Environ. Sci. Technol. 2005, 39, 4707-4713
Biogeographic Provinces of Total and Methyl Mercury in Zooplankton and Fish from the Beaufort and Chukchi Seas: Results from the SHEBA Drift G . A . S T E R N * ,† A N D R . W . M A C D O N A L D ‡ Department of Fisheries and Oceans, Freshwater Institute, 501 University Crescent, Winnipeg, Manitoba, Canada, R3T 2N6, and Department of Fisheries and Oceans, Institute of Ocean Sciences, P.O. Box 6000, Sidney, British Columbia, Canada, V8L 4B2
Samples of copepods (Calanus hyperboreus) and arctic cod (Boreogadus saida) were collected along the SHEBA (Surface HEat Budget of the Arctic) drift track, which commenced in the Canada Basin (October 1997) and finished in the Mendeleev Basin (October 1998). Here, we report total mercury (HgT) and CH3Hg concentrations in these biological samples and examine concentration variability along the drift track in the context of trophic variation, inputs from land, spring mercury depletion events (MDEs), and oceanographic provinces. We find background concentrations of HgT in C. hyperboreus as low as 0.02 µg/g (dw), with the Canada Basin samples exhibiting approximately 2-fold higher mercury concentrations than those from the Chukchi Plateau and Mendeleev Abyssal Plain. This eastto-west trend in mercury concentration is punctuated by two and possibly three intervals of elevated mercury (HgT, 0.10-0.12 µg/g (dw); CH3Hg, 0.023-0.028 µg/g (dw)) along the drift track. One interval of elevated HgT and CH3Hg levels occurred during and shortly after melt. %CH3Hg reached a maximum of 60% during this time period, three times higher than any other time during the drift. This transient rise in C. hyperboreus CH3Hg concentration seems to strongly point to mercury accumulated in snow during MDEs. However, the alignment of elevated mercury samples with oceanographic fronts and the observed regional differences between basins suggest that variation of mercury concentration is primarily a consequence of ocean structure. Given that large animals such as whales selectively forage in regions of higher food concentration such as fronts, recent change in the ice climate of the western Arctic Ocean, perhaps mediated by changes in heat storage, may provide the means to change their exposure to mercury thus explaining observed increases in mercury concentrations in western beluga whales during the 1990s.
Introduction In October 1997, the CCGS Des Groseilliers was beset in the permanent pack of the Beaufort Sea to provide a platform * Corresponding author tel: (204) 984-6761; fax: (204) 984-2403; e-mail:
[email protected]. † Freshwater Institute. ‡ Institute of Ocean Sciences. 10.1021/es0482278 CCC: $30.25 Published on Web 05/20/2005
Published 2005 by the Am. Chem. Soc.
FIGURE 1. Calanus hyperboreus collection locations along the SHEBA drift (Dec 1997-Sept 1998). from which measurements of heat flux could be made throughout an entire Arctic year (SHEBA, Surface HEat Budget of the Arctic) (1-3). During its drift until September 1998, this platform provided an extraordinary opportunity to conduct ancillary science programs focused on biology (4-6), freshwater sources and watermass distributions (7, 8), and contaminants (9). Thin ice and an inordinately fresh surface layer (2, 10) were noted at the outset of SHEBA and these were later inferred to have contributed to shifts in the numbers and taxonomy of plankton (4). Perhaps equally significant to the interpretation of the SHEBA time series was the drift track actually taken (Figure 1). After a few months of drifting slowly westward in the Canada Basin, the station passed rapidly over the Northwind Ridge and Abyssal Plain in early January and onto the Chukchi Plateau where it meandered for much of the remaining period of drift. Eventually, by September 1998, SHEBA drifted back into the deep water of the Mendeleev Basin. As a consequence of its drifting over the complex topography of the Chukchi Plateau, SHEBA experienced considerable changes in water masses, ocean structure, and source water (7, 8, 11). Within the Arctic, mercury has become a focus of international research following the discovery of atmospheric mercury depletion events (MDEs; 12, 13). Although it is clear that photochemically mediated MDEs remove gaseous mercury from the bottom kilometer of the atmosphere after polar sunrise (14), and deposit it to surfaces in a reactive, biologically available form (15), it remains unknown whether MDEs actually result in increased loadings to aquatic systems. The observation of up to 4-fold increases of mercury over two decades in Beaufort Sea marine mammal livers (16, 17) poses a question of enormous significance to human and ecological health. Even though global emissions of mercury have been decreasing lately (15, 18), and even though the arctic atmosphere shows no recent increasing trend in mercury concentration (14), the increasing trends of mercury in marine mammals suggest that something has been changing in the arctic biogeochemical cycle of mercury. Certainly, MDEs as a recent or recently enhanced phenomenon might offer one explanation (15), but there are likely others linked to climate change which has been shown to alter hydrology, organic carbon cycling, and ecosystem structure in the Arctic (19, 20). Indeed, climate change assumed an unanticipated importance in the SHEBA observation program (2, 4, 7, 21). VOL. 39, NO. 13, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4707
Here we present total mercury (HgT) and CH3Hg data for biota representing two trophic levels collected along the drift track: zooplankton (Calanus hyperboreus) and arctic cod (Boreogadus saida). In the Arctic, zooplankton biomass consists predominantly of the large calanoid copepod species Calanus hyperboreus and Metridia longa. In particular, along the SHEBA drift track older stages of C. hyperboreus predominated with increased relative biomass of M. longa observed over the Northwind Ridge and the Chukchi Ice Cap (5). Arctic cod is a “keystone” species in Arctic ecosystems as cod provide the major link between primary producers, zooplankton, and upper trophic levels such as marine mammals. While very little is known about their migration patterns, distribution and abundance habitat and biology, cod are known most commonly to frequent surface water close to shore among ice floes but are also found near the ocean bottom and throughout the water column in open water (22, 23). An objective of this study was to seek a seasonal pattern in aquatic foodweb mercury concentration in the Canada Basin to determine whether there might be a response to springtime MDEs. The drift of SHEBA, which included basin and Chukchi Plateau portions, however, allowed spatial variation also to play a role and we accordingly examine HgT and CH3Hg transect data for both temporal and spatial patterns.
Methods Sample Collection. Arctic cod and C. hyperboreus, along with other pelagic foodweb components, were collected from 30 sites along the SHEBA drift track (Figure 1). All samples were collected from a laboratory (2.5 × 5 m container) set up on the ice about 100 m from the ship (2). This lab was equipped with a large electric-hydraulic winch spooled with 4 km of Kevlar cable deployed through a 1.25 × 1.25 m hole in the floor. Through this floor zooplankton were collected using a vertically towed 1 m2 nested plankton net (inner mesh ) 3 mm; outer mesh ) 500 µm). Bulk zooplankton samples were poured into a flat transparent glass dish and the individual C. hyperboreus were picked out using forceps. Arctic cod were collected using a dip net. Whole-body composites of C. hyperboreus and individual whole Arctic cod samples were placed in plastic (Whirlpak) bags and stored in a freezer (-20 °C) on board the Des Groseilliers until shipment to the Freshwater Institute where they were again stored at -25 °C until processing. HgT and CH3Hg Sample Extraction and Analysis. Whole arctic cod (cod less than one gram were pooled, based on collection date, for combined sample weights of between 0.9 and 3.7 g) and sub-samples of whole body composites of C. hyperboreus were digested with a sulfuric/nitric acid mixture, after which a potassium permanganate solution was added dropwise until a slight pink color just persisted. The resulting digest was analyzed for total mercury by cold vapor atomic absorption spectroscopy (24). For CH3Hg analysis, tissues were extracted according to the procedure of Uthe et al. (25). Wet tissue samples were homogenized with a solution of acidic sodium bromide and copper sulfate and extracted into a toluene phase. CH3Hg was then partitioned into an aqueous thiosulfate solution and subsequently by addition of potassium iodide back-extracted into toluene. CH3Hg in this final extract was measured on a Varian model 3400 gas chromatograph equipped with a 5-m, SPB-5 megabore column and 63Ni electron capture detector. All chemical analyses were conducted at the Department of Fisheries and Oceans, Winnipeg, MB. The detection limits for HgT and CH3Hg were 0.005 and 0.004 µg‚g-1, respectively. Duplicate samples and standard reference materials (SRMs) were run after approximately every eighth sample. The average HgT concentrations measured for the SRMs LUTS-1, TORT-2, and 4708
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 13, 2005
CRM 2976 were 0.0177 ( 0.0001, 0.260 ( 0.000, and 0.062 ( 0.001 µg g-1, respectively. The average CH3Hg concentration measured for DOLT-2 was 0.668 ( 0.051 µg g-1. For HgT, the average relative difference between duplicates of individual Arctic cod and C. hyperboreus samples were 0.004 and 0.006 µg g-1, respectively. For CH3Hg it was 0.002 µg g-1 for both the Arctic cod and C. Hyperboreus. All results are presented on a dry weight basis. Stable Isotope and Trophic Level Calculations. Samples of Arctic cod muscle tissue and sub-samples of whole body composites of C. hyperboreus were dried in an oven for several days at 60 °C and then ground into a fine powder using a mortar and pestle. Dried samples of 1-2 mg were analyzed for 15N/14N and 13C/12C isotope ratios using a Carlo Erba NA1500 elemental analyzer. Sample gases were introduced into a VG Optima automated mass spectrometer with helium carrier gas, and water and CO2 were cryogenically removed using magnesium perchlorate and an Ascarite column, respectively. All samples were standardized against Pee Dee Belemnite (C) or N2 in air and concentrations expressed in the δ notation as parts per thousand (‰) according to the following equation:
δR‰ ) [(Rsample/Rstandard) - 1] × 1000 where δR corresponds to the relative difference between isotope ratios of the samples and standard gases (15N/14N and 13C/12C). Precision of the instrument for nitrogen and carbon isotope analysis, based on several years of use, has been 0.4 ‰ and 0.3 ‰ (2 SD), respectively. Multivariate Statistics. ANCOVA was used to assess the effects of location, length, and δ15N for cod and location and δ15N for C. hyperboreus. Slopes were tested (length*loc, δ15N*loc), nonsignificant interaction terms were removed, and data were adjusted for the effects of significant variables (e.g., model used for arctic cod; log [Hg] ) location, length, δ15N, location*length, and location* δ15N). Differences between collection locations were examined with paired comparisons of length and δ15N adjusted least-squares mean concentrations (26).
Results and Discussion The distribution of HgT in C. hyperboreus collected along the SHEBA drift track (Figure 2G) shows two dominant features. First, there is an approximately 2-fold decrease in concentration on going from the Canada Basin to the Chukchi Plateau and Mendeleev Abyssal Plain (dotted line on Figure 2G). Second, there appear to be at least two and probably three intervals along the track (vertical arrows on Figure 2G) where HgT is elevated above this general trend to values as high as 0.10-0.12 µg/g (dw). A similar general decreasing trend along the drift track was observed for CH3Hg (Figure 2E). However, when displayed as a percentage of HgT, %CH3Hg shows a peak at ∼60% (Figure 2F) in early June. The spatial/temporal variation in C. hyperboreus HgT and CH3Hg concentrations along the SHEBA drift path could be produced in a number of ways including (1) biological processes such as trophic variation or biomass dilution (27, 28), (2) fluxes into the system by rivers (Mackenzie, Yukon) or atmospheric deposition (MDEs), and (3) regional differences in water masses or water sources along the drift path. Below, we examine ancillary data for evidence of which of these processes is likely to have produced the spatial/temporal patterns evident in Figure 2. Trophic Variation versus Riverine Input. Stable isotopic composition (δ13C, δ15N) of C. hyperboreus provides a basis to assess the potential effect of trophic variation as a contributor to mercury variance along the drift track (Figure 2,
FIGURE 3. Carbon isotope ratios for (A) Calanus hyperboreus (black dots correspond to data from Saupe et al. (33)) and (B) Arctic cod (Boreogadus saida).
FIGURE 2. SHEBA drift section as a function of date (1997-1998) showing (A) δ13C in C. hyperboreus; (B) δ15N in C. hyperboreus; (C) mesozooplankton biomass (data are from Table 1 in Ashjian et al. (5) , Roman numerals correspond to zooplankton distributions along the SHEBA transect); (D) total mercury concentration in snowpack (data from Lu et al. (13)); (E) methyl mercury concentration in C. hyperboreus; (F) %CH3Hg in C. hyperboreus: and (G) [HgT] in C. hyperboreus. Panels A and B, and Figure 3). In marine foodwebs, higher trophic levels typically are slightly enriched in 13C (29) and significantly enriched (3-5 ‰ per trophic step) in 15N (27, 30, 31) with the result that the latter has become a method of choice to establish pelagic trophic structure (32). Using copepods and other zooplankton collected along the Alaskan North slope in the 1980s, Saupe and co-workers (33) established an east-to-west trend in δ13C with less depleted (more positive) values toward the west (Figure 3A, closed circles), a distribution that is mirrored in surface sediments along the Alaskan and Canadian Beaufort Shelves
(34-36). Indeed, this regional east-to-west increase in δ13C has proven sufficiently robust in the context of the food web to provide an elegant way to infer bowhead whale ages and foraging patterns along the Beaufort and Bering Sea migratory pathway (37). Our collection of C. hyperboreus along the SHEBA drift (1997-1998) exhibits a similar increase from east-to-west in δ13C (δ13C ) m*longitude + b, m ) 0.0399, b ) -30.2, r2 ) 0.18, p < 0.005, n ) 47; Figure 3A) but no statistical relationship with latitude (not shown). The general increase in C. hyperboreus δ13C along the drift path (Figure 2A) together with the general decrease in HgT concentration (Figure 2G) imply a negative correlation between these two variables (Figure 4A (δ13C ) m*[HgT] + b, m ) -10.1, b ) -23.3, r2 ) 0.11, p < 0.05; n ) 37). The δ13C east-to-west trend appears to indicate a large-scale geographical, recurrent gradient at the bottom of the food web, possibly supported by regional inputs (low δ13C, high mercury) to the Beaufort Sea by the Mackenzie River, rather than a shift in trophic level (33, 35). In support of this hypothesis, our δ15N data for C. hyperboreus (Figure 2B) exhibit no significant correlation with longitude as reflected by the δ15N along the drift track. The δ15N data, which lie within a range of (1.5‰, imply that C. hyperboreus were positioned within a single trophic level and the isotopic data together, therefore, strongly suggest that the large-scale geographical gradient in HgT content of C. hyperboreus is not related to trophic variation but may be supported by the same mechanism that produces geographical variation in δ13C. Relatively high amounts of Mackenzie River water in the initial portion of the SHEBA drift track (7) adds weight to the hypothesis that the concurrent δ13C enrichments and HgT elevations are supported by regional terrestrial inputs. Despite our conclusion that trophic variation does not contribute to the geographic variance discussed above, the statistically significant correlation between Log10[HgT] and δ15N in C. hyperboreus (Figure 5A) does imply that a portion of the variance VOL. 39, NO. 13, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4709
FIGURE 4. Linear relationship between δ13C and [HgT], dry weight, for (A) Calanus hyperboreus and (B) Arctic cod. in mercury in C. hyperboreus along the drift track is accounted for by trophic variation within the species (r2 ) 0.12). Biomass Dilution. Algal blooms can reduce the uptake of mercury into the food web through dilution (28). The envelope of mesozooplankton biomass measured by Ashjian et al. (5) (Figure 2C) shows a tripling between late 1997 (winter) and mid-May to mid-June 1998. Thereafter, biomass decreases almost monotonically until late September 1998. The increase in biomass following spring bloom could contribute to the decline in C. hyperboreus HgT concentration which started after day 102 (April 12, 1998). However, biomass clearly started to increase well before this time (∼midFebruary) with no accompanying decrease in HgT. In contrast, CH3Hg levels showed a significant decline starting in early February but then increased dramatically in midApril as the biomass reached its maximum. The decline in biomass observed after mid- to late-June is not reflected in a general return of either HgT or CH3Hg to the higher levels observed in the Canada Basin. The mesozooplankton biomass and the HgT concentrations observed in C. hyperboreus suggest that the standing stock of HgT held by mesozooplankton accounts for ∼100-500 ng m-2. Assuming total dissolved Hg ((HgT)D) in surface water in the Canada Basin to be ∼2 pmol (38) implies that the euphotic zone (∼50 m) contains (HgT)D ∼20 000 ng m-2; in other words, the mesozooplankton contain but a small fraction of HgT. We know of no CH3Hg measurements for Arctic Ocean surface waters, but values are likely to be extremely low (
