411.

Hammond, D.E., H.J. Simpson and G. Mathieu. 1977. Radon-222 distribution and transport across the sediment-water interface in the Hudson River estuary. J. Geophys. Res. 82, 3913-3920.

Harr, M.E., 1962. Groundwater and Seepage. McGraw-Hill, New York, 315 p.

Hartman, B. and D.E. Hammond. 1984. Gas exchange rates across the sediment-water and air-water interfaces in the South San Francisco Bay. J. Geophys. Res. 89, 3593-3603.

Hesslein, R.H. 1976. An in situ sampler for close interval porewater studies. Limnol. Oceanogr., 21, 912-914.

Horrigan, S.G., J.P. Montoya, J.L. Nevins and J.J. McCarthy, 1990. Natural isotopic composition of dissolved inorganic nitrogen in Chesapeake Bay. Est. Coastal Shelf Sci. 30: 393-410.

Huettel, M. and G. Gust. 1992. Solute release mechanisms from confined sediment cores in stirred benthic chambers and flume flows. Mar. Ecol. Prog. Ser. 82, 187-197.

Hunt, J. and W. Herrnkind, 1993. Sponge and lobster research. Final report to the Florida DNR, Contract C-8077. 9pp.

Jahne, B., K. O. Munnich, R Bosinger, A. Dutzi, W. Huber, P. Libner. 1987. On the parameters influencing air-water gas exchange. J. Geophys. Res. 92: 1937-1950.

Johannes, R.E., 1980. The ecological significance of the submarine discharge of groundwater. Mar. Ecol. Prog. Ser. 3: 365-373.

Key, R.M., N.L. Guinasso, Jr., and D.R. Schink. 1979. Emanation of radon-222 from marine sediments. Mar. Chem. 7, 221-250.

Kohout, F.A., 1960. Cyclic flow of salt water in the Biscayne Aquifer of southeastern Florida. Jour. Geophys. Res. 65: 2133-2141.

Kump, L., 1998. EPA Final Report, Contract #X994870-96-0.

Lapointe, B.E. and J. O'Connell, 1989. Nutrient-enhanced growth of Cladophora prolifera in Harrington Sound, Bermuda: eutrophication of a confined, phosphorus-limited marine ecosystem. Est. Coast. Shelf Sci. 28: 347-360.

Lapointe, B.E., J.D. O'Connell, and G.S. Garrett. 1990. Nutrient couplings between on-site sewage disposal systems, groundwaters, and nearshore surface waters of the Florida Keys. Biogeochemistry 10: 289-307.

Lapointe, B., and M. Clark, 1992. Nutrient inputs from the watershed and coastal eutrophication in the Florida Keys. Estuaries 15: 465-476.

Lee, D.R, 1977. A device for measuring seepage flux in lakes and estuaries. Limnol. Oceanogr. 22: 140-147.

Libelo, E. L. and W. G. MacIntyre, 1994. Effects of surface-water movement on seepage-meter measurements of flow through the sediment-water interface. Applied Hydrogeology 2: 49-54.

Light, S.S. and J.W. Dineen, 1994. Water control in the Everglades: a historical perspective, p. 47-84. In: S.M. Davis and J.C. Ogden (eds.), Everglades: The Ecosystem and Its Restoration. St. Lucie Press, Delray Beach, Florida.

Macintyre, S., R. Wanninkhof, and J.P. Chanton. 1995. Trace gas exchange across the air-sea interface in freshwater and coastal marine environments. In: P.A. Matson, R.C. Harriss (eds.) Biogenic Trace Gases: Measuring Emissions from Soil and Water, Blackwell Science Ltd. 52-97.

Martens, C.S., G.W. Kipphut, and J.V. Klump. 1980. Sediment-water chemical exchange in the coastal zone traced by in situ radon-222 flux measurements. Science, 208, 285-288.

Mathieu, G., P. Biscayne, R. Lupton, and D. Hammond. 1988. System for measurements of ²²²Rn at low levels in natural waters. Health Physics 55, 989-992.

McAuliffe, C. 1971. Gas chromatographic determination of solutes by multiple phase equilibrium. Chem. Technol. 1, 46-51.

Moore, W.S., 1999. The subterranean estuary: A reaction zone of groundwater and seawater. Mar. Chem. Marine Chemistry: 1-24.

Nixon S.W. and M.E.Q. Pilson, 1983. Nitrogen in estuarine and coastal ecosystems. In: Carpenter, E.J. and D.G. Capone (Eds.) Nitrogen in the Marine Environment (pp 565-648) New York: Academic Press.

Oberdorfer, J. A., M. A. Valentino, S. V. Smith. 1990. Groundwater contribution to the nutrient budget of Tomales Bay, California. Biogeochemistry 10, 199-216.

Paul, J.H., J.B. Rose, J. Brown, E.A. Shinn, S. Miller, and S.R. Farrah, 1995. Viral tracer studies indicate contamination of marine waters by sewage disposal practices in Key Largo, Florida. Applied and Environmental Microbiology 61: 2230-2234.

Paul, J.H., J.B. Rose, S.C. Jiang, X. Zhou, P. Cochran, C. Kellog, J.B. Kang, D. Griffin, S. Farrah, and J. Lukasik, 1997. Evidence for groundwater and surface water contamination by waste disposal wells in the Florida Keys. Water Research 31: 1448-1454.

Peng, T.H., T. Takahashi, W.S. Broecker. 1974. Surface radon measurements in the north Pacific station Papa. J Geophy. Res. 79, 1772-1780.

Perkins, R.D., 1977. Pleistocene depositional framework in south Florida. In: Enos, P. and R.D. Perkins, Quarternary sedimentation in south Florida: Geological Society of America Memoir 147: 131-198.

Phlips, E. J., T.C. Lynch, and S. Badylak. 1995. Chlorophyll a, tripton, color, and light availability in a shallow tropical inner-shelf lagoon, Florida Bay, USA. Mar. Ecol. Prog. Ser. 127, 223-234.

Phlips, E. J. and S. Badylak. 1996. Spatial variability in phytoplankton standing crop and composition in a shallow inner-shelf lagoon, Florida Bay, Florida. Bull. Mar. Sci. 58: 203-216.

Phlips E. J., S. Badylak, and T.C.Lynch. 1999. Blooms of the picoplanktonic cyanobacterium Synechococcus in Florida Bay, a subtropical inner-shelf lagoon. Limnol. Oceanogr. 44: 1166-1175.

Powell, G.V.N., J. Kenworthy, and J.W. Fourqurean, 1989. Experimental evidence for nutrient limitation of seagrass growth in a tropical estuary with restricted circulation. Bull. Mar. Sci. 44: 324-340.

Rona, E. 1917. Diffusionsgrösse und atomdurchmesser der radiumemanation. Zeitschrift für Physikalische Chemie. 92, 213-218.

Rudnick, D., Z. Chen, D. Childers, J. Boyer, and T. Fontaine. 1999. Phosphorous and nitrogen inputs to Florida Bay: The importance of the Everglades watershed. Estuaries. 22, 398-416.

Runnels, D.D., 1969. Diagenesis, chemical sediments, and mixing of natural waters. J. Sed. Petrol. 39: 1188-1201.

Rutgers van der Loeff, M., L.G. Anderson, P.O.J. Hall, A. Iverfeldt, A.B. Josefson, B. Sundby and S.F.G. Westerlund. 1984. The asphyxiation technique: An approach to distinguishing between molecular diffusion and biologically mediated transport at the sediment-water interface. Limnol. Oceanog. 29, 675-686.

Sansone, F.J., G.W. Tribble, C.C. Andrews, J.P. Chanton, 1990. Anaerobic diagenesis within recent, Pleistocene, and Eocene marine carbonate frameworks. Sedimentology 37: 997-1009.

Sansone, F.J., J.P. Chanton, P.R. Haberstroh, 1993. Methane cycling in coral reef frameworks. In: Trends in Microbial Ecology, R. Guerrero and C. Pedros-Alio (eds.), Spanish Society for Microbiology, 157-162.

Shaw, R.D. and E.E. Prepas, 1989. Anomalous, short-term influx of water into seepage meters. Limnol. Oceanogr. 34: 1343-1351.

Shinn, E.A., R.S. Reese, C.D. Reich, 1994. Fate and pathways of injection-well effluent in the Florida Keys. U.S. Geological Survey Open-File Report 94-276.

Simmons, G.M. and F.G. Love, 1987. Water quality of newly discovered groundwater discharge into a deep coral reef habitat. In: Cooper, R.A. and A.N. Shepard (Eds) Science applications of current diving technology on the U.S. continental shelf. NOAA Symposium Series for Undersea Research 2: 155-163.

Simmons, G.M., 1992. Importance of submarine groundwater discharge (SGWD) and seawater cycling to material flux across sediment/water interfaces in marine environments. Mar. Ecol. Prog. Ser. 84: 173-184.

Strickland, J.D.H. and T.R. Parsons, 1972. A Practical Handbook of Seawater Analysis. 2nd ed. Fisheries Research Board of Canada, Ottawa, 310p.

Sweeny, R.E., E. Kahil, and I.R. Kaplan, 1980. Characterization of domestic and industrial sewage in S. Cal. coastal sediments using N, C, S, and U tracers. Mar. Env. Res. 3: 225-248.

Tribble, G.W., F.J. Sansone, S.V. Smith, 1990. Stoichiometric modeling of carbon diagenesis within coral reef and surface sea water. Geochim. Cosmochim. Acta 54: 2439-2449.

Ullman, W.J. and R.C. Aller. 1982. Diffusion coefficients in nearshore marine sediments. Limnol. Oceanog. 27, 552-556.

USEPA, 1996. Water quality protection program for the Florida Keys National Marine Sanctuary. First Biennial Report to Congress.

Valiela, I., J.M. Teal, S. Volkmann, D. Shafer and E.J. Carpenter, 1978. Nutrient and particulate fluxes in a salt marsh ecosystem: Tidal exchanges and inputs by precipitation and groundwater. Limnol. Oceanogr. 23: 798-812.

Valiela, I. and J.M. Teal, 1979. The nitrogen budget of a salt marsh ecosystem. Nature 280: 652-656.

Valiela, I., J. Costa, K. Foreman, J.M. Teal, B. Howes, and D. Aubrey, 1990. Transport of water-borne nutrients from watersheds and their effects on coastal waters. Biogeochemistry 10: 177-198.

Wanninkhof, R.. 1992. Relationship between wind speed and gas exchange over the ocean. J. Geophys. Res. 97, 7373-7382.

Younger, P.L., 1996. Submarine groundwater discharge. Nature 382: 121-122.

Zieman, J.C., 1972. Orgin of circular seagrass beds of Thalassia (Spermatophyta: Hydrocharitaceae) in South Biscayne Bay, Florida, and their relationship to mangrove hammocks. Bull. Mar. Sci. 22: 559-573.

Modeling Approach

Diffusive Benthic Fluxes

Benthic Flux Experiments

Diffusive Flux Experiments

Sediment Equilibration

Peeper

Sinks (dpm m-2 min-1)

Sinks (nmoles m-2 min-1)

Benthic Chamber Experiments

Sinks (dpm m^-2 min^-1)

Sinks (nmoles m^-2 min^-1)