River influence on mercury bioaccumulation in the coastal food web of Eeyou Istchee, James Bay, Canada

A regional community-driven research network (Arctic Eider Society 2024) participated in a study of metal concentrations in coastal food webs of four communities on east Hudson Bay (Sanikiluaq, Kuujjuarapik, Umiujaq, and Inukjuak) and the community of Chisasibi on east James Bay (Fig. 1). The southern portion of the study area is within Eeyou Istchee, the traditional territory of the Cree of Quebec, while the northern portion comprises Inuit homelands of Nunavik and Nunavut. Hudson Bay and James Bay are polar marine ecosystems with winter ice cover and ice-adapted marine animals typical of more northern latitudes including polar bear, Arctic cod (Boreogadus saida), and ringed seal (Stewart and Barber 2010). The region is also important for migratory birds, which use coastal habitats during the spring and summer seasons. More details about the environmental characteristics of the Hudson Bay complex can be found in Stewart and Lockhart (2005), Stewart and Barber (2010), and Keller et al. (2014).

Spatial patterns of mercury bioaccumulation were investigated using two bioindicator species found throughout the study area: blue mussel (Mytilus edulis) and common eider (Somateria mollissima). Blue mussel is a sedentary filter-feeding invertebrate that feeds predominately on pelagic rather than sediment carbon (Rosa et al. 2018), and it is widely studied around the world to monitor pollution (Beyer et al. 2017). Common eider is a diving sea duck that feeds on molluscs (including blue mussel), crustaceans, and echinoderms in benthic habitats (Ouellet et al. 2013). Common eiders in east Hudson Bay and east James Bay are the sedentaria subspecies, which are non-migratory and remain year-round by overwintering in open water of polynyas and leads of ice particularly around the Belcher Islands (Mallory et al. 2004; Goudie et al. 2020). In spring, the eiders disperse to breeding grounds on islands and in coastal areas of James Bay and Hudson Bay, where they remain for much of the open-water season to breed, rear their young, and molt (Goudie et al. 2020).

Blue mussels and common eiders were collected by local hunters near each of their communities. The collections were completed during the open water season (June to early December) between 2014 and 2017, though additional common eiders were also collected at Chisasibi in 2021 to increase the sample size. Blue mussels were collected by hand or trawl net from several near-shore sites per community. Ten individual mussels were collected per site to form a composite sample, and the number of composite blue mussel samples varied from 6 to 20 per community, representing sampling of multiple sites over a 2- or 3-year period. A total of 59 composite blue mussel samples were obtained for the study. Blue mussels were frozen whole in polyethylene zip bags and shipped to the National Wildlife Research Centre (NWRC, Ottawa, Canada) for laboratory processing. Common eiders were captured with a firearm from sites near each community, with sample sizes ranging from 14 to 24 birds over a 2- or 3-year period, except for Inukjuak where only 4 birds were collected in 1 year. Breast muscle and the liver were removed from each bird after collection, frozen in polyethylene zip bags, and shipped to NWRC for laboratory processing. A total of 77 common eiders were obtained for the study. The repeat sampling of blue mussel and common eider at each community incorporated both local spatial variation and inter-annual variation.

In the laboratory, biotic samples were homogenized and freeze-dried prior to chemical analysis. Blue mussels were de-shelled, and the whole-body soft tissue of 10 individuals from a site was homogenized with an Omni Mixer with wide window shafts to form a composite sample. Individual common eider livers were also mechanically homogenized while eider muscle samples were manually homogenized in their containers after drying. All samples were freeze-dried for a minimum of 48 h.

Total mercury (THg) was measured in blue mussel, eider liver, and eider muscle with a Direct Mercury Analyzer (Milestone Inc., Shelton, Connecticut, USA) at NWRC. For a small subset of samples (liver and muscle of 14 eiders collected in 2021), THg was measured following wet digestion by atomic absorption spectrometry at RPC laboratories (Fredericton, New Brunswick, Canada) due to lab closure at NWRC. Analytical duplicates were measured every 10 samples with a relative percent difference (RPD) of <10% (n = 14). During the project, six types of certified reference materials (CRMs) were analyzed showing mean recoveries of 86%–102% (n = 108; see Rohonczy et al. 2024, for details). An inter-laboratory comparison of eider mercury results from RPC with those from NWRC showed high precision (mean RPD = 10%, n = 10).

Methylmercury was also measured on a subset of blue mussel samples (n = 44) with a cold-vapour atomic fluorescence spectrometer at NWRC or Flett Research Ltd (Winnipeg, Manitoba, Canada). Samples were digested prior to analytical detection with nitric acid at NWRC and with KOH-methanol at Flett Research Ltd. Duplicate analytical measurements had a mean RPD = 4% (n = 7). Recoveries of CRMs averaged 90% (range of 85%–97%) for both DORM-3 fish protein (n = 4; National Research Council of Canada) and NIST 2976 mussel tissue (n = 6; National Institute of Science and Technology).

A suite of 24 elements was measured in blue mussel samples (n = 59) with an inductively coupled plasma mass spectrometer (ICP-MS) at NWRC or the Alberta Institute of Technology Futures (Vegreville, Alberta, Canada). Only elements that were above analytical detection (Ag, As, Ba, Cd, Co, Cu, Fe, Li, Mn, Mo, Ni, Pb, Se, Sr, U, V, and Zn) are reported here. Analytical duplicates were measured every 10 samples with deviations generally <10%. Mean recoveries of individual elements ranged from 92% to 105% for measurement of 2–8 types of CRMs, with the exception of Pb, which had a lower mean recovery (77%). Details on the quality assurance and quality control of measurements for individual elements are provided in the supplementary materials (Table S1).

Blue mussels and common eider muscle were analyzed for nitrogen, carbon, and sulfur isotope ratios on a DeltaPlus XP isotope ratio mass spectrometer interfaced to a Vario El III elemental analyzer via a Conflo II at the Ján Viezer Stable Isotope Laboratory at the University of Ottawa (Ottawa, Ontario, Canada). Isotope ratios are reported in delta notation as the per mil (‰) deviation from N₂ (air) for nitrogen, Vienna Pee Dee Belemnite for carbon, and Vienna Canyon Diablo Troilite for sulfur. Analytical precision is typical <0.2‰ for carbon and nitrogen stable isotopes and <0.3‰ sulfur stable isotopes.

Mercury stable isotope ratios were measured on a subset of common eider muscle samples (n = 25) with a Nu Plasma II multi collector-ICP-MS instrument in the Water Quality Centre at Trent University (Peterborough, Ontario, Canada). Samples were digested using 3 mL of aqua regia in a 40 mL amber glass vessel by heating on a hot plate at 120 °C for 4 h. An ESI hydride ICP hydride generation system was used to generate a Hg0 vapor by the quantitative reduction of Hg(II) in solution using stannous chloride (3% w/v in 1 M HCl). Mercury isotope ratios are reported as the per mil deviation from the xxxHg/198 Hg ratio of the NIST SRM 3133 mercury standard. Results for replicate measurements of the RM-8610 cinnabar standard are provided in the supplementary materials (Table S2).

The raw concentration data for THg, MeHg, and other elements as well as carbon, nitrogen, and sulfur isotope ratios are available online in an open-access database (Government of Canada 2022), and summarized data for common eiders and blue mussels from Hudson Bay are also reported in Rohonczy et al. (2024). Spatial patterns of mercury concentrations in biota were examined by comparing community means using one-way Analysis of variance (ANOVA) tests. Data were log-transformed, as required, and probability values of pair-wise comparisons were Holm-adjusted to correct for the experiment-wise error rate. Diet influences on mercury concentrations of common eider were examined by testing Pearson correlations with carbon, nitrogen, and sulfur stable isotope ratios. Blue mussels are sedentary bio-indicators of metal bioaccumulation near the base of the food web, and therefore, their element concentrations were examined for potential spatial patterns in metal exposure and links with mercury bioaccumulation. A principal component analysis (PCA) of blue mussel element concentrations was performed in R program (R Core Team 2021) using the built-in prcomp function and plotted with the ggbiplot package. Pearson correlations were tested for associations of blue mussel THg and MeHg concentrations with other tissue element concentrations. Mercury sources and exposure pathways to common eider were investigated by comparing mercury stable isotope ratios between communities with one-way ANOVAs.

A clear spatial pattern was evident of higher mercury concentrations in blue mussels from near Chisasibi (Fig. 2). Those blue mussels had statistically higher THg and MeHg concentrations compared to mussels collected near four communities in east Hudson Bay (one-way ANOVAs, F_4,58 = 12.6 (THg), F_4,43 = 36.4 (MeHg), p < 0.001). There was approximately a twofold difference in mean THg and MeHg concentrations of blue mussels between Chisasibi and Sanikiluaq, the community where the lowest mercury concentrations were observed. In addition, the mean (± standard error (SE)) percent MeHg content of blue mussels was higher near Chisasibi (32% ± 1%) compared with three of the four other communities (Umiujaq, Sanikiluaq, and Inukjuak), where percent MeHg means ranged from 19% to 21% (Supplemental Fig. S1). The higher THg, MeHg, and percent MeHg of blue mussels near Chisasibi indicated greater mercury exposure near the base of the food web in that area of east James Bay.

Common eiders, which feed on blue mussels, showed a corresponding spatial pattern in THg concentration (Fig. 3). Both liver and breast muscle of eider near Chisasibi had approximately fivefold higher THg compared to the four communities in east Hudson Bay (one-way ANOVAs, F_4,76 = 23.4–33.2, p < 0.001). Concentrations of THg in eider were similar among the four Hudson Bay communities. For all birds, liver THg concentrations were higher than in muscle. Note that the THg of muscle is mainly in the form of MeHg while liver contains variable amounts of both inorganic mercury and MeHg (Chételat et al. 2020).

The higher THg concentrations of blue mussels near Chisasibi (mean ± SE = 0.28 ± 0.01 µg/g) were above median and mean values, though within the range, of published data reported in other geographic regions for this widely distributed species of marine bivalve. In the Estuary and Gulf of St. Lawrence in eastern Canada, THg concentrations of blue mussels ranged from 0.06 to 0.51 µg/g with a grand mean of 0.17 µg/g and a median of 0.16 µg/g (Cossa and Tabard 2020). At 20 coastal sites in the North Atlantic subarctic and Arctic, blue mussel THg concentrations were typically <0.1 µg/g and the maximum site mean was 0.23 µg/g (Jörundsdóttir et al. 2014). In the Baltic Sea, median THg concentrations of blue mussels were ≤0.08 µg/g though maximum values reached ∼0.3–0.5 µg/g (converted from wet weight values assuming 87% moisture content) (Dietz et al. 2021). Blue mussels on the coast of Massachusetts and Maine in the United States had mean THg concentrations 0.13 and 0.07 µg/g, respectively (Meattey et al. 2014). Blue mussels are widely used as biosentinels of coastal marine pollution around the globe and their mercury burdens reflect local environmental contamination, although other biological and environment factors such as growth and seasonal variability can also be influential (Beyer et al. 2017; Cossa and Tabard 2020).

For common eider, the THg concentrations in liver and muscle from Chisasibi are among the highest reported for the circumpolar subarctic and Arctic (Fig. 4). Published THg concentrations were available from 23 sites for common eider liver and from 12 sites for muscle (data sources are provided in Supplemental Table S3). Much of these data were compiled from previous reviews and spatial studies of mercury concentrations in common eiders (Braune et al. 1999; Mallory et al. 2004; Mallory et al. 2017). Published site means ranged from 0.15 to 5.81 µg/g for liver and from 0.06 to 1.71 µg/g for muscle. In comparison to those values, the eiders collected near Chisasibi had the highest mean THg concentration for liver (mean ± SE = 6.31 ± 0.61 µg/g) and the second highest mean for muscle (mean ± SE = 1.37 ± 0.12 µg/g). Thus, the regionally elevated mercury concentrations of those common eiders are also exceptional more broadly in the circumpolar Arctic. While biological factors such as sex can influence mercury concentrations of common eider, differences between males and females are small relative to the large spatial variation observed in this study (Provencher et al. 2016). The eiders were collected near Chisasibi during their period of residency on James Bay, and their mercury burdens likely reflect local dietary exposure. The pattern is consistent with the elevated mercury in blue mussels (an important diet item for eider) from the same study area. The mercury concentrations of liver and muscle were compared to established toxicity risk thresholds for birds (calculated using blood equivalent concentrations, see Ackerman et al. 2016), which suggested no to low risk of mercury toxicity to the eiders (Chastel et al. 2022).

Carbon, nitrogen, and sulfur isotope ratios indicated an influence of diet on the THg concentrations of common eiders. Only minor differences were observed in mean δ¹³C values of eiders, ranging from −20.2 to −19.1‰ among the five communities (Supplemental Fig. S2). Despite the small range of δ¹³C values, carbon isotope ratios were negatively correlated with log-transformed THg concentrations of eider (muscle: Pearson r = −0.35, p = 0.002; liver: Pearson r = −0.47, p ≤ 0.001; n = 77). Little variation was observed in the mean δ¹⁵N values of eiders among communities (10.3–11.3‰), indicating a similar trophic position among birds collected in the study. There was no association between eider δ¹⁵N values and their log-transformed THg concentrations (Pearson correlation, p ≥ 0.10, n = 77). A larger range of sulfur isotope ratios was observed, with eider mean values ranging from 12.9 to 17.2‰ between communities and a similar range of variation (∼5–8‰) within most collection areas (Supplemental Fig. S2). The sulfur isotope ratios were not correlated with log-transformed THg concentrations of eider liver or muscle (Pearson correlation, p ≥ 0.14, n = 77). Thus, only the dietary carbon source of eiders was associated with THg concentrations in their tissues.

The carbon isotope values of common eiders were consistent with diving sea ducks feeding on a benthic diet in the study region (Rohonczy et al. 2024). Blue mussels, which primarily filter organic matter from the water column, had a mean δ¹³C (± SD) of −22.7 ± 0.6‰ (n = 59), while sea urchin, which consume organic matter at the ocean floor, had a mean δ¹³C of −18.6 ± 0.9‰ (n = 19) (as reported in Rohonczy et al. 2024). Considering those two end members, the negative correlation between eider THg concentrations and carbon isotope ratios suggests higher exposure to mercury occurred through greater consumption of marine invertebrates supported by water column organic matter, such as filter feeders. Other Arctic marine food web studies have similarly found that greater feeding of pelagic-derived carbon was associated with higher mercury concentrations in consumers (Cardona‐Marek et al. 2009; Hilgendag et al. 2022; Rohonczy et al. 2024). There was a notable absence of a freshwater organic matter signal in the stable isotope content of eider muscle. The δ¹³C values (>−21‰) did not suggest high consumption of terrestrial organic matter from river inputs, which has lower values close to −27‰ (Godin et al. 2017). Likewise, the δ³⁴S values of eider muscle were >10‰, which are higher than values typically observed in freshwater biota (Chételat et al. 2020).

The elemental composition of blue mussels suggested environmental conditions varied among study sites. A PCA of tissue concentrations of 18 elements in blue mussels showed broadly similar burdens (as indicated by overlapping 95% confidence ellipses) for mussels near three Hudson Bay communities (Inukjuak, Umiujaq, and Sanikiluaq) and differences from mussels at Chisasibi and Kuujjuarapik (Supplemental Fig. S3). While many elements contributed to the PCA gradients in tissue concentrations, the THg of blue mussels was most strongly correlated with cobalt, iron, uranium, and vanadium (Pearson r = 0.67–0.71, p < 0.001, n = 59; Supplemental Fig. S4). Those elements were also strongly correlated with MeHg concentrations of blue mussels (Pearson r = 0.61–0.81, p < 0.001, n = 44). Iron was noteworthy in its strong association with THg and MeHg concentrations of blue mussels (Fig. 5). This co-variation of tissue element burdens provided an indication of possible environmental processes associated with greater mercury bioaccumulation in blue mussels. We hypothesize that the positive associations of mercury with iron and other crustal elements reflect the influence of river inputs on mussel exposure to metals.

Rivers transport vast amounts of materials to oceans, linking crustal weathering processes in terrestrial environments to ocean shelves (Andersen et al. 2007; Krachler and Krachler 2021). On a global scale, riverine sources of many elements (including cobalt, iron, vanadium, and uranium) are among the most important fluxes in ocean budgets (Andersen et al. 2007; Boyd and Ellwood 2010; Swanner et al. 2014; Schlesinger et al. 2017). River water contains organic matter and trace elements, which are distributed between dissolved, colloidal, and particulate phases. Estuaries are sinks for terrestrial organic matter and crustal elements transported by rivers because of particle settling and flocculation (Boyle et al. 1977; Demarty et al. 2021). The colloidal fraction of river water, rich in iron and carbon, is a dominant carrier of trace elements, including mercury (Jilbert et al. 2018; Demarty et al. 2021; Saniewska et al. 2022). When river water colloids interact with saline water in estuaries, flocculation occurs, causing a large fraction of trace metals and organic matter to be efficiently removed to sediments in the mixing zone (Jokinen et al. 2020; Khoo et al. 2022). Though few empirical measurements exist, high removal rates have been reported for riverine mercury entering estuaries due to flocculation and particle settling (Soerensen et al. 2016; Demarty et al. 2021; Saniewska et al. 2022). Model estimates suggest that 94% of mercury discharged from rivers is sequestered in estuary and coastal shelf sediments (Zhang et al. 2015). Together, the covariance of mercury with iron and other crustal elements in blue mussels across sites may be explained by exposure to river-transported elements that settle out on the coast. However, this hypothesis requires further study, and other factors could be influencing the patterns including differences in benthic substrates between Hudson Bay and James Bay, particularly the presence of tidal mud flats in James Bay.

Mercury isotope values of eiders from Chisasibi were different from other communities in the study area. The δ²⁰²Hg values of a subset of eider muscle samples (n = 25) from the five communities ranged from −0.29 to 0.85‰, and the Δ¹⁹⁹Hg values ranged from 0.42 to 2.22‰. On average, eiders from Chisasibi had significantly higher δ²⁰²Hg values than those from the other four communities (Fig. 6; one-way ANOVA, p < 0.001, n = 25, Holm’s p ≤ 0.044). Similarly, the Δ¹⁹⁹Hg values of Chisasibi eiders were significantly higher than those from three of the other communities (one-way ANOVA, p < 0.001, n = 25, Holm’s p ≤ 0.014 for Sanikiluaq, Kuujjuarapik, and Inukjuak). This spatial variation of mercury isotope ratios indicated different origins of bioaccumulated mercury in eiders from Chisasibi.

Variation in Δ¹⁹⁹Hg values is caused by mass-independent fractionation during photoreduction of inorganic mercury and photochemical breakdown of MeHg in aquatic environments (Bergquist and Blum 2007; Tsui et al. 2020). The slope of Δ²⁰¹Hg versus Δ¹⁹⁹Hg is commonly used to evaluate whether inorganic mercury or MeHg photochemistry is driving the odd mass-independent fractionation (Tsui et al. 2020). For eiders in the study area, the slope of 1.42 ± 0.06 (Supplemental Fig. S5; linear regression R² = 0.96, p < 0.001, n = 25) was close to the experimentally-derived slope of 1.36 ± 0.02 for fractionation due to MeHg photodemethylation (Bergquist and Blum 2007), indicating the higher Δ¹⁹⁹Hg values of Chisasibi eiders (mean ± SD = 1.84 ± 0.42‰) were likely due to a source of more photo-degraded MeHg. The average Δ¹⁹⁹Hg values of eiders from the other communities ranged from 0.86 to 1.26‰. However, three individual eiders from Umiujaq or Kuujjuarapik had higher Δ¹⁹⁹Hg values, suggesting exposure to more highly photo-degraded MeHg was not limited to the Chisasibi area (Supplemental Fig. S5). Published data for estuarine and coastal marine fish in other geographic areas typically have lower Δ¹⁹⁹Hg values between ∼0 and 1‰, which has been interpreted as minor photodemethylation of MeHg in coastal environments (Senn et al. 2010; Gehrke et al. 2011; Kwon et al. 2014; Rua-Ibarz et al. 2019; Tsui et al. 2020). The only published mercury isotope data for common eider (to our knowledge) are from estuarine food webs in the northeastern United states where Δ¹⁹⁹Hg values of the birds (0.68–0.82‰) were similar though slightly lower than eiders from our sites in east Hudson Bay (Kwon et al. 2014). In contrast, the higher Δ¹⁹⁹Hg values of Chisasibi eiders were comparable to freshwater fish of the Churchill River in subarctic eastern Canada (∼1–3‰), which had higher values than coastal marine fish downstream, presumably due to greater photodegradation of MeHg in the river system (Li et al. 2016).

The mercury isotope ratios of eiders indicated greater mercury bioaccumulation was associated with a distinct exposure pathway. The THg concentration of eider muscle was positively associated with its Δ¹⁹⁹Hg value; thus, eiders with higher mercury burdens were exposed to more highly photo-degraded MeHg (Fig. 7; linear regression R² = 0.61, p < 0.001, n = 25). Further, a negative correlation between the δ¹³C and Δ¹⁹⁹Hg values of eider muscle indicated that a diet supported by pelagic carbon led to greater exposure to photo-degraded MeHg (Fig. 8; linear regression R² = 0.45, p < 0.001, n = 25). Together, these results suggest that a source of more highly photodegraded MeHg in the water column contributed to elevated mercury in the common eiders.

Multiple lines of evidence suggest elevated mercury in the coastal food web of east James Bay near Chisasibi was due to mercury loading from the La Grande River. First, blue mussels and common eiders were collected within the coastal plume of the La Grande River (Supplemental Fig. S6), a large system that discharges up to 5 000 m³/s of water to James Bay. The impact of this river discharge on seawater salinity is evident up to ∼100 km north along the coast during winter and ∼60 km during summer (Peck et al. 2022). Second, the mercury burden of blue mussels was positively associated with crustal elements (iron, vanadium, cobalt, and uranium), suggesting a link between river export and coastal accumulation. Finally, the mercury stable isotopes of eiders from James Bay near Chisasibi were distinct from eiders in Hudson Bay and the isotopic signature was consistent with photodegraded MeHg of freshwater origin (Li et al. 2016).

We hypothesize that riverine MeHg is entering the coastal food web via uptake by suspension feeding invertebrates consuming MeHg of freshwater origin. Methylmercury concentrations of water in the La Grande River (0.07–0.23 ng/L) are relatively low in comparison to other rivers draining into east James Bay (Fink‐Mercier et al. 2022). However, the La Grande River exports the largest annual load of MeHg to east James Bay, an estimated 5.79 kg/year (Fink‐Mercier et al. 2022), which likely settles along the coast due to flocculation in saline waters (Demarty et al. 2021). This annual MeHg export from the La Grande River is also much higher than estimates for major rivers in east Hudson Bay (Great Whale River and Innuksuac River) (Sonke et al. 2018). Methylmercury in water of the La Grande River is predominately in the dissolved fraction (Fink‐Mercier et al. 2022), and therefore transport of particulate MeHg is likely less important. In addition, associations of pelagic feeding (based on δ¹³C) with THg concentrations and Hg isotope values of eiders suggest that the mercury originated from the water column rather than sediment. Estuarine sediments can be sites of high MeHg production, though the relative importance of sediment versus water column sources of MeHg varies greatly among sites (Chen et al. 2014; Jonsson et al. 2014; Kwon et al. 2014; Taylor et al. 2019; Jonsson et al. 2022). It is possible that environmental characteristics of sediment in James Bay may be more amenable to MeHg production than in Hudson Bay. It is also possible that in situ water column methylation along the James Bay coast may be stimulated by inputs of terrestrial organic matter and nutrients, as observed in another boreal subarctic system (Schartup et al. 2015). Further research is needed to characterize coastal processes leading to enhanced mercury bioaccumulation in the marine food web of east James Bay. In addition, more detailed measurements are needed to determine the spatial extent of enhanced mercury bioaccumulation on the James Bay coast and the degree of exposure to other marine species connected to the pelagic food web.

HydroQuebec has implemented an extensive mercury monitoring program of the La Grande hydroelectric complex for decades (Bilodeau et al. 2017), though downstream effects were primarily considered in the context of a sequential system of reservoirs and the river. To our knowledge, few baseline data were collected for levels of MeHg in the coastal marine food web near Chisasibi during the initial construction and operation of reservoirs in the 1970s–1990s (Lucotte et al. 1999). It was assumed that the export of mercury to the James Bay coast was limited and localized (Schetagne and Verdon 1999), though pre-1996 data for fourhorn sculpin (Myoxocephalus quadricornis) collected within the coastal summer plume of the La Grande River had THg concentrations (0.31–0.55 µg/g wet weight, n = 106) that were several times higher than sculpin at sites outside the plume (0.10–0.22 µg/g wet weight, n = 60) (Schetagne and Verdon 1999). Beyond those observations, no long-term monitoring of the potential impact of reservoir discharge to the marine environment was developed.