Environ. Sci. Technol. 2007, 41, 5895-5901
Elimination of Mercury by Yellow Perch in the Wild J I L L I A N L . A . V A N W A L L E G H E M , †,‡ P A U L J . B L A N C H F I E L D , * ,‡ A N D HOLGER HINTELMANN§ Department of Zoology, University of Manitoba, Z320 Duff Roblin Building, Winnipeg, Manitoba, R3T 2N2, Canada, Freshwater Institute, Fisheries and Oceans Canada, 501 University Crescent, Winnipeg, Manitoba, R3T 2N6, Canada, and Department of Chemistry, Trent University, 1600 West Bank Drive, Peterborough, Ontario, K9J 7B8, Canada
The rate of methylmercury (MeHg) elimination by fish is important in determining the extent of bioaccumulation and for predicting recovery times of MeHg-contaminated fisheries. Rates of MeHg elimination remain uncertain in existing bioaccumulation models due to a lack of field studies. We addressed this problem by monitoring fish that had naturally accumulated isotopically enriched MeHg (spike MeHg) during a whole-ecosystem experiment. We transported yellow perch (Perca flavescens) from the experimental lake to an untreated lake and monitored spike total mercury (THg, most of which was MeHg) losses over 440 d. Spike THg was distributed among fish tissues in a similar way as ambient THg (background non-spike THg). We observed rapid loss of spike THg from liver and other visceral tissues (∼90 d) followed by a plateau. Subsequently, there was prolonged redistribution of spike THg into muscle (180 d). Loss of spike THg from the whole fish occurred >5 times slower (half-life of 489 d) than in past laboratory studies using this species. We determined that MeHg bioaccumulation models with laboratory-based elimination rates produced faster losses than those observed in wild fish. The present findings provide support for refining elimination rates in MeHg models and show the importance of examining biological processes under natural conditions.
Introduction Methylmercury (MeHg) is a neurotoxin that affects humans (1) and wildlife (2) and remains the primary cause of fish consumption advisories across North America (3). The efficient bioaccumulation of MeHg by aquatic biota and biomagnification with trophic level results in high concentrations in large predatory fish (4). The amount of MeHg present in fish is the net result of uptake (from diet and respiration) and elimination rates (5). Relatively fast uptake rates coupled with slow elimination rates, which are typical for large fish, cause increasing MeHg levels over time (6, 7). Correctly predicting fish MeHg bioaccumulation using models requires accurate estimates of elimination. In particular, elimination rates are important for predicting recovery timelines for MeHg-contaminated systems. For example, if * Corresponding author phone: 204-984-4524; fax: 204-9842404; e-mail:
[email protected]. † University of Manitoba. ‡ Fisheries and Oceans Canada. § Trent University. 10.1021/es070395n CCC: $37.00 Published on Web 07/07/2007
2007 American Chemical Society
anthropogenic releases of Hg into the environment were stopped completely, the time required for MeHg levels in fish to decline will depend upon losses from environmental pools (e.g., sediment), and the rate of MeHg loss from fish tissues (as suggested by 8). Accurate estimates of elimination rates are also required to predict increases in fish MeHg concentrations due to industrial projects, such as reservoir flooding (8). Although an understanding of MeHg elimination rates is important, results of prior studies may have limited applicability to natural populations. Nearly all studies on MeHg elimination have been conducted in laboratories and typically involved acute exposure to high MeHg concentrations (10076 500 ng g-1 wet weight (w.w.) 9, 10), administered through either force-feeding or injection (e.g., 11, 12). In contrast, wild fish typically experience chronic exposure to lower MeHg concentrations through natural prey (typically 3-440 ng g-1 w.w. depending on prey, 13). Acute exposure may cause more of the dose to be present in biological compartments that lose MeHg quickly than compartments that tightly bind MeHg (14, 15). Further, most studies have used fish held in aquaria where activity levels, feeding patterns, and water temperatures differ substantially from those in nature (16). Approximately half of laboratory studies had a duration of less than 90 d, which commonly results in overestimation of elimination rates (7). The two field studies that have examined MeHg elimination from fish have both involved moving highly contaminated fish to systems with lower MeHg levels (17, 18). The applicability of these studies is somewhat limited due to continuous exposure to low MeHg levels during the period of elimination. We took a novel approach to estimating rates of MeHg elimination from fish in a natural setting. We moved yellow perch (Perca flavescens) from a lake where they had accumulated isotope-enriched MeHg (referred to as spike MeHg), as part of a whole-lake Hg addition experiment, to an enclosure in a lake containing no spike Hg. Losses of spike THg and MeHg from the fish were then monitored for 440 d. Our study provided unique and realistic conditions including (i) chronic accumulation of environmentally relevant concentrations of spike MeHg via natural routes of exposure; (ii) long study duration; and (iii) lake habitat. The primary objective of this study was to quantify the elimination of spike MeHg from the whole fish and its tissues. We hypothesized that MeHg elimination rates in the field would be different from those estimated by laboratory studies due to factors such as different dosing patterns, distribution among tissues, water temperatures, and fish behavior. We tested this hypothesis by comparing our findings to MeHg models developed from laboratory studies.
Experimental Section Study Site. This research took place at Lakes 658 and 240 (49° 39′ 15′′ N, 93° 43′ 35′′ W) within the Experimental Lakes Area (ELA) in northwestern Ontario, Canada. The two small oligotrophic lakes have similar fish community composition, with northern pike (Esox lucius) as the top predator and yellow perch as the dominant forage fish species. Lake 658 is the site of the Mercury Experiment To Assess Atmospheric Loading In Canada and the United States (METAALICUS). Beginning in 2001, mercury enriched in 202Hg (to 90.8%; termed spike Hg) was added directly to surface waters biweekly for the entire open water season as HgCl2 diluted with 5% nitric acid. Spike Hg was added at approximately 6 times (22 µg m-2 yr-1) the rate of ambient (background or non-spike) Hg deposition (19, 20). Ambient Hg has a different VOL. 41, NO. 16, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5895
ratio of Hg isotopes that can be analytically distinguished from spike Hg. Fish in Lake 658 accumulated spike MeHg (21) reaching 45% of the ambient MeHg concentrations in muscle of yellow perch at the beginning of the present study. Yellow perch were moved from Lake 658 to a large enclosure (12 m × 20 m) that bordered the north shore of Lake 240. The enclosure consisted of wood and wire mesh (0.64 cm) walls (maximum depth of 2 m) and was open to the sediment allowing fish to access prey. We covered the enclosure with nylon mesh (3 cm) to protect fish from predators. Water temperature within the enclosure was recorded every 30 min using Stowaway Tidbit temperature loggers (Onset Computer Corp., Bourne, MA). Transfer and Collection of Fish. On June 17, 2004, yellow perch (58-80 mm fork length, 1.8-6.8 g) were collected from Lake 658 with hoop nets and a trap net. Fish were transported to the ELA field station in coolers containing aerated water. Perch were anesthetized using 0.05 g L-1 MS222 (tricaine methane sulfonate). Wet weight (to the nearest 0.1 g) and fork length (mm) of each perch were measured before subcutaneous injection of pink elastomer, and implantation with a unique decimal coded wire tag (1.1 mm length × 0.25 mm diameter; both fish tags from Northwest Marine Technology Inc., WA). Fish were then placed in a holding pen (91 cm × 102 cm × 183 cm with 0.64 cm mesh) in Lake 240. After 2 d in the holding pen, 140 surviving yellow perch were transferred to the enclosure in Lake 240. Ambient MeHg concentrations in Lake 658 were approximately double those in Lake 240 zooplankton and age-1 yellow perch during the course of this study (P. Blanchfield and M. Paterson, unpublished data). As a result, transferred yellow perch were adjusting to reduced exposure to both ambient and spike MeHg. We added 30 yellow perch (57-77 mm fork length) from Lake 240 to the enclosure to control for spike Hg potentially recycled from excretion by Lake 658 perch or loss from their carcasses. Lake 240 perch were anesthetized, tagged, and marked with yellow elastomer dye to distinguish them from Lake 658 perch held in the same enclosure. These Lake 240 perch were collected 90 d (n ) 9) and 180 d (n ) 1) after beginning the study, and accumulated no appreciable amounts of spike THg. Thus, recycling of spike Hg in the enclosure was negligible and did not influence the observed results. Yellow perch originally from Lake 658 were collected from the enclosure at 0 d (n ) 15; 1.8-6.8 g), 15 d (n ) 10), 30 d (n ) 10), 60 d (n ) 8), 90 d (n ) 9), 135 d (n ) 8), 180 d (n ) 3), and 240 d (n ) 1) after transfer. Fish were captured using a seine net (either a 4 m × 1.5 m pole seine or a 16.8 m × 2.4 m beach seine) in the summer and minnow traps or gill nets (8-10 mm mesh) under the ice in winter. Ice damaged the enclosure, causing all perch to escape; however, some escapees were captured on days 365 (n ) 3), 415 (n ) 1), and 440 (n ) 1). All collected perch were euthanized in 0.5 g L-1 MS222. Fork length and wet weight were recorded and the fish were frozen. Decimal coded wire tags were removed before further processing. Processing of Fish Tissues. Tools and surfaces were rigorously cleaned using 95% ethanol and Kimwipes (Kimberly Clark Professional) between all samples. The liver and approximately 0.2 g of dorsal muscle tissue were removed from each frozen fish. The contents of the gastrointestinal tract were removed before the remaining tissue (referred to as the carcass) and samples of muscle and the liver were freeze-dried. The carcass was ground in an electric stainless steel grinder and approximately 0.15 g was sub-sampled. Each tissue sample was placed in an acid-washed scintillation vial. Total mercury (THg) was measured in the muscle tissue of all fish collected. Both THg and MeHg concentrations were 5896
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 16, 2007
measured in muscle, liver, and carcass of up to five individual fish from each sampling day. Hg species were quantified by inductively coupled plasma mass spectrometry (ICP-MS) at Trent University, Peterborough, Ontario (22). THg was measured in samples after digestion with HNO3/H2SO4 (7:3 v/v) and heating at 80 °C until brown NOx gases no longer formed. THg of sample digests was reduced by SnCl2 and determined by ICP-MS (Thermo-Finnigan Element2) using a continuous flow cold vapor generation technique. MeHg in samples was solubilized by treatment with 20% (w/v) KOH/ MeOH solution at 50 °C and measured after aqueous phase ethylation using NaBEt4 (22). Volatile Hg species were purged and trapped onto Tenax and MeHg was measured after thermodesorption and GC separation using ICP-MS detection (Micromass Platform). To correct for procedural recoveries, all samples were spiked with 201HgCl2 or Me201HgCl prior to sample analysis. Method blanks and certified reference materials were measured for each batch of samples. Results for THg and MeHg in DORM-2 (measured mean ( SD, 4730 ( 100 and 4320 ( 190 ng g-1; certified, 4640 ( 260 and 4470 ( 320 ng g-1 for THg and MeHg, respectively) were not statistically different from certified values. The detection limit for ambient Hg was 1.0 ng g-1 d.w. and 2.0 ng g-1 d.w. for THg and MeHg, respectively. The detection limit for spike Hg is not absolute, but relative to the ambient Hg concentration in each sample. A minimum of 0.5% of the overall Hg must originate from the spike to be detectable. In general, all THg was in the form of MeHg for both ambient Hg (92-108%) and spike Hg (102-115%) in muscle (n ) 38) and carcass samples (n ) 38). These values were more variable for liver, ranging from 56-114% MeHg for ambient and 56-100% for spike. Data for all liver samples collected on day 135 and one liver sample from day 60 are not presented owing to analytical errors with these specific samples only. We present THg data in this study as most THg in fish tissues was in the form of MeHg. Calculation of Body Burden and Elimination Rate. There was no relationship between fish size and muscle THg concentration for the day 0 yellow perch. Overall, there was little growth during the first 365 d of the study (range: -0.71.9 g). To account for individual differences in weight and possible impacts of growth on THg concentrations, we calculated burdens to indicate the total mass of THg within fish and tissues. THg burdens in yellow perch tissues were calculated by multiplying THg concentration (ng g-1 d.w.) by the dry weight of that tissue. To determine the weight of muscle, all muscle tissue was removed from 14 yellow perch (0.6-13.9 g) that were not analyzed for THg. We used the resulting relationship between fish weight (minus stomach contents) and dry muscle weight (y ) -0.0195 + 0.0813x; r2 ) 0.99) to estimate muscle mass and burden in perch analyzed for THg. The carcass samples analyzed for THg contained residual muscle. We subtracted the THg burden of residual muscle tissue from the carcass, which we termed remaining fish (RF). The burden in the whole fish was determined by summing the tissue burdens of muscle, liver, and RF. We applied the exponential decay models that provided the best fit to spike THg burden data: Y ) ae-kt or Y ) y0 + ae-kt; where Y is burden, y0 is the asymptote, t is days, and a and k are constants. The half-life of the THg burden was calculated with
half-life (days) )
ln (2) k
Model Simulations. We compared the predicted elimination rates of two different MeHg models to observed losses of spike THg. The Wisconsin Fish Bioenergetics model version
TABLE 1. Mean ((1 SD) Percent of THg Present in Muscle and the Remaining Fisha (RF) for Spike Hg (Enriched in 202Hg) and Ambient (Background) Hg in Yellow Perch (Perca flavescens) over 440 days After Being Transferred from Lake 658 to Lake 240 spike
ambient
day
n
muscle
RF
muscle
RF
0 15 30 60 90 135 180 365 415 440
4 5 5 5 5 5 3 3 1 1
67 ( 4 62 ( 2 67 ( 5 73 ( 3 75 ( 2 85 ( 5 86 ( 3 73 ( 2 58 59
32 ( 4 37 ( 2 31 ( 5 27 ( 2 24 ( 2 14 ( 5 13 ( 3 26 ( 2 41 40
82 ( 11 70 ( 2 71 ( 4 78 ( 3 76 ( 4 93 ( 6 80 ( 4 69 ( 7 65 60
17 ( 11 29 ( 2 28 ( 4 21 ( 3 23 ( 4 6(6 19 ( 4 30 ( 7 34 39
a The remaining fish consists of all tissues excluding muscle and liver.
3.0 (23) includes metabolic equations linked to estimates of MeHg uptake and elimination. The constants used in this model were initially developed from laboratory studies (5), and later adjusted to fit field data (24). The Trudel and Rasmussen model (7) (referred to as the TR model) does not include a bioenergetic component. This model is based on four laboratory studies (providing 21 rate estimates) and one field study (providing 4 rate estimates), all longer than 90 d in duration. The TR model has different elimination rates for chronically and acutely exposed fish, and both rates were tested. Measured growth and water temperatures experienced by yellow perch during this study were entered into the two models (inputs given in 25). Statistical Analyses. Nonlinear curves and least-squares linear regressions describing the relationship between time and observed THg burden were fitted and analyzed using Systat within SigmaPlot version 9.0 or SAS version 9.0. Equations of nonlinear curves are shown both including and excluding the sampling days where n ) 1. A one-way analysis of variance (ANOVA) was used to determine differences in muscle burden among sampling days (when n g 3). Data were log transformed to meet assumptions of normality and homogeneity of variance. Tukey’s post-hoc tests were used when significant differences were encountered. To assess the overall fit of model predictions, we statistically compared the relationship between observed and predicted values to a slope of 1 and an intercept of 0 (i.e., equality between observed and predicted data).
Results and Discussion Spike and Ambient THg. On average, concentrations of spike THg in yellow perch muscle tissue were 42% that of ambient THg (range: 37-45%). At the beginning of the study, spike and ambient THg concentrations in muscle were positively correlated (y ) 71.2 + 0.355x: r2 ) 0.87) and similarly distributed among fish tissues (Table 1). Together, these findings suggest that yellow perch in Lake 658 accumulated spike and ambient THg in a similar manner. Therefore, we expect that spike Hg behaves the same way as ambient Hg in the fish. Redistribution of THg among Fish Tissues. Most spike THg was transferred from liver and RF within 90 d followed by a plateau. The estimated half-lives for the fast phase of spike THg redistribution were 24 and 44 d for liver and RF tissues, respectively. In liver, the spike THg burden decreased by 60% before leveling off at approximately 1.3 ng (Figure 1). When the single fish captured on days 415 and 440 were excluded, the equation describing the movement of spike
THg from liver changed slightly (y ) 1.35 + 1.74e-0.032t; r2 ) 0.33, half-life ) 22 d). Spike THg burden in RF tissues decreased over the first 90 d and then leveled off at approximately 50% of the initial burden (Figure 1). Excluding data from days 415 and 440 weakened the fit to the exponential decay model for RF tissue (y ) 43.1 + 50.1e-1.16t; r2 ) 0.06). Asymptotic loss of THg from visceral tissues has occurred more quickly in past studies, typically leveling off within 5-35 d following exposure (11, 26-28). The timing of contaminant relocation among tissues may partially depend on metabolic rate (26) or species (29). Rates of transfer from liver and RF may have been slower in our study than past studies because of field conditions affecting metabolic rate and chronic exposure (as suggested by 14). There was no significant difference in spike THg burdens in muscle among sampling days (F7,58 ) 1.3, p ) 0.29). Nevertheless, mean burdens declined over the first 90 d, followed by a 60% increase until day 180, and another decrease by 15% of peak values (Figure 1). In addition, the percentage of spike THg burden contained in muscle increased significantly with time (linear regression: F1,30 ) 123.3, p < 0.0001, r2 ) 0.80) until day 180 when a peak of 86% was reached (Table 1). The percentage stored in RF tissue followed the opposite pattern of the percentage stored in muscle (Table 1), and liver consistently accounted for 1% of the burden (data not shown). These results imply that THg was transferred from liver and RF tissues to muscle tissue. A net decline in muscle THg was possible only after transfer to the muscle had slowed and peak storage was reached. The changes observed in muscle burden were similar to the findings of a 2-year field study on THg elimination by 100 g yellow perch (17). Laboratory studies have also shown that MeHg from various tissues is transferred via blood to muscle (accounting for 71-86% of the burden) (27, 30), although these studies usually ended before net loss from muscle could occur. MeHg injected directly into muscle tissue is slowly lost from the whole fish, suggesting that elimination from muscle is possible (10, 12). To our knowledge, only studies longer than 250 d in duration have found loss of MeHg from muscle tissue of orally exposed fish (18, 31). If spike THg was transferred via blood directly from RF tissues to muscle, increases in muscle burden would occur simultaneously with decreases in RF burden. Our data did not show this pattern; rather, the muscle burden of spike THg did not increase until 45 d after the RF burden leveled off. Our ability to provide a mass balance of spike Hg may be limited by unequal numbers of samples among tissues and over time. The regression used to determine muscle and RF tissue masses may have varied seasonally, increasing variability in the data. Loss of THg from the Whole Fish. Mean burdens of spike THg in whole fish followed a declining trend (Figure 1), which remained similar when single fish were excluded (y ) 195e-0.0013t; r2 ) 0.12). One year after transfer from Lake 658 to Lake 240, the mean spike THg burden in yellow perch was 57% of initial values. The estimated half-life of spike THg in the whole fish was 489 d (including days 415 and 440; 95% confidence range: 314-787 d), which is more than 5 times slower than predicted by past laboratory studies on yellow perch. Further, the rate of elimination that we found is 1.830 times slower than laboratory estimates for other small fish (
