Trophic Transfer of Mercury in a Subtropical Coral Reef Food Web

______

A Thesis

Presented to

The Faculty of the College of Arts and Sciences

Florida Gulf Coast University

In Partial Fulfillment

Of the Requirement for the Degree of

Master of Science

______

By

Christopher Tyler Lienhardt

2015

APPROVAL SHEET

This thesis is submitted in partial fulfillment of

the requirements for the degree of

Master of Science

______

Christopher Tyler Lienhardt

Approved: July 2015

______

Darren G. Rumbold, Ph.D.

Committee Chair / Advisor

______

Michael L. Parsons, Ph.D.

______

Ai Ning Loh, Ph. D.

The final copy of this thesis has been examined by the signatories, and we find that both the content and the form meet acceptable presentation standards of scholarly work in the above mentioned discipline. i

Acknowledgments

This research would not have been possible without the support and encouragement of numerous friends and family. First and foremost I would like to thank my major advisor,

Dr. Darren Rumbold, for giving me the opportunity to play a part in some of the great research he is conducting, and add another piece to the puzzle that is mercury biomagnification research. The knowledge, wisdom and skills imparted unto me over the past three years, I cannot thank him enough for. I would also like to thank Dr. Michael

Parsons for giving me a shot to be a part of the field team and assist in the conducting of our research. I also owe him thanks for his guidance and the nature of his graduate courses, which helped prepare me to take on such a task. I would also like to thank Dr. Ai Ning

Loh, for her comments and feedback on this thesis, and as someone who pushed me to pursue graduate school after working in her geochemistry laboratory as an undergraduate.

I would also like to thank my friends and fellow graduate students Rheannon Ketover and

Lacey Rains, your backing to apply to the program and support throughout it has been irreplaceable. Thanks are also owed to Alex Leynse, Amanda Ellsworth, Ashley Brandt,

Nicole Fronczkowski, Megan Conkling, Adam Catasus, and Jeff Zingre (all from FGCU) for their assistance in collecting and processing samples. Finally, I am especially grateful to my mother Jamie Lienhardt and my best friend Lillie Simmons, for their moral support and constant encouragement throughout the many ups and downs that are the graduate school experience. ii

Abstract

Mercury is a widespread and damaging toxic metal that is trophically transferred through food webs. Coral reefs present an interesting dynamic in that they are comprised of complex food webs containing a high number of lateral or horizontal links that do not always end in top predators, thus possibly reducing trophic transfer. Therefore, the objective of the study was to assess the efficiency of trophic transfer in a coral reef food web, using mercury as the tracer. Concentrations of mercury and stable isotopes of nitrogen

(δ15N) and carbon (δ13C) were measured in fish from two sites near the coastal waters of

Long Key, Florida. The relationship between mercury and δ15N can be used to estimate biomagnification across the food web (i.e., trophic magnification slope, food web magnification factor). Using mercury and stable isotopes of nitrogen and carbon as tracers assisted in quantifying the efficiency at which coral reef ecosystems transfer these and other bioaccumulative toxins (e.g., ciguatoxins, etc.) through the food web, while also increasing our understanding of the associated flow of energy in the system. A total of 242 samples were collected from April 2012 through December 2013 using spear guns, hook and line, and hand collection techniques. Individual Hg concentrations ranged from 17.33

µg/kg in a gray angelfish (Pomacanthus arcuatus) to 3,317 µg/kg in a great barracuda

(Sphyraena barracuda) at Long Key Hard Bottom, and 19.01 µg/kg in a rock beauty angelfish (Holacanthus tricolor) to 6,842 µg/kg in a porkfish (Anisotremus virginicus) at

Tennessee Reef. Variability in both fish size and δ15N increased the variance in tissue Hg concentration both intra- and inter-specifically. As observed in other systems, the log transformed Hg concentrations in the food web, pooled across species, were significantly iii

related to δ15N. The trophic magnification slope (i.e., slope of Log [Hg] regressed on

δ15N), as an estimate of the biomagnification rate of Hg in the subtropical coral reef food web was 0.23 ± 0.03 (±95% confidence interval) at Tennessee Reef and 0.16 ± 0.04 for

Long Key Hard Bottom. When δ15N was translated to trophic level, the food web magnification factor (calculated from slope of Log [Hg] regressed on trophic level) were

7.8 and 3.4 for Tennessee Reef and Long Key Hard Bottom, respectively. Although there was some evidence to support significant differences between the two sites (which could have been due to differences in water quality), this difference in slopes could also be a result of unbalanced sampling design. Nonetheless, these results clearly demonstrate that

Hg is biomagnified through subtropical coral reef ecosystems and that the transfer efficiency (i.e., slopes) were consistent with previous reports for marine ecosystems.

iv

Table of contents

Acknowledgments ...... i

Abstract ...... ii

Table of contents ...... iv

List of Tables...... v

List of Figures ...... vi

Introduction ...... 1

Objectives ...... 9 Significance ...... 9 Methods ...... 10

Study Area ...... 10 Sample Collection and Processing ...... 11 Mercury Analysis ...... 12 Stable Isotope Analysis ...... 13 Data Analysis ...... 14 Results ...... 15

Variability in Mercury Concentration as a Function of Fish Size ...... 16 Variability in Mercury Concentration as a Function of Location ...... 17 Variability in Mercury Concentration in Relation to Stable Isotopes of Carbon and Nitrogen ...... 19 Trophic Magnification Slope (TMS) and Food Web Magnification Factor (FWMF) .. 21 Discussion ...... 23

Conclusions ...... 38 Literature Cited ...... 39

v

List of Tables

Table 1. Summary of mercury concentration, total length, stable isotopes of δ13C and

δ15N, and trophic level of invertebrate and finfish taxa collected from Tennessee Reef

Light (TRL)………………………………………………………………………………56

Table 2. Summary of mercury concentration, total length, stable isotopes of δ13C and

δ15N, and trophic level of invertebrate and finfish taxa collected from Long Key Hard

Bottom (LKH)…………………………………………………………………………....57

Table 3. Coefficient of determination (r2) values for regression models for various relationships involving fish species at Tennessee Reef…………………………………...58

Table 4. Coefficient of determination (r2) values for regression models for various relationships involving fish species at Long Key Hard Bottom………………………….58

Table 5. Values for trophic magnification slopes of Total Hg biomagnification through food webs of different ecosystems as reported in the published literature……………….59

Table 6. Water quality conditions based on quarterly sampling from May 2010 – September

2013 at the two nearby sites (adapted from the FIU SERC Florida Keys National Marine

Sanctuary). ………………………………………………………………………………60

vi

List of Figures

Figure 1. Map of study area where samples were collected off Long Key, FL………….61

Figure 2. (a) Relationship between log Hg (µg/kg) and total length (cm) in individuals at

Tennessee Reef, (b) log Hg (µg/kg) and total length (cm) for all individuals at Long Key

Hard Bottom……………………………………………………………………………...62

Figure 3. Relationship between stable isotope values of δ13C and δ15N (‰), for Tennessee

Reef (top panel) and Long Key Hard Bottom (bottom panel)…………………………...63

Figure 4. Relationship between (a) δ15N (‰) and total length (cm) with linear regression line in all individuals at Tennessee Reef; (b) δ15N (‰) and total length (cm) with linear regression line for all individuals at Long Key Hard Bottom; (c) δ15N (‰) and total length

(cm) with linear regression line for species means at Tennessee Reef; and (d) log δ15N (‰) and log total length (cm) for species means at Long Key Hard Bottom……………..….64

Figure 5. Relationship between log Hg concentration (µg/kg, wet weight) and δ15N (‰) from a subtropical coral reef food web for individuals with linear regression lines for (a)

Tennessee Reef and (b) Long Key Hard Bottom………………………………………..65

Figure 6. Relationship between log Hg concentration (µg/kg, wet weight) and δ15N (‰) from a subtropical coral reef food web based on species means (mean ± SD) with linear regression line for Tennessee Reef (a) and Long Key Hard Bottom (b)………………..66

Figure 7. Relationship between log Hg concentration (µg/kg, wet weight) and trophic level

(calculated using δ15N, see text), from a subtropical coral reef food web for all individuals caught, with linear regression line for (a) Tennessee Reef and (b) Long Key Hard

Bottom…………………………………………………………………………………...67 vii

Figure 8. Relationship between log Hg concentration (µg/kg) and trophic level for species means at Tennessee Reef (a) and Long Key Hard Bottom (b)…………………………….68 1

Introduction

Mercury (Hg) is a persistent toxic metal that many humans, fish, and other wildlife are exposed to and is the cause of a variety of harmful side effects. Mercury was recognized as an important environmental toxicant following years of poisoning in which a large number of people consuming fish exhibited extreme irreversible neurological damage and teratogenic birth defects in the Japanese village of Minamata during the 1950s (Leonard et al. 1983, De Flora et al. 1994). The number of Hg monitoring programs has increased dramatically across the United States in places like the Great Lakes, California and Florida, due to the increased Hg levels reported in both freshwater and marine fishes (Adams and

McMichael 2001, Strom and Graves 2001, Adams et al. 2003). High Hg levels resulted in the issuance of consumption advisories and closures of over 60 fish species, including highly sought after game species like the largemouth bass (Micropterus salmoides) and any species of shark growing larger than 43 inches long (FDOH 2015). High tissue concentrations (≥1.5 ppm) can present particular health threats to fish and their consumers, including humans and wildlife (Scheuhammer et al. 2007, Adams et al. 2010, Mergler et al. 2007).

Since the early 1990s, considerable time and effort has been devoted to determining sources of Hg and locating areas of high concentration or “hot spots” in South Florida (Fink et al. 1999). Extensive study of Hg commenced in the Florida Everglades after the enactment of the Everglades Restoration Project, where levels observed in sport fish are the highest in the state (Fink et al. 1999). The high levels prompted Florida to undertake the Atmospheric Mercury Study; a 5-year endeavor that determined atmospheric 2

deposition of mercury in the Florida Everglades was approximately double the rate compared to rural Wisconsin, accounting for ninety-five percent of the total mercury inputs to the region (Guentzel et al. 1995). Most of the mercury found in fish tissues is thought to originate from global atmospheric inputs, concentrations of which have increased over the past hundred years, mostly due to anthropogenic activity (Slemr and Langer 1992, Mason et al. 1994, Fitzgerald 1995). Some of these anthropogenic sources include emissions from incinerators, chlor-alkali plants, and fossil fuels, especially coal fired power plants (U.S.

EPA 1997, Boening 2000). Some naturally occuring sources also occur through crustal discharge from volcanic eruptions and oceanic release of mercury through hydrothermal vents (Boening 2000).

Because of its long residence time, anywhere from 100 days up to a year depending on the form (Radke et al. 2007), atmospheric mercury can easily be transported and later redeposited thousands of miles from the original source (U.S. EPA 1997). Monitoring conducted by the National Mercury Deposition Network has shown that Florida has one of the higher atmospheric deposition concentrations relative to the rest of the country and

Canada (Prestbo and Gay 2009). It has been determined that the source of atmospheric mercury in Florida is a combination of long range transport (Guentzel et al. 2001) and local point sources. These two factors make up the majority of wet deposition (Dvonch et al.

1999), and highest concentrations are usually observed during seasons of high rainfall such as summer (Dvonch et al. 2005). High deposition or “hot spots” have been shown to be occuring in Tampa, Miami and South Florida as a result of local anthropogenic emissions and high amounts of rainfall (Myers et al. 2006). This is troubling because some sources can emit water-soluble and easily transported divalent mercury (II) (Carpi 1997, Dvonch 3

et al. 1999). Once it is deposited, inorganic mercury can be transformed by methylation into its toxic organic form, methyl-mercury (MeHg). This is facilitated primarily by anaerobic bacteria that methylate the mercury in the water column and in sediments, typically under anoxic conditions (Boening 2000, Hammerschmidt and Fitzgerald 2006).

Kang et al. (2000) found that runoff from the Florida Everglades was discharging to Florida

Bay and the Florida Keys at a rate between 4-160 µg Hg m-2 yr-1. If deposited into South

Florida waters, natural sheet flow could ultimately transport these toxicants to Florida Bay and affect the food webs of the Florida Keys. Concerns have been raised about this potential for transport of methylmercury (MeHg), to estuarine and marine environments (Kang et al.

2000, Strom and Graves 2001). Mercury is absorbed by organisms in its organic form,

MeHg, mostly through diet (Boening 2000). Because of its long half-life in tissues and its high dietary absorbtion, methylmercury is the form of mercury that biomagnifies, increasing in concentration as it is passed from primary producers all the way up to apex predators such as sharks and humans (Trudel and Rasmussen 1997). Methylmercury is considered a dangerous neurotoxicant (U.S. EPA 1997).

The negative effects of methylmercury exposure observed in humans include neurological and cardiovascular impairments in adults and developmental and neurological issues in young children (NRC 2000, Mergler et al. 2007). Consumption of fish and shellfish from marine and estuarine waters is the primary route of exposure to MeHg in humans (NRC 2000). As a result of this, many research endeavors are now taking place in

North America that focus on MeHg in the muscle tissue of commonly consumed fish and shellfish, including Florida (Adams et al. 2003), the Great Lakes (Bhavsar et al. 2010), the 4

Western United States (Peterson and Van Sickle 2007), and the Northeastern United States and Canada (Evers et al. 2009, Kamman et al. 2005).

The Florida Fish and Wildlife Conservation Commission’s Florida Marine

Research Institute has conducted surveys to support existing Florida Department of Health consumption advisories (Adams and McMichael 2001, Adams et al. 2003). The studies conducted by FWRI targeted species of game fish and other harvestable species during a mercury study conducted from 1989-2001 at multiple study sites along the Florida coasts, including Florida Bay and the Florida Keys. The study indicated that mean mercury levels were below the 0.3 ppm criteria for the protection of human health, established by the

USEPA (2001) for most fish; however, the observed concentrations were highly variable, demanding more research be done to examine the possible differences between habitats and food webs. Recent advisories published by the Florida Department of Health included a few commonly recognized reef species: black grouper (Mycteroperca bonaci), great barracuda (Sphyraena barracuda), gray snapper (Lutjanus griseus), hogfish

(Lachnolaimus maximus), lookdown (Selene vomer), white grunt (Haemulon plumieri), and yellowtail snapper (Ocyurus chrysurus) (FDOH 2015).

Although most of the concern has focused on fish as biovectors to humans, manipulative experiments involving freshwater fish species have demonstrated that high tissue concentrations of mercury are sometimes associated with sub-lethal effects in the fish themselves (Matta et al. 2001, Hammerschmidt et al. 2002, Drevnick and Sandheinrich

2003, Webber and Haines 2003, Houck and Cech 2004). The observed effects included lower growth rates, lower reproductive success, behavioral differences, and histopathological effects on tissues (Houck and Cech 2004, Hammerschmidt et al. 2002, 5

Drevnick and Sandheinrich 2003). Similar studies on marine species are lacking and should be made a priority.

Factors controlling the biomagnification of methylmercury and levels occurring in apex predators are highly variable as evidenced by the variation in Hg levels between individuals and among populations (Chen et al. 2012). Beyond the variation in Hg loading described above, differences in biology and community dynamics (i.e.-food chain length, linkage strength), can play a role in the observed variation. The characteristics of a habitat such as its chemistry, water depth and hydroperiod affect the methylation rate; changing how much mercury becomes bioavailable to organisms at the base of the food web

(Snodgrass et al. 2000). It has been demonstrated that among individuals a relationship often exists between increased MeHg concentration, size and age of fish (Lowery and

Garrett 2005, Adams and Onorato 2005, Adams and McMichael 2007, Bank et al. 2007,

Cai et al. 2007). This can be explained by observing that these older, larger individuals are exposed to MeHg over a longer period of time and the likelihood that they will increase trophic position as they mature, causing a shift in prey as a function of mouth or body size

(Mittelbach and Persson 1998, Scharf et al. 2000) and because of this shift, predators will commonly consume larger prey items that also have an elevated MeHg concentration

(Bowles et al. 2001, Power et al. 2002). Differences in MeHg between species can also be attributed to biological variables such as growth rates, bioenergetics, clearance rate, trophic position, or any combination of the factors (Simoneau et al. 2005, Luoma and Rainbow

2005, Wang 2002, Bank et al. 2007, Cai et al. 2007).

Differences in community dynamics can also lead to elevated Hg even if other factors such as loading and methylation rate are the same. Efficiency of mercury transfer 6

through food webs is highly influenced by the community structure and, in turn, food web complexity. Primary and secondary production, availablity of prey items, food chain length and linkage strength, are all influential factors of biomagnification (Cabana et al. 1994,

Futter 1994, Watras et al. 1998, Pickhardt et al. 2002). Primary production can influence the transfer of mercury by acting as a diluting agent. Productive ecosystems have higher sedimentation rates and a greater biomass dilution factor, lowering the contaminant transfer

(Larsson et al. 1992, Pickhardt et al. 2002). Food chain length also influences transfer of mercury to top predators. Mercury concentrations in top predators of a pelagic food chain in Ontario lakes was significantly influenced by presence or absence of specific forage species that linked top predators with zooplankton, where a presence of that forage species

(longer food chain) increased mercury concentrations and the absence (shorter food chain) had much lower concentrations (Cabana et al. 1994, Rasmussen et al. 1990). Linkage strength or connectedness of a food web can affect the route of mercury transfer as well. A highly connected web can introduce horizontal links between low, intermediate and high trophic levels whether the feeding relationships exist in simple or complex food webs, ultimately decreasing mercury that ends in top predators (Vander Zanden and Rasmussen

1996, Kling et al. 1992). Biomagnification of contaminants tends to increase as food chain length increases. This occurs because more transfers are taking place within a long food chain; generating higher concentrations when compared to simple, short food chains

(Rasmussen et al. 1990, Kidd et al. 1995).

Tropical nearshore environments, including coral reef habitats represent some of the most diverse marine ecosystems. Numerous adaptations and symbioses among autotrophs and mixotrophs have increased the primary production and generated complex 7

food webs in coral reef ecosystems (Odum and Odum 1955, Sorokin 1995). These complex food webs contain a high number of lateral or horizontal links in the food chain that do not always end in top predators, therefore reducing trophic transfer (Stemberger and Chen

1998). Food web complexity has been shown to generally increase with water temperature, suggesting that top down control of productivity becomes less common in increasingly complex food webs, an important variable when comparing biotransfer between subtropical and temperate food webs (Frank et al. 2007). A study of trace metals in Barrier

Reef fishes has demonstrated a clear dependency between mercury concentration in axial muscle tissue and trophic level, again exemplifying the importance of understanding food chain length (Denton and Burdon-Jones 1986). The complex relationship between herbivorous fish, omnivorous fish, apex predators and the complexity of the ecosysten will be an important variable when determining the route of mercury transfer through the coral reef food web.

Stable isotope analysis (SIA) has been employed to assess trophic position and examine the structure of food webs (Peterson and Fry 1987). Stable isotope signatures of carbon-13 (13C) and nitrogen-15 (15N) are used to determine diet habits (Minagawa and

Wada 1984, Peterson and Fry 1987). Isotopic ratios of nitrogen increase approximately 3-

5‰ (mean 3.4‰) percent when transferred from prey to predator (DeNiro and Epstein

1981, Minagawa and Wada 1984, Peterson and Fry 1987, Cabana and Rasmussen 1994,

Post 2002). Isotopes of carbon are less reliable when determining trophic position due to their less distinct (0-1) percent increase between trophic level, but have a distinct signature for primary producers and therefore is a good indicator of sources of carbon to the ecosystem (Haines and Montague 1979, Peterson and Fry 1987). Stable isotopes can be 8

used in ecotoxicology studies to determine variability of toxicant concentrations between individuals, populations, and species (Kidd et al. 1995, Houde et al. 2008, Cai et al. 2007,

Bank et al 2007). Stable Isotopes of nitrogen are particularly useful when defining food web interactions (Atwell et al. 1998, Kidd et al. 2003), and forecasting mercury concentrations in aquatic fish (Kidd et al. 1995). A relatively new approach to using SIA is to assess biomagnification factors across an entire food web (Jardine et al. 2006) rather than the classical biomagnification factor (BMF), which is defined as the ratio of the chemical concentration in an organism to that in its diet at steady state. Instead, the food web magnification factor (FWMF) or trophic magnification factor assesses the increase of a chemical concentrating across multiple trophic levels in a food web (Jardine et al. 2006).

The food web magnification factors aid in determining the degree and severity of a contaminant’s biomagnification in the food web (Fisk et al. 2001, Hop et al. 2002).

Effectively, these data can quantify mercury transfer and alert us to exposure risks to top level consumers, while also allowing us to study the complex food web dynamics within and between ecosystems. From this calculation the FWMF can be compared for the coral reef food web to those already calculated for other ecosystems such as estuarine food webs, freshwater food webs, arctic marine food webs, and a subtropical coastal food web. This is important because very little research has been conducted to measure the magnification factors in tropical or subtropical marine environments and none have characterized the food web magnification factor or basal methylmercury levels for coral reef ecosystems.

9

Objectives

The objective of this study is to assess trophic transfer of mercury in a coral reef food web. Mercury and stable isotope signatures of nitrogen (δ15N) and carbon (δ13C) were measured in fish and their prey in the coastal waters in proximity to Long Key (Figure 1) to accomplish this. The relationship between mercury and δ15N was used estimate biomagnification integrated across the entire food web, which was compared to similar work done in the Caloosahatchee and Shark river estuaries, and near-coastal non-reef food webs. The δ13C signatures assisted in determining whether or not the food web is more closed or open to allochthonous input sources. For example, high variation in δ13C signatures from carbonate algae when compared to plankton, or demersal compared to pelagic fish.

Significance

The data generated from this study is especially relevant from an economic, consumption, and public health viewpoint. The significance of this approach coupled to the traditional SIA methods essentially uses mercury as another tracer to assess the efficiency of trophic transfer across a coral reef food web, which is important because of its direct effect on natural resource management, fish recruitment and reef productivity.

Understanding the biomagnification factor of an ecosystem is pivotal when trying to recognize the relative importance of aspects (e.g., physical, biogeochemical or food web dynamics) influencing mercury levels in top predators. Mercury is a toxicant that poses a real threat to the health of humans, fish and other wildlife. As long as humans continue to consume fish as a source of food, it will be important to monitor the tissue concentrations 10

of mercury. The harmful neurological, developmental and cardiovascular side effects of mercury exposure to humans highlight the importance of long term monitoring. Previous sampling efforts have shown that mercury levels in estuarine and marine fish vary by species collected, size of fish, and sampling location. More studies are needed to determine whether a possible latitudinal gradient exists for mercury biomagnification. Determination of basal mercury entering the food web is important because of its role in determining consumption advisories that are issued by the governmental departments of health.

Continued study of these variables needs to carry on to determine the cause of the variations between ecosystems.

Methods

Study Area

Samples of biota were collected from the Atlantic coastal waters of Long Key,

Florida (Figure 1). This region lies along the Florida Straits and separates the Atlantic

Ocean from the Gulf of Mexico. Sitting between latitudes of 23.5 and 25.5 degrees north the Florida Keys are considered subtropical but, are characterized by tropical climate weather. Habitat types are a shallow, exposed hard substrate (Long Key Hard Bottom) site characterized by soft corals and algal cover, and a barrier reef site characterized by sandy bottom and hard corals (Tennessee Reef). Both sites were selected due to their designation as long term research sites by the Florida Keys National Marine Sanctuary Coral Reef and

Monitoring Project (FKNMS CREMP). 11

Sample Collection and Processing

Sampling targeted a representative food web covering all existing trophic levels of species that inhabit the study sites including invertebrates, herbivorous fishes, omnivorous fishes, carnivorous fishes, and apex predators. Fish and invertebrate samples were collected between April 2012 and December 2013 by as part of a larger project investigating bioaccumulative ciguatoxins. Sampling methods included hook and line, and free hand capture via SCUBA diver (spear-gun and hand collection). Sampling stations were sampled monthly and were located each visit using GPS coordinates. Once biota were returned to the boat they were appropriately bagged and then labeled and immediately stored on wet ice until return to the laboratory at Florida Gulf Coast University were they were stored in a -20°C freezer for further processing and later analysis.

Organisms were identified to species or to the lowest practical taxonomic level using dichotomous key or identification guides (Hoese and Moore 1998). The fish were then weighed and measured. Invertebrate samples were also measured and weighed depending on the type; carapace length or width of crab species, or shell length of mollusks.

These measurements were taken prior to the removal of soft inner tissues from the hard exoskeleton or shell. Once measured, the tissue was removed from the shell or exoskeleton.

Upon completion, samples were frozen at -20 °C and processed within 6 months. Tissues were taken for analysis from the dorsal muscle section of the left side of fish, above the lateral line and anterior to the dorsal fin. Thus, this would be representative of a fillet, if collected for consumption by humans. Forage fish or secondary consumers were treated in an identical fashion with removal of small muscle same representative of a fillet (Bowles et al. 2001, Riget et al. 2007). An effort was made to also sample only muscle tissue (rather 12

than organs or homogenate) of soft bodied invertebrates. All equipment was thoroughly rinsed with tap water then dried with a paper towel to remove excess tissue between samples.

Mercury Analysis

Samples were analyzed for total mercury content (THg) (mg/kg wet weight) by thermal decomposition, amalgamation and atomic absorption spectrophotometry (using a

Nippon Model MA-2000; EPA method 7473). Quality control was incorporated by running a check standard every 20 samples using continuing calibration verifications, blanks, duplicates, and certified reference materials: DORM-3, DORM-4 (Fish protein), DOLT-3

(Dogfish Liver) (National Research Council Canada) and, in a limited number of instances, a hair CRM from International Atomic Energy Agency (IAEA). The correlation coefficient of initial calibration averaged 0.9989 (n=12); percent recovery of continuing calibration verification check samples was 103% ±15.4% (n=60); relative percent difference (RPD) between laboratory duplicate analyses was 7.3±9% (n=31). Because the majority of the mercury in muscle tissue of fish is present in the methylated form (Grieb et al. 1990, Bloom

1992, Kannan et al. 1998, Sveinsdottir and Mason 2005), fish tissues are typically analyzed for total mercury, which is straightforward and less costly procedurally than methylmercury, and interpreted as being equivalent to the analysis of methylmercury. In the case of invertebrate tissue samples, where the percentage of MeHg can vary, THg values were adjusted to percentage MeHg as reported for similar species of bivalves and crustacean by Thera and Rumbold (2014). That study utilized the FDEP’s Central

Laboratory (Tallahassee, FL) where MeHg was analyzed using purge and trap with atomic 13

fluorescence detection (FDEP SOP, HG-003, 2009); THg was analyzed by cold vapor atomic absorption spectrophotometry (FDEP SOP, HG-007, 2009). The average ratio of

THg:MeHg reported by FDEP (i.e., 60%) was then used to adjust THg values generated from the Nippon MA-2000 for the invertebrate samples. From this point forward, results listed as Hg represents total mercury in fish and methyl mercury, converted from total mercury, in invertebrates.

Stable Isotope Analysis

Stable isotope analysis for carbon and nitrogen followed previously published method protocols (Jepsen and Winemiller 2002, Evers et al. 2009). A subsample of tissue was measured into an acetone cleaned, baked aluminum weigh boat. The tissue sample was then inserted into a drying oven at a minimum of 60 °C for 48 hours or until a consistent dry weight is achieved. Dry samples were then be ground with mortal and pestle into a fine powder. Once achieved, the powder was placed into a scintillation vial and dried again inside the oven at 60 °C for 24 hours.

Aliquots of the final powder between 0.6-1.2 milligrams were next weighed into tin capsules (Costech, Valencia, CA) by use of a MX5 automated-S microbalance (MX5,

Mettler Toledo, Columbus, OH). Once folded, these tin capsules were placed into well trays for transport. Replicates were processed every 10 samples. Samples were shipped to

University of California Davis Stable Isotope Facility, Davis, California, USA, for analysis of isotope ratios (13C/12C and 15N/14N) and total carbon and nitrogen content.

Measurements were taken from a continuous flow isotope ratio mass spectrometer (IRMS). 14

Results are expressed as δ13C and δ15N relative to standards (C: PeeDee belemnite; N: atmospheric nitrogen) using the following formula:

훿푋 = [푅푠푎푚푝푙푒/푅푠푡푎푛푑푎푟푑 − 1] × 1000

Where X is 13C or 15N and R is 13C/12C or 15N/14N.

Data Analysis

Unless otherwise noted, total-Hg concentration (hereafter designated as [Hg]) was reported in µg/kg on a wet weight basis. One way analysis of variance (ANOVA) was used to assess interspecific patterns in [Hg], size, and stable isotopes separately. Mercury levels are known to be variable among individuals within a population for reasons like differences in size, age, gender and primarily, bioenergetics (Evans et al. 2015). Assumptions of normality and equal variances were tested by Kolmorogov–Smirnov and Levene median test, respectively. Where necessary, [Hg] was natural-log transformed (ln) to achieve normality or homogeneity of variance. Simple linear regression was used to examine how well the continuous variables, e.g., total length, δ15N, and δ13C, explained the variance in

[Hg] (based on the coefficient of determination [r2]). Trophic levels were calculated using the formula

15 15 푇푟표푝ℎ𝑖푐 퐿푒푣푒푙푐표푛푠푢푚푒푟 = ((훿 푁푐표푛푠푢푚푒푟 − 훿 푁푟푒푓푒푟푒푛푐푒 ) / 3.4) + 2 (Post

15 15 2002, Jardine et al. 2006) where δ Nreference represents the δ Nmean of the primary consumer for each site (frond oyster: TRL, Atlantic wing oyster: LKH), 3.4 is the

“enrichment factor,” representing the increase in δ15N from one trophic level to the next, and 2 represents the assumed trophic level of the reference. Once trophic level is determined, it can be used in tandem with the log mercury concentration to calculate a food 15

web magnification factor (FWMF) for the coral reef food web. This was achieved by taking a regression of the relationship between trophic level and log mercury concentration;

(푙표𝑔 퐻𝑔 푐표푛푐푒푛푡푟푎푡𝑖표푛 = 푏 + (푚 ∗ 푇퐿)) the antilog value of the slope (m; or 10m) being the food web magnification factor (Jardine et al. 2006). Mercury availability at the bottom of the food web was then calculated from the antilog of the y-intercept (b; or 10b). A two sample t-test for comparing means was also utilized to compare species means of total length, isotopic enrichment, and Hg concentration for species that occurred at both test sites to see if a significant difference between different populations existed. The significance of difference between slopes (e.g.,

TMS and FWMF) at the two sites was tested using a two-tailed Student's t-test (α=0.05), assuming unequal variance. Data analyses were performed using Sigmaplot for Windows

Version 11 software (Systat Software).

Results

A total of 131 samples from Tennessee Reef (hereafter designated as TRL) comprising 33 species (2 invertebrate, 31 fish), and a total of 111 samples from Long Key

Hard Bottom (hereafter designated LKH) comprising 33 species (4 invertebrate, 29 fish), were collected and analyzed for mercury. Measurable levels of mercury were recorded in all samples and varied within populations (size dependent), between populations and between species (trophic level dependent).

16

Variability in Mercury Concentration as a Function of Fish Size

Concentrations of mercury were variable within populations (i.e., intraspecifically) with coefficient of variation (CV) ranging from 0.9% to 65.8% at LKH (Table 2) and 0.04% to 126.7% at Tennessee Reef (Table 1). Much of this variability appeared to be related to size with those populations showing the greatest size range (total length, TL) exhibiting the most intraspecific variability in [Hg]: (Table 1, Table 2). An exception to this general rule was observed in porkfish of relatively similar size, which exhibited an unusually wide variation in [Hg] (Table 2).

The relationship between fish size (TL) and mercury concentration was assessed among individuals of the same species by site (where sample size ≥5 individuals). The intra-species relationship between log Hg and increasing length was statistically significant

(and positive) for only two of seven fish species caught at TRL (Table 3): blue stripe grunt

(r2=0.733, p=0.03, n=6) and great barracuda (r2=0.542, p=0.004, n=13). Mercury levels varied depending on size in three of the six fish species caught at LKH (where n≥5, Table

4), with significantly increasing [Hg] with size: Bermuda chub (r2=0.397, P=0.05, n=10), porkfish (r2=0.732, p=0.05, n=5), and yellowtail snapper (r2=0.342, p=0.05, n=12).

A significant positive relationship was observed between [Hg] and increasing size when all fish were pooled, regardless of species, at TRL (r2=0.131, p<0.01, n=114), and

LKH (r2=0.141, p<0.01, n=105) (Figure 2). Species means of fish caught also showed positive relationships but could not be assessed statistically due to failed assumptions of normality and homogeneity.

It is worth noting that a narrow size range of collected fish hampered the assessment of size effect on [Hg] for several species. Three of the six fish species 17

(lookdown, white grunt, yellowtail snapper) from LKH showed narrow size ranges (< 10 cm), and three from TRL (blue tang, blue stripe grunt, white grunt) showed a size range less than 10 cm. Alternatively, certain species exhibited a good size distribution but no obvious relation between [Hg] and size, including black grouper (range = 20.8 cm, r2=<0.01) and porkfish (range = 12.8 cm, r2=0.007) caught at TRL and hogfish (range =

24.5 cm, r2=0.023) caught at LKH.

Variability in Mercury Concentration as a Function of Location

A total of five species of fish were caught at both sites and were, thus, categorized as “cross-over samples” (where n ≥ 3 individuals) that would potentially allow comparing

[Hg] between locations. These five species were made up of secondary consumers including: Bermuda chub, blue stripe grunt, hogfish, porkfish, and white grunt. Although blue striped grunt, porkfish and Bermuda Chubs exhibited the only statistically significant relationships between [Hg] and size, assessing the others was hampered by small sample size at one or both sites, narrow size range or both. Nonetheless, these comparisons must be done cautiously and consider size distribution of the fish caught at the two sites. Mercury concentration in blue stripe grunts (Table 1, 2) did not differ between sites (mean difference

=125.5 µg/kg, t=1.83, df=7, p=0.110); they also did not differ in size (t=0.94, df=7, p=0.380). Similarly, [Hg] and size of porkfish (Table 1, 2) did not differ between sites

(Median =520 and 530 µg/kg, Mann-Whitney U= 26.0, df=5, p = 0.554), (Mann-Whitney

U=67.5, p=0.200). Although Bermuda chub at TRL had higher Hg levels (74.4 µg/kg) than chub at LKH (46.7 µg/kg), the difference was not statistically significant, and chub at TRL were statistically larger (35.5 cm). 18

Differences in size (TL) was also compared for the other cross over species that were caught at both TRL and LKH. When comparing TL of fish between the sites, two of the five crossover species showed a significant difference in size: the aforementioned

Bermuda chub, and white grunt (mean difference = 5.05 cm, t=5.90, df=20, p<0.01). The remaining crossover species were analyzed non-parametrically using Mann-Whitney rank sum test. Hogfish (U=67.5, df=10, p=0.393) and porkfish (U=19, df=5, p=0.200) did not show a statistically significant difference in size between the two sites.

When analyzed for Hg two of the five crossover species met the assumptions necessary for statistical analysis by t-test: blue stripe grunt (mean difference = 125.5 µg/kg, t=1.83, df=7, p=0.110) and hogfish (mean difference = 12.71 µg/kg, t=0.603, df=25, p =

0.552). Those that did not meet assumptions were similarly analyzed under the Mann-

Whitney rank sum test. Neither Bermuda chub (U=9, df=3, p=0.353), porkfish (U=26, df=5, p=0.554), or white grunt (U=42, df=6, p=0.685) showed statistically significant differences in Hg between the two sites. Thus, no species showed a significant difference in [Hg] between sites.

Between-site differences in [Hg] was also assessed in primary consumers: frond oysters from TRL and Atlantic wing oysters from LKH. It was particularly important to assess differences in the bivalves as primary consumers because, as discussed later, they were used to standardize basal δ15N and as an empirical indication of basal [Hg] (i.e., as opposed to y-intercept of FWMF). Mercury concentrations in the bivalves from the two sites did not differ significantly (mean difference = 5.25 µg/kg, t = -1.07, df=8, p=0.32).

19

Variability in Mercury Concentration in Relation to Stable Isotopes of Carbon and

Nitrogen

Stable isotopes of carbon (δ13C) were analyzed in all samples to assess potential differences in carbon source (nearshore v. offshore, benthic v. pelagic) that might explain some of the observed variability in Hg concentrations. δ13C enrichment varied within species, with CV ranging from 0.02% to 12.1 at LKH and 0.2% to 8% at TRL (Table 1, 2).

Isotope analysis showed some grouping between invertebrates and fish with respect to δ15N and δ13C. Some bivalves had comparable levels of δ13C and δ15N to that of herbivorous grazer fish while spiny lobsters also showed δ13C and δ15N enrichment comparable to secondary consumer fish (Figure 3). Sampled populations from the two sites showed very different relationships between Hg and δ13C. Only two of the 13 fish species analyzed

(where n≥5) showed a statistically significant relationship between δ13C and Hg: the blue tang (r2=0.319, p=0.04, n=13, negative slope) at TRL and the Bermuda chub (r2=0.428, p=0.04, n=10, positive slope) at LKH (Table 3, 4). The lookdown fish was another species like the Bermuda chub that appeared to have a strong positive relationship between Hg and

δ13C; however, it was not statistically significant (r2=0.602, p=0.12, n=5). Five other species, besides the aforementioned blue tang, had regressions that returned a negative slope, (albeit not significant); indicating that as δ13C became more enriched (i.e., less negative), Hg concentration decreased. A few of the other species also hinted at a negative relationship but returned weaker r-square values (mean r2=0.155) that were not significant.

The distribution of δ13C appeared to be similar at both sites but a large range of carbon is evident when regressed against δ15N (Figure 3). 20

To assess the variability of mercury in relation to trophic position, stable isotopes of nitrogen (δ15N) were analyzed in all samples. Invertebrates tended to be depleted in δ15N when compared to fish. The exceptions to this were spiny lobster and the Florida Horse conch (Figure 3). The majority of intraspecific relationships in fish had a narrow δ15N range

(<2‰) at both sites (Table 1, 2), exceptions to this were white grunt (range = 5.20‰), hogfish (range = 4.15‰) and great barracuda (4.10‰).Similar to the relation observed in

[Hg] concentration and size, larger fish within a population tended to be more enriched in

δ15N than small fish (Figure 4). This was not surprising because of the expected ontogenetic shifts in diet as size increased. However, of the 13 species analyzed (where n≥5), only blue tang at TRL showed a statistically significant relationship between TL and δ15N, (r2=0.617, p=0.01, n=13). Nine species exhibited non-significant negative slope, denoting a decrease in δ15N enrichment as size increased (Table 3, 4). The relationship between δ15N and TL for all individuals’ pooled (raw data), could not be assessed statistically due to failed assumptions, but did exhibit a positive relationship between increasing δ15N and TL. A regression of δ15N on TL, pooling species means was significant at TRL (r2=0.279, p<0.01, n=31), and showed a positive relationship at LKH, but violated the assumption of normality

(Shapiro-Wilk, p<0.01).

Intraspecific relationships between log Hg and δ15N were statistically significant in three of the 14 species analyzed. At TRL, great barracuda and porkfish showed significant positive relationships; (r2=0.428, p=0.015, n=13) and (r2=0.560, p=0.003, n=13) respectively (Table 3). Interestingly lookdown fish at LKH exhibited a strong negative relationship between log Hg and δ15N, (r2=0.858, p<0.01); however this must be viewed cautiously with a small sample size of 5 (Table 4). 21

As expected, δ15N exhibited obvious interspecific groupings that were consistent with presumed trophic position (Table 1, 2). Primary consumers with low δ15N enrichment consisted of filter feeding bivalves (TRL – frond oyster, n=5, mean = 3.37‰, LKH –

Atlantic wing oyster, n=3, mean = 4.46‰), and herbivorous grazers (parrotfishes, surgeonfishes and damselfishes; TRL- mean = 6.60‰, n=28, LKH- mean = 7.10‰, n=5).

Secondary consumers were made up of mostly opportunistic omnivores (grunts, snappers, and carangids etc.; TRL – mean = 9.16‰, n=62, LKH – mean = 9.47‰, n=88), while the tertiary consumers consisted of predatory fish (groupers and barracudas; TRL – mean =

11.34‰, n=25, LKH – mean = 9.97‰, n=9). A significant difference was found in δ15N enrichment in primary consumers (i.e. basal δ15N) collected at the two sites when bivalves were pooled for each site regardless of species (mean difference= 1.48 ‰, t= 4.63, df =11, p<0.01).

Trophic Magnification Slope (TMS) and Food Web Magnification Factor (FWMF)

A statistically significant, positive relationship between Log [Hg] and δ15N was evident at both locations, based on individual data points pooled across species (Figure 5).

The analysis revealed a significant increase in [Hg] with δ15N enrichment at TRL [Log Hg

= 0.229 + (0.227 * δ15N), p<0.01, r2=0.639], and LKH [Log Hg = 0.774 + (0.155 * δ15N), p<0.01, r2=0.418] (Figure 5). The Trophic Magnification Slope (i.e., TMS), as an estimate of the biomagnification rate of Hg in the subtropical coral reef food web was 0.23 ± 0.03

(±95% confidence interval) for TRL and 0.16 ± 0.04 for LKH. The relationship between

Log [Hg] and δ15N was also statically significant when based on species means (Figure 6).

It is noteworthy that when species means were used to derive the relationship (as opposed 22

to raw data) that the TMS remained unchanged at LKH but the TMS decreased from 0.23 to 0.20 at TRL. The slopes at the two sites were compared statistically using a t-test for independent populations, and were found to differ if based upon individual data points

(t=3.03, df=238, p<0.01), but show no significant difference when based on species mean

(t=1.18, df=62, p=0.244). At TRL, porkfish and great barracuda had Hg concentrations that were elevated when compared to their δ15N enrichment (i.e., they fell above the line, Figure

5a). Likewise the [Hg] for the single Great Barracuda caught at Long Key Hard Bottom also fell above the regression (Figure 5b).

Because basal δ15N can be shifted in areas with anthropogenic nitrogen inputs, the

δ15N is translated into trophic levels by first normalizing individual δ15N value to the average δ15N in a primary consumer at each of the sites (which is arbitrarily assigned a trophic level of 2) and then dividing by the diet tissue discrimination factor. The resulting estimated trophic levels ranged from 2.48 in a stoplight parrotfish to 5.17 in a great barracuda at TRL. Similarly, trophic levels at LKH ranged from 2.44 in an ocean surgeon to 5.21 in a pigfish at Long Key Hard Bottom; the trophic level of the one great barracuda caught at LKH was 4.56. When individuals were pooled by species, the mean trophic levels for Tennessee Reef ranged from 1.99 in frond oyster to 4.57 in great barracuda, and from

1.99 in Atlantic wing oyster to 3.94 in lookdown at Long Key Hard Bottom (Table 1 and

2, Figure 5 and 6). Within all species that had more than one sample collected (n=44), only two species from each site occupied more than one trophic level (i.e., trophic range was >

1.0); great barracuda and hogfish at Tennessee Reef, in addition to leatherjack and white grunt at Long Key Hard Bottom (Table 1 and 2). 23

The inter-species relationship between log [Hg] and increasing trophic level was statistically significant at both sites, TRL: Log [Hg] = -0.832 + (0.824 *Trophic level), p<0.01, r2=0.586, LKH: Log [Hg] = 0.353 + (0.545 * Trophic level), p<0.01, r2=0.608

(Figure 8). From the slope of these relationships a Food Web Magnification Factor

(FWMF) was calculated for each site; Tennessee Reef = 6.7 (100.824), Long Key Hard

Bottom = 3.5 (100.545). The basal Hg concentration (i.e., the amount of Hg available at the base of the food web), calculated from the y-intercept, was determined as 0.15 µg/kg (10-

0.832) for TRL and 2.25 µg/kg (100.353) for LKH.

Discussion

The mercury levels reported were within the range of concentrations that have been observed in the limited number of previous Hg surveys done in the Florida Keys (Strom et al. 1992, Adams et al. 2003, Huge et al. 2014, Tremain et al. 2014). However, comparability was hampered by differences in analytical methods, difference species sampled, and differences in size ranges of captured fish. A survey done back in 1988 by

Strom et al. (1992), employed a different analytical method (e.g. membrane-probe method), and sampled only sediments (coral sand) and two taxa: producers (turtle grass) and consumers (sponges). The [Hg] concentrations in primary consumers collected in the present study were higher than those reported in primary consumer sponges observed by

Strom et al. (1992). Other surveys included species in common with the present study.

Adams et al. (2003) for example, targeted fish species that were “commonly harvested fishes from the Florida Keys National Marine Sanctuary.” Because their study employed wet acid digestion rather than the thermal decomposition used in the present study, Adams 24

et al. (2003) data were adjusted by multiplying values by 1.183, as suggested by Lowery and Garrett (2007), to improve comparability between analytical methods. Following this adjustment, average [Hg] observed in red groupers, black groupers, hogfish, gray snapper and yellowtail snappers in the present study were similar or lower than levels in these species sampled throughout the Keys from 1989-2001 by Adams et al. (2003). However, the fish in the present study were much smaller in size than fish caught by Adams et al.

(2003). Conversely, barracuda caught in the present study were larger than barracuda caught by Adams et al. (2003) and contained higher [Hg]. Two recent surveys focusing on the invasive lionfish (Huge et al. 2014, Tremain et al. 2014), both using thermal decomposition for Hg determination reported low [Hg] (averaging less than 0.2 ppm), consistent with the present study.

An interesting and unexpected observation in the present study was the elevated levels of Hg in porkfish (Anisotremus virginicus) that were comparable or exceeded levels observed in the top predators, the great barracuda; though their δ15N enrichment and subsequent trophic level was lower than other species of grunts (Haemulidae) caught.

Furthermore, these porkfish were relatively small in size compared to the barracuda and other grunts. Moreover, when the influence that size had on variance in [Hg], smaller porkfish tended to have higher concentrations, i.e. a negative slope in the regression (albeit not statistically significant). Therefore, smaller fish were not following the typical biomagnification trend in which higher Hg and δ15N concentrations are found in bigger, older fish. Porkfish have been observed participating in cleaner activities during their juvenile stage, but eventually change their feeding strategies (Bohlke and Chaplin 1968,

Brockman and Hailman 1976, Sazima et al. 2010). This was substantiated in the present 25

study when what appeared to be parasitic copepods were found in the stomach contents of the one porkfish. One could speculate that while acting as a cleaner fish, the juvenile porkfish may have increased Hg exposure which would explain the high levels observed.

However the δ15N of these fish was not markedly higher than that of the larger porkfish and so does not support these speculations. Despite near identical δ15N enrichment and a similar size range in individuals collected from both sites, the difference in Hg between

TRL and LKH was 3 fold. This might indicate a few variables; higher level of cleaner activities, increased presence of top predators visiting cleaner stations at TRL, or an extended period of cleaning behavior (beyond juvenile stage) occurring at TRL.

The [Hg] levels reported here were similar to or higher than Hg concentrations observed in other surveys of coral reef food webs. In general, invertebrates in the present study had Hg levels within the ranges of concentrations reported in other surveys (Hong et al. 2013, Kehrig et al. 2013, Voegborlo and Akagi 2007). However the one species of scavenger type invertebrate, the Caribbean spiny lobster, did show elevated levels of Hg when compared to previous literature (Plessi et al. 2001, Voegborlo and Akagi 2007).

Spiny lobster are known to be scavenger species that will feed upon detritus and other dead organisms they come across. Likewise, Morrison et al. (2015) reported [Hg] much lower than that of the present study in surgeonfish and barracuda on fringe reefs in the remote

Pacific near the American Samoan island of Tutuila. The difference was an order of magnitude despite the fish being similar size. Alternatively, goatfish in Tutulia had higher mean [Hg] than those collected in the present study; however, fish from that study were markedly larger in size. Likewise, Chouvelon et al. (2009) reported Hg concentrations 26

much lower than that of the present study in chubs (Kyphosidae) and groupers (Serranidae) collected along the coast of New Caledonia.

These results are therefore consistent with the observation that South Florida is a

MeHg hotspot (Evans and Crumley 2005, Rumbold et. al 2008, Chen et al. 2012, Thera and Rumbold 2014, Evans et al. 2015). As previously discussed MeHg hot spots may be the result of numerous drivers acting individually or in combination, including: variability in mass loading of inorganic Hg, bioavailability of inorganic Hg, rate of net methylation, activity of methylating bacteria and subsequent substrate availability (i.e. electron acceptor), and the bioavailability of the newly methylated Hg and its entry into the food web via primary consumer. Previous work has shown that Hg concentrations in wet deposition to the Florida Keys (Little Crawl Key) is only slightly lower than concentrations being deposited in the South Florida mainland (Guentzel et al. 2001). Furthermore, Hg loading from atmospheric deposition to the eastern portion of Florida bay, also a recognized hotspot, has been reported to be greater than the Hg loading originating in freshwater runoff the mainland (Rumbold et al. 2011). Equally important, Rumbold et al.

(2011) reported mercury methylation in the sediments of Florida Bay. At this point there is no information on whether or not methylation is occurring in the calcium carbonate based sediments of the Florida Keys; however, it cannot be ruled out nor can it be ruled out that methylation might be occurring in the water column or even within sponges (cf. Hoffmann et al. 2009).

As reasoned previously, a coral reef, which are thought to be very complex and extremely efficient at retaining and recycling nutrients (and possibly other substrates like

Hg), would seem an excellent choice to test the influence that community dynamics has on 27

Hg biomagnification. Additionally, there was a desire to move beyond simply comparing

Hg levels in a few taxa at different locations (which, as discussed above, can be hampered by differences in methods, taxa collected and size) and focus on a metric of this trophic transfer efficiency.

The trophic magnification slope (TMS), as a measure of Hg biomagnification through the entire food web, was 0.23±0.03 (slope ± 95% confidence interval) at TRL and

0.16±0.04 at LKH, when based on individual data points (Figure 6). When designing a study to explore the efficiency and rate of transfer of persistent pollutants like Hg, one must decide whether or not more weight should be put on the raw data (individuals) versus the average data (species means) (Figure 5 vs. Figure 6). In a review by Borgå et al. (2012), it is recommended that regression models used to calculate these biomagnification factors be based on raw data (individuals) rather than reducing to species means. Generally when you use species means, a loss of statistical power occurs due to the reduction of total sample size. In the present study there was a need to assess the influence of the unbalanced sample design. The TMS derived from species means differed markedly from the slope based on raw data at least at TRL where it went from 0.23±0.03 to 0.20±0.07 supporting the idea that the large number of upper-level predators (e.g., porkfish and barracuda) skewed the slope of the raw data. Conversely the two slopes did not differ at LKH except that the confidence intervals were wider, as expected, when based on species means due to smaller sample size, which again is consistent with Borgå et al. (2012). The difference between the two sites suggests a location effect. Statistical analysis (t-test for independent populations) revealed a significant difference between slopes when based on individual data points but not when based on species means. The slope of the regression with more statistical power 28

(i.e., individuals) had confidence intervals that did not overlap. The possibility that unbalanced sampling design and fish size might have skewed the slopes and played a role in the apparent difference of two sites was supported by the fact that the TMS based on species means (which were also standardized for fish size, at least for barracuda) did not significantly differ between sites. Thus, the observed difference in TMS may not be ecologically significant. Alternatively, a local or regional factor, like water chemistry or physical characteristics of the specific sites could be truly affecting Hg biomagnification

(i.e., TMS) (Kidd et al. 2012, Clayden et al. 2013). For example, TMSs can be lower in ecosystems with high nutrient loads due to increased productivity (biomass dilution) and increased growth rates (biodilution) (Clayden et al. 2013). Therefore, it is possible that the productivity of LKH is higher than that of the TRL site and could help explain why the slopes are slightly different.

To test whether or not there might be real differences in the primary production and nutrients between the two sites, this study employed the data of the long term water quality monitoring in the Florida Keys National Marine Sanctuary by Florida International

University’s Southeast Environmental Research Center (Table 6). Although only total organic carbon and light attenuation were statistically significant the monitoring results revealed higher levels of total nitrogen, total phosphorus, total organic carbon, chlorophyll- a, and light attenuation at LKH when compared to levels at TRL (Table 6). This physicochemical difference between the two sites could explain the differences in TMS, as dissolved organic carbon (DOC) has been shown to strongly bind to MeHg, making it unavailable or unable to cross biological membranes (Wiener et al. 2006), and reduce trophic transfer in other ecosystems that exhibit high DOC levels by reducing the amount 29

entering the food web (Rolfhus et al. 2011, Dittman and Driscoll 2009). Rolfhus et al.

(2011) also found that depending on the structure of the DOC, increased DOC can lead to increased MeHg concentration through mediated transport, but bioaccumulation factors

(i.e., Hg in biota/Hg in water) decreased in these areas of increased DOC, while TMS remained the same. This suggests that Hg in upper trophic level organisms was ultimately defined by how much Hg was entering the base of the food web in those systems. The present study and reviews by Lavoie et al. (2013) support the hypothesis that spatial differences in chemistry or physical characteristics may have a stronger influence on biomagnification than community structure.

To further assess ecosystem differences between the two sampling locales, the present study also looked at stable isotopes of carbon. Carbon isotopes are used as a tracer of sources of primary production and have been used for differentiation between nearshore and marine systems (Peterson and Fry 1987, Chasar et al. 2005, Coleman 2012) and benthic and pelagic ecosystems (Power et al. 2002, Kidd et al. 2003). This analysis may also provide a measure of the degree of openness (subsidies from other systems) in the two systems. The ranges of δ13C at the two sampling sites suggested that taxa collected occupy a broad range of habitats (Figure 3). This is not surprising due to the proximity of numerous habitats and the rapid shifts in habitat type (patch reefs, hard bottom, seagrass beds) observed in the Florida Keys area. The range of δ13C enrichment was consistent between the two study sites. Isotopic results in the present study can also be evaluated against commonly used benchmarks found in the literature. For example, Fry and Sherr (1984) recommend the following benchmarks for carbon source based on δ13C: phytoplankton: -

23‰ to -18‰, benthic: -18‰ to -13‰, and seagrass: >-13‰. Based upon these 30

benchmarks, it would appear that the majority of species collected at TRL derived their carbon from benthic sources, likely algae and coral (zoozxanthellae) sources, with the exception of the primary consumer frond oyster, which appeared to draw from planktonic sources. Primary consumers collected in this study were two different species of bivalve that showed slightly different concentrations of Hg and δ15N. The Atlantic wing oyster from LKH had higher mean Hg (43.0 µg/kg) and δ15N (4.46‰) in comparison to the frond oyster collected at TRL, (34.3 µg/kg) and (3.37‰) respectively. Similarly, a single frond oyster was collected from LKH and had greater Hg and δ15N concentrations. It would therefore appear that there may be a slight difference in Hg and nitrogen availability between the two sites, as evidenced by primary consumers.

LKH, which is located nearshore, likely had additional sources of carbon that was borne out by a much larger range of δ13C values when compared to that of TRL. The grazer type species collected in the present study had higher enrichment of carbon and nitrogen for comparable species of surgeonfish and parrotfish surveyed in the upper Florida Keys

(Lamb et al. 2012), and had trophic levels only slightly above that of the primary consumer bivalves collected (Table 1,2). This could be due to the differences in available food sources. The differences in coral cover versus algal cover for reefs in the marine protected areas during the study by Lamb et al. (2012) may have provided more available coral for grazer fish, possibly changing the amount of enrichment and decreasing comparability to this study.

At the outset of this project there was an expectation that, while recognizing some differences between the two sites as examples of coral reef food webs, their TMS would still be more similar than compared to temperate, polar or freshwater systems. When 31

assessing physical and chemical factors that can influence TMS on a regional basis the present study encountered variations that made it difficult to draw definitive conclusions comparing sites. Other studies (Kidd et al. 2012, Clayden et al. 2013) suggest that physical and chemical differences can overwhelm food web dynamics when influencing the TMS.

It is likely that meta-analyses of large data sets, similar to the one conducted by Lavoie et al. (2013) are necessary, to overcome variations that might occur regionally and locally and detect the influence of food web dynamics on TMS.

Interestingly many of these same physical and chemical factors have been known for some time to affect bioavailability and methylation, and consequently basal MeHg.

Early comparisons concluded that differences in basal MeHg (Riget et al. 2007) was the dominant factor driving differences in TMS and FWMF. Now it appears these factors can influence both basal Hg and the efficiency of biomagnification. The basal or background concentration of Hg is the bioavailable portion that is the first step in the chain of bioaccumulation. Although earlier studies suggested the possibility of using the y-intercept of the log Hg: trophic level regression as an estimate for basal Hg (Jardine et al. 2006), recently concerns have been raised about how much the intercept depends on the slope and the uncertainties embedded within it, i.e. diet tissue discrimination factor (Borgå et al.

2012, Lavoie et al. 2013).

The present study did not survey [Hg] in phytoplankton, nor did it assess [Hg] in zooxanthellae-containing coral species. While a previous report suggests very low concentrations (Strom 1992), there are reports of [Hg] ranging as high as 50 ppb (dry weight) in Caribbean corals (Guzman and Garcia 2002, Berry et al. 2013). Parrotfish, an herbivorous (primary consumer) species that typically feeds upon coralline algae and the 32

coral polyps that contain symbiotic zooxanthellae were only caught at TRL. The stoplight parrotfish species (Sparisoma viride) was the closest fish to the primary consumer bivalves, with respect to trophic level (2.50), and had mean Hg concentrations that were lower than that of the frond oysters collected.

Although stable isotope analysis has been used for some time to assess trophic position and more recently combined with toxicant analyses to assess biomagnification, a significant source of uncertainty stems from the fact that nutrient inputs from both natural and anthropogenic sources, can vary across systems and affect the baseline isotopic ratios

(for review, see Cabana and Rasmussen 1996, Fry 1999, Jardine et al. 2006). Isotopes of nitrogen and carbon have been observed to differ in species of snapper, jacks, and pinfish residing in Florida Bay, often separated by less than 25 km (Chasar et al. 2005). That study found significant differences in carbon, nitrogen and sulfur isotopes collected in fish, seston and seagrass from 5 stations in the Florida Bay area. Fish that have low site fidelity and, consequently, be deriving their δ15N (and Hg) signatures from multiple ecosystems would add to the confusing patterns. This was especially relevant in the present study due to the differences in observed δ15N in primary consumers which was higher in the Atlantic wing oyster collected at LKH. Lapointe et al. (2004) reported that anthropogenically influenced discharges from the mainland can affect δ15N signatures in the waters of the

Florida Keys. This is especially relevant because of recent changes to the Everglades

Restoration plan and the funneling of massive amounts of water and nutrients south from

Lake Okeechobee, ultimately draining into the Florida Keys (Chasar et al. 2005, Evans and

Crumley 2005, Wozniak et al. 2012). It is for these reasons that the basal nitrogen is normalized to local primary consumers for site specific adjustment. 33

Once δ15N was normalized to the primary consumer for each site, trophic levels were calculated using a nitrogen diet-tissue discrimination factor of 3.4 and the stable isotope δ15N data from each organism (Jardine et al. 2006). Then the relationship between

Hg and trophic level was assessed and eventually used to calculate a food web magnification factor (FWMF), also known as a trophic magnification factor (TMF) (Borgå et al. 2004, Jardine et al. 2006, Lavoie et al. 2013). Food web magnification factors have increased interpretability over the TMS because they provide a measure of biomagnification efficiency integrated across the entire food web (Borgå et al. 2012,

Jardine et al. 2006). The adjustments a FWMF makes for differences observed in δ15N enrichment at the base of the food web allows seamless comparisons across systems, and independent study of biomagnification rates (Borgå et al. 2004, Jardine et al 2006).

However, several authors have recently raise concerns about the heavy dependence that

FWMF has on the diet tissue discrimination factor (DTDF, i.e., change in δ15N with the change in trophic level), which may vary across latitudes (due to temperature dependence) and among taxa, particularly endotherm and ectotherms and certainly sharks (Jardine et al.

2006, Borgå et al. 2012, Rumbold et al. 2014). In past studies of persistent organic pollutants, it was generally accepted that the inclusion of mammals and sea birds increased the trophic magnification slope and food web magnification factors of a food web when compared to one that included only fish and invertebrates (Fisk et al. 2001, Hallanger et al.

2011). Based on their extensive review of literature, Lavoie et al. (2013) found that neither the species composition nor the percentage of endotherms in the food web affected the trophic magnification slope. Interchangeable DTDF also allow systems with unrelated species and broad ranges to be comparable (Jardine et al. 2006). The DTDF used in the 34

present study to estimate trophic level was 3.4‰. This was based on mean values derived from early laboratory research (DeNiro and Epstein 1981, Minagawa and Wada 1984), and has been the most frequently used DTDF in reviewed studies (Lavoie et al. 2013). Slopes calculated for the species means of the regression between log Hg and trophic level were

0.82±0.26 (Slope± 95% CI) for TRL and 0.55±0.16 for LKH, equivalent to FWMFs of 6.7 for TRL and 3.5 for LKH. Statistical analysis of the two slopes revealed no statistical difference (t=1.89, df=62, p=0.06). The confidence intervals for TRL and LKH also encompassed an overlapping range of values, meaning the present study cannot conclude that there is a significant difference between the slopes.

As previously mentioned, the biomagnification that takes place in a food web is influenced by a multitude of factors like primary and secondary productivity, community structure, availability of prey, food chain length, and linkage strength (complex or simple food web) (Cabana et al. 1994, Futter 1994, Watras et al. 1998, Pickhardt et al. 2002,

Lavoie 2013).

The TMS, as our measurement of biomagnification in this coral reef food web, was

0.23 at TRL and 0.16 at LKH (Table 5) and, thus, averages 0.20 for the two. This is identical with the average trophic magnification slope for all marine sites of 0.20 ± 0.10

(mean ± SD) as reported by Lavoie et al. (2013) in their metaanalysis. Lavoie et al. (2013) found considerable variation in TMS with a range of -0.19 to 0.48 based on THg for this global meta-analysis. Such a wide variance highlights the unknown master mechanism that controls TMS. “Biomagnification increased with latitude and this phenomenon was likely due to a combination of interdependent variables related to temperature,” (Lavoie et al.

2013, p. 13391). This too highlights the possibility that physical and chemical variables 35

may have a stronger influence on TMS over community structure. Slopes of total Hg have also been reported in this range for freshwater lakes and streams (Lavoie et al. 2013,

Campbell et al. 2003, Kidd et al. 2003, Campbell et al. 2004, Chasar et al. 2009, Churmchal and Hambright 2009), freshwater arctic systems (Power et al. 2002, Swanson and Kidd

2010), marine arctic systems (Atwell et al. 1998, Campbell et al. 2005, Riget et al. 2007,

Swanson and Kidd 2010), and tropical and sub-tropical marine systems (Jarman et al. 1996,

Al-Reasi et al. 2007, Di Beneditto et al. 2012, Lavoie et al. 2013, Thera and Rumbold

2014). Although recent studies have begun placing less weight on food web magnification factors due to the uncertainty of diet-tissue discrimination factors, the FWMF’s for these coral reef food web sites were 7.8 for TRL and 3.4 for LKH when based on all individuals

(Figure 7). This indicates that Hg was increasing by factors of 7.8 and 3.4 with each increase in trophic level. Those factors are within the range of FWMFs reported in the literature for comparable benthic and coastal food webs in the tropical southeast of Brazil

(Muto et al. 2014, Kehrig et al. 2013), an estuary in North Queensland, Austrailia (Jardine et al. 2012), a coastal food web in the Gulf of Oman (Al-Reasi et al. 2007), an Arctic marine food web in Greenland (Riget et al. 2007), a coastal ecosystem in the South China sea (Zhu et al. 2013) and a coastal ecosystem off the coast of Southwest Florida (Thera and Rumbold

2014) (Table 5). The FWMFs were slightly lower than the highest FWMFs found in literature reviewed: an estuary ecoystem (FWMF = 8.3) in Sarasota Bay, FL (Hong et al.

2013) and a deep sea system (FWMF = 11.3) in Suruga Bay, Japan (Sakata et al. 2015)

(Table 5), and also fall within the range reported by Lavoie et al. (2013) of 6.2 ± 4.1.

Based on observed similarities in slopes, Riget et al. (2007) concluded biomagnification was similar across regions and ecosystems. This lead them and others to 36

suggest that spatial variation in Hg accumulation in top predators was primarily the result of differences in the basal Hg entering the food web. The more recent review by Lavoie et al. (2013) concluded that despite intensive comparative studies, no general consensus had emerged with respect to the main variables affecting Hg biomagnification in aquatic ecosystems. That study explored the physical, chemical and biological factors that could explain the observed variability in trophic magnification slopes on a global scale. That review found a latitudinal gradient with average TMS at polar and temperate sites (0.21 and 0.17, respectively) consistently higher than tropical sites (mean = 0.13), attributing the differences to drivers like temperature, and ecosystem productivity variables (e.g. total phosphorous and Chl-a). Warmer temperatures were thought to stimulate higher growth rates in organisms which can decrease the amount of Hg in the tissues and biodilute, when compared to colder temperatures where growth is suppressed. Colder temperatures also lead to lower excretion rates of Hg and higher accumulation in organisms across the food web (Lavoie et al. 2013). Food webs at higher lattitudes are also very simple and exhibit a low level of biodiversity, which leads to higher bioaccumulation when compared to the complex and highly diverse food webs of the tropical and subtropical marine systems.

Differences in productivity and its effects were also observed in a study by Swanson and

Kidd (2010) in an arctic marine system (0.08) and an arctic freshwater system (0.16-0.26)

(Table 5). Marine systems typically have a wide variety of available prey items within the food web due to the openess of the systems compared to restricted freshwater systems, possibly leading to lower biomagnification (Gray 2002).

All of the factors discussed above were hypothesized to effectively decrease the efficiency with which Hg biomagnifies through a coral reef food web. The present study 37

originally hypothesized a relationship that was more similar at the two sites as compared to other ecosystems. However, this study had mixed results that neither completely supported or refuted this hypothesis. Although there was some evidence to support signficant differences in the TMS between the two sites (which could have been due to differences in water quality), this difference in slopes could also be a result of unbalanced sampling design. Nonetheless, these results clearly demonstrate that Hg is biomagnified through subtropical coral reef ecosystems and that the transfer efficiency was identical to the global average reported for marine ecosystems (Lavoie et al. 2013). The present study also highlights the need to include variance estimates for the slopes. The difficulty in comparing across systems was compounded by the fact that even some of the most recent studies fail to include variance terms (i.e., standard error or confidence interval of slope), despite recommendations in recent reviews (Borgå et al. 2012, Lavoie et al. 2013).

The results from this study should serve as reference for future studies in coral reef ecosystems and other marine food webs. The implications of this study should be considered by future studies of biomagnification of other toxicants and natural toxins.

Continued research of these complex systems is needed to assess whether or not latitudinal gradients exist when assessing Hg transfer. Differences in study designs (examination of large number of sites, characterization of baseline by primary consumer, decision on diet- tissue discrimination factor for δ15N) also need to be standardized to improve comparability. There is also a need for continued monitoring of Hg levels in the biota of the Florida keys, especially in species that are targeted recreationally and where the present study found [Hg] well above the limit set for safe consumption (i.e. great barracuda). The 38

suprisingly high Hg levels and apparent inverse relationship between Hg and size observed in the porkfish also warrants attention for future studies to see if there might be a link between the cleaner behavior of small juveniles and Hg tranfer from parasites picked from large top predators.

Conclusions

The relationships between the biomagnification of Hg, δ15N, and trophic level were within ranges reported from previous studies of other marine food webs. The TMS, as a gauge of mercury biomagnification, was slightly higher at TRL, likely influenced by the unbalanced sampling design, as well as the high [Hg] observed in barracuda and porkfish.

There was also some evidence of differences in water quality between locations but was only statistically significant with respect to light attenuation and TOC. The results of the present study did not definitively support the original hypotheses that the relationship between Hg and δ15N or trophic level would be relatively similar at the two sites in the

Florida Keys and, more importantly, that the complex food web within the coral reef, with its high number of lateral or horizontal links, would decrease the efficiency Hg biomagnification as compared to other ecosystems.

39

Literature Cited

Adams, D. H. and R. H. McMichael, Jr. 2001. Mercury levels in marine and estuarine fishes of

Florida. Florida Marine Research Institute Technical Report TR-6. 35 pp.

Adams, D. H., R. H. McMichael, Jr., and G. E. Henderson. 2003. Mercury levels in marine and

estuarine fishes of Florida 1989–2001. Florida Marine Research Institute Technical Report

TR-9. 2nd ed. rev. 57 pp.

Adams, D. H. and R. H. McMichael, Jr. 2007. Mercury in king mackerel, Scomberomorus cavalla,

and Spanish mackerel, S. maculatus, from waters of the south-eastern USA: regional and

historical trends. Marine and Freshwater Research 58:187-193.

Adams, D. H. and G. V. Onorato. 2005. Mercury concentrations in red drum, Sciaenops ocellatus,

from estuarine and offshore waters of Florida. Marine Pollution Bulletin 50:291-300.

Adams, D. H., C. Sonne, N. Basu, R. Dietz, D.H. Nam, P. S. Leifsson, and A. L. Jensen. 2010.

Mercury contamination in spotted seatrout, Cynoscion nebulosus: An assessment of liver,

kidney, blood, and nervous system health. Science of the Total Environment 408:5808-

5816.

Bank, M. S., E. Chesney, J. P. Shine, A. Maage, and D. B. Senn. 2007. Mercury bioaccumulation

and trophic transfer in sympatric snapper species from the Gulf of Mexico. Ecological

Applications 17:2100-2110.

Bhavsar, S. P., S. B. Gewurtz, D. J. McGoldrick, M. J. Keir, and S. M. Backus. 2010. Changes in

mercury levels in Great Lakes fish between 1970s and 2007. Environmental Science and

Technology, 44(9), 3273-3279. 40

Blaber, S. J. M. 1982. The Ecology of Sphyraena-barracuda (Osteichthyes, Perciformes) in the

Kosi system with notes on the Sphyraenidae of other natal estuaries. South African Journal

of Zoology, 17(4), 171-176.

Bloom, N. S. 1992. On the chemical form of mercury in edible fish and marine invertebrate tissue.

Canadian Journal of Fisheries and Aquatic Sciences. 49:1010-1017.

Boening, D. W. 2000. Ecological effects, transport, and fate of mercury: a general review.

Chemosphere 40:1335-1351.

Bohlke, J. E., and C. C. G. Chaplin. 1968. Fishes of the Bahamas and adjacent waters. Livingston

Publication Co. Wynnewood, Pennsylvania.

Borgå, K., A. T. Fisk, P. F. Hoekstra, and D. C. G. Muir. 2004. Biological and chemical factors of

importance in the bioaccumulation and trophic transfer of persistent organochlorine

contaminants in arctic marine food webs. Environmental Toxicology and Chemistry

23:2367-2385.

Borgå, K., K. A. Kidd, D. C. Muir, O. Berglund, J. M. Conder, F. A. Gobas, J. Kucklick, O. Malm,

and D. E. Powell. 2012. Trophic magnification factors: considerations of ecology,

ecosystems, and study design. Integrated Environmental Assessment and Management,

8(1), 64-84.

Bowles, K. C., S. C. Apte, W. A. Maher, M. Kawei, and R. Smith. 2001. Bioaccumulation and

biomagnification of mercury in Lake Murray, Papua New Guinea. Canadian Journal of

Fisheries and Aquatic Sciences 58:888-897.

Broman, D., C. Rolff, C. Näf, Y. Zebühr, B. Fry, and J. Hobbie. 1992. Using ratios of stable

nitrogen isotopes to estimate bioaccumulation and flux of polychlorinated dibenzo‐p‐ 41

dioxins (PCDDs) and dibenzofurans (PCDFs) in two food chains from the Northern Baltic.

Environmental Toxicology and Chemistry, 11(3), 331-345.

Cabana, G. and J. B. Rasmussen. 1994. Modelling food chain structure and contaminant

bioaccumulation using stable nitrogen isotopes. Nature 372:255-257.

Cabana, G., A. Tremblay, J. Kalff, and J. B. Rasmussen. 1994. Pelagic food chain structure in

Ontario lakes: a determinant of mercury levels in lake trout (Salvelinus namaycush).

Canadian Journal of Fisheries and Aquatic Sciences, 51(2), 381-389.

Cabana, G., and J. B. Rasmussen. 1996. Comparison of aquatic food chains using nitrogen

isotopes. Proceedings of the National Academy of Sciences, 93(20), 10844-10847.

Cai, Y., J. R. Rooker, G. A. Gill, and J. P. Turner. 2007. Bioaccumulation of mercury in pelagic

fishes from the northern Gulf of Mexico. Canadian Journal of Fisheries and Aquatic

Sciences 64:458-469.

Carpi, A. 1997. Mercury from combustion sources: A review of the chemical species emitted and

their transport in the atmosphere. Water, Air, and Soil Pollution 98:241-254.

Chasar, L. C., J. P. Chanton, C. C. Koenig, and F. C. Coleman. 2005. Evaluating the effect of

environmental disturbance on the trophic structure of Florida Bay, USA: multiple stable

isotope analyses of contemporary and historical specimens. Limnology and Oceanography,

50(4), 1059.

Chasar, L. C., B. C. Scudder, A. R. Stewart, A. H. Bell, and G. R. Aiken. 2009. Mercury cycling

in stream ecosystems. 3. Trophic dynamics and methylmercury bioaccumulation.

Environmental Science & Technology, 43(8), 2733-2739.

Chen, C. Y., C. T. Driscoll, and N. C. Kamman. 2012. Mercury Hotspots in Freshwater

Ecosystems: Drivers, Processes, and Patterns. In Mercury in the Environment: Pattern and 42

Process, Bank, M. C., Ed.; University California Berkeley Press: Berkeley, California,

2012.

Chouvelon, T., M. Warnau, C. Churlaud, and P. Bustamante. 2009. Hg concentrations and related

risk assessment in coral reef crustaceans, molluscs and fish from New Caledonia.

Environmental Pollution, 157(1), 331-340.

Clayden, M. G., K. A. Kidd, B. Wyn, J. L. Kirk, D. C. Muir, and N. J. O’Driscoll. 2013. Mercury

biomagnification through food webs is affected by physical and chemical characteristics

of lakes. Environmental Science and Technology,47(21), 12047-12053.

Coleman, D. C. (Ed.). 2012. Carbon isotope techniques. Academic Press.

De Flora, S., C. Bennicelli, and M. Bagnasco. 1994. Genotoxicity of mercury compounds. A

review. Mutation Research/Reviews in Genetic Toxicology 317(1), 57-79.

DeNiro, M. J. and S. Epstein. 1981. Influence of diet on the distribution of nitrogen isotopes in

animals. Geochimica et Cosmochimica Acta 45:341-351.

Denton, G. R. W., and C. Burdon-Jones. 1986. Trace metals in fish from the Great Barrier

Reef. Marine Pollution Bulletin, 17(5), 201-209.

De Sylva, D. P. 1963. Systematics and life history of the great barracuda, Sphyraena barracuda

(Walbaum). University of Miami Press, Institute of Marine Science.

Dittman, J. A., and C. T. Driscoll. 2009. Factors influencing changes in mercury concentrations in

lake water and yellow perch (Perca flavescens) in Adirondack

lakes. Biogeochemistry, 93(3), 179-196.

Drevnick, P. E. and M. B. Sandheinrich. 2003. Effects of dietary methylmercury on reproductive

endocrinology of fathead minnows. Environmental Science and Technology 37:4390-4396. 43

Dulvy, N. K., R. P. Freckleton, and N. V. Polunin. 2004. Coral reef cascades and the indirect

effects of predator removal by exploitation. Ecology letters, 7(5), 410-416.

Dvonch, J. T., J. R. Graney, G. J. Keeler, and R. K. Stevens. 1999. Use of elemental tracers to

source apportion mercury in South Florida precipitation. Environmental Science and

Technology 33:4522-4527.

Dvonch, J. T., G. J. Keeler, and F. J. Marsik. 2005. The influence of meteorological conditions on

the wet deposition of mercury in Southern Florida. Journal of Applied Meteorology

44:1421-1435.

Eggleston, D. B., C. P. Dahlgren, and E. G. Johnson. 2004. Fish density, diversity, and size-

structure within multiple back reef habitats of Key West National Wildlife Refuge. Bulletin

of Marine Science, 75(2), 175-204.

Evans, D. W., and P. H. Crumley. 2005. Mercury in Florida Bay fish: Spatial distribution of

elevated concentrations and possible linkages to Everglades restoration. Bulletin of Marine

Science, 77(3), 321-346.

Evans, D. W., M. Cohen, C. Hammerschmidt, W. Landing, D. G. Rumbold, J. Simons, and S.

Wolfe. 2015. White Paper on Gulf of Mexico Mercury Fate and Transport: Applying

Scientific Research to Reduce the Risk from Mercury in Gulf of Mexico Seafood. NOAA

Technical Memorandum NOS NCCOS 192. 54 p

Evers, D. C., R. T. Graham, C. R. Perkins, R. Michener, and T. Divoll. 2009. Mercury

concentrations in the goliath grouper of Belize: an anthropogenic stressor of concern.

Endangered Species Research 7:249-256.

Fink, L., D. G. Rumbold, and P. Rawlik. 1999. "The Everglades mercury problem." In Everglades

Interim Report. South Florida Water Management District, West Palm Beach, FL. 44

Fisk, A. T., K. A. Hobson, and R. J. Norstrom. 2001. Influence of chemical and biological factors

on trophic transfer of persistent organic pollutants in the Northwater Polynya marine food

web. Environmental Science and Technology 35:732-738.

Fitzgerald, W. F. 1995. Biogeochemical cycling of mercury: Global and local aspects. In M.

Martin (ed). U.S. Environmental Protection Agency, National Forum on Mercury in Fish:

Proceedings. U.S. Environmental Protection Agency, Washington, D.C.

Frank, K. T., B. Petrie, N. L. Shackell. 2007. The ups and downs of trophic control in continental

shelf ecosystems. Trends in Ecology and Evolution 22:236-242.

Fry, B. 1999. Using stable isotopes to monitor watershed influences on aquatic trophodynamics.

Canadian Journal of Fisheries and Aquatic Sciences, 56(11), 2167-2171.

Fry, B. and E. B. Sherr. 1984. δ13C measurements as indicators of carbon flow in marine and

freshwater ecosystems. Contributions in Marine Science 27:13-47.

Futter, M. N. 1994. Pelagic food-web structure influences probability of mercury contamination

in lake trout (Salvelinus namaycush). Science of the Total Environment 145:7-12.

FDEP (Florida Department of Environmental Protection). 2009. Analysis of methyl mercury in

sediment and tissue by purge-and-trap/GC/AFD. FDEP mercury SOP HG-003. Available:

http://www.dep.state.fl.us/labs/cgi-in/sop/sop3.asp (Nov 2012)

FDEP (Florida Department of Environmental Protection). 2009. Trace level total mercury analysis

in tissue by cold vapor atomic fluorescence CVAF) spectroscopy. FDEP mercury SOP

HG-007. Available: http://www.dep.state.fl.us/labs/ cgi-bin/sop/sop3.asp (Nov 2012)

FDOH (Florida Department of Health). 2015. Florida Fish Advisories. Tallahassee, FL. 45

Grieb, T. M., C. T. Driscoll, C. L. Schofield, G. L. Bowie, and D. B. Porcella. 1990. Factors

affecting mercury accumulation in fish in the upper Michigan peninsula. Environmental

Toxicology and Chemistry. 9:919-930.

Guentzel, J., Landing, W. M., Gill, G. A., and Pollman, C. D. 1995. Atmospheric deposition of

mercury in Florida: The FAMS project (1992–1994). Water, Air, and Soil Pollution, 80(1-

4), 393-402.

Guentzel, J. L., W. M. Landing, G. A. Gill, and C. D. Pollman. 2001. Processes influencing rainfall

deposition of mercury in Florida. Environmental Science and Technology 35:863-873.

Haines, E. B. and C. L. Montague. 1979. Food sources of estuarine invertebrates analyzed using

13C/12C ratios. Ecology 60:48-56.

Hallanger, I. G., N. A. Warner, A. Ruus, A. Evenset, G. Christensen, D. Herzke, G. W. Gabrielsen,

and K., Borgå. 2011. Seasonality in contaminant accumulation in arctic marine pelagic

food webs using trophic magnification factor as a measure of

bioaccumulation. Environmental Toxicology and Chemistry, 30(5), 1026-1035.

Hammerschmidt, C. and W. Fitzgerald. 2006. Bioaccumulation and trophic transfer of

methylmercury in Long Island Sound. Archives of Environmental Contamination and

Toxicology 51:416-424.

Hammerschmidt, C. R., M. B. Sandheinrich, J. G. Wiener, and R. G. Rada. 2002. Effects of dietary

methylmercury on reproduction of fathead minnows. Environmental Science and

Technology 36:877-883.

Hoffmann, F., R. Radax, D. Woebken, M. Holtappels, G. Lavik, H. T. Rapp, M. L. Schlappy, C.

Schleper, and M. M. Kuypers. 2009. Complex nitrogen cycling in the sponge Geodia

barretti. Environmental Microbiology, 11(9), 2228-2243. 46

Hop, H., K. Borgå, G. W. Gabrielsen, L. Kleivane, and J. U. Skaare. 2002. Food web magnification

of persistent organic pollutants in poikilotherms and homeotherms from the Barents Sea.

Environmental Science and Technology 36:2589-2597.

Hoese, H., and R. Moore. 1998. Fishes of the Gulf of Mexico, Texas, Louisiana, and adjacent

waters (2nd ed.). College Station: Texas A and M University Press

Houck, A. and J. J. Cech, Jr. 2004. Effects of dietary methylmercury on juvenile Sacramento

blackfish bioenergetics. Aquatic Toxicology 69:107-123.

Huge, D. H., P. J. Schofield, C. A. Jacoby, and T. K. Frazer. 2014. Total mercury concentrations

in lionfish (Pterois volitans/miles) from the Florida Keys National Marine Sanctuary, USA.

Marine Pollution Bulletin, 78(1), 51-55.

Jardine, T. D., I. A. Halliday, C. Howley, V. Sinnamon, and S. E. Bunn. 2012. Large scale surveys

suggest limited mercury availability in tropical north Queensland (Australia). Science of

the Total Environment, 416, 385-393.

Jardine, T. D., K. A. Kidd, and A. T. Fisk. 2006. Applications, considerations, and sources of

uncertainty when using stable isotope analysis in ecotoxicology. Environmental Science

and Technology 40:7501-7511.

Jepsen, D. B. and K. O. Winemiller. 2002. Structure of tropical river food webs revealed by stable

isotope ratios. Oikos 96(1), 46-55.

Kamman, N. C., N. M. Burgess, C. T. Driscoll, H. A. Simonin, W. Goodale, J. Linehan, R.

Estabrook, M. Hutcheson, A. Major, A. M. Scheuhammer, and D. A. Scruton. 2005.

Mercury in freshwater fish of northeast North America–a geographic perspective based on

fish tissue monitoring databases. Ecotoxicology, 14(1-2), 163-180. 47

Kang, W. J., J. H. Trefry, T. A. Nelson, and H. R. Wanless. 2000. Direct atmospheric inputs versus

runoff fluxes of mercury to the lower Everglades and Florida Bay. Environmental Science

and Technology 34:4058–4063.

Kannan, K., R. G. Smith, Jr., R. F. Lee, H. L. Windom, P. T. Heitmuller, J. M. Macauley, and J.

K. Summers. 1998. Distribution of total mercury and methyl mercury in water, sediment,

and fish from South Florida estuaries. Archives of Environmental Contamination and

Toxicology 34:109-118.

Kehrig, H. A., T. G. Seixas, O. Malm, A. P. M. Di Beneditto, and C. E. Rezende. 2013. Mercury

and selenium biomagnification in a Brazilian coastal food web using nitrogen stable

isotope analysis: a case study in an area under the influence of the Paraiba do Sul River

plume. Marine Pollution Bulletin, 75(1), 283-290.

Kidd, K. A., R. H. Hesslein, R. J. P. Fudge, and K. A. Hallard. 1995. The influence of trophic level

as measured by δ 15 N on mercury concentrations in freshwater organisms. Water, Air, and

Soil Pollution 80(1), 1011-1015.

Kidd, K. A., H. A. Bootsma, R. H. Hesslein, W. Lyle Lockhart, and R. E. Hecky. 2003. Mercury

concentrations in the food web of Lake Malawi, East Africa. Journal of Great Lakes

Research 29:258-266.

Kidd, K. A., M. Clayden, and T. Jardine. 2012a. Bioaccumulation and biomagnification of

mercury through food webs. Environmental chemistry and toxicology of mercury. Wiley,

Hoboken, 455-499.

Kidd, K. A., D. C. Muir, M. S. Evans, X. Wang, M. Whittle, H. K. Swanson, T. Johnston, and S.

Guildford. 2012b. Biomagnification of mercury through lake trout (Salvelinus namaycush) 48

food webs of lakes with different physical, chemical and biological characteristics. Science

of the Total Environment, 438, 135-143.

Kling, G. W., B. Fry, and W. J. O'Brien. 1992. Stable isotopes and planktonic trophic structure in

arctic lakes. Ecology, 561-566.

Lamb, K., P. K. Swart and M. A. Altabet. 2012. Nitrogen and carbon isotopic systematics of the

Florida reef tract. Bulletin of Marine Science, 88(1), 119-146.

Lapointe, B. E., P. J. Barile, and W. R. Matzie. 2004. Anthropogenic nutrient enrichment of

seagrass and coral reef communities in the Lower Florida Keys: discrimination of local

versus regional nitrogen sources. Journal of Experimental Marine Biology and Ecology,

308(1), 23-58.

Larsson, P., L. Collivin, L. Okla, and G. Meyer. 1992. Lake productivity and water chemistry as

governors of the uptake of persistent pollutants in fish. Environmental Science and

Technology, 26, 346-352

Leonard, A., P. Jacquet, and R. Lauwerys. 1983. Mutagenicity and teratogenicity of mercury

compounds. Mutation Research/Reviews in Genetic Toxicology 114(1), 1-18.

Lewis, M. and C. Chancy. 2008. A summary of total mercury concentrations in flora and fauna

near common contaminant sources in the Gulf of Mexico. Chemosphere 70:2016-2024.

Lowery, T., and E. S. Garret III. 2005. A Synoptic Survey of Total Mercury in Recreational Finfish

of the Gulf of Mexico. NOAA Fisheries.

Lowery, T. A., R. S. Winters, and E. S. Garrett III. 2007. Comparison of total mercury

determinations of fish fillet homogenates by thermal decomposition, amalgamation, and

atomic absorption spectrophotometry versus cold vapor atomic absorption

spectrophotometry. Journal of Aquatic Food Product Technology, 16(2), 5-15. 49

Luoma, S. N. and P. S. Rainbow. 2005. Why is metal bioaccumulation so variable? Biodynamics

as a unifying concept. Environmental Science and Technology 39:1921-1931.

Mason, R. P., W. F. Fitzgerald, and F. M. M. Morel. 1994. The biogeochemical cycling of

elemental mercury: Anthropogenic influences. Geochimica et Cosmochimica Acta

58:3191-3198.

Mergler, D., H. A. Anderson, L. H. M. Chan, K. R. Mahaffey, M. Murray, M. Sakamoto, and A.

H. Stern. 2007. Methylmercury exposure and health effects in humans: a worldwide

concern. AMBIO: A Journal of the Human Environment, 36(1), 3-11.

Matta, M. B., J. Linse, C. Cairncross, L. Francendese, and R. M. Kocan. 2001. Reproductive and

transgenerational effects of methylmercury or Aroclor 1268 on Fundulus heteroclitus.

Environmental Toxicology and Chemistry 20:327-335.

Minagawa, M. and E. Wada. 1984. Stepwise enrichment of 15N along food chains: Further

evidence and the relation between δ15N and animal age. Geochimica et Cosmochimica

Acta 48:1135-1140.

Mittelbach, G. G. and L. Persson. 1998. The ontogeny of piscivory and its ecological

consequences. Canadian Journal of Fisheries and Aquatic Sciences 55:1454-1465.

Morrison, R. J., P. J. Peshut, R. J. West, and B. K. Lasorsa. 2015. Mercury (Hg) speciation in coral

reef systems of remote Oceania: Implications for the artisanal fisheries of Tutuila, Samoa

Islands. Marine Pollution Bulletin, 96 (1) 41-56.

Muto, E. Y., , L. S. Soares, J. E. Sarkis, M. A. Hortellani, M. A. Petti, and T. N. Corbisier 2014.

Biomagnification of mercury through the food web of the Santos continental shelf,

subtropical Brazil. Marine Ecology Progress Series, 512, 55-70. 50

Myers, T.C., Y. Wei, A. B. Hudischewskyj, J.L. Haney, and S.G. Douglas. 2006. Model-based

analysis and tracking of airborne mercury emissions to assist in watershed planning.

Prepared for the U.S. EPA Office of Water by ICF International, San Rafael, CA.

Available: http://www.epa.gov/owow/tmdl/pdf/final300report_10072008.pdf

NRC (National Research Council). 2000. Toxicological effects of methylmercury. Board on

Environmental Science and Toxicology. National Academy Press, Washington, DC.

Odum, H. T. and E. P. Odum. 1955. Trophic structure and productivity of a windward coral reef

community on Eniwetok Atoll. Ecological Monographs 25:291-320.

O’Toole, A. C., A. J. Danylchuk, T. L. Goldberg, C. D. Suski, D. P. Philipp, E. Brooks, and S. J.

Cooke. 2011. Spatial ecology and residency patterns of adult great barracuda (Sphyraena

barracuda) in coastal waters of The Bahamas. Marine Biology, 158(10), 2227-2237.

Peterson, B. J. and B. Fry. 1987. Stable isotopes in ecosystem studies. Annual Review of Ecology

and Systematics 18:293-320.

Peterson, S. A., J. Van Sickle, A. T. Herlihy, and R. M. Hughes. 2007. Mercury concentration in

fish from streams and rivers throughout the western United States. Environmental Science

and Technology, 41(1), 58-65.

Pickhardt, P. C., C. L. Folt, C. Y. Chen, B. Klaue, and J. D. Blum. 2002. Algal blooms reduce the

uptake of toxic methylmercury in freshwater food webs. Proceedings of the National

Academy of Sciences 99:4419-4423.

Plessi, M., D. Bertelli, and A. Monzani. 2001. Mercury and selenium content in selected seafood.

Journal of Food Composition and Analysis, 14(5), 461-467.

Post, D. M. 2002. Using stable isotopes to estimate trophic position: Models, methods, and

assumptions. Ecology 83:703-718. 51

Power, M., G. M. Klein, K. R. R. A. Guiguer, and M. K. H. Kwan. 2002. Mercury accumulation

in the fish community of a sub-Arctic lake in relation to trophic position and carbon

sources. Journal of Applied Ecology 39:819-830.

Prestbo, E. M., and D. A. Gay. 2009. Wet deposition of mercury in the US and Canada, 1996–

2005: Results and analysis of the NADP mercury deposition network (MDN). Atmospheric

Environment, 43(27), 4223-4233.

Radke, L. F., H. R. Friedli, and B. G. Heikes. 2007. Atmospheric mercury over the NE Pacific

during spring 2002: Gradients, residence time, upper troposphere lower stratosphere loss,

and long-range transport. Journal of Geophysical Research, 112(D19), D19305.

Rasmussen, J. B., D. J. Rowan, D. R. S. Lean, and J. H. Carey. 1990. Food chain structure in

Ontario lakes determines PCB levels in lake trout (Salvelinus namaycush) and other pelagic

fish. Canadian Journal of Fisheries and Aquatic Sciences, 47(10), 2030-2038.

Riget, F., P. Moller, R. Dietz, T. G. Nielsen, G. Asmund, J. Strand, M. M. Larsen, and K. A.

Hobson. 2007. Transfer of mercury in the marine food web of West Greenland. Journal of

Environmental Monitoring 9:877-883.

Rolfhus, K. R., B. D. Hall, B. A. Monson, M. J. Paterson, and J. D. Jeremiason. 2011. Assessment

of mercury bioaccumulation within the pelagic food web of lakes in the western Great

Lakes region. Ecotoxicology, 20(7), 1520-1529.

Rumbold, D. G., D. W. Evans, S. Niemczyk, L. E. Fink, K. A. Laine, N. Howard, D. Krabbenhoft,

and M. Zucker. 2011. Source identification of Florida Bay's methylmercury problem:

Mainland runoff versus atmospheric deposition and in situ production. Estuaries and

Coasts, 34(3), 494-513. 52

Rumbold, D. G., T. R. Lange, D. M. Axelrad, and T. D. Atkeson. 2008. Ecological risk of

methylmercury in Everglades National Park, Florida, USA. Ecotoxicology, 17(7), 632-641.

Rumbold, D., R. Wasno, N. Hammerschlag, and A. Volety. 2014. Mercury Accumulation in

Sharks From the Coastal Waters of Southwest Florida. Archives of Environmental

Contamination and Toxicology, 67(3), 402-412.

Sazima, C., P. R. Guimarães, S. F. Dos Reis, and I. Sazima. 2010. What makes a species central

in a cleaning mutualism network?. Oikos, 119(8), 1319-1325.

Scharf, F. S., F. Juanes, and R. A. Rountree. 2000. Predator size - prey size relationships of marine

fish predators: interspecific variation and effects of ontogeny and body size on trophic-

niche breadth. Marine Ecology Progess Series 208:220-248.

Scheuhammer, A. M., M. W. Meyer, M. B. Sandheinrich, and M. W. Murray. 2007. Effects of

environmental methylmercury on the health of wild birds, mammals, and fish. AMBIO: A

Journal of the Human Environment 36:12-18.

Simoneau, M., M. Lucotte, S. Garceau, and D. Laliberté. 2005. Fish growth rates modulate

mercury concentrations in walleye (Sander vitreus) from eastern Canadian lakes.

Environmental Research 98:73-82.

Slemr, F. and E. Langer. 1992. Increase in global atmospheric concentrations of mercury inferred

from measurements over the Atlantic Ocean. Nature 355:434-437.

Snodgrass, J. W., C. H. Jagoe, A. L. J. Bryan, H. A. Brant, and J. Burger. 2000. Effects of trophic

status and wetland morphology, hydroperiod, and water chemistry on mercury

concentrations in fish. Canadian Journal of Fisheries and Aquatic Sciences 57:171-180.

Sorokin, Y. I. 1995. Coral reef ecology. Springer, Berlin. 53

Springer, V. G., and A. J. McErlean. 1961. Tagging of great barracuda, Sphyraena barracuda

(Walbaum). Transactions of the American Fisheries Society, 90(4), 497-500.

Stemberger, R. S., and C. Y. Chen. 1998. "Fish tissue metals and zooplankton assemblages of

northeastern US lakes." Canadian Journal of Fisheries and Aquatic Sciences, 55, (2): 339-

352.

Strom, R. N., R. S. Braman, W. C. Jaap, P. Dolan, K. B. Donnelly, and D. F. Martin. 1992. Analysis

of selected trace metals and pesticides offshore of the Florida Keys. Florida scientist, 55,

1-13.

Strom, D. G. and G. A. Graves. 2001. A comparison of mercury in estuarine fish between Florida

Bay and the Indian River Lagoon, Florida, USA. Estuaries, 24:597-609.

Sveinsdottir, A. Y. and R. P. Mason. 2005. Factors controlling mercury and methylmercury

concentrations in largemouth bass (Micropterus salmoides) and other fish from Maryland

reservoirs. Archives of Environmental Contamination and Toxicology. 49:528-545.

Swanson, H. K., and K. A. Kidd. 2010. Mercury concentrations in Arctic food fishes reflect the

presence of anadromous Arctic charr (Salvelinus alpinus), species, and life

history. Environmental Science and Technology, 44(9), 3286-3292.

Thera, J. C., and D. G. Rumbold. 2014. Biomagnification of mercury through a subtropical coastal

food web off Southwest Florida. Environmental Toxicology and Chemistry, 33(1), 65-73.

Tremain, D. M., and K. E. O'Donnell. 2014. Total mercury levels in invasive lionfish, Pterois

volitans and Pterois miles (Scorpaenidae), from Florida waters. Bulletin of Marine Science,

90(2), 565-578.

Trudel, M. and J. B. Rasmussen. 1997. Modeling the elimination of mercury by fish.

Environmental Science and Technology 31:1716-1722. 54

U.S. EPA (Environmental Protection Agency). 1997. Mercury study report to Congress. EPA-

452/R-97-003-010. U.S. Environmental Protection Agency, Washington, D.C., USA.

U.S. EPA (Environmental Protection Agency). 2001. Water quality criterion for the protection of

human health: Methylmercury. EPA-823-R-01-001. Washington, DC.

Vander Zanden, M. J., and J. B. Rasmussen. 1996. A trophic position model of pelagic food webs:

impact on contaminant bioaccumulation in lake trout. Ecological Monographs, 451-477.

Voegborlo, R. B., and H. Akagi. 2007. Determination of mercury in fish by cold vapour atomic

absorption spectrometry using an automatic mercury analyzer. Food Chemistry, 100(2),

853-858.

Wang, W.-X. 2002. Interactions of trace metals and different marine food chains. Marine Ecology

Progress Series 243:295-309.

Watras, C. J. W., R. C. Back, S. Halvorsena, R. J. M. Hudson, K. A. Morrison, and S. P. Wente.

1998. Bioaccumulation of mercury in pelagic freshwater food webs. Science of the Total

Environment 219:183-208.

Webber, H. M. and T. A. Haines. 2003. Mercury effects on predator avoidance behavior of a forage

fish, golden shiner (Notemigonus Crysoleucas). Environmental Toxicology and Chemistry

22:1556-1561.

Wiener, J. G., B. C. Knights, M. B. Sandheinrich, J. D. Jeremiason, M. E. Brigham, D. R.

Engstrom, L. G. Woodruff, W. F. Cannon, and S. J. Balogh. 2006. Mercury in soils, lakes,

and fish in Voyageurs National Park (Minnesota): importance of atmospheric deposition

and ecosystem factors. Environmental Science and Technology,40(20), 6261-6268. 55

Wilson, S. K., D. T. Wilson, C. Lamont, and M. Evans. 2006. Identifying individual great

barracuda Sphyraena barracuda using natural body marks. Journal of Fish Biology, 69(3),

928-932.

Wozniak, J. R., W. T. Anderson, D. L. Childers, E. E. Gaiser, C. J. Madden, and D. T. Rudnick.

2012. Potential N processing by southern Everglades freshwater marshes: Are Everglades

marshes passive conduits for nitrogen?. Estuarine, Coastal and Shelf Science, 96, 60-68.

Zhu, A., W. Zhang, Z. Xu, L. Huang, and W. X. Wang. 2013. Methylmercury in fish from the

South China Sea: Geographical distribution and biomagnification. Marine Pollution

Bulletin, 77(1), 437-444.

56

Table 1. Summary of mercury concentration, total length, stable isotopes of δ13C and δ15N, and trophic level of invertebrate and finfish taxa collected from Tennessee Reef Light (TRL). 57

Table 2. Summary of mercury concentration, total length, stable isotopes of δ13C and δ15N, and trophic level of invertebrate and finfish taxa collected from Long Key Hard Bottom (LKH). 58

Table 3. Coefficient of determination (r2) values for regression models for various relationships involving fish species (where n≥5) at Tennessee Reef. Bold values are significant at p≤0.05.

Table 4. Coefficient of determination (r2) values for regression models for various relationships involving fish species (where n≥5) at Long Key Hard Bottom. Bold values are significant at p≤0.05. 59

Table 5. Values for trophic magnification slopes of Total Hg biomagnification through food webs of different ecosystems as reported in the published literature. 60

Table 6. Water quality conditions (mean ±1SD) based on quarterly sampling (n = 14) from May 2010 – September 2013 at the two nearby sites (adapted from the FIU SERC Florida Keys National Marine Sanctuary).

Site Salinity Total Phosphorus (TP, Soluble Reactive Chlorophyll a Total Light Temp

(PSU) Nitrogen mg/L), Phosphorus (CHLA, μg/L). Organic atten. (Kd, (°C)

(TN, mg/L) (SRP, ug/L) Carbon m‐1)*

(TOC,

mg/L)*

Long Key 36.2 ±0.75 0.217 ±0.11 0.0055 ±0.002 1.4 ±0.75 0.348 ±0.16 2.2 ±1.3 0.48 ±0.82 26.78

Channel ±4.14

Tennessee Reef 36.1 ±0.18 0.165 ±0.09 0.0049 ±0.003 1.0 ±0.64 0.334 ±0.11 1.5 ±1.1 0.14 ±0.14 26.72

±2.9

* Median value of parameter differed between sites based on Mann Whitney test (p<0.05) 61

Figure 1. Map of study area where samples were collected off Long Key, FL. 62

Figure 2. (a) Relationship between log Hg (µg/kg) and total length (cm) in all individuals at

Tennessee Reef, Species-specific regression lines are shown where they were statistically significant; (b) log Hg (µg/kg) and total length (cm) for all individuals at Long Key Hard

Bottom. 63

Figure 3. Relationship between stable isotope values of δ13C and δ15N (‰, individual data points) for Tennessee Reef (top panel) and Long Key Hard Bottom (bottom panel). 64

Figure 4. Relationship between (a) δ15N (‰) and total length (cm) with linear regression line in all individuals at Tennessee Reef; (b) δ15N (‰) and total length (cm) with linear regression line for all individuals at Long Key Hard Bottom; (c) δ15N (‰) and total length (cm) with linear regression line for species means at Tennessee Reef; and (d) log

δ15N (‰) and log total length (cm) for species means at Long Key Hard Bottom.

65

Figure 5. Relationship between log Hg concentration (µg/kg, wet weight) and δ15N (‰) from a subtropical coral reef food web for individuals with linear regression lines for (a)

Tennessee Reef and (b) Long Key Hard Bottom.

66

Figure 6.Relationship between log Hg concentration (µg/kg, wet weight) and δ15N (‰) from a subtropical coral reef food web based on species means (mean ± SD) with linear regression line for TRL (a) and LKH (b). 67

Figure 7. Relationship between log Hg concentration (µg/kg, wet weight) and trophic level (calculated using δ15N, see text), from a subtropical coral reef food web for all individuals caught, with linear regression line for (a) TRL and (b) LKH. 68

Figure 8. Relationship between log Hg concentration (µg/kg, wet weight) and trophic level

(calculated using δ15N, see text) from a subtropical coral reef food web for species means

(mean ± SD) with linear regression line for TRL (a) and LKH (b).