Ebullition rates and mercury concentrations in St. Lawrence river sediments and a benthic invertebrate


  • N. Roxanna Razavi,

    1. Department of Biology, Queen's University, Kingston, Ontario, Canada
    Search for more papers by this author
  • Jeffrey J. Ridal,

    1. School of Environmental Studies, Queen's University, Kingston, Ontario, Canada
    2. St. Lawrence River Institute of Environmental Sciences, Cornwall, Ontario, Canada
    Search for more papers by this author
  • William de Wit,

    1. St. Lawrence River Institute of Environmental Sciences, Cornwall, Ontario, Canada
    Search for more papers by this author
  • M. Brian C. Hickey,

    1. School of Environmental Studies, Queen's University, Kingston, Ontario, Canada
    2. St. Lawrence River Institute of Environmental Sciences, Cornwall, Ontario, Canada
    Search for more papers by this author
  • Linda M. Campbell,

    1. Department of Biology, Queen's University, Kingston, Ontario, Canada
    2. School of Environmental Studies, Queen's University, Kingston, Ontario, Canada
    3. Environmental Science, Saint Mary's University, Halifax, Nova Scotia, Canada
    Search for more papers by this author
  • Peter V. Hodson

    Corresponding author
    1. Department of Biology, Queen's University, Kingston, Ontario, Canada
    2. School of Environmental Studies, Queen's University, Kingston, Ontario, Canada
    • Department of Biology, Queen's University, Kingston, Ontario, Canada
    Search for more papers by this author


Ebullition, the release of gas from anaerobic decomposition in sediments, was recorded in a mercury-contaminated depositional zone (Zone 1) of the St. Lawrence River Area of Concern in Cornwall, Ontario, Canada. The aim of the present study was to test if this disturbance affected the bioavailability of total mercury (THg) and methylmercury (MeHg) in surficial sediments to a benthic invertebrate (Echinogammarus ischnus). Ebullition rates ranged from <1 to 2,800 ml/m2 daily, with methane gas comprising 29 to 84% of the total. No direct effects of ebullition were found on either abiotic (sediment or pore water THg or MeHg concentrations) or biotic (amphipod THg or MeHg concentrations) variables measured. Instead, amphipod MeHg concentrations were best predicted by pore water THg and MeHg concentrations, organic matter of surficial sediments, and water depth and location. Trend surface analyses demonstrated that a shallow, southwestern part of Zone 1 was most contaminated with pore water mercury, which decreased in a gradient toward the northeast. Further study is needed to determine if the amount of sediment resuspended by ebullition affects the spatial distribution of mercury. Environ. Toxicol. Chem. 2013;32:857–865. © 2013 SETAC


Mercury (Hg) inputs from industries have left legacy contamination in freshwater sediments of rivers, lakes, and reservoirs. The biomagnifying properties of neurotoxic methylmercury (MeHg) threaten human and ecosystem health 1, 2. To improve risk assessments at these sites, the processes that facilitate Hg transport out of historically contaminated sediments must be understood. Sediments can act as a source of contaminants through transport processes (e.g., resuspension, diffusion, bioturbation) that can elevate contaminant exposure to biota, especially in depositional environments 3. Disturbance events can alter sediment biogeochemical conditions that control Hg release; for example, methylation potential increased in sediments mixed by tidal resuspension 4, although these consequences are not yet fully understood 5.

Ebullition is the release of gas that is formed in organic sediments undergoing anaerobic decomposition 6. Ebullition may act as a transport mechanism for contaminants through sediment resuspension or bubble migration that increases pore water circulation 7. Gas bubbles can transport Hg(0) rapidly through sediments 8, and solute release of Hg was best explained by ebullition, not diffusion 9. Despite these indications that Hg availability from resuspended sediment and pore water can be elevated due to mechanical disturbance and mixing, very few studies have measured whether in situ Hg concentrations in the lower food web (i.e., benthos) are increased by MeHg released from contaminated sediments undergoing ebullition.

At the St. Lawrence River at Cornwall, Ontario, Canada, a Great Lakes Area of Concern, acoustic mapping revealed extensive ebullition from Hg-contaminated sediments in one depositional zone (Zone 1, Fig. 1), downstream from a pulp and paper mill, now closed. Ebullition was much less evident in other zones farther downstream. The gas is likely generated from the decomposition of buried wood chips 10, with methane production stimulated by highly anoxic conditions 11. The Hg concentrations in biota from Zone 1 are also significantly higher than in other zones 12, despite apparently lower mean Hg sediment concentrations 13, 14. Heterogeneous Hg concentrations in fish within Zone 1 at a fine-scale resolution 15 suggest that an internal source or some distinct aspect of Zone 1 may be contributing to the higher rate of Hg transfer from sediments to the food web. The only previous measurement of ebullition in Zone 1 found negligible gaseous Hg released to the atmosphere 11. The present study is the first to investigate the possibility that in situ ebullition may be disturbing the accumulation of sediments and enhancing the availability of Hg to benthos from contaminated pore water and sediments. A parallel study characterized the sediment–water interface in Zone 1 and assessed differences in Hg biogeochemistry due to ebullition (M. Fathi, 2009, Master's thesis, University of Ottawa, Ottawa, ON, Canada). Our main objective was to compare Hg concentrations in amphipods (Echinogammarus ischnus) to ebullition rates and to Hg concentrations in surficial sediment and pore water to determine if ebullition is a potential driver for Hg accumulation. Water depth and bulk sediment characteristics are hypothesized to have no effect on Hg concentrations in the abiotic or biotic compartments measured or on ebullition rates.

Figure 1.

The St. Lawrence River at Cornwall, Ontario, Canada. The study area is highlighted in gray.


Study area

Mercury contamination of St. Lawrence River sediments near Cornwall is the result of excessive inputs by local industries 13, 14 including a chlor-alkali plant (ICI Forest Products, formerly Canadian Industries Limited, and Cornwall Chemicals), a textile mill (Courtaulds), and a pulp and paper mill (Domtar Fine Papers). Modifications to river morphology for navigation created depositional zones downstream of old industrial discharges. These are now considered Hg hot spots as reduced water flows allow the settling of particulate matter 10. Mean sediment Hg concentrations in these hot spots have regularly exceeded the province of Ontario's Lowest Effect and Severe Effect Levels of 200 and 2,000 ng/g dry weight, respectively 14, 16, 17. In depositional zones, the distribution of total Hg (THg) concentrations in surface sediments shows that older, deeper sediments are more contaminated but that MeHg concentrations remain elevated compared to preindustrial sediments 18. The availability of MeHg to the food web in these hot spots is reflected in higher Hg concentrations in benthos and fish compared to reference sites 18. Zone 1 fish have significantly higher Hg concentrations compared to other contaminated zones, coincident with greater pore water MeHg availability 18. Extensive methane production was only observed in Zone 1 and could explain elevated Zone 1 fish Hg concentrations. Since the extent and rate of ebullition within Zone 1 were unknown and sediments were highly heterogeneous 10, a random sampling design was employed.

Sample collection

Zone 1 was divided into a grid of 100 squares (8 × 8 m). Ten of these grid squares were selected randomly before every sampling event, which occurred four times over the 2007 field season (June 14–26, July 17–31, August 9–22, and September 9–22). No square was sampled more than once for the entire study. All study squares were characterized by location, water depth, ebullition rate, and amphipod abundance. Physicochemical parameters of the overlying water, such as pH (8.4–8.6), dissolved oxygen (8.5–9.7 mg O2/L), and temperature (18–23°C), were stable throughout the water column (complete mixing assumed 18) and across Zone 1 (outside of the near-shore vegetated areas) where sampling took place. Reducing conditions (–116 to –213 mV) were measured in surface sediments 19. Zone 1 sediment pore water concentration profiles for sulfide, sulfate, and reduced iron showed that redox processes did not affect the dissolved MeHg concentrations at the sediment–water interface (M. Fathi, 2009, Master's thesis, University of Ottawa, Ottawa, ON, Canada).

Amphipod collection

Because of their close association with sediments, relatively limited mobility, and importance as a food source for fish, benthic invertebrates are ideal for testing the effects of localized disturbance on Hg availability. Amphipods are robust to handling and frequently used for biomonitoring worldwide. Amphipods were the most consistent prey item of Zone 1 juvenile yellow perch (56% of 330 analyzed stomach contents), and THg concentrations in amphipods collected from perch stomachs were significantly correlated with yellow perch THg (log perch THg = 0.39 log amphipod THg + 2.15, n = 19 r2adj = 0.22, p = 0.04; L.E. Yanch, 2007, Master's thesis, Queen's University, Kingston, ON, Canada). Artificial substrates were used to collect amphipods to avoid possible recontamination of samples associated with dredging. Gravel rock baskets, 20 × 20 × 20 cm, were used as artificial substrates as they are readily colonized by amphipods in the St. Lawrence River (F.C.J. van Herpen, 2006, Master's thesis, Wageningen University, Wageningen, The Netherlands); basket materials were not a source of Hg (Supplemental Data). Though two other species of amphipods are found in this area (Gammarus fasciatus and Hyalella azteca), baskets discriminated for the invasive E. ischnus (Supplemental Data). Echinogammarus ischnus is tolerant to a wide range of habitats but in the upper St. Lawrence River was found to be associated primarily with gravel-sized sediment 20. In each square sampled, two replicate baskets were anchored to one cement block and left for a two-week period to ensure that sufficient biomass was collected for tissue Hg analyses. After retrieval, baskets were dismantled immediately, amphipods were enumerated as a measure of abundance and placed in foil over ice, and each sample was stored at –20°C for Hg analyses.

Gas collection and analysis

The gas collector employed was adapted from Huttunen et al. 21. An 80-cm-long polyvinylchloride pipe of 2.5 cm diameter overlapped the narrow end of a 30-cm-diameter plastic, down-facing funnel. The collector was suspended from a float approximately 1 m above the sediment surface and kept upright by a weight. A 60-ml syringe was fitted to the upper end of the polyvinylchloride pipe to collect and measure gas volume. On average, volume was measured every other day over the period of deployment. Five milliliters of gas were taken whenever sufficient volume was available, and the methane content was analyzed by gas chromatography (GC Varian 3300, stainless steel column packed with Haysep Q Mesh). One milliliter of sample was used for each run, blanked after each duplicate, and a standard gas (1.99% CO2, 38% N2, and 60.01% CH4) was analyzed every 10 samples.

Sediment and pore waters

Triplicate sediment cores were taken at each site using a gravity corer 9, and the top 5 cm of each core was collected directly into a nitrogen-purged bag and mixed manually. Ideally, we would have sectioned at a finer scale (1 cm), but this was not feasible given our sample size and requirements for pore water Hg analysis. All sediments and pore waters were subsequently handled in a glove box under nitrogen atmosphere. Sediment subsamples were centrifuged for 15 min at 1,575 g to separate the solid and aqueous phases. The solid phase was preserved immediately by freezing. The pore water (aqueous phase) was centrifuged at 1,575 g for 5 min before passing through a nitrogen-purged Whatman syringe filter (25 mm GD/X, polyethersulfone membrane, 0.45 µm pore size). Sufficient volume was collected for duplicate analyses. Pore water was preserved for THg (1% BrCl) and MeHg (0.5% HCl) analysis and refrigerated at 4°C. Sediment bulk density (g/cm3), percentage of water content, and loss-on-ignition (LOI) were also determined (Table 1) 9, 22. Sediment–pore water distribution coefficients (Kd, liters per kilogram) were derived from the ratio of sediment Hg (nanograms per kilogram) to pore water Hg (nanograms per liter) for both THg and MeHg.

Table 1. Range of values for environmental variables measured as site descriptors in Zone 1 (St. Lawrence River at Cornwall, ON, Canada) between June and September 2007
Measured parameterRangeArithmetic meanGeometric meann
Water temperature (°C)18–22.720.54
Water depth (m)5.5–118.938
Bulk density (g/cm3)0.46–2.551.3823
% water content41.6–80.360.126
% loss-on-ignition (LOI)5.5–4316.625
Bubbling rate (ml/m2 daily)>1–2,80079037
Amphipod abundance (N/baskets)9–81610937

Mercury analyses

Frozen amphipods were dried at 50°C for 18 to 24 h and homogenized. Amphipods were analyzed in replicates (samples from one homogenate) for Hg (nanograms THg per gram dry wt) using a Nippon Automated Mercury Analyser SP-3D (Nippon Instruments), with a detection limit of 0.01 ng and an average coefficient of variation or precision of 15.9% (range 1.9–48.5%, n = 39). Following 10 runs, a standard was analyzed and blanks were added after highly contaminated samples to ensure the removal of any Hg residues. After each analytical session, a certified reference material (CRM; DOLT-3 from the National Research Council of Canada) was analyzed. The average THg in DOLT-3 samples was 3.33 ± 0.64 µg/g (n = 6), 98.7% of the expected 3.37 ± 0.14 µg/g. Methylmercury in amphipods was analyzed using capillary GC coupled with atomic fluorescence spectrometry (Hewlett Packard, GC model number HP 6890 series with HP 7683 injector) at a method detection limit of 0.02 ng/L 23. The average precision was 9.5% (range, 1.4–30.1%, n = 39), and the average MeHg and standard deviation of DOLT-3 was 1.49 ± 0.58 (n = 5), 93.7% of the expected 1.59 ± 0.12 µg/g.

Sediment and pore water THg and MeHg concentrations were analyzed following methods described in Delongchamp et al. 24. The sediment THg coefficient of variation was 6.9% (range, 0.4–14%, n = 35), and CRM recovery (MESS-3 from the National Research Council of Canada) was 91 ± 13 (n = 19), 99% of the expected 91 ± 9 µg/g. Sediment MeHg precision was 12.1% (range, 0.7–47.5%, n = 23 pairs). The recovery of spiked samples was 91.3% for MeHg using the IAEA-405 CRM (n = 5, estuarine sediment; International Atomic Energy Agency) and 105.3% using the ERM-CC580 CRM (n = 4, sediment; Institute for Reference Materials and Measurements). Procedural blanks for pore water analyses revealed no contamination during THg or MeHg extraction or analysis, and mean recoveries of spiked water samples were always above 95%.

Statistical analyses

Data were analyzed using the statistical package JMP 7.0 (SAS Institute). Regression analyses were performed to assess covariation among measured environmental variables. The random sampling design removed the possibility of bias created by earlier sampling events but had the disadvantage of making direct temporal comparisons invalid. Nevertheless, we report any temporal trends observed (using analysis of variance [ANOVA]) because we believe they are biologically relevant. Data for two squares were removed as outliers from all analyses. These squares were in shallow areas (3 m) where the habitat (dense vegetation) was considerably different from that of the other sampling squares in deeper water (5.5–10 m), where vegetation was very sparse or absent. Transformations were used to ensure the normality of residuals from the regression, and model II regressions were fitted to all graphs. Null hypotheses were rejected at the 0.05 alpha level. Amphipod MeHg concentrations were predicted from multiple linear regression models, and models were compared by Aikake's information criterion (Supplemental Data). While correlations were strong, they do not prove causal relationships.

Surface representation and analysis


To represent the data spatially, we used inverse distance weighting, which estimates the values at unsampled locations (Z) by the equation

equation image

where Z(xj) are the data points and dij is the distance between each point and a point at an unsampled location. Inverse distance weighting calculates a local mean by giving more weight to measured data that are near than those that are further away, assuming that proximity is a gauge of similarity 25. Inverse distance weighting is a useful screening technique to obtain a rough contour map of Hg concentrations with small sample sizes because of its low computational complexity and ease of implementation, although it does not estimate interpolation error 25.

Trend surface analysis

To assess whether our data contained any major feature or trend, trend surface analysis was applied. A trend is a large-scale systematic change occurring predictably from one end of the study zone to the other 25. Trend surface analysis generates a surface based on a polynomial expansion using locational coordinates (X, Y) as the independent variables. The coefficients of the polynomial function were calculated by the SPSS package (SPSS Inc.) and used to determine the height (Z) of the surface or the dependent variable as follows

equation image

where r and s are the polynomial coefficients, p is the order of the trend surface, and brs is the constant. This technique has merit mostly as an exploratory tool 25 and provided a rough generalization about the shape of our field.



Ebullition rates ranged from <1 to approximately 2,800 ml/m2 daily, with arithmetic median and mean rates of approximately 660 and 790 ml/m2 daily, respectively. The records did not follow a gaussian distribution: 32 of the 37 records were <1,500 ml/m2 daily and nearly half of those were <500 ml/m2 daily; a square root transformation of ebullition rates was necessary to meet assumptions of normality.

Percentage of CH4 in gas ranged from 29 to 84% and increased with increasing ebullition rates. The relationship was significant when an outlier was removed (%CH4 = 0.87√ebullition + 37.8, r2 = 0.37, n = 22, p = 0.0028). Ebullition rates decreased significantly with increasing water depth (p = 0.0153; Table 2).

Table 2. Results of regression analyses, with slope direction from model II regressiona
Response variablePredictor variablenDirection of sloper2p
  • a

    Significance level at α = 0.05. Comparisons of bulk sediment characteristics (i.e., % water content, % loss-on-ignition, and bulk density) to pore water MeHg concentrations were not possible because those parameters were not collected in June and pore water MeHg was above the detection limit in only two samples in all other months.

  • MeHg = methylmercury; THg = total mercury; LOI = loss-on-ignition.

Log amphipod MeHgWater depth23Negative0.340.0035
Log pore water THg21Positive0.260.0190
Log pore water MeHg8Positive0.680.0113
Log %LOI14Negative0.320.0351
Log amphipod THgLog pore water THg34Positive0.130.0386
√Ebullition rateWater depth37Negative0.160.0153
Log abundanceWater depth37Positive0.160.0148
√Ebullition rate36Negative0.120.0380
Log pore water MeHgWater depth9Negative0.540.0246
Log pore water THgLog sediment THg34Positive0.190.0106

Mercury in sediments and pore water

All surficial sediment Hg concentrations (Table 3) exceeded the Provincial Sediment Quality Guidelines for the Lowest Effect Level of 200 ng/g THg dry weight 17. The Severe Effect Level of 2,000 ng/g was exceeded in 69% of the samples. The proportion of MeHg in sediments was always below 1% (Table 3); sediment THg and MeHg concentrations showed a significant and positive relationship (log sediment MeHg = 0.69 log sediment THg - 1.64, r2 = 0.42, n = 21, p = 0.0016). Pore water concentrations (Table 3) of THg were significantly correlated to sediment THg (p = 0.0106; Table 2). Pore water MeHg concentrations were significantly and negatively correlated with water depth (p = 0.0246; Table 2 and Fig. 2A) but not with sediment THg or MeHg. Values of log Kd ranged from 3.5 to 5.6 for THg (median 5.3) and ranged from 2.3 to 3.3 for MeHg (median 2.9). No significant relationships were found between log Kd and ebullition rate, water depth, sediment organic matter (i.e., %LOI), or other sediment bulk characteristics for either THg or MeHg.

Table 3. Geometric means, ranges and sample sizes (n) of mercury concentrations in amphipods, sediments, and pore water of Zone 1 (St. Lawrence River at Cornwall, ON)a
 THgMeHg% MeHgb
MatrixGeometric meanRangenGeometric meanRangenGeometric meanRangen
  • a

    Sediment and pore water concentrations are from surficial sediments (0–5 cm).

  • b

    Arithmetic means.

  • THg = total mercury; MeHg = methylmercury.

Amphipods (ng/g dw)172.960.2–596.13863.427.6–146.92341.29.6–85.123
Sediments (ng/g dw)5000768–17,911348.32.7–18.8210.20.04–0.621
Pore water (ng/L)72.54.1–944.1349.32.6–22.99142–5821
Figure 2.

(A) Amphipod methylmercury (MeHg) concentrations versus water depth, and pore water MeHg concentrations versus water depth. (B) Amphipod MeHg concentrations versus pore water total mercury (THg) concentrations, and amphipod MeHg concentrations versus pore water MeHg concentrations. (C) Amphipod MeHg concentrations versus log percentage of loss-on-ignition (%LOI). Percentage of LOI was not sampled in June, so the associated amphipod MeHg concentrations were excluded. See Table 2 for n, r2, and p values.

Mercury in amphipods

Amphipod THg was only significantly positively correlated to pore water THg (p = 0.0386; Table 2). In contrast, amphipod MeHg was positively correlated to pore water THg and MeHg concentrations and negatively correlated to water depth and organic matter (%LOI; Table 2). The variance explained by pore water THg concentrations was greater for amphipod MeHg than amphipod THg concentrations (p = 0.0190 and p = 0.00386, respectively; Table 2). Variation in amphipod MeHg concentrations was best explained by pore water MeHg concentrations (p = 0.0113; Fig. 2B). However, this regression was based on a small sample size (n = 8) because all the samples of pore water taken in July and 80% of the samples taken in August were below detection limits for MeHg. Amphipod MeHg decreased significantly with water depth (p = 0.0035; Table 2 and Fig. 2A) and sediment carbon content (%LOI, p = 0.0351; Table 2 and Fig. 2C). Sediment THg and MeHg concentrations and ebullition rates did not explain the variation in amphipod THg or MeHg concentrations.

Increasing ebullition rates corresponded to significantly fewer amphipods collected (p = 0.0380; Table 2), while water depth showed a significant positive correlation with amphipod abundance (p = 0.0148; Table 2).

Spatial trends

The trends of ebullition rates and amphipod and pore water MeHg concentrations with water depth described above suggested a spatial component to MeHg availability. Model surfaces created using inverse distance weighting showed similar distributions of amphipod MeHg and pore water MeHg concentrations (Fig. 3). Trends in water depth and amphipod and pore water MeHg concentrations were confirmed with regression models using latitude and longitude as predictor variables (Table 4). For instance, 53% of the variation in amphipod MeHg concentrations was explained by the location variables in the linear trend surface equation (Table 4). Variations in pore water THg and MeHg concentrations were also explained by the linear trend surface equation (38 and 93%, respectively). The trend surface showed a decrease in MeHg concentrations in amphipods and pore waters from the southwest to the northeast of the zone (Table 4 and Fig. 3). A decrease in pore water THg concentrations was observed following a west-east gradient only (Table 4 and Fig. 3). These patterns may be a reflection of the bathymetry of Zone 1 as variance in water depth was also significantly explained by the linear trend surface equation (Table 4). The sampling squares toward the eastern part of the study area were deeper than the squares where the higher Hg concentrations in amphipods and pore water were observed. No other variables showed the same spatial patterns as those observed for amphipod MeHg, pore water Hg, and water depth.

Figure 3.

Zone 1 depth profile and study area outlined in black (image modified from Biberhofer and Rukavina 10). Surface representations of ebullition rates (milliliters per meter squared daily), pore water total mercury (THg) and methylmercury (MeHg) concentrations (nanograms per liter), and amphipod THg and MeHg concentrations (nanograms per gram dry wt) were created using inverse distance weighting (see text for details).

Table 4. Summary of regression models and coefficients using geographic location, namely, Easting (18T521158m E to 18T521428m E) and Northing (4984191m N to 4984403m N), as independent variablesa
RegressionRr2SSdfMean squareFpModelUnstandardized coefficientsp
SlopeStandard error
  • a

    The general equation for the regression models is y = constant + (B1 × Easting) + (B2 × Northing).

  • MeHg = methylmercury; THg = total mercury.

MeHg amphipod0.730.538,31424,15711.10.001Constant−63,18955,3330.267
       Easting (B1)−
       Northing (B2)
THg pore water0.620.38329,0952164,5479.80.000Constant550,805363,5210.140
       Easting (B1)−1.490.340.000
       Northing (B2)
MeHg pore water0.970.93352217640.40.000Constant−23,4386,3060.010
       Easting (B1)−
       Northing (B2)0.010.0010.000
Water depth0.620.3921.7210.911.20.000Constant2,4162,7070.378
       Easting (B1)0.010.0020.000
       Northing (B2)−0.0010.0010.008

Temporal trends

Although this was not explicitly tested by our study design as described in Materials and Methods, we observed highest MeHg concentrations in amphipods in June compared to July and August (ANOVA, F ratio = 8.45, p = 0.002, n = 23). As mentioned above, pore water MeHg concentrations were highest in June, and we also observed significantly higher pore water THg in June and August than July and September (ANOVA, F ratio = 8.29, p = 0.0004, n = 34). These differences were also reflected in the log Kd values for THg (significantly higher in June; ANOVA, F ratio = 11.48, p < 0.001, n = 34) and MeHg (significantly higher in June, t test, t ratio = 3.90, p = 0.08, n = 8).

Multiple linear regression models

Several models to explain amphipod MeHg were biologically important based on the Akaike information criterion with correction, but only the single variable model using water depth as a predictor variable was within the confidence set based on Akaike weights (Supplemental Data).


We investigated whether ebullition was associated with contaminated surficial sediments and pore waters and with increased Hg exposure of a benthic invertebrate (E. ischnus). Variable rates of ebullition were not correlated directly with Hg concentrations in abiotic or biotic compartments. Ebullition rates were correlated inversely with water depth and with amphipod abundance, which may indicate that amphipods prefer deeper water or avoid high ebullition rates. Spatial predictors of Hg in amphipods were apparent (water depth and location within zone), and correlations with pore waters Hg (both THg and MeHg) and organic matter of sediments were also found. Future work is proposed to test other possible indirect effects of ebullition.

Factors affecting ebullition rates

Previous work measured a total gas flux rate in Zone 1 of approximately 89 ml/m2 daily (n = 1 11), falling within the rates recorded in the present study (<1 to approximately 2,800 ml/m2 daily, n = 37). This broad range demonstrates that many measurements are needed to characterize flux within even a relatively small spatial scale (6,400 m2). The variability in rates was not surprising given the differences in grain size, organic matter content, and bottom type within Zone 1 10 as well as the range of water depths over which we sampled. Hydrostatic pressure along with organic carbon and sediment strength regulate bubble formation. In the present study, ebullition rate decreased significantly with increasing water depth. Similarly, in the Penobscot River in Maine, USA, ebullition from sediments rich in organic matter was maximal above 6 m water depth 26. We did not find a correlation between organic content of the surficial sediments and ebullition rates, despite the large range in carbon content (5–43%). Freshwater river sediments did not exhibit ebullition when carbon content averaged <1% 26, so it is possible that the relatively high carbon content (average = 16.6%) in sediments masked the expected trend. The lack of a clear relationship between ebullition rate and organic content in surficial sediments illustrates that the material being degraded is not always found in the upper 5 cm of sediments. The most likely sources of material undergoing degradation are the wood chips, of which we saw visible differences in the quantity and size, even within replicate cores. We and others 24 observed that the amount of accumulated sediments on top of these deposits was not always consistent (from 0 to more than 10 cm). These variations in the surficial sediment cover suggest the potential resuspension of particulates by ebullition and different rates of net sedimentation within Zone 1. Hence, older, more contaminated sediments are not always out of reach of surface-dwelling animals.

Factors correlated with amphipod mercury

We were interested in testing whether ebullition caused increased Hg in surficial sediments and pore waters and, thus, elevated exposure to a benthic receptor. Ebullition can release pore waters from sediments to overlying water in a ratio of 1:1 with gas release and can be important in solute release at high ebullition rates 27. However, we found that ebullition had no significant relationship with either possible vector of Hg to amphipods (i.e., pore water or sediment) and no significant relationship with amphipod THg or MeHg. Similarly, the bioavailability of THg and MeHg to clams did not appear to increase under laboratory mesocosm conditions mimicking tidal resuspension 28. Recent work highlights the role of methanogens in Hg methylation 29, so a link between MeHg concentrations in pore water was expected with increasing ebullition rates. However, experiments using Zone 1 sediments showed that methanogens were also active demethylators and may explain why no increase in methylation was found at sites of high ebullition 30. The effect of ebullition may not be visible in the top 5 cm of sediment we sampled as the active sediment layer (where methylation occurs) was found to increase to 12 to 15 cm in sediments undergoing tidal resuspension versus 3 to 5 cm in depositional areas 4. Ebullition rates were inversely correlated with amphipod abundance, which suggests that ebullition may be acting indirectly on Hg uptake in amphipods by controlling their distribution. However, this possibility is confounded by the trend of both ebullition and amphipod abundance with water depth.

In Zone 1, maximum sediment THg concentrations were much higher than previously reported, which we attributed to the larger sample size of the present study compared to previous ones (n = 34 vs n = 1–6 13). We did not find a relationship between bulk sediment Hg and amphipod Hg concentrations. Nevertheless, our mean THg concentrations in amphipods were approximately twice the average found in an upstream reference zone (L.E. Yanch, 2007, Master's thesis, Queen's University, Kingston, ON, Canada). Amphipods also had a range of %MeHg (10–85%) comparable in magnitude to 34 to 67% found in Gammarus sp. in the Idrijca River, one of the most highly Hg-contaminated rivers in the world 31. Carnivory (and cannibalism), known to be high in E. ischnus (http://nas.er.usgs.gov/queries/factsheet.aspx?SpeciesID=23), could account for the higher %MeHg observed. Since 95% of Hg in fish is MeHg, these proportions are consistent with a trophic transfer of MeHg from amphipods to fish in Zone 1, suggested previously by a strong relationship between yellow perch THg concentrations and THg in amphipods collected from perch gut contents 18.

Amphipods such as E. ischnus are epibenthic. Under stable sediment conditions they should not be directly exposed to contaminated pore water 32. However, we found that pore water Hg (both THg and MeHg) concentrations were among the best predictors of MeHg concentrations in amphipods. Others have found evidence that Hg is quickly remobilized into pore water in this system and elevated pore water MeHg bioavailability in Zone 1 19, 24. However, we did not measure all potential MeHg exposure routes to E. ischnus, such as zebra mussel deposits that they consume. Food, in the form of fresh organic matter, is often the largest determinant of Hg uptake, although under certain conditions (i.e., the absence of fresh algal food), exposure from Hg in pore water increased in an estuarine amphipod 33. Characterizing the exposure of benthos to Hg in situ is a challenge: exposure routes of Hg are numerous and require equal consideration 33. Future work is needed using approaches such as biodynamic modeling 34, which takes into account the uptake and efflux rates, assimilation efficiency, as well as ingestion and growth rate constants to predict accumulated concentrations in a given species.

In addition to pore water, organic matter (i.e., %LOI) of surficial sediments and spatial determinants were predictors of E. ischnus MeHg concentrations. Sediment organic matter is known to limit MeHg bioaccumulation from particles to benthic invertebrates 35. Despite the control that organic matter can also have on MeHg partitioning into pore water 36, we did not find the same spatial trends in organic matter to account for the spatial trends in amphipod MeHg and pore water Hg. The spatial trends could indicate that the muddy and shallower sediments in the western part of Zone 1 have more bioavailable Hg. Previous work 19 from this system showed that Hg speciation and partitioning in surface sediments are strongly influenced by formation of organic sulfur compounds and that these chemodynamics may largely determine Hg bioavailability at the sediment–water interface.

Future work

This experiment indicates no direct correlations between ebullition and Hg in surficial sediments and pore water. An analysis of Hg cycling within Zone 1 surface sediments showed that diffusion of THg and MeHg at the sediment–water interface was minimal and did not differ between a location expected to have high ebullition and a control site (M. Fathi, 2009, Master's thesis, University of Ottawa, Ottawa, ON, Canada). Still, uncertainties remain about how resuspended particulates are affected by ebullition in situ and how these processes may undergo seasonal changes.

Resuspension can increase significantly as a result of ebullition 37. A study is needed in Zone 1 to determine the quantity and frequency, as well as the spatial distribution, of resuspended particulates associated with ebullition. The process of redistribution of sediments by resuspension of particulates from wind and wave action is termed sediment focusing. In Zone 1, patterns of water flow (back eddies) may bring resuspended particles from areas of ebullition to shallower parts of the zone where currents are slower, thus facilitating a transfer of contaminated particulates or dissolved forms of Hg resident in amphipods to the southwest area. Significant relationships were observed elsewhere between water depth and sediment, as well as surficial pore water MeHg concentrations, which was attributed to sediment focusing 38. As the amount of sediment resuspended is affected by bubble size, bubbling frequency, and the duration of an ebullition event 37, this remains an important research goal.

Finally, there are seasonal variations in amphipod feeding, the activity of methylating sulfate-reducing bacteria, and ebullition rates 6, 26. We found that MeHg concentrations in amphipods were highest in June compared to July and August, mirroring the same trend in amphipods collected from yellow perch stomachs 18, pore water MeHg concentrations, and sediment–pore water distribution coefficients. The period of highest pore water MeHg concentrations most likely reflected favorable surface sediment redox conditions including higher concentrations of pore water sulfate. Sulfate was rapidly consumed in Zone 1 sediments in June and was not detectable in July and August (M. Fahti, 2009, Master's thesis, University of Ottawa, Ottawa, ON, Canada). Clearly, seasonal effects are pronounced in this system and deserve additional study.


No direct links between rates of ebullition and Hg concentrations in surficial sediments, pore water, and amphipods were observed. An indirect link between ebullition rates and amphipod abundance was found that may affect amphipod exposure to Hg. Amphipod MeHg concentrations were best explained by pore water THg and MeHg concentrations, sediment organic matter, water depth, and location. We recommend that future research examine all possible effects of ebullition, including sediment resuspension and seasonal trends in these processes. Biodynamic modeling is also needed to measure all Hg sources to a benthic receptor to aid the understanding of causes of elevated Hg concentrations in fish and to determine future susceptibility of biota to Hg exposure from legacy contamination.


The Supplemental Data contains three sections of text and a table. (107 KB DOC).


C. MacLean, J. Martin, J. Szwec, and C. Veenstra provided technical support. Mercury analyses were supervised by D.R.S. Lean and E. Yumvihoze (University of Ottawa), and spatial analyses were supervised by G. Barber (Queen's University). L. Campagna assisted with figures. Funding was from an Ontario Ministry of the Environment Best-in-Science grant to P.V. Hodson, J.J. Ridal, J.M. Blais, and D.R.S. Lean and an Ontario Graduate Scholarship to N.R. Razavi. The authors confirm that they have no conflict of interest, financial or otherwise, related to this publication.