Evaluation of zebra mussels (Dreissena polymorpha) as biomonitors of mercury contamination in aquatic ecosystems



Zebra mussels have invaded many lakes in the United States and could be a useful tool for monitoring responses of aquatic biota to changes in mercury loading. The goal of the present study was to evaluate zebra mussels for use as a biomonitor of mercury contamination by comparing zebra mussel mercury concentrations between a lake with only indirect atmospheric mercury contamination (Otisco Lake, NY, USA) and a lake that was directly contaminated by mercury discharges (Onondaga Lake, NY, USA). Zebra mussels were sampled in both the spring and fall of 2004 and 2005. Total mercury (THg) concentrations in zebra mussels were approximately seven times greater in Onondaga Lake than in Otisco Lake, and water column mercury concentrations differed by an order of magnitude between the two lakes. Seasonal differences resulted in significantly higher zebra mussel THg concentrations during the fall for both lakes. There was also significant variation among different sampling sites in Onondaga Lake. Mussel methylmercury concentrations averaged 53% of THg concentrations but were highly variable. Strong relationships between water column THg and zebra mussel THg suggest that zebra mussels are a good indicator of aquatic mercury concentrations and could be used as an effective biomonitor of mercury contamination in aquatic ecosystems. Environ. Toxicol. Chem. 2013;32:638–643. © 2012 SETAC


Mercury (Hg) is a ubiquitous, persistent environmental pollutant that poses a contamination risk to aquatic ecosystems all over the world. Local source Hg pollution from industrial production, mining waste, and wastewater discharge have directly contaminated many waters 1, and virtually all waters have been contaminated by atmospheric emissions (e.g., coal combustion, waste incineration) and the subsequent long-range atmospheric transport and deposition 2. Most Hg is deposited in the ionic form (Hg [II]), but within aquatic ecosystems, Hg(II) can be converted into more bioavailable methylmercury (MeHg) by anaerobic microbes 3. MeHg bioaccumulates in animal tissue, increasing exposure and posing a health risk to wildlife or humans that consume organisms from contaminated ecosystems 4, 5. Despite recent controls on Hg emissions, every state in the United States has issued Hg advisories for at least one body of water, and Hg currently accounts for more fish consumption advisories than all other contaminants combined 6. Currently, the U.S. Environmental Protection Agency (U.S. EPA) has advanced a rule that would further decrease Hg emissions through controls on electrical generating facilities 7.

Despite the widespread nature and considerable toxicity of Hg, few national programs monitor Hg in the environment 6. The Mercury Deposition Network has approximately 80 sites in the United States and Canada where Hg is monitored in precipitation. The Atmospheric Mercury Monitoring Network includes approximately 20 sites where concentrations of species of Hg in air are measured and dry deposition is estimated. However, there are no national programs to monitor Hg in aquatic ecosystems. Consistent monitoring of Hg is complex because of the multimedia nature of the contaminant. Analysis of aqueous samples for Hg requires specialized equipment and ultraclean sampling techniques and is very expensive. For this reason, sampling of organisms as an indicator of Hg contamination has been proposed. For monitoring, biological sampling has advantages over aqueous sampling because biological samples are a more direct measure of Hg exposure and are more difficult to contaminate than aqueous samples. Another advantage is that solid samples can be analyzed by using U.S. EPA method 7473 8, which is more time-efficient and less costly than the aqueous analysis U.S. EPA method 1631 9. Moreover, organisms accumulate Hg over time and therefore integrate Hg exposure. However, many organisms are not suitable for biomonitoring.

Organisms that make good biomonitors are abundant, widely distributed, relatively immobile, and reasonably easy to sample. Zebra mussels (Dreissena polymorpha) have invaded numerous waters in the eastern United States 10, and their sessile life cycle and widespread range make them a potentially useful biomonitoring species. In addition, they are heavily concentrated in lakes and rivers in the midwestern and northeastern United States, areas that have traditionally shown elevated Hg concentrations in a variety of media 11, 12. Previous studies have explored the use of zebra mussels as indicators of Hg and concluded that zebra mussels are suitable monitoring organisms 13–15. However, these studies did not attempt to correlate water-column Hg concentrations with mussel Hg concentrations or account for temporal variability.

The purpose of the present study was to assess Hg contamination of zebra mussel populations in two lakes, one with historic point-source Hg pollution and one contaminated only by atmospheric deposition. We also evaluated relationships between zebra mussels and water column Hg and explored sources of variability to determine whether zebra mussel biomonitoring is suitable for assessing recovery of both severely and moderately mercury-impacted aquatic ecosystems.


Sampling sites

Two central New York, USA, lakes were chosen as study sites: Onondaga Lake (N 43° 06.520′, W 76° 14.050′) and Otisco Lake (N 42° 50.913′W 76° 16.340′) (Fig. 1). Onondaga Lake is a 12-km2 lake on the northwest side of Syracuse, New York. The watershed for Onondaga Lake is highly developed and urbanized, and the lake is plagued by many water quality problems 16. More than 75 metric tons of Hg was directly discharged into the lake before 1970 from a former chlor-alkali facility, and as a result, the lake waters, sediment, and biota experience chronic Hg contamination. Zebra mussels were first observed in Onondaga Lake in 1992 but did not successfully colonize in large densities until 1999 17.

Figure 1.

Map showing location of sampling lakes and sampling sites within lakes.

Otisco Lake is located approximately 30 km southwest of Syracuse, New York, and is the smallest of the Finger Lakes. Unlike Onondaga Lake, Otisco Lake has no direct sources of Hg contamination. Otisco Lake has similar morphometric features to Onondaga Lake (Table 1), and the outflow (Nine Mile Creek) is a major inflow into Onondaga Lake. Otisco has been used as a reference lake for previous studies of Onondaga Lake, and the rationale for using the two lakes for comparison studies is outlined in Denkenberger et al. 18.

Table 1. Characteristics of study lakes
  • a

    Mean concentrations for all samples collected. Number of samples analyzed indicated in brackets.

    THg = total mercury; MeHg = methylmercury.

Surface area (ha)1,200900
Average depth (m)1110
Max depth (m)2123
Length (km)88.7
Width (km)1.51.2
Residence time (months)4∼12
Unfiltered THga (ng/L)2.77 [12]0.28 [6]
Unfiltered MeHga (ng/L)0.22 [12]0.03 [6]
Mussel THga (ng/g dry wt)440 [223]61 [87]
Mussel MeHga (ng/g dry wt)217 [34]29 [27]

Two sites were chosen for sampling in Otisco Lake (Fig. 1), both of which were on the west lake shore in areas that have significant zebra mussel populations. A spatial sampling design was implemented in Onondaga Lake to examine variation in zebra mussel Hg concentrations near different tributaries and inflows, and four sampling sites were chosen (Fig. 1). Site 1 is located at the northeast shore of the lake away from major inflows. Site 2 is located near the inlet of Nine Mile Creek, the site of elevated Hg inflows from the former chlor-alkali facility. Site 3 is located adjacent to the in-lake waste deposit, which is waste material that has been shown to supply much of the contemporary Hg that is available in Onondaga Lake 19. Site 4 is located near another inlet, Ley Creek.

Sampling methods

Zebra mussels were collected from Onondaga Lake and Otisco Lake on four separate dates in 2004 and 2005. Spring samples were collected on April 1, 2004, and May 23, 2005, and fall samples were collected on November 12, 2004, and November 15, 2005. Selected sampling sites were near the shoreline in 1 to 1.5 m water. Sites were approached by boat, and to avoid disturbing the sediments and water column, water chemistry samples were collected before collecting mussels. Water samples were collected in 1-L Teflon bottles using clean techniques, and one bottle was collected from each sampling site during each sampling season, with the exception of fall 2005, when no water chemistry samples were collected (total n = 6 for Otisco, 12 for Onondaga). Mussels were then collected by removing large pieces of mussel-covered substrate, returning to the boat, and severing the byssal threads with a knife. Extracted mussels were placed in mesh bags and transported to the laboratory on ice.

Laboratory methods

Once in the laboratory, mussels were thoroughly rinsed with deionized water and allowed to air dry. Shell length (posterior to anterior axis) was measured for each mussel using calipers, and mussels were sorted into four length categories: 15 to 20 mm, 20 to 25 mm, 25 to 30 mm, and 30+ mm. Wet tissue was removed from each mussel by dissection and freeze dried for 72 hours. Five to 10 mussels were combined to form composite samples, and the pooled dry tissue was crushed into a fine powder for total Hg (THg) and MeHg analysis. All mussel composite samples were run in duplicate and reported as the mean concentration in terms of dry weight. Analysis of THg was performed on a LECO AMA direct combustion analyzer (Leco Corporation) or a Milestone Direct Mercury Analyzer 80 (Milestone), both of which use catalytic combustion, gold amalgamation, thermal desorption, and atomic absorption spectroscopy 8. Unfiltered water samples were analyzed for THg by oxidizing samples with bromine monochloride to convert all Hg to Hg(II), reduction via stannous chloride to Hg(0), then purging with nitrogen onto gold traps followed by detection with cold vapor atomic fluorescence spectroscopy (Tekran 2600, Tekran Instruments) 9. Composite mussel samples were digested for MeHg analysis using sodium hydroxide and methanol. Digested mussels and unfiltered water samples were analyzed for MeHg by purging with nitrogen to release volatile Hg, which was subsequently adsorbed to traps. Traps were desorbed thermally (350 ATD, Perkin Elmer), and Hg species were separated by gas chromatography (Clarus 500, Perkin Elmer). MeHg was then detected using cold vapor atomic fluorescence spectroscopy (Tekran 2500). Instruments were calibrated with certified standards (National Research Council Canada TORT-2, 270 ± 60 ng/g, and National Institute of Standards and Technology 1515, 40 ± 4 ng/g), and certified mussel tissue was used as an external quality control (National Institute of Standards and Technology 2976, 61 ± 3.6). All quality control standards were within ± 10% of certified value. Method detection limits (MDL) for solid THg, aqueous THg, digested MeHg, and aqueous MeHg were 0.1 ng/g, 0.2 ng/L, 0.64 ng/g, and 0.02 ng/L, respectively. All solid samples used for data analysis were above the MDL. Aqueous samples below the MDL were assigned the MDL concentration value for data analysis.

Data analysis

Mean zebra mussel THg concentrations were calculated for each combination of sample site and sampling date. Least-squares regression was used to examine relationships between water chemistry variables and mean zebra mussel Hg concentrations. Bioconcentration factors were calculated for MeHg using the formula equation image 20. A two-way nested analysis of variance (ANOVA) was used to compare Hg concentrations between lakes and among the sampling seasons. Because of unequal sample sizes from the lakes, Satterthwaite approximation was used to estimate degrees of freedom for the nested ANOVA 21. To evaluate spatial effects, a one-way ANOVA was used to compare Hg concentrations among sampling sites within Onondaga Lake. Comparisons in MeHg measurements between the two lakes were assessed using Student's t test. Datasets that were non-normally distributed or were highly influenced by outliers were log-transformed before data analysis. Statistically significant effects were defined at α ≤ 0.05. All statistical tests were computed using IBM SPSS 19.0 statistical software (IBM).


Lake comparison

Mercury concentrations were significantly higher in Onondaga Lake than Otisco Lake for all types of samples analyzed. In unfiltered water column samples, THg was approximately 10 times higher in Onondaga Lake, and MeHg was more than 7 times higher (Table 1). Average zebra mussel THg concentrations and MeHg concentrations were also approximately 7 times higher in Onondaga Lake. The percentage of THg occurring as MeHg in zebra mussels was also higher in Onondaga Lake (58%) than Otisco Lake (47%); however, the difference was not statistically significant (t = 1.29, p = 0.20, degrees of freedom = 59). Simple linear regression between unfiltered water column THg and mean zebra mussel THg showed a significant, strong relationship among all lake sites (p < 0.001, r2 = 0.68; Fig. 2). A significant relationship also existed between unfiltered water column MeHg and mean zebra mussel THg, although the relationship with MeHg was somewhat weaker than the relationship with THg (p < 0.001, r2 = 0.55).

Figure 2.

Relationships between water column total mercury (THg) and mean zebra mussel THg (r2 = 0.68) and water column methylmercury (MeHg) and mean zebra mussel THg (r2 = 0.55). Circles represent samples taken from Onondaga Lake, and triangles represent samples taken from Otisco Lake. Open symbols represent THg, and closed symbols represent MeHg. Samples below the method detection limit were assigned the method detection limit value. One sample from Onondaga Lake was excluded because it was an extreme outlier (28 ng/L).

Size effect

The effect of zebra mussel size on THg concentrations was examined at selected sites with adequate sample size in the spring 2004 and fall 2004 sampling periods for both lakes. Within Onondaga Lake, comparisons were made on mussels collected from the same sampling location to prevent bias from spatial variability. One-way ANOVA was used to evaluate the effect of size on THg concentrations. At each site and for each season analyzed, no significant effect of shell length on zebra mussel THg concentrations was seen, with the exception of Onondaga Lake in spring 2004 (Table 2). However, post hoc testing using Tukey's test revealed no logical pattern among size classes and THg concentrations for this sampling period. Because there was no statistically discernible pattern among the size classes, all size classes were pooled together for analyses that compare means between lakes and among sampling sites.

Table 2. Mean and standard deviation total Hg values (ng/g dry wt) for separate size classes of zebra mussels at sites within Onondaga Lake (OND) and Otisco Lake (OTI)a
 OND1 spring 2004OND4 spring 2004bOTI spring 2004OND1 fall 2004
  • a

    Number of composite samples analyzed are indicated in parentheses.

  • b

    Significant size effect, F3,38 = 6.16, p = 0.002.

15–20 mm247 ± 74 (10)331 ± 78 (10)62 ± 13 (10)462 ± 77 (11)
20–25 mm242 ± 73 (10)472 ± 87 (10)63 ± 19 (16)433 ± 72 (10)
25–30 mm289 ± 53 (12)372 ± 98 (10)66 ± 22 (10)522 ± 146 (6)
30+ mm298 ± 61 (11)490 ± 121 (9)529 ± 78 (5)

Seasonal and spatial variability

Seasonal variability was explored by comparing average zebra mussel THg concentrations between the two lakes and among the sampling seasons. This was accomplished by using a two-way ANOVA with seasons nested within lakes. For this comparison, log-values of THg concentrations were used to distribute the data more normally and dampen effects of extreme values. Significant differences were detected between lakes (F1,6 = 80.4, p < 0.001) and among sampling seasons (F6,302 = 20.5, p < 0.001). In both lakes, average fall zebra mussel THg concentrations were significantly higher than spring concentrations in both 2004 and 2005 (Fig. 3). Calculated MeHg bioconcentration factors were also slightly higher in the spring than in the fall in Otisco Lake (3.4 × 105 vs 3.0 × 105) and at the Onondaga Lake site 1 (1.7 × 105 vs 1.6 × 105), which were the only sites with adequate sample size to calculate MeHg bioconcentration factor.

Figure 3.

Box and whisker plots of log-transformed seasonal concentrations of zebra mussel total mercury (THg) for Onondaga Lake (OND) and Otisco Lake (OTI). Concentrations of THg were significantly higher in fall samples.

Spatial variability of zebra mussel THg was examined in Onondaga Lake in the fall 2004 sampling period. Results of one-way ANOVA indicated significant differences in average zebra mussel THg among the sampling sites (F3,80 = 31.85, p < 0.001). Pairwise comparisons using Tukey's method revealed that sites 1, 2, and 4 had similar THg concentrations, whereas site 3 accounted for most of the differences in mussel THg concentrations (Fig. 4).

Figure 4.

Box and whisker plots of zebra mussel total mercury concentrations for different sites within Onondaga Lake.


Mean zebra mussel THg concentrations in Otisco Lake (61 ng/g) are similar to concentrations reported from lakes that are not contaminated from point sources (mean values from literature 50–70 ng/g) 13, 22, whereas concentrations in Onondaga Lake are among the highest ever reported. The highest mean THg concentration found in the literature was 525 ng/g dry weight and was found at a site that, like Onondaga Lake, was contaminated by discharge from a chlor-alkali facility 15. This compares favorably to our mean THg value of 440 ng/g in Onondaga Lake, although site 3 near the waste deposit was considerably higher (807 ng/g).

The results of the present study clearly show that Hg concentrations in zebra mussels are much higher in a site with point-source contamination than in a site experiencing contamination from atmospheric Hg deposition. The marked differences in Hg concentrations between mussels from contaminated and uncontaminated sites indicate that zebra mussels continue to feed and accumulate toxins even in highly polluted areas, making them well-suited indicators of biological contamination in polluted areas. The relationship that was observed between THg concentrations in mussels and THg concentrations in the water column suggest that THg in zebra mussels is reflective of the THg that is available in the adjacent environment, meaning that zebra mussels could be an equally effective biomonitor of Hg in both contaminated and uncontaminated sites.

Several issues must be addressed when developing biological monitoring programs for contaminants. One such issue that needs to be resolved is the importance of organism size. Some species, such as fish, show increases in Hg concentration with increasing size and weight 23. We attempted to evaluate the role of size in THg accumulation in zebra mussels but found little effect on mussel THg concentrations, a conclusion that is supported by some other research 13, 24, 25. Some studies, however, have indicated that mussel size does play a role in THg concentration. Carrasco et al. 15 found that mussel THg concentrations decreased with increasing size, whereas Wiesner et al. 26 found that THg concentrations increased with increasing size. Because some disagreement exists among the different studies, effective design of a biomonitoring program should focus on mussels within similar size classes to eliminate this variable as a potential confounding factor.

Seasonal differences also had a strong influence on mussel THg concentrations in both Onondaga Lake and Otisco Lake. Higher concentrations recorded in the fall could be the result of numerous factors. In spring, zebra mussels are typically exposed to much colder temperatures than in the fall. Cooler water leads to lower metabolic demands, and even moderate increases in temperature can increase metabolism and respiration rates in zebra mussels 27, so the difference could possibly be a simple difference in feeding rates. Another possibility is that reproductive cycles influence Hg accumulation. Zebra mussels often spawn when water reaches 12°C 28. It is possible that Hg is moved from somatic tissues into reproductive tissues, which would be expelled from the mussels during spawning. However, because water temperatures in the spring sampling periods were approximately 10°C, spawning is unlikely to explain the lower THg concentrations. A third possibility is that lower dissolved oxygen levels and increasing water temperature lead to an increase in Hg availability throughout the summer, and samples collected in the fall simply reflect a higher availability of mercury in the environment. Todorova et al. 29 reported elevated concentrations of THg and MeHg in Onondaga Lake in the fall as a result of mixing of hypolimnetic waters after turnover. Because zebra mussels filter water and seston, the increases in Hg concentrations in the water column after turnover would likely lead to direct increases in Hg concentrations in zebra mussels. This hypothesis is supported by the seasonal MeHg bioconcentration factor calculations, which were nearly identical between the spring and autumn in both lakes. This would suggest that the seasonal differences in mussel Hg reflect differences in availability within the water column. Regardless of the explanatory cause, it is evident that timing of sampling events can have a significant influence on THg concentrations of zebra mussels, and any biomonitoring project should consider this factor when designing a sampling program.

Spatial variability among the different sites in Onondaga Lake led to significantly different THg concentrations in zebra mussels. There was no significant difference among sites 1, 2, and 4, whereas the THg concentrations at site 3 were high in comparison. Site 3 is located within an area known as the in-lake waste deposit, which before 1970 received chemical contamination from industrial activities. Water column THg at site 3 was five times higher than other sites in autumn 2004, and the zebra mussel THg concentrations were the highest of all sites from any sampling season. Sampling of sediments has shown that this area has the highest sediment THg concentrations of any area in the lake (M. Hall, State University of New York College of Environmental Science and Forestry, Syracuse, NY, USA, unpublished data,). The other sampling sites had sediment Hg concentrations that were less than half the value of concentrations at site 3. Shallow water and prevailing westerly winds often resuspend sediment particles near site 3, leading to turbid water conditions 19. Moreover, the in-lake waste deposit is the focus of dredging activities that are part of the remediation plan for Onondaga Lake. Because zebra mussels are filter feeders, they feed indiscriminately and ingest sediment particles along with plankton and other food sources. The proximity of site 3 to the in-lake waste deposit and the availability of contaminated sediment particles in the water column is likely the explanatory cause for the elevated THg concentrations. This demonstrates how spatial heterogeneity can influence THg uptake by zebra mussels and illustrates the need for consistent sampling locations when using zebra mussels as a biomonitor.

When monitoring Hg concentrations, one must evaluate whether THg or MeHg concentrations should be measured. Although MeHg is the more bioavailable form of Hg that accumulates in food chains and ultimately results in exposure to humans and wildlife, it is much more expensive and time consuming to monitor. As a result, the benefits of measuring MeHg should be clear to justify the additional time and cost. For the present study, little justification seems available for a need to monitor MeHg. Based on the subset of samples that were analyzed for MeHg, the percentage of THg that was MeHg was approximately 53%. This value is comparable to other zebra mussel studies that observed ratios of approximately 60% and 50% 14, 15. In addition, the coefficient of variation among MeHg samples (0.98) was much greater than the coefficient of variation among THg samples (0.79). Because the variability among MeHg concentrations is large and the fraction of MeHg among multiple studies is similar, biomonitoring projects may want to focus on THg concentrations when using zebra mussels as an indicator organism.

Overall, zebra mussels were good indicators of Hg contamination in both highly polluted and less polluted sites. Results from the present study indicate that zebra mussel THg concentrations reflect concentrations in the water column and that they could serve as an effective Hg biomonitoring organism. However, seasonal and spatial variability led to significant differences in Hg concentrations within zebra mussels and should be considered if zebra mussels are to be used as a biomonitor of mercury. This information could be used to initiate regional programs to assess the effectiveness of environmental controls in lowering Hg contamination within aquatic ecosystems and also could be used to monitor recovery of Hg-contaminated ecosystems. Zebra mussels have also been used as monitors of contamination from organic pollutants 30, 31 and other metals 22, so the use of zebra mussels could satisfy the requirements of monitoring programs for multiple pollutants, making them an even more cost-effective monitoring organism. The use of mussels as biological monitors of contamination is a time-efficient and cost-effective method of ecological monitoring and is a tool that aquatic ecosystem managers should consider when designing mercury monitoring programs.


The present study was supported by the U.S. Environmental Protection Agency and the Syracuse Center of Excellence through the Collaborative Activities for Research and Technology Innovation (CARTI) program. We appreciate the support of S. Effler and the Upstate Freshwater Institute.