Environ. Sci. Technol. 2007, 41, 4952-4958
Experimental Evidence of a Linear Relationship between Inorganic Mercury Loading and Methylmercury Accumulation by Aquatic Biota D I A N E M . O R I H E L , * ,†,‡ MICHAEL J. PATERSON,‡ PAUL J. BLANCHFIELD,‡ R . A ( D R E W ) B O D A L Y , ‡,| A N D HOLGER HINTELMANN§ Clayton H. Riddell Faculty of Environment, Earth, and Resources, University of Manitoba, Winnipeg, Manitoba, Canada R3T 2N2, Freshwater Institute, Fisheries and Oceans Canada, 501 University Crescent, Winnipeg, Manitoba, Canada R3T 2N6, and Department of Chemistry, Trent University, 1600 West Bank Drive, Peterborough, Ontario, Canada K9J 7B8
Developing effective regulations on mercury (Hg) emissions requires a better understanding of how atmospheric Hg deposition affects methylmercury (MeHg) levels in aquatic biota. This study tested the hypothesis that MeHg accumulation in aquatic food webs is related to atmospheric Hg deposition. We simulated a range of inorganic Hg deposition rates by adding isotopically enriched Hg(II) (90.9% 202Hg) to 10-m diameter mesocosms in a boreal lake. Concentrations of experimentally added (“spike”) Hg were monitored in zooplankton, benthic invertebrates, and fish. Some Hg(II) added to the mesocosms was methylated and incorporated into the food web within weeks, demonstrating that Hg(II) deposited directly to aquatic ecosystems can become quickly available to biota. Relationships between Hg(II) loading rates and spike MeHg concentrations in zooplankton, benthic invertebrates, and fish were linear and significant. Furthermore, spike MeHg concentrations in the food web were directly proportional to Hg(II) loading rates (i.e., a percent change in Hg(II) loading rate resulted in, statistically, the same percent change in MeHg concentration). This is the first experimental determination of the relationship between Hg(II) loading and MeHg bioaccumulation in aquatic biota. We conclude that changes in atmospheric Hg deposition caused by increases or decreases in Hg emissions will ultimately affect MeHg levels in aquatic food webs.
that Hg levels in the environment have increased on a global scale since the onset of industrialization (2). Currently, the largest anthropogenic source of Hg emissions to the atmosphere is the combustion of coal in power plants and in residential and commercial boilers (3). Mercury emitted from these sources can be dispersed long distances in the atmosphere and subsequently returned to the Earth’s surface via wet deposition of oxidized species [Hg(II)], although dry deposition may also be important in some areas (4). Natural processes in lakes and their watersheds convert atmospherically derived Hg(II) to the potent toxin methylmercury (MeHg) (5). Methylmercury is transferred through the diet and strongly biomagnifies in aquatic food webs, which results in high concentrations in fish and other top predators (5). Proposals to reduce anthropogenic Hg emissions to the atmosphere are intended to lower MeHg levels in fish, but it is unclear whether these policies will be effective in achieving their goal. The amount by which Hg emissions need to be reduced to achieve an acceptable level of MeHg in fish remains a complex, unresolved issue. Developing effective policies on Hg emissions requires models that can predict how changes in atmospheric Hg deposition affect levels of MeHg in fish. This relationship is difficult to quantify because many factors affect Hg cycling in aquatic ecosystems. Fish MeHg levels are correlated with several environmental factors, including watershed characteristics, lake morphometry, water chemistry, and food web structure (6-8). Because this suite of factors inevitably varies from one ecosystem to the next, lakes receiving similar rates of Hg deposition may have quite different fish MeHg levels (9). We tested the hypothesis that MeHg concentrations in aquatic biota are related to atmospheric Hg deposition by simulating different rates of Hg deposition in a boreal lake. We isolated sections of the lake in large mesocosms and added a different amount of isotopically enriched Hg(II) to each mesocosm. Adding Hg(II) enriched in a stable isotope allowed us to monitor the methylation and subsequent bioaccumulation of the experimentally added (“spike”) Hg. In previous papers on this experiment (known as the MESOSIM project), we concluded that major biogeochemical processes, including methylation, were Hg(II)-limited and responded linearly to increases in Hg(II) loading (10, 11). In this paper, we examine the bioaccumulation of spike MeHg in the food webs of the mesocosms. First, we describe how quickly spike MeHg was incorporated into the food web. Second, we analyze the dose-response relationship between rates of Hg(II) loading and concentrations of spike MeHg in zooplankton, benthic invertebrates, and fish. Specifically, we test the hypothesis that the bioaccumulation of spike MeHg in the food web was directly proportional to the rate of Hg(II) loading. By this, we mean that a percent change in Hg(II) loading rate resulted in the same percent change in the spike MeHg concentrations in biota.
Introduction
Experimental Section
During the last century, more than 200 million kilograms of mercury (Hg) were emitted to the atmosphere from anthropogenic activities (1). Lake sediments and peat bogs reveal
Study Design. Mesocosms were installed in a 2-m deep littoral zone of Lake 240 at the Experimental Lakes Area (ELA; Ontario, Canada). The construction and physicochemical characteristics of the mesocosms were described previously (10, 12). Briefly, mesocosms were 10 m in diameter, open to the atmosphere and sediments, and contained oligotrophic, circumneutral water and sandy sediments. As described in ref 10, isotopically enriched Hg(II) (90.9% 202Hg) was added to the mesocosms once a week for 8 weeks, beginning on June 26, 2002. Each mesocosm was randomly assigned a loading rate of 1×, 2×, 3×, 4×, 5×, 6×, 7×, 8×, 12×, or 15×
* Corresponding author phone: 204-984-8751; fax: 204-984-2404; e-mail:
[email protected]. † University of Manitoba. ‡ Fisheries and Oceans Canada. § Trent University. | Current affiliation: Penobscot River Mercury Study (Maine), 115 Oystercatcher Place, Salt Spring Island, British Columbia, Canada V8K 2W5. 4952
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 14, 2007
10.1021/es063061r CCC: $37.00
2007 American Chemical Society Published on Web 06/19/2007
the wet deposition rate of total Hg at the ELA in the 1999 water year (7.1 µg Hg m-2 year-1) (13). All references to “Hg(II) loading rate” refer to the rate of experimental Hg(II) addition. Zooplankton and Benthic Invertebrate Sampling. Between June 26 and September 4, 2002, zooplankton were collected every 2 weeks between 13:00 and 16:00 with a sweep net (150-µm mesh). Because some non-zooplankton material was inadvertently collected, the term “zooplankton” refers to all particles in the water column greater than 150 µm. Between September 6 and 11, 2002, benthic invertebrates were collected using a petit ponar dredge and light traps. Benthic invertebrates were sampled only in September to avoid disturbing the sediments during the experiment. Dredge samples were collected from six different locations in each mesocosm and then sieved as described in ref 12. Light traps (12) were set on the sediment surface after dusk and collected at dawn. In the laboratory, benthic invertebrates from dredges and light traps were sorted and combined into major taxonomic groups. Eight benthic invertebrate taxa were common in all mesocosms: Amphipoda (scud; generalist), Gomphidae (dragon fly larva; predator), Hexagenia (mayfly nymph; collector-gatherer), Hydracarina (water mite; predator), Oligochaeta (aquatic worm; detritivore), Planorbidae (ramhorn snail; herbivore/detritivore), Tanypodinae (midge fly larva; predator), and other Chironomidae (midge fly larva; herbivore/detritivore). Three taxa (Amphipoda, Gomphidae, and Hydracarina) were analyzed in all mesocosms. The other five taxa were only analyzed in mesocosms 2×, 5×, and 12×. Zooplankton and benthic invertebrate samples were stored in sealed plastic bags and frozen within 3-6 h of collection. Samples were freeze-dried, homogenized, weighed to the nearest 0.001 mg in aluminum weighing vessels, and transferred to Teflon vials for MeHg analysis. Fish Stocking and Sampling. On June 24-26, 2002, yellow perch (Perca flavescens) were captured from Lake 240 using a beach seine, marked with a caudal clip, and stocked in the mesocosms (40 fish per mesocosm). We only stocked fish with fork lengths between 50 and 60 mm, which is the typical size range for age 1 yellow perch in Lake 240 in early summer (14). Gut contents of yellow perch of a similar size range stocked in enclosures in an adjacent lake contained, by weight, 66% zooplankton and 31% benthic invertebrates (15). Mesocosms were covered with fine plastic netting (5-cm mesh) to deter predators. Yellow perch were collected after 5 weeks (July 31-August 2, 2002) and 10 weeks (September 4-16, 2002) using minnow traps and gill nets. A beach seine was also used in September to remove any remaining fish. At each sampling period, 5-10 fish were typically sampled from each mesocosm. Fish were killed by a lethal dose of clove oil, measured (fork length) to the nearest 1 mm, and weighed to the nearest 0.1 g. A skinless section of muscle tissue (ca. 0.2 g) was dissected from each fish, weighed to the nearest 0.0001 g, transferred to a 20 mL glass vial, and stored frozen until analysis. Although the mesocosms were fished with minnow traps and gill nets (6-10 mm mesh) before stocking, some youngof-the-year yellow perch were captured in some mesocosms in September. Saggital otoliths were examined (as described in ref 12) to verify the ages of all fish captured during this sampling period. Mercury Analyses. Concentrations of Hg species were determined by inductively coupled plasma mass spectrometry (ICP-MS) at Trent University (Peterborough, Ontario, Canada) (16). Methylmercury was measured in composite samples of zooplankton and benthic invertebrates, and total Hg was measured in muscle samples of individual fish. Total Hg was measured in fish samples because most of the Hg in fish muscle is MeHg (17) and this is the standard protocol
in contaminant monitoring programs (e.g., 9). We confirmed that the large majority of total Hg in muscle tissue of yellow perch was MeHg (mean ( SE: 84 ( 1%; n ) 13). Samples of zooplankton and benthic invertebrates were digested by adding 5 mL of 4 M HNO3 and heating at 55 °C for 24 h (18). Aliquots of sample digests (50-5000 µL) were added to 100 mL of Milli-Q water in gas-wash bottles for purging. After adding 0.2 mL of 2 M acetate buffer, solutions were neutralized with 20% KOH to obtain a pH of 4.9. Methylmercury was ethylated with tetraethyl borate (1%, w/v in 1% KOH, w/v), purged and concentrated on Tenax traps, thermodesorbed, and determined by GC-ICP-MS (Micromass Platform). Fish muscle samples were digested by adding 10 mL of HNO3/H2SO4 (7:3 v/v) and heating at 80 °C until brown NOx gases no longer formed. Total Hg of sample digests was reduced by SnCl2 and determined by ICP-MS (ThermoFinnigan Element2) using a continuous flow cold vapor generation technique. Method blanks and certified reference materials were measured for each batch of samples. Results for MeHg in oyster tissue (measured, 13.5 ( 1.7 ng g-1; certified, 13.2 ( 0.7 ng g-1) and total Hg in DORM-3 (measured, 4680 ( 240 ng g-1; certified, 4640 ( 260 ng g-1) were not statistically different from certified values. No isotope enrichment was detected in samples collected from the control mesocosm. The detection limit of ambient Hg (i.e., Hg not added experimentally) was 0.02-1.0 ng g-1 dry weight (dw) for MeHg in zooplankton and benthic invertebrates (depending on the sample mass available for measurement) and 0.2 ng g-1 wet weight (or 1.0 ng g-1 dw) for total Hg in fish. Mercury concentrations in fish were measured in wet muscle tissue and converted to a dry weight basis assuming a moisture content of 80% based on the average value determined for a subset of fish (mean ( SE: 80.1 ( 0.2%; n ) 24). In early September, ambient Hg concentrations across the ten mesocosms that received Hg(II) additions ranged as follows: from 15.4 to 84.0 ng MeHg g-1 dw for zooplankton; from 20.7 to 40.4, 56.2 to 81.8, and 14.8 to 104.6 ng MeHg g-1 dw for the benthic invertebrate taxa Amphipoda, Gomphidae, and Hydracarina, respectively; and from 146 to 1057 ng Hg g-1 dw for age 1 yellow perch. The detection limit of spike Hg (i.e., Hg added experimentally) was 0.5% of each sample’s ambient Hg concentration. If spike Hg was not detected in a sample, the concentration was estimated as one-half of the detection limit. Data Analysis. All data analyses were performed with STATISTICA 6.1 (StatSoft, Inc.). Model I linear regression was used to quantify the relationship between Hg(II) loading rates and spike MeHg concentrations in biota. For fish data, we used the mean spike Hg concentration of each mesocosm because concentrations were not correlated with fish size. All variables were log10-transformed and residuals were examined for linearity, homoscedasticity, and normality. For each regression model, an F-test was performed to determine whether the relationship between the variables was significant (Ho: the slope is zero) and a two-tailed t-test was performed to determine whether the relationship between the variables was directly proportional (Ho: the slope is one). Because a directly proportional relationship (Y ) aX) also takes the form logY ) loga + 1logX, a slope of one in a log-log relationship indicates that the relationship between the variables is directly proportional (i.e., a percent change in the independent variable results in the same percent change in the dependent variable).
Results Zooplankton. Two weeks after Hg(II) loading began, zooplankton contained detectable levels of spike MeHg in mesocosms receiving higher loading rates (Figure 1). By 4 weeks, spike MeHg was detected in zooplankton from all VOL. 41, NO. 14, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4953
FIGURE 1. Spike MeHg concentrations in the food web of each mesocosm over time. Each point represents a composite sample of zooplankton or benthic invertebrates, or a mean of individual fish samples. Concentrations in fish were measured in wet muscle tissue (as total Hg) and converted to a dry weight basis assuming a moisture content of 80%. In panel A, sample from mesocosm 8× was lost. Values below detection limits were estimated as one-half the detection limit and marked with an “e”. mesocosms. Maximum concentrations of spike MeHg in zooplankton ranged from 1.5 ng g-1 dw in mesocosm 1× to 25 ng g-1 dw in mesocosm 15×. After 10 weeks, spike MeHg concentrations in zooplankton were significantly related to Hg(II) loading rates to the mesocosms (F ) 6, p ) 0.049, R 2 ) 0.45; Figure 2A). The slope of this relationship was not significantly different from one (t ) 0.31, p ) 0.76). Therefore, we failed to reject the null hypothesis that spike MeHg concentrations in zooplankton were directly proportional to Hg(II) loading rates. Examination of the residuals revealed 4954
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 14, 2007
that some of the variability could be attributed to differences in zooplankton community composition. Mesocosms with negative residuals (i.e., mesocosms 1×, 3×, 5×, and 6×) tended to have a lower total zooplankton biomass and a higher percentage of copepod nauplii than other mesocosms (12). Benthic Invertebrates. In all mesocosms, spike MeHg was consistently detected in the three taxa of benthic invertebrates examined, except for Hydracarina in mesocosm 1× (Figure 1). Concentrations of spike MeHg ranged from
FIGURE 2. Relationship between Hg(II) loading rate and spike MeHg concentration in the food web. Samples were collected September 4-16, 2002. Both variables are log transformed, axes are scaled so that a slope of one has an angle of 45°, and regression lines are shown with 95% confidence bands. Each panel provides the results from the F-test, which was performed to determine whether the relationship between the variables was significant (Ho: the slope is zero). Comments in the caption for Figure 1 also apply to this figure. 0.7 to 9.0 ng g-1 dw in Amphipoda, from 0.6 to 10.9 ng g-1 dw in Gomphidae, and from