Supplemental Data 2: Review of Mercury Site Studies
CRITICAL REVIEW OF MERCURY SQVS FOR THE PROTECTION OF
BENTHIC INVERTEBRATES
Sediment toxicity and/or benthic invertebrate community data from 23 studies of 14
mercury sites are described below. These results are summarized in Table 2 of the printed paper.
References are numbered in accordance with the printed paper. Sites are presented below in
alphabetical order.
Androscoggin River (New Hampshire, USA). Chalmers et al. [78] evaluated sediment
toxicity in the Androscoggin River downstream from a former chloralkali facility. Total mercury
concentrations in sediment ranged from < 0.1 to 0.3 mg/kg. This study also incorporated several
measures of mercury bioavailability, including porewater analyses, mercury speciation, and
sequential extraction procedures. Methylmercury was present in bulk sediment at concentrations
up to 2.0 µg/kg and in porewater at concentrations up to 0.003 µg/L. Total mercury
concentrations in porewater ranged up to 0.017 µg/L. Chronic tests using Hyalella azteca and
Chironomus dilutus indicated no toxicity, and the benthic invertebrate community was
determined to be unimpaired based on a regionally calibrated benthic index of biotic integrity
(unbounded NOEC of 0.3 mg/kg).
Augusta Bay (Italy). Concentrations of total mercury in Augusta Bay sediments were
found to be elevated near a chloralkali plant source, but bioaccumulation tests and porewater
concentrations indicated that mercury was largely unavailable to benthic invertebrates [94]. For
the amphipod Rhepoxynius abronius, a mercury concentration of 37 mg/kg was associated with
67% mortality (LOEC), while no significant mortality was associated with 22 mg/kg mercury
(NOEC) [79]. For the amphipod Leptocheirus plumulosus, 38 mg/kg mercury caused no
significant mortality (NOEC), and no amphipods survived exposure to sediment containing 373
mg/kg mercury (LOEC). For the amphipod Ampelisca abdita, statistical power was low due to
very high variability; therefore, data for the other two amphipod species are preferable for
interpreting mercury effects at this site. Toxicity identification evaluation (TIE) indicated
multiple inorganic and organic contaminants were contributing to observed toxicity, and
concentrations of PAHs in porewater were sufficiently elevated to explain the toxicity observed
in some Augusta Bay sediment samples. Although porewater and tissue analyses indicated
limited mercury bioavailability and significant PAH exposures ([94] and J.M. Conder,
unpublished data), the observed LOECs are included in the present analysis as potential effects
of mercury, in the interest of conservatism.
Berry’s Creek/Eight Day Swamp (New Jersey, USA). Multiple contaminants complicate
the interpretation of benthic invertebrate survey results for Eight Day Swamp. At this site,
benthic community diversity varied widely but showed no relationship with total mercury in
sediment up to a concentration of 68 mg/kg, with low diversity occurring more frequently at
mercury concentrations above this level [43]. Diversity was also correlated with several other
metals [43]. In this study, the highest concentrations without mercury-related effects were
considerably elevated above SQVs, but the cause of effects at lower concentrations could not be
1
clearly determined. Without measurements of bioavailable mercury exposures, the possibility of
effects due to bioavailable mercury associated with lower total mercury concentrations cannot be
definitively excluded based on this study. Therefore, this study is considered as supporting
information but is not included in Table 2 for further quantitative analysis.
Brunswick Estuary (Georgia, USA). Portions of salt marsh in the Brunswick estuary are
contaminated with mercury and Aroclor 1268 due to past operations at the LCP Chemicals of
Georgia Superfund site [85]. Sediments containing 18 to 25 mg/kg mercury did not cause
mortality in freshwater amphipods (Hyalella azteca; unbounded NOEC of 25 mg/kg), although
decreased leaf consumption was observed (unbounded LOEC of 18 mg/kg) [85]. The ecological
significance of amphipod feeding rate is uncertain. Horne et al. [86] observed no significant
mortality or behavioral effects on estuarine amphipods (Leptocheirus plumulosus) at mercury
concentrations up to 170 mg/kg mercury in 14-d exposures (unbounded NOEC). When exposed
for a longer test duration (28 d), the same species exhibited 73% mortality when exposed to 972
mg/kg of mercury (LOEC), as well as 27,000 µg/g OC total PCBs, but only 6% mortality when
exposed to 551 mg/kg mercury (NOEC) [87]. At some locations, the benthic community showed
increased dominance by polychaetes and a shift toward surface feeders [86], but co-varying
gradients in physicochemical conditions (e.g., organic carbon) and PCB concentrations, as well
as small sample size (n = 4) and unmeasured chemicals, complicate interpretation of this result.
Wall et al. [88] found only “subtle indications” of impairment of salt marsh function at lower
trophic levels, including a disruption in the size versus-fecundity relationship for grass shrimp
(Paleomonetes pugio); absolute fecundity was not affected. The authors did not provide
sufficient chemistry data to identify a mercury or methylmercury threshold concentration for this
effect [88]. Following removal of the most contaminated sediments at the site, additional
toxicity monitoring was conducted using L. plumulosus and P. pugio, measuring a variety of
chronic endpoints including reproduction [89]. Because “toxicity” was frequently observed in
apparently uncontaminated reference sediments (suggesting methodological issues) and
concentration-response relationships were not clearly evident for mercury or other chemicals,
this monitoring effort does not provide a strong basis to judge mercury toxicity to invertebrates.
Overall, the Brunswick estuary data indicate limited mercury-related effects on benthic
invertebrates. Therefore, the toxicity test results for amphipod mortality and feeding rate are
reasonably representative of benthic invertebrate responses to mercury at this site, for purposes
of the present analysis.
Calcasieu Estuary (Louisiana, USA). In the Calcasieu estuary (Louisiana, USA),
sediment toxicity near a mercury source was found to be caused primarily by
hexachlorobutadiene, but a follow-up investigation targeted sample locations with elevated
mercury concentrations and lower levels of organic contaminants [96]. Amphipod mortality and
growth impairment were observed in samples containing 0.27 to 4.1 mg/kg, but the observed
effects showed no relationship with total mercury concentrations in sediment. Interpretation of
these results is limited by a lack of information on mercury bioavailability, small sample size (n
= 3), control performance issues in some replicate tests, and the presence of multiple
contaminants [96]. Because the possibility of effects due to bioavailable mercury associated
with lower total mercury concentrations could not be definitively excluded, this study is
2
considered to provide supporting information but is not included in Table 2 for further
quantitative analysis.
Clear Lake (California, USA). The abandoned Sulphur Bank Mercury Mine released a
large quantity of mercury into Clear Lake, resulting in total mercury concentrations up to 1,200
mg/kg in sediment. Potential confounding factors included grain size and depth effects, as well
as extensive past applications of DDD and methyl parathion to control midges. Although
preliminary data collection suggested potential relationships between sediment mercury
concentrations and abundance of a few species,1 these findings were not borne out by subsequent
sampling efforts [73,74]. The authors concluded that no population- or community-level effects
were detected at the site [73], indicating an unbounded NOEC of 1,200 mg/kg. The lack of
effects may be due to lack of mercury bioavailability, as neither total mercury nor
methylmercury were notably elevated in chironomid tissues [73].
Lake Maggiore and Toce River (Italy). Bizzotto et al. [79] evaluated sediment toxicity in
a deep subalpine lake and a major tributary downstream from a chemical manufacturing facility.
Both DDT and mercury were present in sediments. In a Sediment Quality Triad study, no effects
on amphipod (Hyalella azteca) reproduction or midge (Chironomus dilutus) survival, growth, or
emergence were observed. Amphipod survival and growth were more variable in both the
upstream (i.e., uncontaminated) reference areas and the study area, but no relationship with
mercury exposures was discernable [79]. Benthic sampling revealed that amphipods were absent
from lacustrine habitat and uncommon in riverine habitat at reference and study area locations,
reflecting possible physical unsuitability of the sediment and habitat for amphipods (P.C.
Fuchsman, unpublished data). Observations made using sediment profile imagery indicated that
habitat conditions (e.g., water depth, deposition and scour patterns, organic loading), rather than
mercury exposures, likely influenced community structure. The benthic community observations
were thus consistent with the midge toxicity test results. An unbounded NOEC of 5.2 mg/kg was
identified based on these results.
Lavaca Bay (Texas, USA). Parametrix, Inc. [77] performed a Sediment Quality Triad
study for Lavaca Bay, where mercury contamination originated from a chloralkali plant.
Chronic toxicity testing was conducted using the amphipod Leptocheirus plumulosus and the
polychaete Neanthes arenaceodentata. No significant differences between study area and
reference locations were observed with respect to amphipod survival or polychaete growth or
survival. Consistent with the sediment toxicity test results, benthic community composition was
related to physical factors but not mercury exposures. Unbounded NOECs from this study are
4.6 mg/kg for total mercury and 2 µg/kg for methylmercury in sediment [77].
1
Suchanek TH, Richerson PJ, Holts LJ, Lamphere BA, Woodmansee CE, Slotton DG, Harner EJ, Woodward LA.
1995. Impacts of mercury on benthic invertebrate populations and communities within the aquatic ecosystem of Clear
Lake, California. Water Air Soil Pollut 80:951-960.
3
Mabubi River (Tanzania). Chibunda et al. [95] evaluated effects on development rate,
growth, and emergence of midges (Chironomus riparius) exposed to sediments collected
downstream of the Mugusu artisanal gold mine. A significant reduction in growth (by
approximately 30%), but not survival or percent emergence, was associated with a mercury
concentration of 2.3 mg/kg (LOEC for growth and unbounded NOEC for emergence). Delayed
development rates were associated with a mercury concentration of 1.6 mg/kg, although no
growth effects were observed at this level (LOEC development rate, NOEC for growth). A lack
of significant effect on development rate was reported in sediment with a concentration of 0.23
mg/kg (NOEC). The presence of arsenic at a concentration of 72 mg/kg suggests that both the
NOEC and LOEC are conservative with respect to mercury-only toxicity (i.e., arsenic may have
contributed to toxicity). While other confounding factors cannot be ruled out (e.g., potential
organic contaminants such as PAHs were not analyzed), mercury bioavailability in the Mabubi
River could be relatively high because the mercury source is recent and ongoing. Thus, mercury
bioavailability in this system might be more similar to that observed in spiked sediment toxicity
tests, compared to sites contaminated by historical sources.
New York Harbor (New York, USA). Rubinstein et al. [80] conducted toxicity tests with
worms (Nereis virens), clams (Mercenaria mercenaria), and grass shrimp (Palaemonetes pugio)
in sediments collected from New York Harbor. Although the locations of the sediment
collections were not specified, it was assumed that the locations were impacted by a mercuryrelated historical source due to the high range of concentrations of mercury in sediment (3 to 35
mg/kg). Co-contaminants include PCBs and other metals. In 100-d toxicity tests, none of the
sediments tested resulted in more than 10% mortality, yielding an unbounded NOEC of 35
mg/kg.
Peninsula Harbour (Ontario, Canada). Environment Canada applied a standardized
benthic assessment of sediment (BEAST) methodology to evaluate sediment toxicity and benthic
invertebrate community quality in Peninsula Harbour, located on the north shore of Lake
Superior [75]. The main contaminants at this site are mercury, PCBs, and wood waste. Toxicity
test endpoints included survival and growth of amphipods (Hyalella azteca) and mayflies
(Hexagenia spp.) following chronic exposures, survival and growth of midges (Chironomus
riparius) following subchronic exposures, and survival and reproduction of tubificid worms
(Tubifex tubifex) following chronic exposures. The study found no evidence of toxicity due to
mercury [75]. Significant adverse effects were noted at only 4 of 33 stations, and effects in two
of these samples were hypothesized to be due to physical unsuitability of the sediment (i.e., hard,
compact material). The cause of apparent toxicity in the remaining 2 of 33 samples was not
determined, but sediment mercury concentrations were 0.16 and 0.37 mg/kg, lower than in 28 of
the 29 non-toxic site sediments tested. Total concentrations of mercury as high as 19.5 mg/kg
were non-toxic (unbounded NOEC). Non-toxic concentrations of methylmercury were as high
as 22.6 µg/kg.
Ria de Aveiro (Portugal). Reduced benthic invertebrate abundance and species richness
were found at sample locations with elevated mercury concentrations closest to a former
chloralkali plant [91]. Salinity varied greatly across sample locations and over time, due to
4
reduced water renewal with distance from the coast, high evaporation in summer and increased
rainfall in winter. Sample locations with high concentrations of mercury also contained high
concentrations of nitrite and nitrate, as well as high organic matter content, which fluctuated
seasonally. Ordination analysis showed mercury to be a strong determinant of benthic
community structure for the two locations closest to the mercury source (LOEC of 7 mg/kg),
while other locations appeared to be affected primarily by salinity (NOEC of 3 mg/kg).
Subsequent studies focused on population dynamics for several ecologically important species in
the estuary [90,92,93]. The mysid Mesopodopsis slabberi was found to be tolerant of mercury
exposure, with higher densities and biomass in the most contaminated areas (unbounded NOEC
of 200 mg/kg) [92]. Population age structure of the snail Peringia ulvae [93] was adversely
affected at the station with the highest concentration of mercury (LOEC of 82 mg/kg, NOEC of
24 mg/kg), although density, biomass, and growth productivity were not adversely affected at
any of the stations (unbounded NOEC of 82 mg/kg). Cardoso et al. [90] evaluated the dynamics
of the epibenthic crustacean community and statistically analyzed results according to three main
taxa groups (mysids, isopods, and amphipods). Amphipods were the least mercury-tolerant of
these taxa, with statistically significant negative correlations observed between amphipod density
and sediment mercury concentrations observed at the two stations with the highest
concentrations of mercury in sediment (11 to 206 mg/kg). Productivity at these stations was
more than 20% lower than at other stations (LOEC of 11 mg/kg, NOEC of 2.0 mg/kg). In
contrast, isopod and mysid density were positively correlated with mercury at the station with the
highest concentrations of mercury in sediment (86 to 206 mg/kg), yielding unbounded NOECs of
206 mg/kg for these species groups. Sediment toxicity testing with the amphipod Corophium
multisetosum was also performed in Ria de Aveiro2 but was not accompanied by sediment
chemical analyses. Within the area investigated in the benthic community studies described
above, amphipod survival was moderate to high, with moderate to severe effects on growth and
fecundity observed at some locations. Although no specific quantitative relationship can be
drawn between these toxicity test results and mercury exposures, the observation of adverse
effects in the laboratory tests is in agreement with the adverse effects noted in the previous
studies.
South River (Virginia, USA). Flanders et al. [83] conducted a Sediment Quality Triad
evaluation in a stream impacted with mercury from a former textile mill and found no adverse
ecological effects at any of the five study locations, which contained sediment with
concentrations of mercury up to 18.9 mg/kg (unbounded NOEC). Toxicity testing included 10-d
growth and survival tests with amphipods (Hyalella azteca) and midges (Chironomus dilutus)
and a variety of benthic colonization and benthic community census measurements. Measured
concentrations of mercury in sediment porewater, which were generally below 20 μg/L, were
well below concentrations in sediment porewater associated with effects on growth for C.
2
Castro H, Ramalheira F, Quintino V, Rodrigues AM. 2006. Amphipod acute and chronic sediment toxicity
assessment in estuarine environmental monitoring: An example from Ria de Aveiro, NW Portugal. Mar Pollut Bull
53:91-99.
5
riparius (LOEC of 140 μg/L) exposed to mercury-spiked sediment [70], even though the bulk
sediment mercury concentrations were lower in the spiked sediment test than in the South River
sediments. This finding is consistent with observations that bulk sediment concentrations in
spiked sediment tests generally do not realistically reflect bioavailability in the field. In a
separate experiment with South River sediments, Bundschuh et al. [84] observed a 35%
inhibition of Hyalella azteca feeding rate in sediments containing 10.3 mg/kg mercury (LOEC)
and no detectable inhibition in feeding rate (compared to control) in sediments containing 5.9
mg/kg (NOEC). Feeding rate may be less directly related to community-level effects than more
standard toxicity test endpoints but is nevertheless included as a toxicity endpoint in the present
analysis.
St. Clair River (Ontario, Canada). In a study of St. Clair River sediments using the same
protocols described above for Peninsula Harbour, Milani et al. [75] observed that benthic
community characteristics upstream and downstream of a chloralkali plant were not different and
were not related to mercury in sediment. Benthic community characteristics at a station
exhibiting the highest concentration of mercury (49 mg/kg) were the most similar to reference
conditions among all the stations studied (unbounded NOEC). No dose-response relationship
with mercury concentrations in sediment was evident. Milani et al. [76] observed toxicity in
three sediments where total mercury concentrations ranged from 1.4 to 16.2 mg/kg, while nine
other sediments containing 1.4 to 3.8 mg/kg total mercury were not toxic (the sediment
containing 49 mg/kg was not tested for toxicity). Concentrations of mercury and methylmercury
in field-collected invertebrates from multiple sediments indicated no relationship between
bioavailable mercury exposures and toxicity test results. Therefore, the observed effects in a
small number of tested samples were considered to be due to factors other than mercury.
Sudbury River (Massachusetts, USA). Sediment bioaccumulation tests conducted with
the mayfly Hexagenia sp. yielded no significant mortality in sediments collected near a
hazardous waste site primarily contaminated with mercury, which exhibited concentrations up to
22.1 mg/kg (unbounded NOEC) [54]. Sediment from six of seven stations, each sampled during
two seasons, also did not affect mayfly growth. Differences in growth were unrelated to mercury
concentrations based on several measures of exposure (i.e., total and methylmercury in sediment
and mayfly tissue, as well as methylmercury in overlying water). Growth of mayflies exposed to
the sediment exhibiting the highest concentration of mercury was within 10% of the highest
growth result in the entire experiment. Naimo et al. [54] hypothesized that physical
characteristics of the sediment may have caused the observed growth effects at the single
affected station, although co-contaminants (other metals, PAHs, PCBs, and other persistent
organic pollutants) also may have influenced results.
Whatcom Waterway (Bellingham, Washington, USA). Four investigations were
conducted from 1996 to 2008 at a site contaminated via multiple sources, including a chloralkali
plant [81,82]. Toxicity test endpoints included 2-day survival of larval sand dollars (Dendraster
excentricus) or bivalves (Mytilus sp. or Crassostrea sp.), 20-day growth of polychaetes
(Neanthes arenaceodentata), and 10-day survival of amphipods (Eohaustorius estuarius).
Toxicity was defined based on statistical testing and minimum significant differences, as defined
6
by the Washington State Department of Ecology.3 Toxicity tests conducted in 2002-2008
yielded no adverse effects in 37 of 38 samples, with mercury concentrations up to 2.6 mg/kg
[81,82]. Toxicity tests at this site in earlier years (1996-1998) showed a greater incidence of
effects, but toxicity was correlated with concentrations of phenolic compounds related to wood
waste (which subsequently declined), rather than mercury [81].
3
Washington State Department of Ecology. 2008. Sediment Sampling and Analysis Plan Appendix. Ecology
Publication No. 03-09-043. Sediment Source Control Standards User Manual, Washington Department of Ecology
Sediment Management Unit.
7