Technical Support Document
Revision of December 2000 Regulatory Finding on the Emissions of Hazardous Air
Pollutants From Electric Utility Steam Generating Units and the Removal of Coal-
and Oil-Fired Electric Utility Steam Generating Units from the Section 112(c) List:
Reconsideration
October 21, 2005
-------
Technical Support Document
Revision of December 2000 Regulatory Finding on the Emissions of Hazardous Air
Pollutants From Electric Utility Steam Generating Units and the Removal of Coal-
and Oil-Fired Electric Utility Steam Generating Units from the Section 112(c) List:
Reconsideration
Section 1.
Introduction
Section 2.
Marine (Open Ocean) Cycling
Section 3.
Utility Attributable Mercury in Marine Seafood Diet
Section 4.
Near-Shore Exposure Pathway
Section 5.
Aquaculture Exposure Pathway
Section 6.
Commercial Freshwater Pathway
Section 7.
Joint Consumption
Section 8.
Health Benefits and Costs
Section 9.
Evaluation of NESCAUM Report
Section 10.
2010/2015 CMAQ Modeling Results
Section 11.
Global Source Impact
Section 12.
References
1. Introduction
In the Technical Support Document (Effectiveness TSD) for the Revision of
December 2000 Regulatory Finding on the Emissions of Hazardous Air Pollutants From
Electric Utility Steam Generating Units and the Removal of Coal- and Oil-Fired Electric
Utility Steam Generating Units from the Section 112(c) List (Revision Rule)(US EPA
2005a) the Agency analyzed the mercury (Hg) exposure due to coal-fired electric utility
steam generating units ("power plants" or "utilities") as defined in the Revision Rule
remaining after implementation of the Clean Air Interstate Rule (CAIR) and after
implementation of both CAIR and the Clean Air Mercury Rule (CAMR). This analysis
included a partly quantitative, partly qualitative treatment of the exposure from
recreational and subsistence caught freshwater fish. Other pathways including
commercial freshwater fish, estuarine fish, and marine fish were treated qualitatively.
The quantitative component of the analysis focused on the recreational and
subsistence freshwater fish pathway because, as described below, it is this pathway that
leads to the greatest individual exposure due to utility-attributable mercury emissions.
Based on a qualitative analysis, EPA concluded that the other pathways lead to smaller
individual exposure levels and that the combined exposure from all the pathways would
not be materially different than the exposure due solely to the recreational/subsistence
fish pathway for the individuals most highly exposed to utility-attributable mercury. See
70 FRat 16,102.
1
-------
This document expands on information presented in the Revision of December
2000 Regulatory Finding on the Emissions of Hazardous Air Pollutants From Electric
Utility Steam Generating Units and the Removal of Coal- and Oil-Fired Electric Utility
Steam Generating Units from the Section 112(c) List: Reconsideration regarding the
Agency's assessment of utility attributable mercury concentrations in commercial
freshwater fish, aquaculture, estuarine fish, and marine fish exposure pathways, including
why EPA believes these pathways are not reasonably anticipated to result in a hazard to
public health after reductions in power plant emissions due to CAIR and, independently,
CAMR.
2. Marine (Open Ocean) Cycling
The inclusion of the exposure pathway associated with marine (open ocean) fish
in the original freshwater quantitative analysis supporting the Revision Rule does not
materially change the results because the impact of power plant emissions on mercury
concentrations in open ocean environments is limited. First, over half of the U.S.
commercial fish supply is imported. Second, the majority of domestic commercial fish is
caught in open ocean regions that have not been impacted by anthropogenic mercury
releases to the same extent as the atmosphere, near-shore, or inland systems. This means
that current fish tissue concentrations likely do not reflect present day atmospheric
mercury concentrations or deposition rates (Mason and Gill, 2005; Kraepiel et al., 2003).
EPA could not support extension of the Mercury Maps freshwater modeling framework
to marine systems. Application of the Mercury Maps model to marine environments
would be an extension of the modeling framework beyond the realm for which it was
intended for application and has not been empirically evaluated. The power plant
contribution to mercury in these fish (discussed in more detail below) is difficult to
quantify with confidence at this time and is expected to be relatively small in pelagic
marine species based on the analysis described below.
Predatory marine fish are a significant source of methylmercury exposure for the
U.S. population (Carrington et al. 2004). Exposure is a function of both the amount of
mercury in fish and the quantities of fish consumed in the U.S. High levels of
methylmercury in fish are generally the result of bioaccumulation in larger, older fish.
In the case of marine fish, higher trophic level species tend to have comparable
concentrations to top predators in freshwater ecosystems. In addition, the quantities of
marine fish consumed by humans are larger than the quantities of freshwater fish. Based
on the rationale described in detail below, EPA expects that utility-attributable mercury
in these fish is a small fraction of their overall burden.
It is extremely difficult to determine the response of oceans to changes in mercury
emissions from human sources due to limited scientific understanding at this time.
However, the best available science suggests that the significance of changes in marine
fish mercury concentrations in response to reductions in power plant Hg emissions will
be small and will require on the order of decades to centuries to be achieved. To further
elucidate our rationale, we present the results of a sensitivity analysis below that shows
the relative importance of changes in present day atmospheric deposition on the
2
-------
magnitude and timing of changes in ocean mercury concentrations in the Atlantic and
Pacific Oceans.
In the original analyses supporting the final Section 112 rule, EPA quantified
reductions in methylmercury concentrations in freshwater fish associated with declining
inorganic mercury emissions from coal-fired utilities but concluded that the science of
Hg cycling in marine systems is not sufficiently advanced to allow for a similar
quantification of this exposure pathway. Other studies (NESCAUM 2005, Trasande et al
2005) that did quantify the marine exposure pathway used assumptions for a central
tendency estimate that are not supported by the literature on marine fate and transport of
Hg, likely resulting in an overestimate of the power plant contribution to marine fish by
an unknown but possibly large amount. We briefly discuss some of the assumptions of
the NESCAUM report as they relate to marine fish in Section 9.
When quantifying the relationship between power plant mercury emissions in
deposition over coastal and marine areas the NESCAUM study relied on the REMSAD
model, which appears to over-predict Hg deposition from US power plants. Next, using
the approach applied in NESCAUM study for marine systems would require an
additional assumption that present day concentrations in the surface ocean are tracking
changes in atmospheric concentrations and deposition. This premise is not supported by
the screening analysis detailed below for marine systems. The proportionality
assumption is also not applicable for coastal ecosystems with significant riverine inputs
or watershed areas (most coastal systems).
One of the greatest uncertainties in coastal and marine systems is the rate and
location of methylmercury formation. This is important because methylmercury
formation rates determine bioavailability to fish and shellfish. The NESCAUM study
assumed that changes in methylmercury concentrations in marine fish would be
proportional to changes in deposition of (inorganic) mercury and surface water mercury
concentrations. This actually embodies two assumptions - 1) changes in the surface
water concentration of total mercury are proportional to changes in the air deposition of
mercury and 2) changes in the methylmercury concentration in fish are proportional to
the concentration of total mercury in the water. The literature offers no clear-cut
guidance on how to address the second assumption. However, irrespective of this
relationship between total Hg in water and MeHg in fish, the assumption of
proportionality (first assumption) between changes in deposition and concentrations of
Hg in the surface ocean is not consistent with some basic physical oceanographic
principles that determine the magnitude and reservoir of Hg in the world's oceans. For
example, a number of studies have measured large fluxes of mercury lost through gas
exchange (volatilization) at the ocean surface (Amyot et al., 1997, Mason et al., 2001,
Rolfhus and Fitzgerald, 2004). In addition, monitoring data indicate that concentrations
of mercury in surface waters among the different oceans are not equal, and that deep
ocean water concentrations are very low and likely unaffected by anthropogenic mercury
(Gill and Fitzgerald, 1988, Mason and Sheu, 2002, Laurier et al., 2004). This means that
the large flows of ocean waters (laterally among oceans, oceanic upwelling and deep
water formation) will dilute or add (depending on the flow) to the existing mercury in
3
-------
surface waters. A conceptual model of the various inputs and outputs of mercury, in
addition to atmospheric deposition, is shown in Figure 2-1. A sensitivity analysis
illustrating the magnitude of this uncertainty is presented below.
MARINE BOUNDARY LAYER
K
EVASION
Cair
ATMOSPHERIC
DEPOSITION
RIVERS
~
DEEP WATER
FORMATION
RESERVOIR
M = C V
W wWwW
o
o o
cs>
o
V
PARTICLE
SETTLING
V01
c>
LATERAL
OCEAN
FLOW
02
Figure 2-1. Conceptual model of mercury cycling in the surface waters of the
Atlantic Ocean. Inputs are shown as blue arrows and outputs in grey arrows. The
concentration of mercury in surface waters will be a function of the combined inputs and
losses of mercury over time, with atmospheric deposition representing only one
component of the overall flux of mercury.
Even if a proportional relationship between U.S. utility-attributable mercury
deposition and marine methylmercury fish tissue concentration were a reasonable
representation of the science, the saltwater species would nevertheless contribute only a
very small amount of dietary utility attributable mercury because of the small
contribution of the power plant emissions to the global pool. Annual emissions of
mercury from U.S. utilities account for about 1.0 percent of total global emissions
(including natural and recycled mercury). The total global emissions is estimated to be
between 4400 and 7500 tonnes (US EPA 1997a), while total anthropogenic Hg emissions
were approximately 2269 tonnes per year in 2001 (Pacyna and Munthe 2004). Overall,
anthropogenic emissions from all sources in the United States still comprise less than 3
percent of the global total (UNEP 2002). Even in coastal environments where EPA's
deposition models suggest that there is a higher contribution of Hg deposition by U.S.
power plants, particularly the Atlantic and Gulf Coast area, the utility-attributable
contribution to human mercury exposure in these regions is thought to be small as
explained below.
4
-------
2.1 Model of Mercury Cycling in World's Oceans
Modeling the effect the reduction in Hg deposition from U.S. power plants would
have on Hg concentration in ocean fish is complicated by non-linear relationships
between changes in human emissions, atmospheric Hg deposition over the oceans, water
column concentrations and fish Hg concentrations. In contrast to freshwater systems,
there is a paucity of field data on factors controlling MeHg formation and concentrations
in marine environments and it is not clear where the majority of methylation occurs in the
ocean (see review by Mason and Gill, 2005).
Mason and Sheu (2002) suggest that Hg in the ocean has only increased by 9
percent relative to pre-industrial levels even though air releases have more than doubled.
Several revisions to this estimate, including Mason and Gill (2005), have suggested a
larger anthropogenic contribution to mercury in the world's oceans of up to 60 percent in
the Atlantic. This can be compared to the average global enrichment of the atmosphere
which is between 200-500 percent. These revised estimates all support the premise that
surface ocean concentrations are not tracking atmospheric deposition. Further supporting
this premise are empirical data from the Pacific Ocean for water column mercury
concentrations (Laurier et al., 2004) and tuna (Krapiel et al., 2004), which show no
significant differences in concentrations over the last 20-30 years. Presently, there are no
data to support the assumption that atmospherically deposited Hg would be preferentially
converted to MeHg relative to the large reservoir of legacy Hg in marine environments.
This preferential conversion would be one condition needed to achieve the assumed near-
instantaneous proportional reduction in marine species MeHg concentrations described in
the NESCAUM study. Based on this information, EPA concludes that the linear approach
employed in the NESCAUM study will likely lead to a significant overestimation of
utility attributable exposure and an even greater overestimation of the benefits of
reducing power plant emissions.
At this time there are insufficient empirical and mechanistic data on the rate and
location of MeHg formation in marine environments to constrain a detailed predictive
model for MeHg cycling and accumulation in fish in the open ocean. However, for the
purposes of developing a comparable quantitative model to the NESCAUM report and to
provide a sensitivity analysis, EPA has adapted a recent modeling approach by Mason (in
review) with results discussed in Mason and Gill (2005). This adapted model is used to
investigate the influence of changing atmospheric deposition rates resulting from control
of coal-fired utilities on concentrations of total Hg in the open ocean water. EPA
recognizes that this assumption is a simplification of Hg bioaccumulation in marine
environments and must be viewed as a preliminary assessment of the potential response
of marine environments to changes in atmospheric Hg deposition and must be interpreted
with caution.
We selected the model approach discussed in Mason and Gill (2005) to describe
Hg cycling in marine environments because it represents the interconnectedness of the
5
-------
different oceans through lateral transport of surface waters, upwelling, and deep-water
formation in the North Atlantic. The model is based on the 16-box model for the world's
oceans published by Kahana, et al. (2004) (see Figure 2-2). Kahana et al.'s model is an
expanded version of Stommel's pioneering box model (Stommel, 1961), which originally
described ocean circulation as a function of salinity and temperature. Inputs of Hg from
major rivers globally are characterized using concentrations data from Cossa, et al.
(1996), and flux data from the National Center for Atmospheric Research (NCAR)
website (www.cgd.ucar.edu/cas/catalog/dai/runoff-table2-top50r.html). Losses of Hg
through particulate settling and evasion are characterized using mercury-to-carbon
(Hg:C) ratios for different oceans described in Mason, et al. (1994), and empirically
measured values for different oceans, respectively. Overall, fluxes are constrained using
empirical data on measured fluxes in each compartment and the Mason and Sheu (2002)
global budget. Atmospheric deposition for each ocean was estimated from measured wet
deposition rates over the oceans from existing data at that time, developed by Mason, et
al. (1994), and revisited for the Mason and Gill (2005) analysis to update estimated dry
deposition based on gradients in total gaseous mercury (TGM) concentrations in the
marine boundary layer (MBL).
North
Atlantic
overturning
Atlantic/
Mediterranean
300 m
1500 m
4000 m
Bottom flow into the Atlantic
Figure 2-2. Conceptual model of the major oceanic circulation patterns adapted
from Kahana et al. (2004). Also shown are average mercury concentrations (pM) and
reservoirs (Mmol) of mercury in each of the ocean compartments based on the data and
model discussed in Mason and Gill (2005).
Advantages of this adapted modeling approach compared to previously published
models of Hg cycling in the oceans include:
1. The model treats different oceans and well-mixed components of these oceans
in a discrete manner and constrains concentrations and fluxes using the most recent field-
data.
6
-------
2. The model is based on a simplified oceanographic model allowing for various
depths of the mixed layer determined by the locations of the permanent thermoclines in
each ocean rather than assumption of 100 meter mixed layer for all oceans. This is
important because it determines the reservoir of mercury present in each ocean and
affects the temporal responses of each ocean to changes in deposition.
3. The model includes empirically based fluxes of Hg through evasion from
surface waters, particulate settling and riverine inputs as well as advection resulting from
major ocean circulation patterns.
To model the potential response of marine fish to changes in Hg deposition resulting
from the power plant regulation the following assumptions are necessary:
1. Change in fish mercury concentrations will be proportional to the change in total
mercury concentrations in the water column.
2. Power plant regulation will not significantly affect ocean concentrations other than
the Surface Atlantic and North Pacific. This means that we are assuming utility
attributable mercury in the U.S. will not significantly change concentrations of
mercury in other oceans (e.g., Antarctic and Indian Oceans, etc.)
3. Rate constants describing the various inputs and outputs of mercury in surface
waters (see Figure 2-1) can be reasonably characterized using available empirical
data on measured mercury fluxes. Note that the NESCAUM modeling approach
did not use rate constraints because it assumes that any change in ocean mercury
concentrations will be proportional to changes in atmospheric deposition.
Table 2.1 Comparison of major assumptions used in NESCAUM analysis to this
analysis based on Mason and Gill (2005)
Study Characteristics NESCAUM This
Assume equilibrium between Hg in the ocean and
atmospheric Hg deposition
Assume change in total mercury concentration in the water
is proportional to change in atmospheric mercury deposition
Based on oceanographic circulation data (e.g., to
characterize depth of well-mixed surface waters impacted
by atmospheric deposition)
Evasion of Hg°, particulate transport, lateral advection of
water (inputs and outputs of Hg), deepwater formation,
upwelling included explicitly in model
Assumes change in methyl mercury concentration in fish is
proportional to change in total mercury concentration in
water
Analysis Analysis
Yes No
Yes
No
No
Yes
No
Yes
Yes
Yes
2.2 Results of Ocean Model Sensitivity Analysis
7
-------
Results for EPA's sensitivity analysis are presented in tables 2.2 and 2.3. The
adapted marine cycling model was first run to reflect steady state conditions in both
ocean compartments (Atlantic and Pacific). The empirically constrained fluxes of
mercury for both oceans indicate that neither ocean is currently at steady state. This can
be seen in Figures 2-3 and 2-4 by virtue of the fact that adding all the input and output
fluxes results in a net loss of mercury in the Atlantic and a net gain of mercury in the
Pacific
We can also use the empirical data to estimate the time needed to reach steady
state in each ocean by calculating the sum of rate constants describing inputs and losses
of mercury to each ocean. Rate constants were calculated by dividing the empirically
constrained fluxes by the total reservoir of mercury in each ocean compartment. Using
this methodology, the Atlantic Ocean is expected to reach steady state (ocean
concentrations reflect present atmospheric deposition) in approximately 3 decades with
on the order of a seven percent decline in fish mercury concentrations (Figure 2-3). The
North Pacific has a somewhat larger reservoir of mercury and a deeper mixed layer (see
Figure 2-2 and 2-4). Thus, the time to steady state (ocean concentrations reflect present
atmospheric deposition) for the North Pacific is on the order of 2 centuries and results in
a ten percent increase in mercury in fish if all other factors are constant. Note that these
calculations assume no changes in mercury inputs over time (i.e., present conditions
remain constant).
Baseline Scenario Surface Atlantic
Marine Boundary Layer
TGM = 1.75 ng/m2
Rivers
(Cr=30 pM)
2.48 Mmol/yr
-2.17 Mmol/yr
Surface
Antarctic
(Cw = 0.9 pM)
Well-Mixed
Surface Waters
24.0 Mmol QO
Cw = 2.0pM °
Vw= 1.2 x 10s m3o
oo
-0.59 Mmol/yr
-0.01 Mmol/yr
Sfc. Med.
Particle
Settling
0.32 Mmol/yr
Upwelling
Intermed.
Atlantic
(Cw = 1.5 pM)
Figure 2-3. Model of mercury inputs and outputs in the surface Atlantic Ocean
based on data from Mason and Gill (2005) and Mason (2005).
-------
Baseline Scenario North Pacific
Marine Boundary Layer
TGM = 1. 50 ng/m2
Rivers
(CR=30pM)
1.33 Mmol/yr
Well-Mixed
Surface Waters
73.8 Mmol
Cw = 1.25 pM
-1.13 Mmol/yr
@ ©
©
-0.47 Mmol/yr
SPI
V=5.9x106m3 °
W oo
0.38 Mmol/yr
1500 m
Upwelling Deep
Pacific & Indian
(Cw = 1.0 pM)
-0.05 Mmol/yr
Particle
Settling
Figure 2-4. Model of mercury inputs and outputs in the North Pacific based on
data from Mason and Gill (2005) and Mason (2005).
After running the model to steady state, a baseline scenario was developed to
explore the potential contributions of utilities to the Atlantic and Pacific oceans. To do
this, mercury inputs from atmospheric deposition were adjusted by the relative change in
deposition forecasted in nearshore areas using the CMAQ model after power plant
emissions are removed (2 percent in the Atlantic and 0.65 percent in the North Pacific).
Because the changes in atmospheric deposition were taken from nearshore rather than
offshore areas, they represent an upper bound for changes in deposition that might be
attributable to U.S. coal fired utilities. In addition, changes in deposition resulting from
the power plant regulation could also affect river inputs of mercury from North America
to the Atlantic and Pacific Oceans. However, these changes are likely to be relatively
small given the dominance of freshwater inputs to the ocean basins from South America
and European rivers relative to North America. Such potential changes in river mercury
concentrations have not been quantified at this time; thus, for the purposes of the
sensitivity analysis we explore a range of reductions in riverine mercury inputs between
zero and two percent. The upper bound of this analysis is likely a significant
overestimate of potential changes in riverine mercury inputs to the oceans and should be
caveated appropriately. As a sensitivity analysis, we also explore a range in potential
atmospheric reductions associated with removal of U.S. utilities as a source of mercury.
In the Atlantic ocean the range in atmospheric reductions explored is 0.5 percent and for
the North Pacific it is 0.1 to 1 percent.
9
-------
Table 2.2 Sensitivity analysis for the Atlantic Ocean to assess potential change in
mercury concentrations in marine fish observed at steady state as the result of
control of power plant emissions. Note that changes are expressed as a fraction of
the current fish mercury concentration. The top row shows a hypothetical, upper
bound range for the change in mercury concentrations in riverine inputs to the
ocean, the left column indicates the decrease in atmospheric deposition of mercury.
Hypothetical Decrease in Hg River Inputs
0.0%
0.5%
0.7%
1.0%
1.5%
1.8%
2.0%
0.5%
-0.3%
-0.4%
-0.4%
-0.5%
-0.5%
-0.6%
-0.6%
1.0%
-0.7%
-0.8%
-0.8%
-0.8%
-0.9%
-0.9%
-1.0%
1.5%
-1.0%
-1.1%
-1.1%
-1.2%
-1.2%
-1.3%
-1.3%
Hypothetical
Decrease in
H9 .
Deposition
2.0%
-1.4%
-1.4%
-1.5%
-1.5%
-1.6%
-1.6%
-1.6%
2.5%
-1.7%
-1.8%
-1.8%
-1.9%
-1.9%
-2.0%
-2.0%
3.0%
-2.1%
-2.1%
-2.2%
-2.2%
-2.3%
-2.3%
-2.3%
3.5%
-2.4%
-2.5%
-2.5%
-2.5%
-2.6%
-2.6%
-2.7%
4.0%
-2.7%
-2.8%
-2.8%
-2.9%
-2.9%
-3.0%
-3.0%
5.0%
-3.4%
-3.5%
-3.5%
-3.6%
-3.6%
-3.7%
-3.7%
Note that the ranges in atmospheric deposition used above are meant to explore the
sensitivity of model forecasted water/fish mercury concentrations to changes in
atmospheric deposition. EPA's analysis indicates that a 1% change in atmospheric
deposition over the Atlantic Ocean is a reasonable upper bound for changes in
atmospheric deposition resulting from reductions in utility emissions (see Section 3).
10
-------
Table 2.3 Sensitivity analysis for the North Pacific to assess potential change in
mercury concentrations in marine fish observed at steady state as the result of
control of power plant emissions. Note that changes are expressed as a fraction of
the current fish mercury concentration. The top row shows a hypothetical, upper
bound range for the change in mercury concentrations in riverine inputs to the
ocean, the left column indicates the decrease in atmost pheric deposition of mercury.
0.0%
Hypothetical Decrease in Hg River Inputs
0.1% 0.2% 0.3% 0.4% 0.5%
1.0%
-------
EPA expects that for a given decrease in deposition, the decrease in fish tissue
concentration will be less than the decrease in deposition. However, due to the scientific
uncertainty on this point, EPA feels that it is not appropriate at this time to use the results
from Section 2 to create IDI values for marine fish. Instead, we conduct a bounding
analysis of the utility attributable mercury in marine fish assuming a proportional
relationship between utility-attributable mercury deposition decreases and MeHg fish
tissue concentration. This is the same assumption as was used for the self-caught
freshwater analysis and, for reasons stated in the Section 2, this likely overstates the
utility-attributable contribution. Even with this upper bound assumption, the marine fish
pathway results in only a small contribution to methylmercury consumption due to the
small contribution of the U.S. power plant emissions to open ocean environments.
While the EPA CMAQ model does extend into the Atlantic and Gulf Coast
region, and could potentially be used to determine the utility-attributable portion of
mercury deposition in 2020 after the implementation of CAIR, the modeled deposition in
the open ocean region is quite small. Using CMAQ would indicate that the average post-
CAIR utility attributable portion for the Atlantic and Gulf Coast Ocean in 2020 would be
less than one percent. Another measure of open ocean mercury deposition may be drawn
from the U.S. power plant contribution to the total global mercury pool. U.S.
anthropogenic Hg emissions are estimated to account for roughly three percent of the
global total, and emissions from the U.S. power sector are estimated to account for about
one percent of total global emissions (US EPA 1997a). Since both estimates are in the
range of one percent, EPA believes that changes in power plant Hg emission in this rule
will have a very small impact through this exposure pathway.
3.1 Index of Daily Intake for Average Seafood Consumption by the General U.S.
Population
In order to describe the utility-attributable portion of Hg in the typical seafood
diet, we used the Index of Daily Intake (IDI) values, as was described in the
Effectiveness TSD (U.S. EPA, 2005a). The IDI is an index of exposure to Hg due solely
to power plants. An IDI of 1 or greater indicates that an individual exposure to mercury
from power plants is equal to or exceeds the EPA reference dose (RfD) for mercury due
solely to utility-attributable mercury exposure.
This index uses the RfD for Hg of 0.1 microgram per kilogram body weight per
day (ug/kg-bw/day) as a reference point. The IDI is defined as the ratio of exposure due
solely to power plants divided by the reference dose. The Hg exposure due solely to
power plants is calculated as
Equation 3.1: Exposure due to power plants
f consumption > € methylmercmy i
Exposure \ fa^.e > X \ concentration > X (1.5)
( due to "J - due to power plants •*
\ Power | _ fbodyweight)
plants
12
-------
Consumption rates are generally given in ounces of cooked fish, therefore, the 1.5 in this
equation reflects the increase in the concentration of Hg from cooking. It should be noted
that this implies a concentration factor of .667 of initial fish weight. Carrington et al.
(2004) report less concentration of methylmercury occurring from cooking (i.e. a
concentration factor between 1 and .737) , making the use of a concentration factor of
.667 an upper bound assumption of the amount of methylmercury in cooked fish. This
exposure value is then divided by the RfD of 0.1 to obtain the IDI.
The Effectiveness TSD (EPA 2005a) lists the IDI values for freshwater fish
consumption at various rates of consumption in Table 6-4. To create an equivalent table
for marine fish consumption, we require the concentration of methylmercury in seafood
and consumption rates. Carrington et al. (2004) provide a table (Table 3.2) of the
mercury concentration for 42 commercial finfish and shellfish seafood species. This is a
relatively comprehensive list, containing 99 percent of the market share for commercial
seafood, and includes fish in both open ocean and near shore areas. The Hg
concentration for the seafood diet of the general population can then be estimated by
summing the product of Hg concentration times market share for each species. This
produces an estimate of a mean Hg concentration of 0.104 ppm for the seafood diet of the
general population. Note that this is the Hg concentration from all sources in all
countries; U.S. power plants account for a small fraction of this total.
Consumption rates of marine fish can be obtained from the Exposure Factors
Handbook (US EPA 1997b). The mean intake of all fish species (marine, freshwater,
estuarine, and aquaculture) for the general population is 20.1 grams per day (g/day) with
an estimate of the 95 percent consumption level of 63 g/day. For marine fish alone, the
best estimate of general population consumption is 14.1 g/day. If we assume the
distribution of marine fish is proportional to total fish consumption, then the 95th
percentile of marine fish consumption would be approximately 44 g/day.
The 95th percentile consumption value of 65 g/day of all fish (marine, freshwater,
estuarine, and aquaculture) is generally consistent with Carrington and Bolger (2002)
estimate of seafood consumption, based on data from the U.S. Department of Agriculture
Continuing Survey of Food Intake by Individuals. Adjusting their data to account for the
consumption of high-frequency consumers, they estimate a 95th consumption rate for all
seafood1 of 62.7 g/day. They also estimate consumption rates of 6.8, 41.5, 62.7, 123.2,
141.3, and 206.6 g/day for the median, 90th, 95th, 99th, 99.5th, and the 99.9th percentile,
respectively. This information can be used to estimate the consumption rate at very high
levels of consumption.
Table 3.1 below lists the mercury exposure, in |ig/kg-bw/day, from all sources at
the various rates of consumption described above. Values of 0.1 or greater, which occur
1 Carrington and Bolger (2002) use records of seafood-consumption events. We assume that this includes
consumption from all fish (marine, freshwater, estuarine, and aquaculture). As such, the tables in this
section that include high-end consumption values must be interpreted as if all of the consumer's fish diet
came from marine fish.
13
-------
at consumption levels between 90 and 95 percent, imply that some individuals are
experiencing levels of Hg above the RfD due to fish consumption. It is very important to
note, however, that this is the Hg exposure from all sources. This includes natural
sources as well as man-made sources, both U.S. and from other countries. It also
assumes that the mercury concentration in fish tissue consists entirely of MeHg, biasing
the estimates upward.
Table 3.1 Mercury Exposure from Seafood Consumption from all Sources
(units in jug/kg-bw/day)
Consumption Rate
a/dav
Mean Marine Fish Intake
14.1
0.03
95% Long-Term Marine Fish Intake
44
0.10
Mean Total Fish Intake
20.1
0.05
95% Long-Term Total Fish Intake
63
0.15
Median Carrington & Bolger Seafood Consumption
6.8
0.02
90% Carrington & Bolger Seafood Consumption
41.5
0.10
95% Carrington & Bolger Seafood Consumption
62.7
0.15
99% Carrington & Bolger Seafood Consumption
123.2
0.29
99.5% Carrington & Bolger Seafood Consumption
141.3
0.33
99.9% Carrington & Bolger Seafood Consumption
206.6
0.48
Note: Mean fish tissue methylmercury concentration from all sources is 0.1 ppm.
The scaling method used for freshwater fish assumed a proportional relationship
between reductions in air deposition of Hg and MeHg concentrations in fish. This
relationship has been documented in the MMaps approach (US EPA 1997a, US EPA
2001). The MMaps model assumes that, in steady-state, changes in MeHg
concentrations in fish are proportional to changes in Hg inputs from atmospheric
deposition. This solution only applies to situations where air deposition is the only
significant source of Hg to a water body, and the physical, chemical, and biological
characteristics of the ecosystem remain constant over time. These conditions do not hold
in open ocean systems. As discussed in the previous section, assuming proportionality
would most likely overstate the change in MeHg concentration in marine systems given a
change in the atmospheric deposition of Hg. This means that assuming proportionality to
calculate the marine fish Hg levels due to utilities would likely overstate the contribution
of those utilities. This assumption could be useful, however, to produce an estimate that
likely overstates the utility-attributable Hg levels in marine fish. If this utility attributable
estimate is deemed not to represent a hazard to public health, we can reasonably conclude
that a more realistic estimate, although not currently quantifiable, would also not
represent a hazard to public health. Using the bounding assumption of proportionality
between reductions in Hg deposition and reductions in fish tissue concentrations, then we
can produce an approximation of the IDI for the 2001 base case.
Broadly speaking, global Hg deposition comes from three sources, approximately
one-third comes from natural sources (e.g., volcanoes), one-third comes from the
14
-------
emissions of modern man-made sources, and one-third from redeposition.2 Mercury
emissions from the U.S. power sector are estimated to account for about one percent of
total global emissions (US EPA 1997a). We use this one-percent as the bounding
estimate of the open ocean fish concentration reductions that could be expected from a
100 percent reduction in power plant emissions in 2001.3 This assumes proportionality
between emissions and open ocean fish concentrations, likely an overestimate, and so is
used in this bounding calculation 4 Multiplying the exposure factors in Table 3.1 by this
one percent and dividing by the RFD of 0.1 produces the IDI values for marine fish
consumption in Table 3.2 below. Since all values in Table 3.2 are all well below one, we
can conclude that Hg emissions from U.S. power plants do not cause a health concern
and do not significantly contribute to the U.S. general public's exposure to Hg due to
marine fish consumption.
Table 3.2 IDI values for Seafood Consumption for the General U.S. Population
(units in IDI values) (2001)
Consumption Rate
a/dav
Mean Marine Fish Intake
14.1
0.00
95% Long-Term Marine Fish Intake
44
0.01
Mean Total Fish Intake
20.1
0.00
95% Long-Term Total Fish Intake
63
0.01
Median Carrington & Bolger Seafood Consumption
6.8
0.00
90% Carrington & Bolger Seafood Consumption
41.5
0.01
95% Carrington & Bolger Seafood Consumption
62.7
0.01
99% Carrington & Bolger Seafood Consumption
123.2
0.03
99.5% Carrington & Bolger Seafood Consumption
141.3
0.03
99.9% Carrington & Bolger Seafood Consumption
206.6
0.05
Note: Mean fish tissue methylmercury concentration from all sources is 0.1 ppm. The
utility attributable fraction of that concentration is used for calculating IDI values.
3.2 IDI for Consumption of Species with Very High Methylmercury Concentrations
The results above use a mean fish tissue MeHg concentration of an average
seafood diet, proportional to the market share for commercial seafood. This understates
the Hg exposure from an individual who eats proportionally more high-Hg fish than the
average consumer. As an extreme assumption, we consider the MeHg concentration for a
2 see http://www.epa.gov/mercury/control_emissions/global.htm
3 Another possible measure would be data from the CMAQ model used for the Revision Rule. Using the
difference between Hg deposition in the coastal Atlantic and the Gulf of Mexico region in the base case and
the 2020 zero-out CMAQ run, it is estimated that U.S. power plants contribute an average of less than one-
percent to total deposition in this area. Averaging the entire Atlantic and Gulf Coast ocean area modeled
by CMAQ implies an average change in atmospheric deposition of 0.65 percent in 2020 using a "zero out"
of utility mercury emissions from a post-CAIR baseline. We use the estimate of the contribution of U.S.
power plant emissions to the global pool of one percent rather than the value from the CMAQ model
because it provides an upper bound estimate.
4 By 2020 after CAIR and furthermore after CAMR, this fraction would decline.
15
-------
fish with a concentration of lppm.5 This is one of the highest Hg concentrations from
marine fish and represents a type of fish that a person might potentially eat on a regular
basis.
As an extreme assumption, we consider an individual whose fish consumption
comes exclusively from a marine species with a methymercury concentration of 1.0 ppm.
Table 3.3 uses this MeHg concentration of 1.0 ppm, the one percent change in MeHg
concentration, and the various consumption levels from Table 3.1. Again, the IDI value
never exceeds one, even for a 99.9 percent consumption level. Given that this assumes
someone exclusively eating a fish with one of the highest marine Hg concentrations and
eating at the highest consumption rate, it is reasonable to conclude that U.S. power plants
do not significantly contribute to the U.S. exposure to Hg from marine fish consumption.
Table 3.3 IDI values for Exclusive Consumption of Marine Species with High
Methylmercury Concentration
(units in IDI values)
Consumption Rate
a/dav
Mean Marine Fish Intake
14.1
0.03
95% Long-Term Marine Fish Intake
44
0.10
Mean Total Fish Intake
20.1
0.05
95% Long-Term Total Fish Intake
63
0.15
Median Carrington & Bolger Seafood Consumption
6.8
0.02
90% Carrington & Bolger Seafood Consumption
41.5
0.10
95% Carrington & Bolger Seafood Consumption
62.7
0.15
99% Carrington & Bolger Seafood Consumption
123.2
0.29
99.5% Carrington & Bolger Seafood Consumption
141.3
0.33
99.9% Carrington & Bolger Seafood Consumption
206.6
0.48
Note: High fish tissue methylmercury concentration from all sources is 1.0 ppm. The
utility attributable fraction of that concentration is used for calculating IDI values.
3.4 Conclusions Regarding the Marine Pathway
In summary, EPA believes that the utility-attributable portion of Hg in the
seafood diet is not significant. However, EPA is soliciting comments on the IDI
constructed above as well as other methods to account for commercially important
marine fish that have relatively high Hg concentrations.
4. Estuarine Near-Shore Exposure Pathway
5 Carrington, Montwill, and Bolger (2004) report a mean Hg concentration of 0.97 ppm and a median
concentration of 0.86 ppm for swordfish.
16
-------
Despite the lack of a comprehensive quantitative analysis, EPA finds that the
available data support our position that the utility attributable U.S. population wide
exposure to methylmercury from estuarine fish and shellfish will likely be small relative
to that from recreational self-caught freshwater fish described in previous sections of the
Effectiveness TSD. In general, coastal finfish species of the same age/size and trophic
level as freshwater fish have relatively lower mercury concentrations (Mason and Gill,
2005).
Overall methylmercury exposure from coastal fish and shellfish is likely smaller
than exposure from a comparable amount of freshwater fish. First, coastal fish of the
same age/size and trophic level as freshwater fish generally have relatively lower
mercury concentrations (Mason and Gill, 2005). Second, shellfish tend to have a lower
fraction of methylmercury relative to their total mercury burden than freshwater fish
(Bloom et al. 1992, Joiris et al. 2000, Mikac et al. 1985), which further lowers
methylmercury exposure from all coastal species.
When compared to overall landings of marine fish in the U.S., domestically
caught estuarine fish and shellfish (defined as those harvested within 3 miles from shore)
make up only 38 percent of the total 2001 commercial fish and shellfish landings in the
U.S. (NMFS 2002) Based on the average mercury content of estuarine fish and shellfish,
a conservative (high) estimate of total mercury exposure is approximately 23 percent of
the total mercury intake from the U.S. Commercial Seafood Market (described below). A
comparably larger fraction of the seafood consumed in the U.S. originates from open
ocean/marine areas discussed in section 2. The utility attributable fraction of this 23
percent will also be very small as indicated by data showing that: 1) Utility attributable
mercury deposition in near-shore areas is small relative to inland areas where recreational
fishing occurs; 2) Legacy mercury sources and inputs from watershed areas in most
coastal systems will further lower the percentage of utility attributable mercury in
estuarine fish and shellfish.
The following sections outline EPA's rationale for concluding that utility
attributable human exposure resulting from consumption of estuarine fish is small. First,
we discuss the major scientific uncertainties in the estuarine exposure pathway. Second,
we review the data on consumption of estuarine fish and shellfish in the United States
and the relative exposure to mercury from this source compared to other pathways.
Finally, we review the utility attributable deposition estimates in near-shore area of the
United States and discuss how this may affect concentrations in estuarine species.
4.1 Uncertainty in the Near-Shore Fish Exposure Pathway
EPA believes that the state of the science currently does not support a national-
scale quantitative analysis for this component of the exposure pathway. The most
important difficulty in conducting this analysis is determining the response of coastal and
estuarine systems to changes in atmospheric mercury deposition. Some studies have
assumed an instantaneous proportional relationship between declines in deposition and
concentrations in estuarine fish to complete their exposure analysis. However, such an
17
-------
assumption has not been endorsed by EPA or the scientific community as an appropriate
method for characterizing the effects of emissions reductions on estuarine fish
concentrations.
A proportional relationship between mercury deposition and the methylmercury
concentration in fish was developed and has been evaluated for application for certain
freshwater water bodies across the U.S. if air deposition is the only significant source of
mercury to a water body (US EPA 1997a, US EPA 2001). As described above, such a
relationship is thought to be an overestimate for marine systems so can be used in a
bounding calculation. Applying such a relationship to estuarine systems extends this
assumption beyond the realm of application for which this model was developed for
freshwater systems and, unlike marine systems, it is not known whether it would likely be
an overestimate for all estuaries. External sources of methylmercury to estuaries, like
other aquatic systems, are typically small such that in situ methylmercury production
accounts for the majority of mercury in fish (Mason and Benoit 2003). Mercury cycling
in estuarine areas differs significantly from both the open-ocean and freshwater
environments. Because estuaries are much shallower than open ocean areas, active
methylmercury production in estuarine sediments is an important source of mercury for
fish and shellfish residing in these regions (Hammerschmidt et al, 2004, Cossa and
Gobeil, 2000, Lawson et al., 1998). Unlike freshwater systems, most estuaries have
significant inputs of tidal water, which generally dilutes incoming mercury from rivers,
streams, and direct atmospheric deposition. In addition, because of the saline
environments found in estuaries, the geochemistry of these systems differs significantly
from inland lakes, affecting the rate and magnitude of methylmercury production (Heyes
et al., 2005). There is, however, a paucity of data and modeling describing mercury
cycling in coastal areas compared to freshwater and marine environments. This limits
our ability to develop a comprehensive quantitative assessment of the exposure pathway
at this time.
Some empirical observations of mercury cycling in coastal ecosystems often show
total mercury concentrations are not necessarily good predictors of ambient
methylmercury concentrations (Benoit et al., 2003). In addition, such an estimate likely
overestimates the response of coastal ecosystems to changes in atmospheric deposition if
all other ecosystem characteristics remain constant. A number of studies have shown that
the production and bioavailability of mercury in coastal ecosystems is a function of
environmental characteristics like total organic carbon, sulfides concentrations in water
and sediments and temperature (Benoit et al, 1999, Sunderland et al, 2005). When the
methylation potential of estuarine systems is limited by some environmental or
geochemical characteristic of the ecosystem, small changes in inorganic mercury
deposition from the atmosphere are unlikely to significantly impact methylation rates and
ultimately fish methylmercury concentrations. Fish methylmercury concentrations in
coastal ecosystems that have been significantly impacted by legacy mercury sources (e.g.,
San Francisco Bay, CA, Lavaca Bay, TX) are also unlikely to be affected by small
changes in atmospheric deposition because they receive the majority of their mercury
from watershed sources and from contaminated sediments (Bloom et al, 1999, Sager,
2002, Conaway et al., 2003).
18
-------
Most coastal areas in the United States are highly populated and therefore receive
significant mercury loadings from land based sources through wastewater effluents and
other industrial contamination. In addition, most coastal regions are likely to have
relatively deep active sediment layers relative to freshwater systems due to mixing
processes that occur when rivers and tidal waters meet. Effectively, a deeper active
sediment layer allows coastal systems to store legacy contaminants and allows
historically released mercury to interact with the water column, undergo methylation and
accumulate in the food-web, slowing the ecosystems response to changes in mercury
inputs (Sunderland et al., 2004). Coastal systems with deep active sediment layers that
contain large amounts of legacy Hg from past anthropogenic contamination and large
contributions from watershed-based Hg sources will respond very slowly to changes in
emissions from utilities, and the magnitude of this response will also likely be small.
Finally, the life cycles of marine species (fraction of time spent in near-shore
relative to pelagic environments) needs to be considered to accurately model mercury
bioaccumulation. Many of the coastal and estuarine species, listed in Table 4.1 below,
spend a portion of their lifecycle in near-shore areas and the remainder in open-ocean
regions. This behavior is likely to further lower the significance of utility-attributable
mercury in the U.S. on methylmercury concentrations in these species.
Based on this, EPA believes that a simple linear relationship that assumes
instantaneous changes in Hg levels in fish and shellfish across all estuarine systems in the
U.S. is proportional to changes in atmospheric deposition does not appropriate reflect the
best available information on mercury cycling in coastal ecosystems. As such, we cannot
conduct a comprehensive, national-scale quantitative analysis of this pathway for all
estuary and coastal regions.6
4.2 Consumption of Near-Shore Species in the United States and Associated
Mercury Exposure
To assess the total mercury exposure from consumption of fish and shellfish in
the United States, we analyzed commercial landings data from the National Marine
6 EPA has distinguished between estuarine and open ocean (marine systems) in this analysis. Bounding
calculations were possible for marine systems because of the relative homogeneity of human influences on
mercury cycling in marine systems compared to estuarine systems. Specifically, open-ocean environments
are isolated from local variability in methylmercury formation rates and biotic concentrations that are
commonly observed in estuaries. This is because open ocean environments do not exhibit the same
variability in geochemical properties affecting methylmercury formation as estuaries (i.e., the deep water
column means that methylmercury production in the sediments has a negligible effect on water column
concentrations). Finally, we believe that the analysis for the open ocean environment is justifiable because
it is a sensitivity analysis taking into account the limited nature of the data on mercury fate and transport in
the open-ocean. Given the regional differences in utility attributable mercury deposition to estuaries in the
US, the fraction of mercury deposited that is converted to methylmercury, and fish methylmercury
concentrations, we felt that it would be unrealistic to attempt to develop a national estimate of changes in
mercury exposure resulting from changes in utility mercury deposition in all estuaries across the US at this
time because of the heterogeneity of estuarine systems and that it would be more appropriate to present a
qualitative assessment.
19
-------
Fisheries Service (NMFS, 2002). For this analysis, we define estuarine species as those
caught within 0-3 miles of the U.S. coastline. Total commercial fish production in 2001
was 4.4 million tonnes, with 3.9 million tonnes as finfish and 0.5 million tonnes as
shellfish. Of the total commercial catch, 63% of the shellfish and 34% of the finfish in
2001 were caught within 0-3 miles from the shore. This estimate includes some species,
such as menhaden, that are also used for animal feed or other purposes. As a sensitivity
analysis we subtracted out menhaden, resulting in a somewhat lower percentage of finfish
(excluding menhaden) coming from the near-shore area.
Using the product of mean mercury concentrations and percent of the commercial
market for each species from Carrington et al. (2004), we are able to rank the Hg
exposure from the top 25 species in the US commercial seafood market (Table 4.1). We
then multiply each finfish or shellfish species by the overall fraction caught in coastal
waters to provide a conservative estimate of the relative mercury exposure from coastal
species. As illustrated in Table 4.1, the fraction of mercury exposure from fish caught in
near-shore waters in the U.S. commercial seafood market is roughly 35 percent of the
total. However, the utility-attributable fraction of mercury exposure from fish and
shellfish will be much smaller as outlined below.
This estimate of total exposure from estuarine species is thought to be
conservative (high) because it is based on total mercury concentrations in fish and
shellfish rather than methylmercury concentrations, the mercury species in fish that is
toxicologically most significant. It is fairly well established that shellfish in general will
have a lower fraction of total mercury present as methylmercury (%MeHg) in their
tissues than predatory fish. For example, Mikac et al. (1985) found that marine mussels
had 5-27%) of mercury in the organic form, while Mason et al. (2000) found between 50-
80%o of the mercury in crayfish was present as MeHg. Thus, it appears that there is
considerable variation in estimates of the fraction of methylmercury contained within the
tissue of invertebrates. This lowers methylmercury exposure typically associated with
certain species by assuming measured total mercury concentrations are equivalent to
methylmercury concentrations.
Table 4.1. Estimated mercury intake from species harvested within three miles from
shore based on the Hg content of the top 25 species in the U.S. Commercial Seafood
Market.
Species
% Hg Intake
by Species1
% Coastal
Estimate
Tuna, albacore (canned)
17.9%
6.1%
Tuna, light (canned)
15.9%
5.4%
Haddock, hake, and monkfish
8.7%
3.0%
Pollock
7.1%
2.4%
Tuna, fresh
6.5%
2.2%
Cod
6.5%
2.2%
Swordfish
3.9%
1.3%
Lobsters, American
3.8%
2.4%
20
-------
Catfish
3.0%
1.0%
Crabs
2.8%
1.8%
Lingcod and scorpionfish
2.5%
0.9%
Salmon
2.2%
0.8%
Flatfish
2.0%
0.7%
Halibut
1.9%
0.6%
Shrimp
1.7%
1.1%
Bass, saltwater
1.5%
0.5%
Anchovies, herring, and shad
1.5%
0.5%
Shark
1.2%
0.4%
Orange Rougy
1.0%
0.4%
Lobsters, spiny
1.0%
0.6%
Grouper
0.9%
0.3%
Snapper, porgy, and sheepshead
0.7%
0.3%
Squid
0.7%
0.2%
Sablefish
0.7%
0.2%
Skate
0.5%
0.2%
Percent Total
96.2%
35.4%
species in Carrington et al., 2004.
2 Percent coastal intake was estimated using 2001 landings data averages of 34% of all
finfish harvested from 0-3 miles from shore (near-shore areas) and 63% of all shellfish
(NMFS 2002).
4.3 Utility Attributable Mercury Deposition in Near-shore Areas
The significance of utility-attributable mercury for estuarine species is expected to
be a small component of their overall body burden because of the prevalence of legacy
contaminants, watershed based mercury sources and other atmospheric mercury sources
in these regions. When analyzing mercury exposure in most coastal regions, it is
apparent that the utility-attributable portion of this exposure will be limited. For
example, the Pacific Coast accounts for over 65 percent of the productions of fish species
listed in table 4-1 (NMFS 2002) and will be negligibly impacted by utility-attributable
mercury deposition. This is reinforced by EPA's deposition modeling, which indicates
that coal-fired utilities will not significantly impact Hg levels on the Pacific Coast. As a
bounding analysis, if we recalculate the potential fraction of total mercury exposure that
can then be affected by changes in utility deposition from Table 4.1, we are left with 35
percent of the original 35 percent of total exposure from the U.S. commercial seafood
market. In other words, only slightly more than 12 percent of the total mercury exposure
from the consumption of marine fish can be from the consumption of coastal and
estuarine fish from the Gulf coast, southeast Atlantic and New England. Only a small
fraction of that 12 percent can be attributed to utility-attributable mercury as discussed
below because utility deposition comprises only a small fraction of the mercury that is
present and available for methylation in the Gulf coast, southeast Atlantic and New
England.
21
-------
EPA's modeling of deposition due to mercury emissions from Utilities in 2001
suggests that Utilities represent between 0 and 10 percent of the total deposition in the
Gulf Coast, 5 to 20 percent in the Southeast Atlantic, 0 to 10 percent in the New England
coastline, and 20 to 30 percent in the Chesapeake Bay area. In 2020 after CAIR (and
furthermore after CAMR) the deposition due to mercury emission from utilities falls to a
range between 1 percent and 12 percent for the Gulf and Southeast Atlantic coasts. Most
Gulf and Southeast Atlantic Coast areas will experience approximately 2 percent of total
deposition from utilities. Most areas of the New England coastline will experience
slightly over 2 percent of total deposition from utilities, with a maximum utility-
attributable deposition of 7 percent. After CAIR in 2020, utility attributable deposition
will be highest close to shore off the Northeastern States but will still be less than 7
percent, compared to up to 19 percent in the highest freshwater fishing areas (see Table
2.6 of the Effectiveness TSD).
The Chesapeake Bay represents 6.5 percent of total U.S. landings (NMFS 2002)
and is likely one of estuarine ecosystems most sensitive to atmospherically deposited
mercury from US power plants because of the significance of coal fired power plants (20-
30 percent of total in 2001) to overall atmospheric mercury deposition rates and its
relatively small watershed to water surface area ratio (resulting in a greater importance of
the atmospheric pathway compared to many other estuary ecosystems). Utility
attributable deposition after CAIR is expected to be in the range of 8.5 percent (and will
be reduced even further following CAMR) of the total deposition. The percent of total
deposition attributable to utilities for the Chesapeake Bay watershed is somewhat less
that the 8.5 percent for the Chesapeake Bay itself. EPA's analysis of projected mercury
deposition rates after implementation of CAIR and CAMR show some of the largest
reductions in mercury deposition are expected to occur in the Chesapeake Bay region.
4.4 Conclusions Regarding the Near-Shore Pathway
As discussed above, populated coastal regions like the Chesapeake Bay and
Baltimore Harbor (Mason and Lawrence, 1999) will receive significant land-based
mercury inputs from wastewater effluents, municipal waste discharges and historical
mercury contamination that are slowly leaching from the watershed. In addition, legacy
mercury stored in the active sediment layers of these systems will continue to supply
coastal systems with inorganic mercury for many decades. These types of inputs to
coastal ecosystems lower the overall significance of small changes in atmospheric
deposition to the overall magnitude of mercury in coastal systems. These estimates of
utility attributable deposition following CAIR, combined with the magnitude of
commercial fish landings data from the Eastern U.S. coastline further reinforces that
exposure from this source is small.
Although we are not currently able to quantitatively estimate IDI values associated with
various levels of consumption of estuarine fish, the available information suggests that
they will be bounded by the freshwater recreational/subsistence IDI values presented in
the Effectiveness TSD (US EPA 2005a). We therefore continue to use the IDI values in
Table 6.4 of the Effectiveness TSD as our estimate of the maximum individual risk. In
22
-------
addition, we will use the utility attributable MeHg in the 85th percentile freshwater fish as
a bound of the utility attributable MeHg in estuarine fish.
5. Aquaculture Exposure Pathway
Due to the unique nature of the aquaculture pathway and gaps in the available
data, it is not possible to conduct a quantitative assessment of the utility-attributable
exposure from mercury in farm-raised fish. However, based on the available information
we are able to conclude that the contribution of aquaculture to utility-attributable mercury
exposure is small. By breaking this potential exposure pathway into three main
components that are discussed below, we outline our rationale for determining that
utility-attributable exposure due to consumption of aquaculture is small.
First, we find that farm-raised fish accounts for only 10 percent of total
commercial fish production in the U.S (NMFS 2002), which limits the relative
importance of consumption of aquaculture species harvested in the US that could
potentially be affected by U.S. power plant mercury deposition compared to other fish
sources. Second, the main aquaculture species are salmon and catfish, which are
generally low in mercury. Third, the fraction of the mercury attributable to power plants
in farm-raised fish is likely much smaller than other wild fish because their diet is
specifically engineered from a number of different protein sources that may not
necessarily originate in the U.S. and are therefore not appreciably affected by utility-
attributable mercury. Further, the fish meal used to make fish feed is usually derived
from smaller fish, which are generally lower in mercury than are larger fish. Because the
mercury residue in fish tissues is mainly the result of dietary biomagnification, uptake of
mercury from the water column (which may potentially be affected by power plant
emissions) by farm-raised is expected to be small. Given this information, it is
reasonable to conclude utility-attributable mercury from this particular pathway will not
add significantly to the overall population body burden.
5.1 Quantity of Fish Produced in Marine and Freshwater Aquaculture
In 2001, total aquaculture production in the U.S. was 371,470 metric tons,
compared to total commercial finfish that were domestically caught in the same year of
3,738,769 metric tons (NMFS 2002).7 Because only U.S. aquaculture is relevant from
the standpoint of assessing utility-attributable mercury exposure, these numbers indicate
that U.S. aquaculture is at most 10 percent of the other fish landings within the country.
In short, potential utility-attributable mercury exposure from U.S. aquaculture is small
relative to other types of fisheries.
5.2 Mercury in Aquaculture Fish from Direct Deposition versus Fish Feed
It is well known that the major pathway of mercury accumulation in fish is
through the diet rather than uptake from water (e.g., see review by Rodgers 1994). The
7 This estimate includes some species, such as menhaden, that are also used for animal feed or other
purposes.
23
-------
process whereby contaminants accumulate in organisms to many times the concentrations
found in the ambient environment as larger organisms eat smaller organisms is known as
"biomagnification." For example, it is not uncommon to see concentrations of mercury in
predatory wild fish that are a million times higher than the methylmercury concentration
in water because of dietary biomagnification. Direct uptake of a contaminant from the
ambient environment is known as "bioconcentration" and both of these processes
together are known as "bioaccumulation."
Because the major pathway of mercury accumulation in fish tissue occurs through
biomagnification of methylmercury, concentrations of mercury in the diet of aquaculture
species are most relevant for determining their ultimate tissue Hg residue. In order for
utility-attributable mercury to be significant in farm-raised fish, a significant fraction of
the diet would need to be composed of fish that are affected by utility-attributable
mercury. This is not the case as illustrated below.
Two key ingredients of fish feed, for which economic data exist, are fish meal and
fish oils. In 2002, approximately 28 percent (148 million pounds) of the U.S. supply of
fish meal was imported and the remaining 72 percent (389 million pounds) was
domestically produced (NMFS, 2003, p. 83). Data on the U.S. supply of fish oils shows
that it is roughly 50 percent imported and 50 percent domestically produced (NMFS,
2003, p. 83). The mercury concentration in domestically produced fish meal and fish oil
depends on where the fish were caught. Only 16 percent of U.S. domestic commercial
landings occur on the Atlantic coast, versus 83 percent on the Gulf and Pacific coasts
(NMFS, 2003, p. 6). Of U.S. commercial landings, only 36 percent occur within 0 to 3
miles from U.S. shores, while 61 percent occur between 3 and 200 miles of U.S. shores
(NMFS, 2003, p. 19). Because Gulf coast, U.S. Pacific coast, internationally landed fish,
and Atlantic coast fish caught greater than 3 miles off shore are expected to see a very
small decrease in exposure to U.S. utility-attributable Hg emissions, only a small fraction
of the U.S. supply of fish meal and fish oil is expected to see a significant change in Hg
exposure. Therefore, while the dynamics of Hg contained in imported fish meal and fish
oil is largely unknown, the mercury contained in this fish is likely from the global pool,
of which US power plants represent about 1 percent.
In contrast to dietary mercury accumulation from fish oil and meal, much smaller
contributions are expected from ambient Hg concentration and wild food sources and
ingredients in fish food other than fish oil and fish meal. Any geographic analysis of
deposition with regard to fish farm location would assume that reductions in atmospheric
deposition would directly impact farm-raised fish Hg loads through bioconcentration,
which is not the case as described in detail above. Regional changes in Hg emissions and
subsequent atmospheric deposition resulting from the Revision Rule are not expected to
have a direct impact on the content of Hg in farm-raised fish.
In summary, EPA finds that it is the location of the fish caught to make fish feed
that is relevant, as opposed to the location of aquaculture farms.
5.3. Summary of the Aquaculture Exposure Pathway
24
-------
In summary, the limited available data indicate that Hg body burdens in
aquaculture fish are less than or equal to their wild counterparts, and mercury in
aquaculture fish is predominantly the result of the mercury content of fish feed. The fish
feed used in US aquaculture comes from a variety of sources including the Gulf Coast,
the Atlantic Ocean, US Pacific Coast, and internationally landed fish - fish that have
small utility-attributable mercury levels. Therefore, while there are insufficient data
available to conduct a full quantitative assessment, EPA believes that the existing
information support the conclusion that that utility-attributable mercury exposure from
aquaculture is small and that all individuals or groups of individuals who consume
aquaculture fish are likely to ingest utility-attributable methylmercury at levels that are
reflective of the levels ingested from the types of marine fish that comprise aquaculture
fish feed.
6. Commercial Freshwater Exposure Pathway
EPA's claim that freshwater commercial fish are not a significant pathway is valid
since 17 million pounds/year (lb/yr) is small when compared to recreational freshwater
fish consumption of 377 million lb/yr, or 22 times the Great Lakes commercial haul.89
Further even though utility attributable deposition is comparatively higher around the
Great Lakes and the bordering areas (including the states of Michigan, Indiana, Illinois,
and Ohio and other surrounding areas) in comparison with the rest of the United States, it
is still only a small percentage of mercury deposition from all sources. The typical
percent of total deposition that is attributable to utilities in these areas is approximately
10 percent. Thus, following the assumptions in Mercury Maps, only approximately 10
percent of the mercury in the fish found in this area is attributable to utilities. In addition,
the areas in the Great Lakes that are affected do not experience a disproportionately high
deposition rate compared to the surrounding land area, where recreational freshwater fish
are caught, so the commercial freshwater pathway is still expected to be small relative to
the recreational/subsistence freshwater pathway.
Because exposure is determined at the population level, the determination for
including this pathway is whether adding it (population exposed times exposure rate) will
significantly alter the cumulative distribution of total exposure rates. As described above,
the commercial freshwater harvest is small compared to recreational freshwater
consumption, the percent of utility-attributable deposition in the primary commercial
freshwater harvesting area is 10 percent, and those levels are not disproportionate to the
areas for recreational freshwater harvest. These facts lead the EPA to conclude that
including the commercial freshwater pathway in the exposure model would result in a
8 Recreational freshwater fish consumption was calculated by multiplying the population of fishers
(27,900,000 (US FWS 2002)) by both the percent of recreational fishers who consume their catch (0.84
(average of values presented in West at al. 1989, Chemrisk 1991, and West et al 1993 as presented in EPA
1997b)), and the number of friends and family with whom the average recreational fisher shares his or her
catch (2.5 (EPA 1997b)) and converting to pounds.
9 The Great Lakes commercial haul is 0.2% of the total commercial haul of finfish (8.2 billion pounds)
(NMFS 2002). The marine haul represents the most significant fraction of the total haul and is discussed
elsewhere.
25
-------
relatively small change in the general population level exposure estimate. In addition,
there is no reason to include this pathway to address its effect on the higher end groups
(e.g. subsistence fishers) because most of the fish they eat is self-caught and it is highly
unlikely, given the nature of their fishing activity, that more than a small fraction, if any,
of their consumption is comprised of commercially caught freshwater fish. As such, we
believe that the IDI values for this pathway are bounded by the freshwater
recreational/subsistence IDI values.
7. Joint Consumption
In order to examine utility-attributable Hg exposure from total fish consumption
quantitatively, it would be necessary to have information on the distribution of
consumption of each type of fish - recreational freshwater, commercial freshwater,
recreational saltwater, etc - as well as utility-attributable MeHg concentrations (either
sufficiently accurate or upper-bound) for each type of fish. If we were able to identify
the consumption of each type of fish as well utility-attributable MeHg concentration for
each type of fish, then the IDI values from each type of fish could be calculated and
added together to arrive at a total IDI value. Currently no such data exists. Regardless,
for the reasons described above, EPA maintains that self-caught freshwater fish
consumption represents the most significant exposure pathway for the populations with
the highest utility-attributable exposure.
At any given total fish consumption rate noted in our analyses, introducing
aquaculture, marine, or estuarine fish into the diet of a self-caught freshwater fish
consumer necessarily implies reducing consumption of self-caught freshwater fish (e.g.,
in order to maintain the same total fish consumption rate). As discussed in previous
sections, because utilities contribute more Hg to freshwater fish species than to any other
fish species, such substitution implies a lower IDI than is associated with consumption of
self-caught freshwater fish alone, supporting the assertion that self caught freshwater
consumption represents the primary source of utility-attributable Hg exposure. As can be
seen in the seventh column of Table 6-4 of the Effectiveness TSD and Table 3.2 of this
document, for any given consumption rate, the estimated IDI for self caught freshwater
fish consumption is higher than the estimated IDI for marine fish consumption (for fish
tissue methylmercury percentiles 50 percent or greater). Hence, for any given
consumption rate, consumption of self-caught freshwater fish alone leads to a higher IDI
than that of any other combination of fish, supporting our decision to focus our analysis
on consumption of self-caught freshwater fish.
Table 6.4 of the Effectiveness TSD (US EPA 2005a) shows an array of
consumption values combined with percentiles of methylmercury concentration in
freshwater fish. Results for 2020 with CAIR indicate that estimated IDIs are all well
below 1 for the first three consumption rates. Estimated IDIs are over one for 99th
percentile recreational fishers and mean subsistence Native Americans only when all of
the fish consumed has MeHg concentrations at the 99th percent level, a convergence of
26
-------
factors which is unlikely to occur.10 See 70 FR 16024. While estimated IDIs for the 95th
(170 g/day) and 99th percentile (295 (g/day) subsistence Native American consumers are
above one for lower percentile MeHg concentration fish, it is unlikely that these
consumers would add significant amounts of non-self-caught freshwater fish to their diets
over the course of a year, but rather would substitute fish, again supporting our focusing
on the consumption of self-caught freshwater fish. Finally, the IDI values for the
combinations of fish consumption rates and MeHg concentrations bordering the
combinations with IDIs above one are sufficiently below one that it is unlikely that a
consumer in these combinations would add a sufficient amount of other fish (with lower
utility-attributable MeHg concentrations than freshwater fish) to their freshwater fish diet
to cause their IDI to exceed one.
Further, we have no evidence that high end consumers of self caught fish also
consume other types of fish. It is highly unlikely that subsistence individuals eat 170
g/day or 295 g/day of self caught freshwater fish and consume significant quantities of
marine fish. Even if we were to assume that these consumers do eat additional fish, the
additional MeHg ingested by these consumers is small as we have shown above.
8. Health Benefits and Costs
8.1 Introduction
Below we describe a bounding analysis that includes the exposure pathways
described in this TSD and self-caught freshwater fish. In this calculation, we use mean or
central tendency estimates of variables when available. However, for several variables we
are not able to provide such an estimate and we therefore use a conservative estimate that
would overestimate the utility-attributable mercury exposure. The final calculation
therefore represents a combination of central tendency estimates for some variables and
conservative estimates for other variables. This analysis is a bounding analysis in the
sense that the final health benefit estimate presented below is very likely to be above the
true health benefits of improved neurological performance associated with reducing
mercury emissions from power plants because of the compounding use of conservative
estimates for certain variables.11 The bounding analysis approach supports our reasonable
belief that the costs of reducing mercury emissions beyond CAIR under section 112 from
power plants outweigh the health benefits of reduced utility-attributable mercury
exposure.
10 In addition to the particular combinations shown in the table, there are a multitude of other combinations
of fish consumption rates and methylmercury concentrations possible. For example, Table 6.4 shows that
the 99th percentile recreational fisher consuming 47 grams per day of fish with the 99th percentile of utility
attributable mercury concentration would have an IDI value of 1.12. An individual consuming slightly less
than 47 grams per day of fish with the 99th percentile of utility attributable mercury concentration would
also have an IDI value greater than 1. Similarly, an individual consuming 47 grams per day of fish with
slightly less than the 99th percentile of utility attributable mercury concentration would also have an IDI
value greater than 1.
11 Note that the assumptions in this bounding analysis are different from those in the marine cycling
section and this section should not be considered as an extension of that analysis.
27
-------
8.2 Outline
The benefit calculation will follow directly from the IDI values presented in Table
6-4 of the Effectiveness TSD (US EPA 2005a) and Table 3.2 of this TSD. These IDI
values represent an estimate of exposure due to power plants and, equivalently, the
reduction in exposure that would occur if power plant mercury emissions were
eliminated. Using a dose-response relationship, we translate these IDI values into
neurological improvements, using intelligence quotient (IQ) points as a surrogate. We
then estimate the monetized value of these IQ point increments and discount these future
monetized benefits to account for the ecosystem response time. These discounted benefits
can then be compared with the discounted costs, taking into account the important
uncertainties described elsewhere.
8.3 IDI values
We use the IDI values from Table 6-4 of the Effectiveness TSD and this TSD. For
the purposes of estimating total benefits of reducing mercury emissions from power
plants, we use the mean fish consumption rate for each pathway. The mean consumption
rate is 8 g/day freshwater fish and is 14.1 g/day for the marine pathway (US EPA 1997b).
The IDI values can be converted to ppm of mercury in hair by first multiplying it
times 5.4, which is a conversion factor between the mercury in blood (in ppb) and
exposure of someone exposed to mercury at the RfD. In other words, an exposure of 0.1
|ig/k g/day of MeHg (i.e., the RfD) is associated with 5.4 ppb in blood (see the one-
compartment model for MeHg in IRIS (US EPA 2002a) for more detail12), and then
dividing by the 4, which is the conversion ratio between blood mercury and hair mercury
(250 ppb blood mercury = 1 ppm hair mercury).13
8.4 IQ decrements
EPA has chosen to focus on quantification of intelligence quotient (IQ)
decrements associated with prenatal mercury exposure as the initial endpoint for
12 The one compartment model converts the concentration of MeHg in blood (in |ig/L) to daily dietary
intake (in |ig/kg/day) using the equation d = (c*b*V)/(A*f*bw), where c is the blood concentration (|ig/L).
b is the elimination constant (days1), V is the volume of blood (L), A is the absorption factor (unitless), f is
the fraction of absorbed dose taken up by blood (unitless), and bw is the body weight (Kg). Using the
recommend values in IRIS (b=0.014, V=5, A=0.95, f=0,059, bw=67), substituting the RfD value of 0.1
|ig/kg/day for d and solving for c, we arrive at a conversion factor of 5.4. Note that others have equated the
RfD to a maternal blood equivalent of 5.8 ppb. This estimate does not take into account the fact that the
RfD was rounded to one significant digit and was not based on a single measure for the RfD critical
endpoint. (see Table 2 at http://www.epa.gov/iris/subst/0073.htm) Rather, EPA based this RfD for this
assessment on several scores from the Faroes measures, with supporting analyses from the New Zealand
study, and the integrative analysis of all three studies. We therefore take the RfD of 0.1 ug/kg/day as the
starting point of this calculation and then apply the one compartment model.
13 http://www.epa.gov/iris/subst/0073.htm
28
-------
quantification and valuation of mercury health benefits.14 The IQ dose-response analysis
uses data from three major prospective studies investigating potential neurotoxicity of
low-level, chronic mercury exposure. Epidemiological studies of prenatal mercury
exposure conducted in the Faroe Islands (Grandjean et al. 1997), New Zealand
(Kjellstrom et al. 1989, Crump et al. 1998), and the Seychelles Islands (Davidson et al.
1998, Myers et al. 2003) have examined neurodevelopmental outcomes through the
administration of tests of cognitive functioning.
A statistical analysis was conducted to integrate data from the three studies to
produce a single estimate of the IQ dose-response relationship. Details of the analysis,
including statistical model formulation, selection of input values, results and sensitivity
analysis are reported in Ryan (2005). For the analysis in this section, EPA is using a
linear model that goes through the origin to fit population-level dose-response
relationships to the pooled data from the three studies. The application of a linear model
should not be interpreted to suggest that any of the three studies used have data showing
health effects from methylmercury exposure at or below the RfD.15 It is also important to
note that the use of a linear model applied to all exposed individuals is done for purposes
of developing an upper bound estimate of the IQ detrimental effect of MeHg. In effect, it
assumes that all exposed individuals are exposed above the RfD. 16 This assumption
14 There is limited evidence directly linking IQ and methylmercury exposure in the three large
epidemiological studies that were evaluated by the NAS and EPA. Based on its evaluation of the three
studies, EPA believes that children who are prenatally exposed to low concentrations of methylmercury
may be at increased risk of poor performance on neurobehavioral tests, such as those measuring attention,
fine motor function, language skills, visual-spatial abilities (like drawing), and verbal memory. Fortius
analysis, EPA is adopting IQ as a surrogate for the neurobehavioral endpoints that NAS and EPA relied
upon for the RfD.
In the Faroes Island Study, a full scale IQ evaluation was not conducted. However, two core
subtests were evaluated (Similarities and Block design) and one supplementary test was conducted (Digit
Span). The Similarities and Block Design tests are reported to be well correlated with the full WISC-R
battery (0.885, see Bellinger (2005)), but how the Digit Span test relates is not reported. In the EPA
analysis, we assume that it relates similarly. In the Faroes study, performance scores on the Similarities and
Block Design tests were not shown to be statistically related to cord blood or maternal mercury levels; the
Digit Span test did show a statistical relationship with cord blood mercury.
Both the New Zealand and Seychelles study administered the WISC IQ test (WISC III in
Seychelles, WISC R in New Zealand). A reanalysis of the New Zealand data found a positive association,
but it was not statistically significant. No significant associations were seen in the Seychelles study. As
displayed in Figure 5 of Ryan (2005), the confidence intervals for full scale IQ in both these studies include
zero. However, Ryan conducted an integrative analysis, combining results from all three studies. When
combined, the statistical power of the analysis increases. While the size of the dose-response relationship
declined relative to past studies with a statistically significant finding, Ryan found a statistically significant
relationship between IQ and mercury. The confidence interval did not include zero.
15 The RfD is an estimate (with uncertainty spanning perhaps an order of magnitude) of a daily exposure to
the human population (including sensitive subgroups) that is likely to be without an appreciable risk of
deleterious effects during a lifetime (EPA 2002). EPA believes that exposures at or below the RfD are
unlikely to be associated with an appreciable risk of deleterious effects. It is important to note, however,
that the RfD does not define an exposure level corresponding to zero risk; mercury exposure near or below
the RfD could pose a very low level of risk which EPA deems to be non-appreciable. It is also important to
note that the RfD does not define a bright line, above which individuals are at risk of adverse effect.
16 From recent data (MMWR November 5, 2004 / 53(43);1018-1020,
http://www.cdc.gov/mmwr/preview/mmwrhtml/mm5343a5.htm), we know that 5.7% of women of
29
-------
produces an upper bound, but is not consistent with the fact that most of the U.S.
population is below the RfD and at these lower levels of exposure EPA believes that
there is not an appreciable risk of deleterious effects occurring.
The model makes use of dose-response coefficients for IQ. These coefficients
express a central estimate of the average reduction in children's scores in tests of IQ (or
other tests of cognitive performance) for a one unit change in the mercury body burden of
the mother during pregnancy. The model also incorporates coefficients for other
cognitive tests conducted in the studies, in an effort to obtain more robust estimates of the
IQ relationship that account for within-study (endpoint-to-endpoint) variability as well as
variability across studies. A Bayesian hierarchical statistical model was used to estimate
the integrated dose-response coefficient. This is similar to the approach used by the NRC
panel to calculate a benchmark dose value integrating data from all three studies (NRC
2000). A more technical description of these same methods has been provided by Coull et
al. (2003).
The statistical analysis produced a dose-response relationship, integrating data
from all three studies, with a central estimate of an IQ change of -0.13 IQ points (95%
confidence interval -0.28, -0.03) for every ppm of mercury in maternal hair.
An DDI value can be converted to an IQ decrement by first converting it to a blood
equivalent mercury level (in ppb) and then converting it to a hair mercury level (in ppm),
as described above. The result can then be multiplied by the estimated dose-response
coefficient of 0.13 to produce an IQ decrement. In other words:
For example, an individual born to a mother exposed above the RfD due to non-US
power plant sources and with an IDI value of 1.0 is estimated to experience a 0.1755
decrement in his or her IQ due to the utility-attributable exposure according to the
equation above.
Applying this equation to all exposed individuals, in effect, assumes that all
individuals are above the RfD due to non-US power plant sources and the utility-
attributable exposure is in addition this level. As mentioned above, this will produce
larger estimate of the IQ decrements attributable to power plants than if each individual's
exposure (above or below the RfD) were known and accounted for. To produce an upper
bound estimate for this analysis, we falsely, but conservatively, assume that all
individuals are above the RfD from non-U.S. power plant sources.
childbearing age have blood Hg levels above the RfD-equivalent level used by the CDC. Thus the
assumption that all exposed individuals are above the RfD is extremely conservative.
IQ Decrement
(IDI) * (Blood equivalent of RfD) * (iQ decrement per 1 ppm hair)
Conversion between blood (ppb) and hair (ppm)
30
-------
8.5 Monetized value of IQ decrements
The valuation approach for assessing losses associated with IQ decrements is
based on an approach used by EPA to assess benefits for reductions in lead exposures
(EPA, 2000a). For that analysis, EPA used results from a study by Salkever (1995) to
estimate the effects of IQ loss on expected future earnings and years of education.
Salkever analyzes data from the National Longitudinal Study of Youth (NLSY) and uses
a three equation regression model to estimate the relationships between IQ levels,
educational attainment, and expected future earnings. The results of this study indicate
that the average effect (for men and women combined) of a one point decrease in IQ is:
1. A 2.379 percent decrease in future earnings; and
2. A 0.1007 decrease in years of schooling.
To estimate the expected monetary value these effects, EPA first estimated the
average present value of future earnings at the time of birth for a person born in the U.S.
Using earnings data from the 1992 Current Population Survey (CPS) and discounting at a
3 percent annual rate, this present value was estimated to be $366,021 in 1992 dollars.
EPA then estimated the average direct and indirect costs associated with one additional
year of schooling. Based on Department of Education data, the average annual
expenditure per student was estimated to be $5,500, and the average annual opportunity
cost (lost income from being in school) was estimated to be $10,925. Assuming that these
costs were incurred at age 19 (based on an average of 12.9 years of education among
those over age 25 in the U.S.) the combined present value of these two costs at time of
birth (discounted at 3 percent) were estimated to be $9,367 per additional year of
schooling in 1992 dollars.
Combining these estimates with the results from the Salkever (1995) study
summarized above implies that the average present value of net earnings losses
associated with a one point decrease in IQ is $7,765 in 1992 dollars. This value is
calculated as the average present value of lost earnings per IQ point loss ($8,708 =
$366,021 * 0.2379) minus the partially offsetting change in average education costs per
IQ point loss ($943 = $9,367 * 0.1007). Corrected for inflation using the GDP deflator,
the average present value of net earnings losses per IQ point loss is $8,807 in 1999
dollars. The value per IQ estimate using a 7 percent discount rate is $1,580 per IQ
point.17 In keeping with our desire to produce an upper-bound estimate and given the
uncertainty in the appropriate value per IQ point, we use the estimate of $8,807.
8.6 Undiscounted Benefits
17 EPA acknowledges that lost earnings from IQ loss is not the conceptually correct metric for valuing
benefits of reduced mercury exposure besides the fact that IQ is being used as a surrogate for other subtle
neurobehavioral endpoints. Ideally, we should use a measure of willingness-to-pay (WTP) to avoid these
endpoints caused by mercury exposure. However, there is a lack of informative research on which to form
an estimate of WTP.
31
-------
Multiplying this value times the IQ decrement for any given DDI value produces
the estimated undiscounted economic value for that IDI. In other words, this is the
economic value that the individual would gain if he or she were no longer exposed to any
utility attributable mercury, which is why it is listed as a benefit. This benefit value can
also be viewed as the economic loss associated with utility-attributable, post-CAIR
mercury exposure.
8.7 Total Benefits
Since there is a lag between the decrease in deposition and the decrease in
methylmercury exposure, this economic value must be discounted over some number of
years to arrive at the economic value of eliminating utility-attributable mercury
deposition. For freshwater systems, this lag is between 10 and 50 years. As a
conservative estimate, we can assume a 15 year lag and a 3% discount rate.
Mathematically, this means that we should divide our undiscounted value by
approximately 1.6 (= 1.03A15) to arrive at the present value of eliminating utility-
attributable mercury emissions18. In other words:
Discount
Monetary Value =
yertod (Monetary Value
¦ie S
v 1 + Discount Rate j
1.03
of IQ Decrementy
($8,807)* (IQ Decrement) = $5,552 * (10 Decrement)
: (10 Decrement)
For example, an individual born to a mother exposed to methylmercury at an IDI
of 1 would suffer an economic loss of, roughly, $1,000, assuming non-power plant
exposures are already above the RfD. Table 8.1 below lists the economic loss associated
with individuals exposed to various levels of utility-attributable mercury and at various
consumption levels.
Table 8.1: Economic Loss from IQ Decrements associated with Mercury Exposure due to
Freshwater Fish Consumption in 2020 after CAIR
2020 with CAIR
Discounted Benefits (in dollars)
MeHg %
PPm
5th
0
10th
0.001
15th
0.002
25th
0.004
50th
0.01
75th
0.02
85th
0.027
90th
0.035
95th
0.052
99th
0.102
g/day \
EPA EFH Mean Recreational Fisher
EPA OW 90th Percentile General Population
EPA EFH 95th Percentile Recreational Fisher
EPA EFH 99th Percentile Recreational Fisher
EPA EFH Mean Subsistence Native American
EPA EFH 95th Percentile Subsistence Native American
EPA EFH 99th Percentile Subsistence Native American
8
17.5
25
47
60
170
295
$ 2
$ 4
$ 7
$ 19
$ 37
$ 50
$ 65
$ 97
$ 190
$ 4
$ 8
$ 16
$ 41
$ 81
$ 110
$ 142
$ 212
$ 415
$ 6
$ 12
$ 23
$ 58
$ 116
$ 157
$ 203
$ 302
$ 593
$ 11
$ 22
$ 44
$ 109
$ 219
$ 295
$ 382
$ 568
$ 1,115
$ 14
$ 28
$ 56
$ 140
$ 279
$ 377
$ 488
$ 725
$ 1,423
$ 40
$ 79
$ 158
$ 395
$ 791
$ 1,067
$ 1,383
$ 2,055
$ 4,032
$ 69
$ 137
$ 274
$ 686
$ 1,372
$ 1,852
$ 2,401
$ 3,567
$ 6,996
At a consumption rate of 8 grams per day, the mean rate for a freshwater
fisherman, the economic loss from IQ decrements associated with utility-attributable,
18 Using a 3 percent discount rate produces a higher benefits estimate, which is desirable to produce an
upper bound estimate. Conducting an additional analysis using a 7 percent discount rate, which would be
standard in a regulatory economic analysis, would require dividing by 2.8.
32
-------
post-CAIR mercury exposure from freshwater fish consumption ranges between zero and
$190, depending on the amount of deposition. For example, in a watershed with the 85th
percentile utility-attributable mercury deposition, self-caught freshwater fish are assumed
to have a utility-attributable methylmercury concentration of 0.027 ppm. This translates
to a discounted economic loss of approximately $50 per birth.
A similar analysis can be done using the IDI values for marine fish consumption
and estuarine and coastal fish consumption. For marine fish consumption, we use the IDI
values reported earlier in this TSD. Since the marine environment produces a longer time
lag between deposition reductions and exposure reductions, the discount factor for this
calculation will be slightly different. We use a 3 percent discount rate over 30 years,
which is the shorter of the two lag times described in Section 2.3, the Marine Cycling
section, of this TSD. The results of this calculation are reported in Table 8.2.
33
-------
Table 8.2: Economic Loss from IQ Decrements associated with Mercury Exposure
2020 with CAIR
Discounted Benefits (in dollars)
MeHg
ppm
Mean
0.1
g/day
Mean Marine Fish Intake
95% Long-Term Marine Fish Intake
14.1
44
s :
s
Mean Total Fish Intake
95% Long-Term Total Fish Intake
20
63
s
S 'J
Median Carrington & Bolger Seafood Consumption
90% Carrington & Bolger Seafood Consumption
95% Carrington & Bolger Seafood Consumption
99% Carrington & Bolger Seafood Consumption
99.5% Carrington & Bolger Seafood Consumption
99.9% Carrington & Bolger Seafood Consumption
6.8
42
63
123
141
207
S 1
S (.
S 'J
S IS
s :i
S ^1
Assuming a maternal fish consumption rate of 14.1 grams per day of marine fish, which
is the mean marine fish intake rate, an individual prenatally exposed to utility-attributable
methylmercury in 2020 would suffer a lifetime economic loss of around $2 from IQ
decrements.
8.8 Aggregate Benefits
As described above, the benefits of eliminating mercury emission from U.S.
power plants in 2020, after the implementation of CAIR, can be estimated by summing
the economic loss associated with their damage. This is done by multiplying the
economic loss value times the numbers of births to mothers in each consumption range.
If we assume that consumption is log-normally distributed, then we can do this by
multiplying the economic loss for the mean consumption rate times the total number of
births.
According to the U.S. Census, the U.S. population in 2001 was roughly 280
million people (BOC 2005), and there were approximately 4 million births according to
the CDC's National Center for Health Statistics (CDC 2002). However, because the
consumption self-caught fish represents the highest level of utility-attributable exposure,
it is important to make a distinction between births to self-caught freshwater fishermen
and those to the general public. Assuming that the ratio of births to individuals for the
general population holds for the 58 million self-caught freshwater fishers described in
CAMR, this implies approximately 830,000 births to self-caught freshwater fishers and
approximately 3.2 million births to the rest of the general public.
The consumption rate for these two groups can be obtained from the U.S. EPA's
Exposure Factors Handbook (US EPA 1997b), which recommends using a mean
consumption rate for the general population of 20.1 grams of fish per day, with 14.1
grams associated with marine fish and 6 grams per day of freshwater or estuarine fish
(including aquaculture consumption). These are the consumption rates we assume for the
general public. The recommended mean consumption value for freshwater anglers is 8
34
-------
grams per day of self-caught freshwater fish and we will assume a value of 14.1 grams
per day from the consumption of marine fish. Multiplying the number of births for each
group times the economic loss value produces an estimate of the economic benefits of
eliminating the post-CAIR mercury emission from U.S. power plants.
The economic loss for an individual consuming self-caught freshwater fish is
given in Table 8.1. It would be reasonable to conduct a population level analysis using
the average MeHg exposure at the average consumption level from Table 8, which would
imply a loss value of $19. However, to maintain a clear upper bound estimate, we use the
economic loss value $50 for the mean consumption value for the watershed with the 85th
percentile of mercury deposition.19 Multiplying $50 times the 830,000 births to
freshwater fishers implies an upper bound aggregate economic loss of $41.5 million for
the consumption of self-caught freshwater fish.
The upper bound estimate of the individual loss value for marine fish consumed
by both freshwater fishermen and the general public is $2, as given in Table 8.2.
Multiplying $2 times the 4 million births from both groups implies an upper bound
aggregate economic loss of $8 million for the consumption of marine fish. $1.6 million
of this accrues to freshwater fishers and $6.4 to the rest of the general public.
Table 8.3 summarizes the fact that the upper bound estimate of the aggregate
economic benefits of reduced IQ decrements from eliminating utility-attributable
mercury exposure in 2020 after CAIR are approximately $50 million plus some
additional amount from the consumption of commercial freshwater, estuarine, and
aquaculture fish by the general public. The best estimate of the individual economic loss
from the consumption of these fish is not known because a best estimate of the IDI values
for these pathways has not been estimated. Therefore a best estimate of the aggregate
economic loss is unknown. However, we are able to calculate an upper bound estimate as
described below.
Table 8.3: Upper Bound Estimate of the Aggregate Economic Benefits of Reduced IQ
Decrements from Eliminating Utility-Attributable Mercury Exposure in 2020 after the
Implementation of CAIR
Population
Births
Consumption
Rate (g/day)
Value
Benefits
Freshwater Fishers
58 million
830,000
Freshwater Fish
8
$50
$41.5 million
Marine Fish
14.1
$2
$1.6 million
19 If fish were consumed equally from all watersheds, it would be appropriate to use the mean utility
attributable methylmercury concentration for this calculation. The mean utility attributable methylmercury
concentration based on available information is .016 ppm. To allow for the possibility that fish are
generally consumed from areas with a higher utility attributable methylmercury concentrations than the
mean, we use a utility attributable methylmercury concentration of 0.027 ppm, corresponding to the 85th
percentile. It is highly unlikely that consumption of freshwater fish would be so skewed towards
waterbodies with such high methylmercury. Therefore this assumption is thought to overestimate the actual
utility attributable exposure from freshwater fish consumption.
35
-------
General Public
222 million
3.2 million
Freshwater, Estuarine,
and Aquaculture Fish
6
$V
B
Marine Fish
14.1
$2
$6.4 million
Total =
$49.5 million + B
As described in Section 4, EPA finds that the utility attributable exposure to
methylmercury from estuarine fish and shellfish will likely be small relative to that from
freshwater fish and so the IDI values for this pathway is bound by the freshwater
recreational/subsistence IDI values presented in the Effectiveness TSD (US EPA 2005a).
Therefore the bound for the IDI value from estuarine consumption for the mean
consumption rate of 6 g/day will be 75 percent (6g/8g) of the IDI value for the
recreational fisher consuming fish from the 85th percentile watershed at the mean
consumption level.
Similarly, as described in Section 5, EPA believes that the utility-attributable
mercury exposure from aquaculture is small and exposure to utility-attributable MeHg
from aquaculture fish will be lower than the levels ingested from the types of marine fish
that comprise aquaculture fish feed. Therefore, the $50 value for 8 g/day of freshwater
fish consumption and the $2 value for 14.1 g/day of marine fish consumption can be used
to create an upper bound our estimate of $V above.
As described above, the U.S. EPA's Exposure Factors Handbook (US EPA
1997b) recommends a consumption rate of 6 g/day of freshwater, estuarine, and
aquaculture consumption for the general public. Table 8.4 lists the possible combinations
of marine equivalent and freshwater equivalent fish that could be used to make up the 6
g/day.
Table 8.4: Upper Bound Estimates of Economic Loss from IQ Decrements associated
with Mercury Exposure due to Freshwater, Estuarine, and Aquaculture Consumption by
the General Public in 2020 after CAIR
Freshwater equivalent consumption (g/day)
0
1
2
3
4
5
6
0
$38
1
$31
Marine equivalent
consumption (g/day)
2
$25
3
$19
4
$13
5
$7
6
$1
For example if all of the consumption were marine and aquaculture fish, then the
economic loss value would be $1. This is because 6 g/day is about 40% of the 14.1 g/day
of marine fish consumption with an associated economic loss of $2. 40% of $2 is
(rounding up) about $1. In contrast, if all of the fish were estuarine or freshwater fish,
then the economic loss would be $38. This is because 6 g/day is 75% of the 8 g/day for
freshwater fishers with an associated economic value of $50. So 75% of $50 is $38.
36
-------
Given this, the highest estimate possible for $V in Table 8.4 is $38. Substituting
this value in Table 8.3 above would produce a total upper bound estimate of $168
million. This, however, is a dramatic overestimate of the upper bound estimate of
benefits. It assumes that the 6 g/day of commercial freshwater, estuarine, and
aquaculture fish consumed by the general public is as contaminated as the 85th percentile
of self-caught freshwater fish. In other words all aquaculture fish and all estuary fish,
including fish from the estuaries on the Pacific coast, are as contaminated as self-caught
fish. It is, however, useful to state that the aggregate economic benefits of reduced IQ
decrements from eliminating utility-attributable mercury exposure in 2020 after the
implementation of CAIR can not be higher than $168 million.
8.9 Costs
The methodology for estimating the total annual monetized benefits of CAMR are
described in the Cost TSD (US EPA 2005b). Under the base-base assumptions, the total
annual monetized cost of the CAMR is estimated to be approximately $750 million in
2020.20
8.10 Summary
As can be seen from this analysis, the total monetized costs of CAMR exceed the
total monetized benefits presented here of eliminating all utility-attributable mercury
emissions remaining in 2020 after the implementation of CAIR. It should again be
pointed out that the analysis estimates the upper bound monetized benefits associated
with the potential neuro-toxicity (represented by IQ decrements as a surrogate) of low
level, chronic mercury exposure. This is the endpoint about which we have the most
certainty and which we can monetize (see Section 8.4 above). Furthermore, these
estimates were based on a number of assumptions intended to produce an upper bound
estimate of the benefits for this endpoint. In contrast, the cost estimate is based on the
estimated cost of the cap-and-trade program of CAMR which does not eliminate all
mercury emissions from U.S. power plant. Furthermore, it is generally accepted that, all
else equal, the cost of cap-and-trade programs are less than other regulatory approaches.
By 2020 CAMR reduces between 16 percent and 32 percent of the remaining mercury
emissions, depending upon the species, but does not eliminate all mercury emissions
from U.S. power plants (See table 3 on page 4 of US EPA 2005c). Given that the
monetized costs of reducing 1/3 of the current emissions (by one of the lowest cost
emission reduction schemes available) exceeds the upper bound of monetized benefits of
reduced IQ decrements, it is reasonable to conclude that the cost of requiring further
20 See table 7-19. The cost associated with monitoring emissions, reporting, and record keeping for
affected sources is not included in these annualized cost estimates, but EPA has done a separate analysis
and estimated the cost to be about $76 million (see final CAMR preamble Section VLB. Paperwork
Reduction Act). Under the sensitivity analysis described in the Cost TSD, the total annual monetized costs
are estimated to be $560 million in 2020 (See Cost TSD). We use the estimate of $750 million in 2020
because it reflects our best estimate, although we note that the conclusions that annual costs exceed annual
health benefits would equally apply if we were to use the sensitivity analysis cost estimate of $560 million.
37
-------
reductions in U.S. power plant Hg emissions beyond CAIR would outweigh the benefits
presented here.
9. Evaluation of NESCAUM Report
Recent analyses have attempted to account for exposures from Hg in marine fish,
including "Economic Valuation of Human Health Benefits of Controlling Mercury
Emissions from U.S. Coal-Fired Power Plants" (docket item OAR-2002-0056-5749) by
Glenn Rice and James K. Hammitt, Harvard Center for Risk Analysis (NESCAUM
Report) (NESCAUM 2005).
EPA's approach to modeling exposure and health benefits of reducing emissions
from power plants differs in some important ways from the approach that NESCAUM
chose. EPA believes that some of these differences simply reflect the large amount of
uncertainty in the underlying science. Other differences reflect situations where the
science and economics are fairly clear and EPA has concerns about the approach that
NESCAUM took.
For example, as noted earlier, the NESCAUM report attempted to quantify the
marine exposure pathway but used assumptions that are not supported by the literature on
marine fate and transport of Hg, likely resulting in an overestimate by an unknown
amount. For example, the NESCAUM study assumes that Hg deposition from the
atmosphere are tracking Hg concentrations in the surface ocean and that the changes in
ocean fish MeHg concentrations will be proportional to changes in total Hg
concentrations in the surface ocean. This proportionality assumption was used in
Section 2 to provide an upper-bound estimate to show that the marine pathway is
probably not a significant contributor to the mercury exposure from U.S. power plants,
but we recognized that this was an upper bound estimate. Also, NESCAUM used
REMSAD modeling which appears to over-predict Hg deposition from US power plants.
The NESCAUM Report focused on estimating the benefits of reductions in
mercury exposure from U.S. power plants and did not attempt to estimate upper-bound
estimate of reducing mercury emissions for power plans. However, for the additional
reasons noted below EPA believes that NESCAUM's approach should be interpreted as
producing an upper-bound estimate of the IQ benefits of reducing Hg emissions from
power plants for two reasons. First, it does not appear that the NESCAUM Report took
into account the timeframe for reduced exposure to MeHg this issue into account. This
omission alone leads to the benefits in the NESCAUM Report being overstated by at least
factor of two. Second, EPA commissioned Harvard researchers Dr. Louise Ryan and Dr.
David Bellinger to perform an integrated analysis of the three major epidemiological
studies (Faroes, Seychelles, New Zealand) and used the resulting relationship between
exposure and neurological problems. NESCAUM relied in part on an unpublished study
and produces estimated neurological benefits four to five times EPA estimates for this
reason alone.
38
-------
9.1 Timeframe for costs and benefits
It is important to consider the time in an economic analysis because future costs
and benefits are discounted. EPA's Revision Rule analyses took into account the
timeframe for costs and benefits. As noted in Section 8, benefits were discounted to take
into account the response times for the freshwater and marine ecosystems. Case studies of
individual ecosystems show that the time necessary for aquatic systems to reach a new
steady state after a reduction in Hg deposition rates can be as short as 5 years or as long
as 50 years or more. The medium response scenarios also varied widely but were
generally on the order of one to three decades. Overall, EPA concludes that the most
likely appropriate response times for freshwater ecosystems to be considered in the
national scale assessment range between 5 and 30 years, while recognizing that some
systems will likely take more than 50 to 100 years to reach steady state. Our preliminary
analysis of the temporal response of marine systems based on the model by Mason and
Gill (2005) indicates that the Atlantic Ocean will take approximately three decades to
reach steady state and the Pacific Ocean will take over two centuries.
At 3 percent and two to three decades, present value benefits would decrease by
about half. At 7 percent, present value benefits would fall even more. This shows the
importance of taking into account the time lag between emissions reductions and
exposure reductions. EPA does not believe that NESCAUM accounted for this important
factor in its analysis and in so doing significantly overestimates benefits.
9.2 IQ dose-response relationship
In the benefits analysis presented above, EPA chose to focus on quantification of
intelligence quotient (IQ) decrements associated with prenatal Hg exposure as the initial
endpoint for quantification and valuation of Hg health benefits. Reasons for this initial
focus on IQ include the availability of well-designed epidemiological studies assessing
IQ or related cognitive outcomes suitable for IQ estimation, and the availability of well-
established methods and data for economic valuation of avoided IQ deficits, as applied in
EPA's previous benefits analyses for childhood lead exposure.
There is limited evidence directly linking IQ and MeHg exposure in the three
large epidemiological studies that were evaluated by the National Academy of Sciences
(NAS) and EPA. Based on its evaluation of the three studies, EPA believes that children
who are prenatally exposed to low concentrations of MeHg may be at increased risk of
poor performance on neurobehavioral tests, such as those measuring attention, fine motor
function, language skills, visual-spatial abilities (like drawing), and verbal memory. For
this analysis, EPA is adopting IQ as a surrogate for the neurobehavioral endpoints that
NAS and EPA relied upon for the RfD.
The NAS identified three well-designed studies (Faroes, Seychelles, New
Zealand) of the neurotoxicological effects of MeHg. "Each of the studies was well
designed and carefully conducted, and each examined prenatal MeHg exposures within
the range of the general U.S. population exposures" (NRC 2000). In order to develop a
39
-------
dose-response relationship that reflects all three studies identified by the NAS, EPA
commissioned Harvard researchers David Bellinger and Louise Ryan to perform an
integrated analysis, combining results from all three studies. When combined, the
statistical power of the analysis increases. While the size of the dose-response
relationship declined relative to past studies with a statistically significant finding, Ryan
found a statistically significant relationship between IQ and Hg. The confidence interval
did not include zero. EPA used Ryan's mean estimate -0.131 as a coefficient to relate
changes in exposure (parts per million in hair) to IQ point changes. EPA also performed
sensitivity analysis using coefficients of -0.108 and -0.233 also based on the work by
Louise Ryan and David Bellinger. (Ryan 2005).21
NESCAUM appears to have used a coefficient of -0.60 based on an unpublished
study (Cohen, et al. full reference unavailable) that is not included in the references. The
paper upon which NESCAUM based its coefficient has not been submitted to the
rulemaking docket, making it difficult for us to assess the NESCAUM approach.
However, the reliance on a coefficient of -0.60 is not consistent with the work done by
Harvard researchers Louise Ryan and David Bellinger or the other existing studies.
9.3 Conclusion
Unlike the analyses conducted for this reconsideration which provide an estimate
of the upper-bound of IQ benefits associated with reduced mercury emissions from power
plants, the NESCAUM Report estimates are presented as "central" or "best" benefits
from marine fish Hg reductions. Further even if presented as upper-bound estimates,
there is a disconnect between the derivation of quantitative results and the summary of
the science since authors relied on a variety of unsupported assumptions.
10. 2010/2015 CMAQ Modeling Results
The Technical Support Document (TSD) for the Final Clean Air Mercury Rule:
Air Quality Modeling, March 2005, describes in Section V.B. that the expected mercury
deposition with CAIR plus CAMR in 2015 is expected to be similar to the mercury
deposition with CAIR plus CAMR in 2020. Since the March 2005 CAMR TSD was
prepared, a combined strategy consisting of the implementation of CAIR, CAMR and
21 Ryan and David develop a linear model that goes through the origin to fit population-level dose-
response relationships to the pooled data from the three studies. The application of a linear model should
not be interpreted to suggest that any of the three studies used have data showing health effects from MeHg
exposure at or below the RfD. Use of a linear model that goes through the origin, rather than one that
reflects a threshold effect is technically more simple and practical. It associates an increment of IQ benefit
with a given reduction in exposure. A linear model allows us to estimate the benefits of reductions in
exposure due to power plants without a complete assessment of other sources of exposure. Other models
would require information on the joint distribution of exposure from power plants and other sources to
estimate the benefits of reducing the exposure due to power plants, which would require much more precise
information about consumption patterns.
40
-------
CAVR has been modeled with CMAQ for 2010, 2015 and 2020. The BART rule
indicates that States that opt into the CAIR trading program do not have to do anything
more for BART for their eligible utility sources. Therefore, CAIR satisfies BART for
utilities for the CAIR States. Thus, in the CAIR States, there would be no further
reductions in utility mercury emissions from BART. It is only, in the States where CAIR
does not apply that the implementation of BART would possibly lead to additional
mercury reductions over those that would occur with CAIR and CAMR.
The total utility mercury emissions for the each of the Clear Skies CAIR-CAMR-
BART scenarios are shown in Table 10.1. More importantly, the utility mercury
emissions of the most readily depositable form of mercury emissions (Reactive Gaseous
Mercury) for each of these scenarios are also shown Table 10.1. It can be seen in Table
10.1 that there is a large (approximately 11 ton) decrease in utility reactive gaseous
mercury (RGM) emissions from 2001 to 2010. The reduction in RGM emissions from
2010 to 2015 is approximately 2 tons and the reduction in RGM emissions from 2015 to
2020 is only approximately 1 ton.
The total modeled mercury deposition for 2010, 2015 and 2020 under
CAIR/CAMR/BART are provided below in Figures 10.1 through 10.3. It can be seen in
Figures 10.1 through 10.3 that the total mercury depositions with CAIR/CAMR/BART
are very similar in 2010, 2015 and 2020. The reduction in total mercury deposition from
2001 to 2010, 2015 and 2020 with the implementation of CAIR/CAMR/BART are shown
in Figures 10.4 through 10.6. It can be seen in figures 10.4 through 10.6 that the
difference in the decrease in total mercury deposition between 2010, 2015 and 2020
relative to 2001 are fairly small. Figure 10.7 provides the reduction in total mercury
deposition between 2010 and 2015 and Figure 10.8 provides the reduction in total
mercury deposition between 2015 and 2020. It can be seen in Figure 10.7 that the
reductions in total mercury deposition from 2010 to 2015 cover scattered areas of the
country with the reductions less than 5 ug/m2. It can be seen in Figure 10.8 that the
reductions in total mercury deposition from 2015 to 2020 cover only scattered small areas
of the country with the reductions generally less than 5 ug/m2. It can be seen by
examining Figures 10.4, 10.7 and 10.8 that there is a much larger reduction in utility
attributable mercury deposition from 2001 to 2010 than from 2010 to 2015. The
reduction in utility attributable mercury deposition from 2015 to 2020 is even smaller
than the reduction from 2010 to 2015.
The mercury deposition reductions shown in Figures 10.4 through 10.6 look very
similar to the reduction in mercury deposition that would occur from 2001 with the
implementation of CAIR and CAMR in 2020, which were modeled under the CAMR
rule, and are shown below in Figure 10.9. The additional reductions in 2020 total
mercury depositions that would occur relative to 2001 with the implementation of BART
in addition to CAIR and CAMR are provided in Figure 10.10. As can be seen in Figure
10.10, the maximum additional reduction with BART is less than 4 ug/m2 in all locations.
The additional reductions with BART are outside the eastern States covered by CAIR.
41
-------
Table 10.1. Utility Mercury Emissions for Clear Skies Act CAIR-CAMR-BART
Scenario (tons/year)
Year
Reactive Gaseous Mercury
Total Mercury Emissions
2001
20.6
48.6
2010
9.7
32.1
2015
7.2
28.7
2020
6.1
25.5
0.000 1
ug/m2
118
January 1,0 0:00:00
Mifl= 3.469 at (71,95), Max= 122.715 at (21,84)
Figure 10.1. Total Mercury Deposition with CAIR/C A MR/BART: 2010
42
-------
I
40.000112
20.000
15.000
10.000
5.000
1.000
u 0.000 1
ug/m2
I
148
January 1,0 0:00:00
Min= 3.451 at (71,95), Max= 142.588 at (21,84)
Figure 10.2. Total Mercury Deposition with CAIR/CA M R/BA RT: 2015
I
40.000112
20.000
15.000
10.000
5.000
1.000
0.000 1
ug/m2
148
January 1,0 0:00:00
Min= 3.441 at (71,95), Max= 163.469 at (21,84)
Figure 10.3. Total Mercury Deposition with C AI R/C A M R/BA RT: 2020
43
-------
ug/mz
148
January 1,0 0:00:00
Min- -3.763 at (23,47), Max= 54.439 at (98,50)
Figure 10.4. Reduction in Total Mercury Deposition: 2010 with
CAIR/C AMR/BART Relative to 2001
I
40.000112
20.000
15.000
10.000
5.000
1.000
0.000
ugfm2
January 1,0 0:00:00
Min= -9.111 at (21,84), Max= 49.665 at (98,50)
Figure 10.5. Reduction in Total Mercury Deposition: 2015 with
CAIR/CAMR/BART Relative to 2001
44
-------
ug/mZ 1 148
January 1,00:00:00
Min= -29.991 at(21,84), Max= 43,705 at(9B,50)
Figure 10.6. Reduction in Total Mercury Deposition: 2020 with
CAIR/C AMR/BART Relative to 2001
40.00«12
20.000
15.000
10.000
5.000
u 0.000 1
ug/m2
January 1,0 0:00:00
Min= -19.874 at (21,84), Max= 4,899 at (117,61)
Figure 10.7. Reduction in Total Mercury Deposition with C AIR/CA M R/BA RT:
2010 to 2015
-------
I
40.000112
20.000
15.000
10.000
5.000
1.000
0.000 1
ug/m2
January 1,0 0:00:00
Min- -20.880 at (21.64), Max= 6.283 at (79,33)
Figure 10.8. Reduction in Total Mercury Deposition with CAIR/CAMR/BART:
2015 to 2020
I
40,000112
20.000
15.000
10.000
5.000
1.000
0.000 1
ug/m2
January 1,0 0:00:00
Min=-30.121 at (21,84), Max= 44.081 at (98,50)
Figure 10.9. Reduction in Total Mercury Deposition with CAIR/CAMR: 2020
Relative to 2001
46
-------
| 40.000(12
' 20.000
15.000
• 10.000
! 5.000
1.000
0.000
itg(tn2
January 1,0 0:00:00
Min* -0.084 at (119,62), Max= 3,726 at (61,46)
Figure 10.10. Reduction in Total Mercury Deposition with BART in addition to
CAIR and CAMR: 2020 Relative to 2001
11. Global Source Impact
A CMAQ 2001 modeling run was performed after the CAMR rule was finalized
to estimate the impact of global sources. In this model run, all mercury boundary
condition input species to CMAQ that were obtained from the GEOS-CHEM global
model were zeroed-out. By comparing this run with the 2001 base case run, which
included the mercury boundary condition species input to CMAQ, the percent of total
mercury deposition attributable to global sources can be estimated. The model estimated
percent of total mercury deposition attributable to global sources is provided below in
Figure 11.1. The scientific understanding of mercury atmospheri c chemistry is still
evolving. Changes in the current understanding of mercury chemistry could possibly
lead to the need to change the mercury chemistry in the global GEOS-CHEM and
regional CMAQ models. Thus, it should be noted that there is considerable uncertainty
associated with the estimates of global source impacts.
47
-------
January 1,0 0:00:00
Min= 14.076 at.(21,84), Max= 99.985 at (148,1)
Figure 11.1. Percent of Total Mercury Deposition Attributable to Global Sources:
2001
48
-------
12. References
Amyot, M, Gill, G.A., and Morel, F.M.M. 1997. Production and loss of dissolved
gaseous mercury in coastal seawater. Environmental Science and Technology, 31:
3606-3611.
Benoit, J., Gilmour, C.C., Mason, R.P. and Heyes, A. 1999. Sulfide controls on mercury
speciation and bioavailability to methylating bacteria in sediment pore waters.
Environmental Science and Technology, 33: 951-957. OAR-2002-0056-6057.
Benoit, J.M., Gilmour, C.C., Heyes, A., Mason, R.P., and Miller, C.L. 2003.
Geochemical and biological controls over mercury production and degradation in
aquatic systems. In: Biogeochemistry of Environmentally Important Trace Metals
(Y. Cai and O.C. Brouds, eds.) ACS Symposium Series. 835. American
Chemical Society, Washington, D.C. 262-297.
Berger, M.C., Blomquist, G.C., Kenkel, D., and Tolley, G.S. 1987. "Valuing Changes in
Health Risks: A Comparison of Alternative Measures." Southern
Economic Journal 53(4):967-984. OAR-2002-0056-5960.
Bloom, NS. 1992. On the chemical form of mercury in edible fish and marine
invertebrate tissue. Canadian Journal of Fisheries and Aquatic Sciences. 49.
1010-1017.
Bloom, N.S., Gill, G.A., Cappellino, S., Dobbs, C., Mcshea, L., Driscoll, C., Mason, R.,
Rudd, J. 1999. Speciation and cycling of mercury in Lavaca Bay, Texas,
sediments. Environmental Science and Technology, 33, 7-13.
BOC (U.S. Bureau of Census). 2005. Data downloaded from the American Factfinder
database at http://factfinder.census.gov/home, accessed October 2005.
Carrington, C.D. and Bolger, M.P. 2002. "An Exposure Assessment for Methylmercury
from Seafood for Consumers in the United States." Risk Analysis, 22(4): 689-
699. OAR-2002-0056-5737.
Carrington, C.D., Montwill, B., and Bolger, P.M. 2004. "An intervention analysis for
the reduction of exposure to methylmercury from the consumption of seafood by
women of child-bearing age." Regulatory Toxicology and Pharmacology, 40:
272-280.
CDC (Center for Disease Control). 2002. "Births: Final Data for 2001" National Vital
Statistics Report Volume 51, Number 2, December 18, 2002
Conaway, C., Squire, S., Mason, R.P., and Flegal, A.R. 2003. Mercury speciation in the
San Francisco Bay Estuary. Marine Chemistry, 80: 199-225.
49
-------
Cossa, D. and Gobeil, C. 2000. Mercury Speciation in the Lower St. Lawrence Estuary.
Canadian Journal of Fisheries and Aquatic Sciences (Suppl. 1), 57: 138-147.
Cossa, D., Coquery, M., Gobeil, C., and Martin, J-M. 1996. Mercury fluxes at the ocean
margins. In Global and Regional Mercury Cycles: Sources, Fluxes and Mass
Balances, Baeyens et al. (eds), Kluwer Academic Press, Dorchester, 229-247,
1996. A-92-55; I-H-447.
Coull B.A., Mezzetti, M., Ryan, L.M. 2003. A Bayesian hierarchical model for risk
assessment of methylmercury. Journal of Agricultural, Biological &
Environmental Statistics, 8(3):253-270.
Crump, K.S., Kjellstrom, T., Shipp, A.M., Silvers, A., and Stewart, A. 1998. Influence
of prenatal mercury exposure upon scholastic and psychological test performance:
Benchmark analysis of a New Zealand cohort. Risk Analysis, 18:701-713. OAR-
2002-0056-5531.
Davidson, P.W., Myers, G.J., Cox, C., Axtell, C., Shamlaye, C., Sloane-Reeves, J.,
Cernichiari, E., Needham, L., Choi, A., Wang, Y., Berlin, M., and Clarkson, T.W.
1998. Effects of prenatal and 9-12 postnatal methylmercury exposure from fish
consumption on neurodevelopment: outcomes at 66 months of age in the
Seychelles Child Development Study. Journal of the American Medical
Association, 280(8):701-7. OAR-2002-0056-5822.
Gill, G.A. and Fitzgerald, W.F. 1988. Vertical mercury distributions in the oceans.
Geochimica et Cosmochimica Acta., 52: 1719-1728.
Grandjean, P., Weihe, P., White, R.F., Debes, F., Araki, S., Yokoyama, K., Murata, K.,
Sorensen, N., Dahl, R., Jorgensen, P.J. 1997. Cognitive deficit in 7-year-old
children with prenatal exposure to methylmercury. Neurotoxicology and
Teratology, 19:417-428. OAR-2002-0056-5183.
Hammerschmidt, C.R. and Fitzgerald, W.F. 2004. Geochemical controls on the
production and distribution of methylmercury in near-shore marine sediments.
Environmental Science and Technology, 38, 1487-1495. OAR-2002-0056-6083.
Harrington, W., and Portney, P.R.. 1987. "Valuing the Benefits of Health and Safety
Regulation." Journal of Urban Economics, 22:101-112. OAR-2002-0056-5963.
Heyes, A., Mason, R.P., Kim, E-H., and Sunderland, E. 2005 (in press). Mercury
methylation in estuaries. Marine Chemistry.
Joiris, C.R., Holsbeek, L., and Otchere, F.A. 2000. Mercury in the bivalves Crassotrea
tulipa and Pernaperna from Ghana. Marine Pollution Bulletin. 40(5): 457-460.
Kahana, R., Bigg, G.R., and Wadley, M.R. 2004. Global ocean circulation modes derived
from a multiple box model. J. Phys. Oceanogr., 34, 1811-1823.
50
-------
Kjellstrom, T., Kennedy, P., Wallis, S., Stewart, A., Friberg, L., Lind, B., et al. 1989.
Physical and mental development of children with prenatal exposure to mercury
from fish. National Swedish Environmental Protection Board Report No. 3642.
OAR-2002-0056-5872.
Kraepiel A.M.L., Keller, K., Chin, H.B., Malcolm, E.G., and Morel, F.M.M. 2003.
Sources and Variations of Mercury in Tuna. Environmental Science and
Technology, 37, 5551-5558.
Laurier, F.J.G., Mason, R.P., Gill, G.A., and Whalin, L. 2004. Mercury Distributions in
the North Pacific Ocean - 20 Years of Observations. Marine Chemistry, 90, 3-19.
Lawson, N.M. and Mason, R.P. 1998. Accumulation of mercury in estuarine food
chains. Biogeochemistry, 40, 235-247. OAR-2002-0056-6064.
Mason, R.P., Fitzgerald, W.F., and Morel, F.M.M. 1994. The biogeochemical cycling of
elemental mercury: Anthropogenic influences, Geochim. Cosmochim. Acta., 58,
3191-3198,1994. A-92-55; I-H-436.
Mason, R.P. and Lawrence, A.L. 1999. Concentration, distribution, and bioavailability
of mercury and methylmercury in sediments of Baltimore Harbor and Chesapeake
Bay, Maryland, USA. Environmental Toxicology and Chemistry, 18, 2438-2447.
Mason, R..P., Laporte, J.-M., Andres, S. 2000. Factors Controlling the Bioaccumulation
of Mercury, Methylmercury, Arsenic, Selenium, and Cadmium by Freshwater
Invertebrates and Fish. Archives of Environmental Contamination and
Toxicology, 38: 283-297.
Mason, R.P., Lawson, N.M. and Sheu, G.-R.. 2001. Mercury in the Atlantic Ocean:
Factors controlling air-sea exchange of mercury and its distribution in the upper
waters. Deep Sea Research II, 2829-2853.
Mason, R.P. and Sheu, G.-R. 2002. The role of the Ocean in the Global Mercury Cycle.
Global Biogeochemical Cycles, 16, Art. # 1093. OAR-2002-0056-5645.
Mason R.P. and Benoit, J.M. 2003. Organomercury compounds in the environment. In:
Organometallics in the Environment (P. Craig, ed.) John Wiley & Sons, New
York, 57-99.
Mason, R.P. (2005). Mercury in the Marine Environment: A Melange of Anthropogenic
Influence, Biogeochemistry, and Microbial Transformation. Presentation made at
the Geological Survey of Canada Short Course #34 Meeting in Halifax, Nova
Scotia, Canada. May 2005.
51
-------
Mason, R.P. and Gill, G.A. 2005. Mercury in the Marine Environment. Mineralogical
Association of Canada Short Course 34 (Eds. M Parsons and J. Percival), Halifax,
Nova Scotia, pp. 1-28.
Mikac, N., Picer, M., Stegnar, P., and Tusek-Znidaric, M. 1985. Mercury distribution in
a polluted marine area, ratio of total mercury, methy mercury, and selenium in
sediments, mussels and fish. Wat. Res., 19:1387-1392.
Myers, G.J., Davidson, P.W., Cox, C., Shamlaye, C.F., Palumbo, D., Cernichiari, E.,
Sloane-Reeves, J., Wilding, G.E., Kost, J., Huang, L.S., and Clarkson, T.W.
2003. Prenatal methylmercury exposure from ocean fish consumption in the
Seychelles child development study. Lancet, 361:1686-1692.
NESCAUM. (Northeast States for Coordinated Air Use Management). 2005. Economic
Valuation of Human Health Benefits from Controlling Mercury Emissions from
U.S. Coal-Fired Power Plants.
http://bronze.nescaum.Org/airtopics/mercurv/rpt050315mercuryhealth.pdf#search
='nescaum%20mercury%20hammitt'. OAR-2002-0056-5749, -5752.
NMFS (National Marine Fisheries Service). 2002. Fisheries of the United States: 2001.
Silver Spring, Maryland.
NMFS. 2003. Fisheries of the United States - 2003. Downloaded from:
http://www.st.nmfs.gov/stl/fus/fus03/index.html
National Research Council (NRC). 2000. Toxicological Effects of Methylmercury.
Committee on the Toxicological Effects of Methylmercury, Board on
Environmental Studies and Toxicology, Commission on Life Sciences, National
Research Council. National Academy Press, Washington, D.C. A-92-55; I-A-
137.
Pacyna, J. and Munthe, J.. 2004. Summary of Research Projects on Mercury Funded by
EC DG Research. Workshop on Mercury, Brussels, Belgium. March 29 - 30
2004.
Rodgers, D.W. 1994. You are What you Eat and a Little Bit More: Bioenergetics Based
Models of Methylmercury Accumulation in Fish Revisited. Ed. C.J. Watras &
J.W. Huckabee. In. Mercury Pollution: Integration and Synthesis. Monterey,
California. Lewis Publishers. Pp. 427-439.
Rolfhus, K.R. and Fitzgerald, W.F. 2004. Mechanisms and temporal variability of
dissolved gaseous mercury production in coastal seawater. Marine Chemistry, 90:
125-136.
Ryan, L.M. 2005. Effects of Prenatal Methylmercury on Childhood IQ: A Synthesis of
Three Studies. Report to the U.S. Environmental Protection Agency.
52
-------
http://docket.epa.gov/edkpub/do/EDKStaffltemDetailView7obiectIcN090007d480
683509. OAR-2002-0056-6048, -6049.
Sager, D.R. 2002. Long-term variation in mercury concentrations on estuarine
organisms with changes in releases into Lavaca Bay, Texas. Marine Pollution
Bulletin, 44, 807-815.
Salkever, D. 1995. "Updated Estimates of Earnings Benefits from Reduced Lead
Exposure of Children to Environmental Lead." Environmental Research 70:1-6.
OAR-2002-0056-5943.
Stommel, H.M. Thermohaline convection with two stable regimes of flow. Tellus, 13,
224-230, 1961
Sunderland, E.M., Gobas, F.A.P.C., Heyes, A., Branfireun, B.A., Bayer, A.K., Cranston,
R.E., and Parsons, M.B. 2004. Speciation and bioavailability of mercury in well-
mixed estuarine sediments. Marine Chemistry, 90, 91-105. OAR-2002-0056-
6096.
Sunderland, E.M., Gobas, F.A.P.C., Branfireun, B.A. and Heyes, A. 2005 (in Press).
Environmental controls on the speciation and distribution of mercury in coastal
sediments. Marine Chemistry.
Trasande, L., Landrigan, P., and Schechter, C. 2005. Public Health and Economic
Consequences of MethylMercury Toxicity to the Developing Brain.
Environmental Health Perspectives, 113(5):590-597.
UNEP (United Nations Environment Programme). 2002. Chemicals, Global Mercury
Assessment. Geneva. OAR-2002-0056-5930, -5931, -5932, -5933.
US Environmental Protection Agency (US EPA). 1997a. Mercury Study Report to
Congress. Vol. 6: An Ecological Assessment for Anthropogenic Mercury
Emissions in the United States. USEPA-452/R-97-008. Downloaded from:
http://www.epa.gov/ttn/oarpg/t3/reports/volume6.pdf. A-92-55; I-A-130.
US EPA. 1997b. Exposure Factors Handbook: Volume II - Food Ingestion Factors.
EPA/600/P-95/002Fa. Washington, D.C. OAR-2002-0056-5955.
US EPA. 2001. Mercury Maps: A quantitative spatial link between air deposition and
fish tissue. Final Report. USEPA/823/R-01/009, US EPA, Washington, D.C.
OAR-2002-0056-6108.
U.S. EPA. 2002a (date of most recent revision of on-line materials; website accessed
March 2005). Integrated Risk Information System (IRIS). Methylmercury. U.S.
EPA Office of Research and Development, National Center for Environmental
Assessment. Oral RfD and inhalation RfC assessments last revised 2-13
53
-------
7/27/2001. Carcinogenicity assessment last revised 5/1/1995. Available online at
http://www.epa.gov/iris/subst/0Q73.htm. OAR-2002-0056-5793.
U.S. EPA. 2002b. Fish Consumption and Environmental Justice. Washington, D.C:
U.S. Environmental Protection Agency. OAR-2002-0056-5952.
US EPA. 2005a. Technical Support Document: Methodology Used to Generate
Deposition, Fish Tissue Methylmercury Concentrations, and Exposure for
Determining Effectiveness of Utility Emission Controls. Downloaded from:
http://www.epa.gov/ttn/atw/utilitv/eff fnl tsd-031705 corr oar-2002-0056-
6301.pdf. OAR-2002-0056-6301.
US EPA. 2005b. Cost and Energy Impacts Technical Support Document
US EPA. 2005c. Technical Support Document for the Final Clean Air Mercury Rule:
Air Quality Modeling. Downloaded from:
http://www.epa.gov/ttn/atw/utilitv/aqm oar-2002-0056-6130.pdf
54
------- |