Mercury, Food Webs, and Marine Mammals: Implications of Diet and Climate Change for Human Health

We modeled the flow of methyl mercury, a toxic global pollutant, in the Faroe Islands marine ecosystem and compared average human methyl mercury exposure from consumption of pilot whale meat and fish (cod, Gadus morhua) with current tolerable weekly intake (TWI) levels. Under present conditions and climate change scenarios, methyl mercury increased in the ecosystem, translating into increased human exposure over time. However, we saw greater changes as a result of changing fishing mortalities. A large portion of the general human population exceed the TWI levels set by the World Health Organization [WHO; 1.6 μg/kg body weight (bw)], and they all exceed the reference dose (RfD) of 0.1 μg/kg bw/day set by the U.S. Environmental Protection Agency (EPA; equivalent to a TWI of 0.7 μg/kg bw). As a result of an independent study documenting that Faroese children exposed prenatally to methyl mercury had reduced cognitive abilities, pregnant women have decreased their intake of whale meat and were below the TWI levels set by the WHO and the U.S. EPA. Cod had approximately 95% lower methyl mercury concentrations than did pilot whale. Thus, the high and harmful levels of methyl mercury in the diet of Faroe Islanders are driven by whale meat consumption, and the increasing impact of climate change is likely to exacerbate this situation. Significantly, base inflow rates of mercury into the environment would need to be reduced by approximately 50% to ensure levels of intake below the WHO TWI levels, given current levels of whale consumption.

Although occurring naturally [United Nations Environment Programme (UNEP) 2002], mercury is a global pollutant and concerns public health when it is elevated above natural background levels, mainly through anthropogenic causes (Boening 2000). The cycling of mercury through the marine environment involves different chemical forms (Morel et al. 1998). In marine organisms, it is most commonly found as monomethyl mercury (CH 3 Hg + ) or as mercury ion (Hg 2+ ; Downs et al. 1998;Morel et al. 1998). Generally, it is monomethyl mercury that is of concern because it bioaccumulates and biomagnifies at all trophic levels in the food web and can have severe toxicologic effects. Methyl mercury first gained notoriety in Minimata, Japan, after causing severe disabilities and death among people eating seafood contaminated through industrial mercury discharge accumulating through the food chain (Fujuki 1980). Mercury concentrates in the marine environment, especially in deep ocean waters, which contain approximately 74% of the global total, compared with approximately 24 and 2% in the shallow part of the oceans and the atmosphere, respectively (Mason and Sheu 2002;Morel et al. 1998). A large portion of mercury in the ocean is transformed to Hg 2+ and becomes available for methylation (Fitzgerald and Mason 1997). Thus, methyl mercury concentrations are primarily a function of methylation and demethylation rates (Morel et al. 1998) and of sedimentation and food chain uptake (Fitzgerald and Mason 1997).
Methylation seems driven by biotic processes (UNEP 2002) and has been linked to sediment-bound sulfate-reducing bacteria (King et al. 2001). However, methylation is also thought to occur throughout the water column (Morel et al. 1998). Significant in light of global climate change, methylation rates are temperature dependent (Downs et al. 1998). Concentrations of mercury measured in the North Atlantic Ocean averaged approximately 1 pM (Mason et al. 1998;Mason R, personal communication), and usually 80-99% of mercury found in fish muscle tissue is methyl mercury, regardless of its concentration in the environment (Downs et al. 1998).
The population of the Faroe Islands (northeast Atlantic, 62°N, 7°W) relies heavily on marine resources, both for consumption and as a key economic activity. Fisheries account for more than 95% of exports and 44.5% of gross domestic product, with demersal species (e.g., Atlantic cod, Gadus morhua) the most important grouping (Zeller and Reinert 2004). There is a long tradition of hunting pilot whales (Globicephala melas), with records back to 1709 (Faroe Government 2004). Today, whale meat accounts for approximately 30% of total meat produced on the islands (Faroe Government 2004) and is a cultural component of the Faroe lifestyle. It is made available using a free, traditional distribution system (Bloch and Zachariassen 1989).
In the 1990s, Grandjean et al. (1992Grandjean et al. ( , 1997 documented cognitive impairment in a cohort of Faroese children who were exposed to elevated levels of methyl mercury prenatally, based on consumption of whale meat during pregnancy. Subsequent studies also provided evidence of attenuated postnatal growth of breast-fed children due to contaminant loading of human milk via maternal seafood diet (Grandjean et al. 2003). The average daily seafood consumption by adults was reported as 12 g whale muscle and 72 g fish (Vestergaard and Zachariassen 1987), with the majority of fish consumed being cod (Grandjean et al. 1992). In response to Grandjean et al. (1992), the diet of pregnant women has since changed to an average daily consumption of 1.45 g whale muscle and 40.2 g fish (Weihe et al. 2003), with an associated decrease in mercury assimilation (Weihe et al. 2005).
We modeled the transfer of methyl mercury through the food web in the marine ecosystem of the Faroe Islands using Ecotracer, a novel routine of the trophic ecosystem modeling approach Ecopath with Ecosim . A published Ecopath model for the Faroe Islands marine ecosystem (Zeller and Reinert 2004) was slightly modified to 21 functional groups plus detritus, with a functional group consisting of either a single species (e.g., cod, Gadus morhua) or a group of species (e.g., plankton). We initiated model simulations with a 100year baseline run (t 0 -t 100 ) using only estimated mercury base inflow rate changes and environmental concentrations based on Mason and Sheu (2002). This baseline simulated a bottom-up approach to reach species/functional group methyl mercury concentrations comparable with field observations. We then performed impact simulations (t 100 -t 200 ) to evaluate changes in fishing mortality rates and Environmental Health Perspectives • VOLUME 113 | NUMBER 5 | May 2005 521 effects of increased sea temperatures due to climate change on methyl mercury bioaccumulation in all species/groups. We evaluated results in the context of human dietary consumption compared with standardized tolerable weekly intake (TWI) limits, based on the World Health Organization (WHO) equivalent [1.6 µg/kg body weight (bw); Food and Agriculture Organization of the United Nations (FAO)/WHO 2003] and the U.S. Environmental Protection Agency (EPA) equivalent [0.7 µg/kg bw converted from the reference dose (RfD) of 0.1 µg/kg bw/day; Electric Power Research Institute (EPRI) 2004]. Furthermore, we derived a functional relationship to predict methyl mercury concentrations in fish species based on growth and life history parameters such as trophic level (TL), consumption to production ratio (Q/P), and the von Bertalanffy growth coefficient (K).

Materials and Methods
Modeling approach. Using Ecotracer , we traced the transfer and bioaccumulation of methyl mercury through all ecosystem components (functional groups, composed of either individual species or species groups) based on diet transfers and direct uptake from the environment. Underlying Ecotracer was a trophic marine ecosystem model modified from Zeller and Reinert (2004;  Within Ecotracer, the concentration of a contaminant in a given species or group of species is expressed as a function of gains from direct uptake from the environment and from the uptake from each food item as defined in the ecosystem model's diet matrix, versus losses due to instantaneous decay rates, unassimilated food, predation, and instantaneous nonpredation death rates . Four contaminant parameters must be provided: a) initial concentrations for each species/group, including environmental concentrations; b) direct uptake rate parameters for each species/group; c) concentrations per biomass in immigrating organisms; and d) metabolism/decay rates for each species/group. Unassimilated food and instantaneous nonpredation death rates are estimated by the modeling routine.
Ecotracer input data. Base inflow rate. The Ecotracer input data derived as described below are presented in Table 1. The base inflow rate is the sum of the methylation rate occurring in the sediments, the demethylation rate by chemical transformation at the sediment-water interface, and the net methylation rate occurring in the water column. Approximately 2% of the total mercury flux is being methylated per year (Fitzgerald and Mason 1997). The present day wet depositional flux is estimated to be approximately 7.7 × 10 -6 g/m 2 for latitudes 30°to 70°N, and approximately 0.68 × 10 -6 g/m 2 for latitudes 70°to 90°N (Downs et al. 1998). Given that the Faroe Islands are located at 62°N, we assumed that the base inflow rate would be intermediate to these two values. A base inflow rate of 0.113 g/km 2 /year, equivalent to a wet depositional flux of 5.6 × 10 -6 g/m 2 , best accounted for empirically measured environmental concentrations and compared favorably with our assumption of an intermediate value. Dry depositions were not considered because they are not deemed to be important in oceans (Mason et al. 1994).
Starting concentrations. The depthaveraged environmental concentration in the entire global ocean increased by approximately 9% between preindustrial periods and the modern, industrial era (Mason and Sheu 2002), whereas concentrations in near-surface waters are thought to have increased 2-to 3-fold (Mason et al. 1994). The modern environmental concentration by volume was based on current measurements of total mercury for the North Atlantic (~1 pM; Mason et al. 1998;Mason R, personal communication), resulting in a methyl mercury volume concentration of 0.02 pM (i.e., 2% of total mercury flux being methylated; Fitzgerald and Mason 1997). Using an average water depth of 838 m [National Oceanic and Atmospheric Administration (NOAA) 2001] and a surface area of approximately 190,200 km 2 for the model area (Zeller and Reinert 2004), we converted the environmental volume concentration into an environmental area concentration of 3.612 g/km 2 for the modern, industrialized period. However, to account for increasing baseline flow rates due to industrialization, we set the initial, preindustrial environmental concentration (t 0 ) of methyl mercury to 3.312 g/km 2 (i.e., a value 9% less than the current measurements, representing the 9% increase between preindustrialized and industrialized periods; Mason and Sheu 2002). Initial species/group concentrations for biota at time t 0 were set at estimated preindustrial period values (Table 1) determined through a prebaseline simulation. In this simulation, the environmental and biota concentrations were initiated at zero and run to the preindustrialized period environmental concentration of 3.312 g/km 2 , while allowing the biological groups to equilibrate, resulting in preindustrialized period estimates for biota.
Direct uptake and demethylation rates. We applied direct uptake rates to phytoplankton, zooplankton, and benthos (Table 1) because this is the dominant entry pathway for mercury accumulation (Canli and Furness 1995;Mason et al. 1996). For higher species and groups, diet accounts for approximately 90% of mercury accumulation (Downs et al. 1998). Therefore, we ignored uptake rates due to respiration.
We used demethylation (decay) rates for marine mammals (Table 1) because they are known to demethylate methyl mercury by forming a mercury selenide complex (Wagemann et al. 1998 marine mammals. However, preliminary simulations without demethylation indicated that the methyl mercury concentrations in marine mammals increased sharply without reaching a limit. Furthermore, although other species may also have demethylation capabilities, no information in this regard is available. Baseline simulations. We followed the bottom-up flow of methyl mercury through ecosystem components (species/groups) with a 100-year baseline simulation (t 0 -t 100 ) and compared end concentrations with published field measurements. Transformation of literature values of mercury to methyl mercury were based on Dietz et al. (1996), Andersen and Depledge (1997), and Joiris et al. (1997b).
Impact simulations. Building on the t 100 baseline, we simulated the effects of changes in fishing mortality rate (F ) on the accumulation of methyl mercury in commercial fish species and pilot whales over a second 100-year period (t 100 -t 200 ). Changing F alters the production to biomass (P/B) ratio of a species/group (and hence mercury accumulation), because P/B can be defined as the total mortality rate, which is fishing mortality plus natural mortality .
We also investigated how methyl mercury concentrations would change with increased seawater temperatures based on global climatechange scenarios. Increases in temperatures would lead to increases in the methylation rate via the Q 10 rule, making increased amounts of methyl mercury available for uptake (Downs et al. 1998). The Q 10 rule relates changes in temperature to changes in the metabolic rates of organisms, whereby a 10°C temperature increase leads to a doubling of the metabolic rate (Randall et al. 2002). Studies on the North Atlantic generally project warming of 0.4-1.0°C per century (International Panel on Climate Change 2001).
Human dietary intake. We modeled the dietary intake (DI) of methyl mercury as where R iw is the daily intake (in grams) of seafood in weight w of species i, and C il is the concentration of methyl mercury (micrograms per gram) in length class l of species i. Length class considerations are deemed important because of age-specific and growthrate-specific increases in contaminant loads (Downs et al. 1998;Joiris et al. 1995). Thus, faster-growing species (higher growth coefficient K) have lower concentrations of methyl mercury than do slower-growing species (lower K), given the same size and environmental conditions.
Predicting methyl mercury concentrations in fish. Methyl mercury concentrations in species are known to increase with TL. However, species at similar TLs may have different life histories and growth patterns, influencing methyl mercury concentrations. We derived a predictive relationship for fish species through a multiple regression analysis using TL, Q/P (both derived from the model), and K (Froese and Pauly 2004).

Results
Baseline simulation. Between t 0 and t 100 , methyl mercury concentrations in all groups increased at a declining rate, whereas concentrations in the environment increased at an average rate of 0.004 g/km 2 per year. Predicted concentrations at t 100 for most species/groups fell within the ranges reported in the literature (Table 2). However, six species/groups (four pooled groups and two individual species, herring and Greenland halibut) differed substantially (± 50% or more) between model output and literature averages. By the very nature of pooling several species into a group, groups often appear to be inadequately represented in the underlying ecosystem input data, whereas herring is poorly represented in the literature and Greenland halibut may represent a unique case.
Conversely, a 20% increase in F on all targeted fish groups and pilot whales had the opposite effect, resulting in decreases of 4.5-27% in biomass for pilot whales, cod, saithe, Greenland halibut, blue whiting, and other deepwater fishes ( Figure 1D). Small increases in biomass were recorded for redfish, other demersal fishes, and squid ( Figure 1E), whereas all other species/groups had an average increase in biomass of 0.9% (range, -1.0% to 2.9%). Increasing F by 20% decreased the methyl mercury concentration in other toothed cetaceans (-5.9%), pilot whales (-12.4%), baleen whales (-4.7%), saithe (-3.8%), other deepwater fishes (-5.4%), and Article | Mercury, food webs, climate change, and human health Environmental Health Perspectives • VOLUME 113 | NUMBER 5 | May 2005 523 Table 2. Predicted concentrations of methyl mercury (g/metric ton) in functional groups after the 100-year simulation (t 0 -t 100 ) compared with field measurements taken from the literature. mackerel (-4.3%), relative to the status quo scenario at t 200 ( Figure 1F). The remaining groups had an average decrease in methyl mercury concentration of -2.4% (range, -3.2% to -0.2%). Changing F by other percentages gave results that were qualitatively the same but differed quantitatively. Climate change scenario. Increases in water temperature resulted in average increases in methyl mercury concentrations of 1.7% (range, 1.6-1.8%) and 4.4% (range, 4.1-4.7%) by t 200 for projected ocean warming rates of 0.4°C and 1.0°C, respectively, per century. Simulations with the combined effects of climate change and changes in fishing mortality indicated that the two effects are cumulative.

Literature mean
Individual species/groups responded to these changes differently, with pilot whales (Figure 2A) displaying a much greater nominal change in methyl mercury concentrations in response to changes in fishing mortality rates and climate, compared with cod ( Figure 2B). However, compared with cod, the trajectory for pilot whales was dampened by the ability of pilot whales to demethylate methyl mercury.
Human dietary intake of methyl mercury. We deemed the human dietary intake of 12 g/person/day of pilot whale meat, as observed in the 1980s (Vestergaard and Zachariassen 1987), to be the representative intake for the general population, because availability based on supply was 14.8 g/person/day in 2000 (Hagstova føroya 2001). Although consumption may have declined in recent years, no study has documented a change in dietary intake for the general population, other than for pregnant women (Weihe et al. 2003). Based on the average dietary intake of whale meat (12 g/person/day) and cod (72 g/person/day), a large portion of the general adult population exceeded the WHO limit under all simulated conditions (Figure 3). At present, individuals with a body weight of < 102 kg are above the TWI level set by the WHO for methyl mercury (FAO/WHO 2003), based on seafood consumption alone. The calculated weekly intake for the general adult population also exceeded the U.S. EPA's limit (EPRI 2004) irrespective of body weight (Figure 3). Ocean temperature changes will increase the number of people above the WHO TWI level to individuals weighing < 105 kg and 107 kg for temperature increases of 0.4°C and 1.0°C, respectively, per century. Simulating a 20% decrease in fishing mortality resulted in members of the general population weighing < 117 kg being above the TWI set by WHO. Significantly, base environmental inflow rates of mercury would need to be reduced by approximately 50% (ignoring climate change scenarios) to ensure that levels of methyl mercury intake fall below the WHO TWI levels for current consumption patterns by the general population ( Figure 3).
Interestingly, pregnant women (if consuming 1.45 g/person/day of whale meat plus 40.2 g/person/day of cod; Weihe et al. 2003) are currently, and under all simulation scenarios, well below the TWI limits set by the WHO. They are also under the limit set by the U.S. EPA, except for the very lightest individuals in cases where F is decreased substantially (Figure 3).

Predictor of methyl mercury in fish.
We derived a multiple regression as where Y i is the concentration of methyl mercury in grams per metric ton in fish species i; TL i is the trophic level of fish species i; Q/P i is the consumption to production ratio of fish species i; and K i is the von Bertalanffy growth parameter for fish species i. Although limited in its scope (df = 6), this relationship has the potential to be a good predictor of methyl mercury concentrations for fish species (p < 0.05; r 2 = 0.905) because it incorporates relevant life history and trophic indicators, such as TL, Q/P, and K.

Discussion
Methyl mercury poses substantial health risks (Fujuki 1980;UNEP 2002), and concerns of non-point-source mercury pollution was highlighted by Grandjean et al. (1992Grandjean et al. ( , 1997, who documented cognitive impairments in young children exposed to elevated levels of methyl mercury prenatally. This exposure was linked to the consumption of whale meat by pregnant women. Fortunately, as a result of the Grandjean et al. (1997) study, the average consumption of whale meat by pregnant women declined by approximately 90% (Weihe et al. 2003), resulting in lower levels of exposure to Article | Booth and Zeller 524 VOLUME 113 | NUMBER 5 | May 2005 • Environmental Health Perspectives   (Weihe et al. 2005). However, our study confirmed that by considering the average seafood diet composition as defined for the general population (Vestergaard and Zachariassen 1987), methyl mercury intakes by the nonpregnant section of the community are likely at or above the TWI levels recommend by the WHO, and substantially above the safe levels recommended by the U.S. EPA. Significantly, the analysis presented here relates to average dietary intakes and therefore does not take into account that a substantial number of people eat more than the average intake reported (i.e., are at the higher end of the intake distribution). They face substantially higher risks, especially persons with lower body weight. Hence, there is also the potential that many pregnant women exceed the U.S. EPA level, and perhaps also the WHO level. Given the present level of consumption by the general population, mercury loading of the environment would need to be reduced by approximately 50% for most of the general adult population to fall below the WHO TWI levels.
In general, simulating decreases in F led to an increasing trend in methyl mercury concentrations, whereas increasing F had the opposite effect. Although our simulations suggested that increasing fishing mortality rates would lower the concentration of methyl mercury in species/groups, it would not be sufficient to decrease methyl mercury concentrations in whale meat substantially. The dominance of methyl mercury exposure through whale meat consumption will remain a problem, irrespective of potential changes to fishing pressures. Therefore, Faroe Islanders should seriously consider reducing whale meat consumption to levels comparable with those of pregnant women (i.e., < 2 g/person/day).
Of additional concern are the likely effects of climate change, resulting in even higher concentrations of contaminants in the marine food supply of Faroe Islanders. The increasing methylation rate due to higher water temperatures will lead to continuous increases in concentrations of methyl mercury. This implies that Faroe Islanders may experience everincreasing exposure levels, unless their dietary habits change to species with lower methyl mercury concentrations.
We have demonstrated that Ecotracer is a capable tool for tracing contaminants through all functional groups of an ecosystem, requiring relatively few toxicologic input parameters to follow the food web flow of a contaminant through all levels of an ecosystem. Significantly, our approach also lends itself to the investigation of other contaminants with significant impacts on global human health, such as dioxins and polychlorinated biphenyls (PCBs).
Although six model groups did not closely predict empirical methyl mercury concentrations, four of these entities were pooled groups (other toothed cetaceans, other deepwater fishes, benthos, and zooplankton). Pooled groups are known to be often poorly represented in the underlying input data and may be missing data for key group components. For example, other toothed cetaceans were missing methyl mercury values for killer whales (Orcinus orca), a species expected to have high levels of methyl mercury because of its position at the top of the food chain. In contrast, model concentrations for herring were below the value found in the literature. This may be due to the empirical data being from near-shore U.K. waters (Dixon and Jones 1994;Rowe et al. 1998) and thus not reflecting the more pelagic environment of the Faroe Islands. Greenland halibut, on the other hand, may represent a special case because the percentage of methyl mercury present in muscle tissue was reported to be between 1 and 53% (Joiris et al. 1997a), which is well below the usual 80-99% reported for most fish species (Downs et al. 1998).
Our simulations have shown that changes in methyl mercury concentrations in ecosystem components not only are due to changes in mercury input (i.e., bottom-up control) but also are influenced by top-down factors (e.g., predation and/or fishing). We also demonstrated that changes in fishing mortality can substantially alter the flow of methyl mercury in an ecosystem, by affecting the P/B ratios and the resulting trophic relations of species/groups. This might explain why tuna caught off Hawaii did not show any significant changes in methyl mercury concentrations between 1971 and 1998, despite increasing environmental loading of methyl mercury (Kraepiel et al. 2003) and growing fishing pressures (Cox et al. 2002).
Methyl mercury affects human health as a result of direct discharges and atmospheric transport. This pollutant is of particular concern to indigenous peoples of the Arctic, who often rely heavily on marine resources, and especially marine mammals, for part of their traditional diets. For example, in Greenland, approximately 43% of blood samples taken from indigenous women of reproductive age had blood mercury levels exceeding guidelines (UNEP 2003). However, increasingly pollutants found in marine resources, such as mercury, PCBs, and dioxins, are of growing concern also to westernized societies, given the growing demand for and consumption of seafood. This is illustrated by the advisory regarding seafood consumption by pregnant women issued by the U.S. Food and Drug Administration in March 2004 (U.S. FDA 2004).
Methyl mercury will continue to be of global concern as long as there are ongoing anthropogenic inputs of mercury. Our ecosystem-scale simulations suggest that substantial reductions in mercury inputs (~50%) would be required to ensure safe exposure levels if people such as the Faroe Islanders wish to continue their cultural dietary traditions. Unfortunately, Article | Mercury, food webs, climate change, and human health Environmental Health Perspectives • VOLUME 113 | NUMBER 5 | May 2005 525 Figure 3. Results of the dietary analysis for the general population and pregnant women compared with the TWI limits set by the WHO (1.6 µg/kg bw) and the U.S. EPA equivalent (RfD of 0.1 µg/kg bw/day expressed as a TWI of 0.7 µg/kg bw). Under all simulations, a substantial portion of the general adult population had dietary intakes greater than the WHO limit, and all members' consumption was greater than the limit set by the U.S. EPA. The TWI level under a simulated 50% reduction in the base inflow rate (BIR) of mercury into the environment is also shown. Pregnant women were well below the limits set by the WHO and the U.S. EPA under all scenarios. Note that the present-day scenarios for both groups at t 100 are not presented because they were only marginally different than the status quo at t 200 .  Figure 2. Effects of climate change and changing fishing mortality (F) on the methyl mercury concentrations in pilot whales (Globicephala melas; A) and cod (Gadus morhua; B). Note the significant difference in scale of the y-axis. Other species showed similar general trends. Increasing F decreased the load relative to the status quo scenario, whereas decreasing F had the opposite effect.

Methyl mercury concentration (g/t)
1.0° C, F -20% 0.4 °C, F -20% F-20% 1.0° C, F +20% 0.4 °C, F +20% the United States in 2002 increased the disposal or release of mercury by 10% more than the previous year (U.S. EPA 2002), whereas China's emissions (~500 metric tons/year), driven primarily by coal combustion, rose by approximately 50 metric tons/year during the early 1990s (Pacyna and Pacyna 2002) and have been tracked across the Pacific Ocean to North America . Thus, anthropogenic pollution with mercury is a global problem that will continue to affect future generations in all regions of the world.