EPA Response to Public Comments on the 2015 Draft Selenium Aquatic Life Ambient Water Quality Criterion


EPA Response to Public Comments on the
2015 Draft Selenium Aquatic Life Ambient
Water Quality Criterion

-------
Contents

Comment Category 1.1 - Comments on Sources of Selenium	1

Comment Category 2.1 - Comments of a General Nature Concerning National Criterion and Primacy
Structure	9

Comment Category 2.2 - Comments of General Nature Concerning Over-Protection or Under-Protection
of Aquatic Life	51

Comment Category 2.3 - Comments Concerning Fish Tissue Criterion Elements	63

Comment Category 2.4 - Comments Concerning Water Column Criterion Elements	105

Comment Category 2.5 - Comments Concerning Items Considered out of Scope	117

Comment Category 2.6 - Comments on Various Implementation Topics	124

Comment Category 2.8 - Comments Concerning the Protection of Threatened or Endangered Species
	143

Comment Category 3.1 - Comments on Use and Calculation of ECio	146

Comment Category 3.2 - Comments on Studies of Fish Reproductive Effects Used in Numeric Criterion
Derivation	152

Comment Category 3.3 - Comments on Deriving Egg-Ovary Criterion Element Value	183

Comment Category 3.4 - Comments Deriving Whole Body and Muscle Criterion Element Values	187

Comment Category 3.6 - Comments Concerning Other Toxicity Data Fulfilling Minimum Data Needs
(N=15)	195

Comment Category 3.7 - Comments on Studies Not Used in Derivation of Fish Tissue Criterion Elements
	198

Comment Category 4.1 - Comments on Translation Equation (Eq. 18)	204

Comment Category 4.2 - Comments on Derivation of Trophic Transfer Function (TTF) Values	213

Comment Category 4.4 - Comments on Derivation of Site-Specific Enrichment Factor (EF) Values	218

Comment Category 4.5 - Comments on EPA-Developed Food-Web Models and Calculation of Trophic
Transfer Function - Composite (TTF ComP°site) Values	230

Comment Category 4.6 - Comments on Classifying Categories of Aquatic Systems	232

Comment Category 4.7 - Comments on Deriving Water Column Criterion Element Concentration Value
	240

Comment Category 4.8 - Comments on Deriving Averaging Period and Intermittent Exposure Water
Criterion Element	246

Comment Category 4.9 - Comments on EPA's Validation of the Translation Equation (Equation 18).... 249

Comment Category 5.1 - Comments on Site Specific Modification of Fish Tissue Criterion Elements ... 251

-------
Comment Category 6.1 - Comments Concerning Corrections and Typos	258

Comment Category 6.2 - Comments Requesting Clarification	260

Comment Category 6.3 - Comments Concerning the Additional Data, Informational Sources, and
Alternative Approaches	262

Comment Category 6.4 - Comments Concerning Failure to Address 2014 Public and/or External Peer
Review Comments	263

Comment Category 7.1 - Comments about Inadequate Time for Review and Extension Requests	273

Comment Category 7.2 - Comments Requesting Delaying Issuance of Criterion	278

Comment Category 8.1 - References	279

Comment Category 8.2 - Additional Data	292

-------
EPA Response to Public Comments on the
2015 Draft Selenium Aquatic Life Ambient
Water Quality Criterion

EPA provided opportunities for the public to provide comments on the 2014 and 2015 drafts of EPA's
recommended criterion. EPA carefully considered the comments submitted and made revisions to the
final recommended criterion document in response to those comments. In addition, EPA prepared this
document which includes EPA's responses to public comments submitted on the 2015 Draft Selenium
Aquatic Life Ambient Water Quality Criterion document. This response document is organized in the
following manner: Public comments are arranged into major categories; this was done to facilitate
considering and responding to public comments overall. For each comment category a summary of
overarching public comments is provided, followed by EPA responses to major subthemes for each
category of comment. Following these EPA responses, specific public comments are provided. Not all,
full text public comments are included here, such as those of a general nature, or those captured under
specific subcategories; comments from individual commenters were divided across identified categories
as noted above. For the full individual public comments, the reader is directed to the public docket.
However, EPA has provided below comprehensive responses to public comments, with reference to
relevant locations in the 2016 Selenium Aquatic Life Ambient Water Quality Criterion document where
public comments have been addressed.

Comment Category 1.1 - Comments on Sources of Sell e mil urn
1.1 Summary

Several commenters provided specific examples of natural selenium sources that contribute to
elevated ambient concentrations. Several commenters described activities, such as mining and
irrigation that contribute to elevated concentrations of selenium in receiving waters. One
commenter asked for an update of the selenium occurrence geographic map.

1.1 Response to comments

1.1 Response to comment on selenium occurrence maps

EPA has replaced the two selenium occurrence geologic maps with one updated figure from the USGS,
(Figure 2.1. Selenium in Surficial Soils and Aquatic Sediments in counties of the Conterminous United
States).

1

-------
1.1 Response to the comment on natural sources of selenium

EPA referred to this issue in the 2015 draft document in section 2.1: "Selenium is a naturally occurring
element present in sedimentary rocks and soils. Where deposits of Cretaceous marine shales occur, they
can weather to produce high selenium soils; such soils are present in many areas of the western U.S.
(Lemly 1993c). Selenium is abundant in the alkaline soils of the Great Plains, and some ground waters in
California, Colorado, Kansas, Oklahoma, South Dakota and Wyoming contain elevated concentrations of
selenium due to weathering of and leaching from rocks and soils." EPA has added additional text to the
document, prior to the description of anthropogenic-related increases in selenium stating that "Natural
weathering of selenium-bearing geologic strata containing selenium can lead to selenium leaching into
groundwater and surface water."

Regarding the comment regarding inclusion of information from the GEI 2014 review on selenium
occurrence, EPA did not use this information in the document as presented in that review. The review's
figure (map with surface water Se concentrations) provided by GEI in their comment which shows the
2014 lentic value (1.3) exceeded at 64% of sites and lotic value (4.8) exceeded at 24% of sites used the
highest selenium value measured at the sites, rather than a concentration that would better reflect the
use of the entire data set, such as a central tendency estimate. EPA discussed potential approaches to
assessing sites with elevated selenium background in Appendix K of the criterion document.

1.1 Response to comments on site-specific criteria and high natural background selenium

The national 304(a) chronic selenium criterion was developed based on both laboratory and
environmentally-exposed organisms and was derived to protect ecosystems and organisms, most
notably oviparous fish, from the adverse effects of selenium, regardless of source. The criterion is based
on a 10 percent reproductive effect level in fish intended to be protective of 95% of organisms in
ecosystems.

States and tribes can develop site-specific criteria, where appropriate. There may be reproducing
populations of some species of resistant fish in some selenium polluted ecosystems. In developing a site-
specific criterion, it will be important to ascertain if these organisms reflect an ecosystem with a healthy
structure and function, prior to concluding that the designated use(s) are being met.

In Appendix K of the criterion document provides recommendations regarding site-specific selenium
criteria development.

1.1 Response to comment suggesting additional references

EPA has reviewed citations recommended by commenters and included new citations as appropriate.
Examples of citations and information added to the 2016 final selenium criterion document include the
Petrov et al. (2012) publication regarding separate forms of selenium in wastewater samples andEPA's
added text regarding selenium in shale deposits in the sources and occurrence section (Section 2.1),

2

-------
which addresses inclusion of the Burau 1985 citation. EPA thanks the commenters for the citations, and
Bennett and Janz 2007.

1.1 Specific comments

EPA-HQ-OW-2004-0019-0415-A1; Idaho Department of Environmental Quality (DEQ); Posted
10/28/2015

Given how much attention phosphate mining and selenium contamination have generated in Idaho it
seems odd to us that Idaho's Phosphate Patch (southeastern Idaho) does not show up on the "Map
indicating deposits of selenium in mining regions" (Figure 2.1, page 6). Also in the caption for this figure,
the units of mg/L do not make intuitive sense to characterize "underlying geology." Perhaps what is
being addressed here are measures of selenium concentrations in ground water influenced by geology?
If so that should be explicitly stated.

EPA-HQ-0W-2004-0019-0401-A1; Colorado Wastewater Utility Council; Posted 10/15/2015
1.1 Background Selenium Sources and Occurrences

EPA provides some discussion of Se sources in Section 3.1 of the 2014 draft selenium criteria document.
EPA discusses natural Se deposits and highlights areas where Se deposits have been brought to the
surface through mining activities and where irrigation in the western United States may cause leaching
from high Se soils. However, in addition to human-induced/ irreversible activities that may contribute to
elevated Se in surface waters, there are also natural processes that may leach Se into groundwater, and
consequently into surface waters, that were not discussed in the draft criteria document. We provided
more in depth information on this topic in our previous review (GEI 2014a).

3.4.2 Appropriateness of Water Column Criteria on a National Basis

In addition, in our previous review (GEI 2014b) we raised concerns regarding naturally occurring Se
deposits that may leach Se into groundwater, and consequently into surface waters (also discussed in
Section 4.1.1). Previous publications and reports have presented the potential for underlying geology,
rich in Se, to contribute to naturally elevated surface water Se concentrations which can be significantly
greater than the default water column values recommended in the 2015 draft Se criteria document (GEI
2013a, b, Herring and Walton-Day 2007, and Burau 1985). Burau (1985) states that of general parent-
rock types, shales have the highest Se content (500-28,000 ng/kg). Due the large areas of Se rich geology
throughout certain regions of the U.S., a water column criteria developed on a regional basis might be
more appropriate.

We conducted a review of surface water data provided through the Water Quality Portal (WQP;

National Water Quality Monitoring Council), accessed May 27, 2014). The information in this database is
composed of data from the USGS National Water Information System and from the USEPA STOrage and
RETrieval Data Warehouse. The complete dataset was filtered for dissolved Se in surface water sampled
from lotic waterbodies (e.g., rivers and streams) in Colorado.

-------
Using the data extracted from the WQP database as described above, Se concentrations were compared
to the draft lotic water quality criterion (3.1 ng/L) as plotted in Figure 1. The samples indicated by red
dots indicate those samples that are in exceedance of the draft lotic criterion. Dissolved Se
concentrations in lotic systems in Colorado exceed the proposed criterion of 3.1 ng/L in 30% of samples.
The majority of these exceedances occur in regions that overly selenium rich geology, or on streams that
have headwaters in regions with shales. Although we do not know the specific circumstances for each
waterbody, it is unlikely that mining or irrigation are causing all of these exceedences.

Original letter contains Figure 1 - Shale-containing geologic formations and dissolved selenium
concentrations in lotic systems across Colorado. See original letter.

In fact, GEI (2013a, b and 2014a) present information on selected stream segments in Colorado that
have approved site-specific Se standards based on elevated ambient concentrations due, in large part,
to the surrounding geology. These ambient site-specific standards were developed after use attainability
analyses were conducted in regions with shale-containing formations and demonstrated that elevated
Se concentrations were present in waterbodies with no (or limited) human influence, and that fish
populations in these regions were reproducing successfully despite elevated Se in both water and their
tissues, or that fish populations were limited due to other factors such as limited flow or habitat. These
ambient standards have been approved by both the Colorado Water Quality Control Division and EPA
Region 8.

If the EPA dataset and methods used for back calculation of water quality criteria (with the updated GEI
data) is narrowed down to just samples collected in CO the resulting lotic water column concentration
would be 5.8 ng/L based on 29 samples. This value is almost two times the proposed nationwide value,
again demonstrating that inclusion or exclusion of only a few data points can cause a significant change
in the calculated criterion, as well as highlighting the importance of consideration of regional and site-
specific water quality characteristics.

EPA-HQ-0W-2004-0019-0400-A1; The National Mining Association (NMA); Posted 10/15/2015
1.1 Background Selenium Sources and Occurrences

EPA-HQ-OW-2004-0019-0411-A2; American Petroleum Institute (API); Posted 10/28/2015
3.1 Background Selenium

-------
EPA provides discussion of Se sources in Section 2.1 of the 2015 draft Se criteria document. EPA
discusses natural Se deposits and highlights areas where Se deposits have been brought to the surface
through mining activities and where irrigation in the western United States may cause leaching from
high Se soils. However, in addition to human-induced/irreversible activities that may contribute to
elevated Se in surface waters, there are also natural processes that may leach Se into groundwater, and
consequently into surface waters, that were not discussed in the draft criteria document.

Previous publications and reports have presented the potential for underlying geology, rich in Se, to
contribute to naturally-elevated surface water Se concentrations which can be significantly greater than
the default water-column values recommended in the 2015 draft Se criteria document (GEI 2013;
Herring and Walton-Day 2007; Burau 1985). Burau (1985) states that of general parent-rock types,
shales have the highest Se content (i.e., 500-28,000 ng/kg).

3.1.1 Incorporation of Background Conditions into Site-specific Criteria

Appendix K of the 2015 draft Se criteria document, which describes methods for deriving site-specific
criteria, does not discuss inclusion/consideration of data from reference or background sites. This is
especially critical at sites with background levels of Se that would be considered elevated, comparatively
speaking. In Appendix A, Examples 1-6 include calculation of site-specific criteria that include
background Se concentrations of 5.0 ng/L. Working through the calculation results in site-specific water-
column criteria that are always lower than the original background concentration. It is unclear how
these calculations could be applied to sites with naturally-elevated background Se and result in a Se
criterion that would be appropriate for that site.

Ambient site-specific criteria may be appropriate in many parts of the country containing underlying
geology with elevated Se levels. This naturally-elevated Se may lead not only to elevated water
concentrations, but also to naturally-elevated fish tissue concentrations, resulting in the need for not
only site-specific water column criteria, but also a need for an approach that acknowledges there could
be site-specific tissue criteria.

One example of this is a study conducted on the St. Charles River near Pueblo, CO in 2011 and 2012 with
the objective of determining the appropriate water-based site-specific Se standard in a short reach of
stream influenced by underlying geology rich in Se. Data were collected seasonally from 2011 — 2012
and included the collection of: water and sediment samples, fish tissue, periphyton tissue, benthic
invertebrate tissue, fish and benthic invertebrate populations, and habitat evaluations. Young-of-year
fish populations of several species were abundant at sites with ambient Se concentrations averaging 121
Hg/L, with populations remarkably similar to those observed at sites with very low Se concentrations,
indicating no effect on juvenile recruitment in these reaches with elevated Se (EPA 2015 Table E-18; GEI
2007). Fish also appeared unaffected by the Se concentrations throughout the study reach with WB Se
concentrations ranging from 6.88 mg/kg dw to 105.88 mg/kg dw. A number of reasons have been
postulated for the persistence offish communities in locations with elevated Se (Canton 2010), which
EPA should consider in the criteria document.

-------
EPA-HQ-0W-2004-0019-0409-A2; OC Public Works; Posted 10/20/2015
10. Natural Background of Selenium

The natural background of selenium is not discussed in the Draft Se Criterion. A large number of waters
across the nation are impacted by naturally-occurring selenium with no or insignificant anthropogenic
influence. Treating natural selenium to perpetuity is clearly not reasonable and not a good use of public
resources.

EPA-HQ-OW-2004-0019-0412-A2; Utility Water Act Group (UWAG); Posted 10/28/2015
2.1 Overview of Selenium Sources and Occurrence

EPA states that: "For example, in scrubbers with forced oxidation systems that produce strong oxidizing
conditions and high temperatures, the majority of discharged selenium is in the form of selenate (Maher
et al. 2010)." Draft Criterion at 8. While this statement is true, UWAG notes that not all flue gas
desulfurization (FGD) scrubbers in operation are the forced oxidation-type. For FGD systems that are the
inhibited oxidation-type, the selenium chemistry is complex and selenate may not be primary form.
Petrov et al. (2012) identified a total of 13 separate forms of selenium in wastewater samples from
these facilities, with selenite, selenate, and selenosulfate being the dominant species in most samples.
We recommend that this citation be included in the final criterion document.

EPA-HQ-OW-2004-0019-0398-A1; Kansas Department of Health and Environmental
Watershed Planning, Monitoring & Assessment Section; Posted 10/15/15

Certain Kansas streams are subject to selenium inputs from underlying marine shales from the
Cretaceous Period, particularly the Pierre Shale and Niobrara Chalk, that manifests as naturally ambient
concentrations that are above our current criterion of 5 jig/I, let alone the proposed 3.1 ng/l criterion.
For example, numerous tributaries to the North and South Forks of the Solomon River flowing into
Waconda Lake in north-central Kansas have selenium concentrations from 6.6 - 16 ng/L. The main
natural source of selenium in the Waconda Lake drainage is from the weathering of upper Cretaceous
bedrock that underlies the drainage basin. The upper Cretaceous bedrock, primarily the Niobrara Chalk,
contains relatively high concentrations of selenium in comparison with other bedrock in Kansas. The
bentonite beds and shales in the Chalk can be especially high in selenium. Soils weathered from the
bedrock can have relatively high selenium content. Rainfall infiltrating through the high selenium soils
and weathered bedrock leaches selenium. Water discharging from the soil and weathered bedrock
transports dissolved selenium into streams. Evapotranspiration consumption of surface and ground
water in the drainage basin then further increases the selenium concentration of the stream water.

Discounting the consumptive use, which concentrates selenium in those streams, we find that our
background levels are persistently over the current and proposed selenium criterion, without an
anthropogenic influence. The proposed use of an intermittent exposure calculation does not provide
any relief because our background concentrations are often far above the 3.1 ng/l value and because
experience has shown that those elevated levels are near constant. Thus the fraction of the 30-day

-------
period used in the calculation will always be at or near 1, rendering the calculated value to be
indistinguishable from the proposed criterion. We urge EPA to work with us and other Midwest and
Western States to develop appropriate assessments and standards that reflect the dominant influence
of geology on selenium levels seen in our streams.

EPA-HQ-0W-2004-0019-0414-A2; Colorado Mining Association (CMA); Posted 10/28/2015

CM A requests consideration of background conditions when calculating site-specific criteria

EPA does not discuss consideration of naturally-elevated background selenium when determining site-
specific criteria. There are areas of Colorado that contain elevated selenium concentrations due to the
geologic conditions (e.g. marine shales, seleniferous soils, etc.). There needs to be a method to derive
criteria for areas that are naturally elevated in selenium. Development of criterion in these areas should
still allow a discharger to operate without imposing overly restrictive limits in the NPDES permit.

EPA-HQ-OW-2004-0019-0371-A2; Wyoming Department of Environmental Quality (WDEQ);
Posted 09/15/2015

The Wyoming Department of Environmental Quality, Water Quality Division (WDEQ/WQD) administers
the Clean Water Act in the State of Wyoming. As such, WDEQ/WQD develops, recommends and
facilitates the adoption of surface water quality standards in the State. WDEO/WQD routinely adopts
EPA's recommended 304(a) criteria to protect human health and aquatic life designated uses of surface
waters in the state. WDEO/WQD is particularly interested in EPA's draft aquatic life criteria for selenium
due to the abundant seleniferous geologic formations in Wyoming that result in naturally high selenium
concentrations in many water bodies.

WDEQ/WQD appreciates the tiered approach outlined by EPA in the draft criteria document; it is
evident that criteria were developed with considerable thought toward implementation and the
difficulties in applying a fish tissue based criteria into discharge permits and TMDLs. Further,
WDEQ/WQD is pleased to see EPA's efforts to include recommendations on how states would adopt
site-specific, fish tissue based criteria (e.g., mechanistic modeling, bioaccumulation factor approaches).
However, WDEQ/WQD is concerned that the criteria may be difficult to adopt and implement in the
majority of the State's waters and that EPA has not provided sufficient guidance and/or flexibility on
specific issues associated with the criteria.

Naturally Occurring Selenium

As outlined in the draft criteria document, selenium is a naturally occurring element present in
sedimentary rocks and soils; weathering of and leaching from these formations and soils can lead to
naturally elevated concentrations of selenium in ground and surface waters. Although EPA describes
these conditions in the draft criteria document, EPA doesn't go any further in describing flexibilities or
recommended approaches for criteria development for states that have naturally elevated selenium.
Therefore, WDEQ/WQD requests that EPA include recommendations on approaches and considerations
for developing site-specific criteria for states that have seleniferous geologic formations.

-------
EPA-HQ-OW-2004-0019-0413-A2; Federal Water Quality Coalition (FWQC); Posted
10/28/2015

2. Background Levels of Selenium

One critical issue with selenium is the presence of high background levels in many areas of the country.
The presence of selenium in bedrock, and its subsequent concentration in ground and surface waters,
varies across U.S. geographic regions, due to geology, climate and topography. Upper Cretaceous to
Oligocene marine sedimentary rocks are variably seleniferous, whereas Pliocene and younger
nonmarine sedimentary rocks are generally nonseleniferous. This is observed in the continental edges of
the Western US, where bedrock geology is primarily from the Tertiary marine sedimentary deposits. As a
result, adjacent surface waters and groundwaters may contain greater residuals of elemental selenium,
particularly if these waters are downstream of irrigated lands.

It is also the case that the magnitude of selenium levels in drainage-affected aquatic ecosystems is
strongly related to the aridity of the area and the presence of terminal lakes and ponds. These selenium
concentrations can be worsened in areas where the evaporation rates are at least 2.5 times that of
precipitation4. Further, erosion aids selenium transport from upland mountains from either irrigation or
natural rainfall. When the runoff meets a terminal stream, pond, or other isolated body of water,
evaporative enrichment of selenium occurs.4 (Two examples of these settings are Kesterson Reservoir
and the Salton Sea.) The concentration may elevate even further if the area or region is arid. Because, in
many cases, elevated concentrations of selenium in surface waters are due to natural conditions,
aquatic organisms in these regions may not be as sensitive to the possible effects of elevated selenium
concentrations as are organisms in other areas of the country. For all of these reasons, EPA should make
it clear that in developing water quality standards for selenium, States should carefully consider factors
such as geology, climate, topography, and the selenium sensitivity of receiving stream biota. Without
consideration of these contributing factors, the water quality standards that are developed will likely be
unrealistic and unattainable.

-------
Comment Category 2.1 - Comments of a General Nature Concerning
National Criterion and Primacy Structure

2.1 Summary

The majority of commenters supported the development of the criteria and hierarchy of the criterion
elements. A large number of commenters supported the fish tissue primacy and criterion structure. A
few commenters expressed general non-support. Many commenters were opposed to the national
water column element but recommended its use as a trigger for fish tissue monitoring, implementation
tool, or use in site-specific criteria. Several commenters had questions about implementation of the
criterion.

2.1 Response to comments

2.1 Response to comments supporting EPAs scientific approach

EPA appreciates the many comments supporting the development and overall structure of the criterion,
including those supporting EPA's:

• focus on the chronic criterion,

• focus on reproductive effects, which provide a more reliable basis for the criteria than non-
reproductive effects such as survivorship and growth endpoints,

• development of tissue-based elements for this bioaccumulative compound,

• the tiered hierarchy of the criteria elements (fish-tissue as the superseding criterion element),

• use of ECioS,

• data quality decisions, such as juvenile survival data including overwinter survival, and

• discussion on development of site-specific criteria.

2.1 Responses to general comments on data quality and the soundness of EPA's methodology for
deriving criterion elements.

The EPA used the best available peer-reviewed science in its development of the selenium water quality
criterion. The Agency reviewed all data to ensure it met the quality standards established in the 1985
Guidelines for derivation of water quality criteria for the protection of aquatic life.

In developing the 2016 final selenium criterion, EPA collected and reviewed a large quantity of peer-
reviewed scientific reports. EPA evaluated 81 studies on selenium toxicity to aquatic organisms,
identified in appendices C, D, and E; and in Section 3 of the criterion document. The 9 fish Genus Mean
Chronic Values (GMCVs) were calculated from 12 Species Mean Chronic Values (SMCVs), which were
calculated from 13 chronic values obtained from 24 studies. An additional 21 non-reproductive toxicity
values were obtained from 20 studies for 10 species, including 5 species that were not used in the SD.
Fish reproductive and non-reproductive toxicity test summaries are included in Appendix C and D of the
2016 criterion document, respectively. An additional 21 toxicity values from 22 studies encompassing 18

-------
species, seven of which were not included among the reproductive or non-reproductive studies listed
above, were evaluated and are included in Appendix E (other data). Three field studies with multiple
species were also evaluated qualitatively to assess the relative sensitivity of Cyprinidae to selenium, and
are included in Appendix E. Over one hundred studies were considered for the determination of fish
tissue conversion factors (CFs). Of these, 21 studies had paired fish tissue Se measurements from two or
more tissues that were used to calculate CFs. Over three hundred studies were considered that had
possible paired selenium measurements in one or more ecosystem compartments (water, algae,
sediment, detritus, invertebrates, and fish). Of these, 19 studies had paired invertebrate and particulate
Se measurements that were used to calculate invertebrate trophic transfer factors (TTFs), 30 studies had
paired fish and invertebrate measurements that were used to calculate fish TTFs, and 21 had paired
water and particulate measurements that were used to calculate enrichment factors (EFs). Over 50
studies were considered that had potential information for the calculation of a trophic transfer factor
based on physiological parameters. Of these, data from nine studies were used to calculate
physiologically-derived TTFs.

The resulting criterion reflects a comprehensive use of the best available science that is both technically
defensible and reflects a level of protection consistent with the protection goals behind other aquatic
life criteria developed by the Agency. The 2016 final selenium criterion database is substantially
expanded and improved over the information used to develop the prior final selenium 1999 criteria,
particularly regarding the number species for which data are available and the spatial representation of
the data.

Specific comments on conclusions regarding inclusion or exclusion of different studies (e.g., brown
trout) from the criterion are discussed in separate responses to Specific comments.

2.1 Responses to general comments on the hierarchy or tiering of the criterion elements

EPA has developed a tiered national 304(a) criterion, and is recommending that States and Tribes adopt
all four recommended elements and tiers. Tiering enhances both the scientific strength and the usability
of the selenium criterion. EPA clearly articulated the application of the tiered criterion in section 4 of the
2015 draft document, and also in the final 2016 document and this is supported by independent expert
peer reviewer comments.

Because the egg/ovary concentrations are the most closely associated and proximate to the adverse
effects, the egg/ovary criterion element is identified as the measurement that has primacy over the
other measurements, where adequate data are available. In the 2016 final criterion document EPA
identified two exceptions to this tiering, 1) for new input conditions when the fish tissue concentrations
may not yet reflect the "steady state" accumulation of selenium in fish tissue because of a lag time for
selenium moving from the water column through the food web into fish, and, 2) for conditions where
fish are absent in the aquatic ecosystem. Where fish tissue data are not available, water column
elements of the criterion remain applicable to protect aquatic life.

The EPA has made some changes to the footnotes associated with the criteria table in the Executive
Summary and Section 4 of the criterion document to clarify the hierarchal relationship among the tissue

10

-------
elements, and between the tissue water elements, as well as assertion of primacy of water over tissue in
fishless waters, and for new discharges until determination of steady state.

EPA paired data water column, detritus/plankton, and fish tissue samples collected within one-year due
to availability of data, and based on statistical analyses that this approach was appropriate for national
criterion development purposes. However, EPA recommends the collection of samples within
appropriately proximate temporal and spatial resolution to ensure the samples represent site conditions
during the collection of data for site-specific criteria development. Technical support documents, under
development, will contain more information on spatial and temporal resolution of samples for site-
specific criteria. Additional information on implementation of the criterion to ensure spatial and
temporal representativeness of samples will be available in technical support documents. These
documents will be made available for public comment after the publication of the final criteria
document.

2.1 Responses to comments on the scientific defensibility of water column criterion elements

The water column criterion element values were derived taking into account accumulation of selenium
in fish tissue, and are therefore intrinsically linked to the toxicity data underlying the egg-ovary criterion
element. EPA used a peer-reviewed scientifically robust model developed by Luoma and Presser (2010)
and worked closely with these developers to ensure that the model was appropriately applied in the
development of the national water quality criterion elements for selenium. EPA also conducted
additional independent external peer review to evaluate the use of the model in the derivation of the
water column criterion elements for selenium.

The water column elements are critical to the construct of the selenium criterion. Because there is
greater variability in the elements for water column than fish tissue, the water column elements are
ranked lower in the tiering hierarchy than those based on fish tissue (see above). It is necessary to
employ certain simplifications when deriving a national default water quality element, since the goal of
304(a) national criteria is to be protective of the designated uses of most waterbodies, most of the time.
EPA has taken the relative variability of the water column into account in deriving the national water
column criterion elements. EPA concluded it is appropriate to include both fish tissue and water column
elements in the criterion to ensure that the criterion as a whole is applicable to the diversity of
waterbodies in the US. In circumstances where water is the only media for which data are available (i.e.,
streams where fish are absent) or where selenium has not reached steady state in the ecosystem, for
example, for a new discharge, the water column element of the criterion provides a scientifically-
defensible translation of the fish tissue national criterion elements.

A few commenters have suggested that EPA conduct additional uncertainty analyses for the water
column criterion elements. Most of the studies that provide the data used to derive the water column
criterion elements do not provide enough information to conduct the analyses recommended. It would
not be possible to develop a national rate-based dynamic model using speciated selenium data because
sufficient data do not exist. One commenter requests an explanation of each parameter in the equation
that describes the kinetics of selenium bioaccumulation. Most of these parameters are measured in

-------
laboratory kinetic studies. All EFs and the majority of the TTFs are field-derived and because steady-
state conditions are assumed kinetic parameters are not relevant. One commenter noted the discussion
of growth rate in App J and requested EPA provide more detail.

• For more information on water column criterion elements, refer to responses to Comment
Category 2.4. For information on translation equation (Equation 18), refer to responses to
Comment Category 4.1.

• For information on deriving water column criterion element concentration value, refer to
responses to Comment Category 4.7.

• For information on modification of water criterion using Equation 18, refer to responses to
Comment Category 5.2.

Topic

Location in 2016
Criterion Document

Location in Appendices

Information on water column criterion elements

Section 3

information on translation equation

Section 3

information on deriving water column criterion
element concentration values

Section 3

Appendices B, H, J, and
K

For information on modification of water
criterion using Equation 18

Appendix K

For information concerning over-protection or under-protection of aquatic life, refer to responses to
Comment Category 2.2.

EPA has provided an expanded, overarching discussion of uncertainty in Section 6 of the criterion
document. Additionally, a comparison of the calculation of a site-specific water criterion using the
mechanistic model approach and the bioaccumulation factor (BAF) approach is included in Appendix K.
Only one study in the water column translation dataset (Saiki et al. 1993) had sufficient data to provide a
side-by-side comparison of the use of the mechanistic model and the bioaccumulation factor (BAF)
approach, which was conducted for four sites. It should be noted, however, that these data are sparse,
with only two selenium measurements per compartment (e.g., algae, detritus, fish species) per site. The
comparison is included in Appendix K to illustrate and compare the two translation methods

2.1 Response to Request for EPA to describe parameters used in the criterion derivation

Some commenters noted that EPA should provide an explanation of parameters used in calculating the
proposed water quality standard. EPA has provided parameter documentation Section 3.2.2.3 to section
3.2.6 of the 2016 criterion document. EPA also discussed the parameters and provides the data for all
parameters in Appendices A, B, H, I, and J of the 2016 selenium criterion document.

-------
2.1 Responses to general comments on the site specific selenium criterion elements.

All 304(a) national criteria are national; however, site-specific criteria can be developed if adequate data
are available. Several commenters provided examples of sites where site-specific criteria were
developed. For example, if a waterbody meets the fish tissue criterion element but not a water column
element, the waterbody may be a candidate for a site-specific water value.

Appendix K of the 2016 selenium criterion document provides information on how a site-specific
criterion elements could be developed.

Additional information on the development and implementation of site-specific criteria will be available
in a WQS Flexibilities Technical Support Document to be published after the publication of the final
selenium criterion.

2.1 Response to comments on effect of timing on spawning strategy and egg-ovary equivalence.

EPA notes that the text in the criterion document actually states: "In this document, concentrations of
selenium in ovaries are considered equivalent to concentrations of selenium in eggs because most
studies measured selenium in the ovaries prior to spawning." Section 6.1 of the 2016 selenium criterion
document notes that researchers who report concentrations of selenium in fish eggs or ovaries,
Osmundson et al. (2007), often found reduced levels of selenium in ovaries after spawning, presumably
due to the loss of selenium through spawning and release of eggs with relatively high concentrations of
selenium. Of the 14 chronic values determined from the maternal transfer reproductive studies, 11
values represent selenium measured in eggs. Three values represent selenium measured in the ovaries:
Schultz and Hermanutz (1990), Hermanutz et al. (1992, 1996) and Carolina Power & Light (1997).
However, information in two of these studies indicates that the concentrations of selenium in the
ovaries were similar to concentrations in eggs. Schultz and Hermanutz (1990) measured selenium in
fathead minnow ovaries at the end of the study from fish that presumably had spawned. The authors
found the concentrations of selenium in the ovaries and embryos of the fathead minnows exposed to
the same treatments to be similar. Hermanutz et al. (1992, 1996) sampled adult female bluegill just prior
to spawning and at the end of the test (post spawning) and found no decreases in the concentration of
selenium in the post-spawned fish. In the Carolina Power & Light (1997) study, the third study, selenium
in ovaries of largemouth bass (Carolina Power & Light 1997) was measured from fish sampled just after
spawning. No comparison to pre-spawning fish or selenium in eggs can be made for the largemouth bass
study, however, the ECio of 26.3 mg Se/kg ovary dw was mid-range of sensitivity indicating this test was
not overly conservative due to lower selenium measurements in post spawning ovaries. Based on the
indications that the selenium concentrations in the ovaries were similar to that in eggs in the Schultz and
Hermanutz (1990) and Hermanutz et al (1992, 1996) study observations stated above, egg selenium and
ovary selenium were considered equal for the toxicity data set. Any potential error resulting from this
assumption would be conservative since the effect of spawning only lowers the selenium concentration
in the ovary. EPA recognizes selenium ovary concentrations may vary in field collected samples due to
fish reproductive cycles and will address such concerns in the implementation information.

-------
The Conley (2014) reference has been reviewed and considered. The agency is aware that spawning
strategy likely plays a role in maternal transfer and deposition of selenium into the eggs. The issue raised
in Conley 2014 will be addressed in the technical support materials under development regarding the
timing and type of sampling (egg-ovary vs whole body vs muscle tissue) based on species spawning
strategy (single batch versus multiple smaller batches) and seasonality (spring [e.g., sunfish, some
salmonids] or fall [e.g. brown trout] versus sporadic spring/summer [e.g., cyprinids]), while also
providing flexibility to states to adapt their monitoring programs to address the additional requirements
for fish tissue monitoring for selenium.

Thank you for your comments. EPA considered all available data for use in the derivation of the water
column criterion elements. EPA applied the QA procedures described in the 2016 criterion document for
the translation effort:

2.1 Specific comments

EPA-HQ-OW-2004-0019-0403-A2; Iowa Department of Natural Resources; Posted 10/15/15

EPA recommends that states adopt all elements of the proposed criteria. However, only fish tissue (egg
ovary) criteria were derived based on toxicity data. All other elements of the proposed criteria were
estimated to protect the fish tissue (egg ovary) criteria using models. For example, EPA derived the 30-
day chronic water column element from the egg-ovary element by modeling selenium bioaccumulation
in food webs of lotic and lentic aquatic systems. The modeling approach is subject to a greater degree of
uncertainty than the toxicity based approach used in the derivation offish tissue (egg ovary) criterion.
As a result, the four elements of the criteria are not scientifically equivalent and thus should not be
applied independently for impairment purposes.

EPA should consider removing the egg/ovary criterion from the options of media/criteria combinations
available to states. As noted in Iowa Comment No.l, the egg/ovary criterion is the only criterion that is
based on toxicity testing.

The remainder of the recommendations for fish tissue (whole body and muscle) and water (30-day
averages and intermittent) are all modeled derivations based on the egg/ovary data. In order to
standardize EPA's criteria recommendations for selenium (i.e., all modeled derivations), the egg/ovary
criterion should be kept separate from the other recommendation. In addition, while there is a long
history of state fish tissue monitoring programs, states do not typically analyze fish eggs or fish ovaries.
EPA's tissue-based criteria recommendations for whole-fish and muscle appear to be sufficient for
assessing potential aquatic life impacts due to selenium.

EPA recommends that states adopt all elements of the proposed criteria with the implication that all
criteria elements be applied to all Clean Water Act programs. Such an approach, however, is
scientifically and technically questionable in that not all four criteria elements are appropriate for
meeting all Clean Water Act requirements. At a minimum, the criteria document should be specific as to
whether EPA expects that all media/criteria combinations be applied to all CWA programs or whether
states can select the most appropriate media/criteria combination to meet its varying CWA
responsibilities. For example, the draft criteria document says that tissue-based criteria for selenium are

-------
given precedence over water-based criteria. EPA should clarify the relationship between tissue-based
and water-based criteria for purposes of identifying Section 303(d) impairments. Iowa's presumption is
that the precedence of tissue-based criteria allows states to identify "full support" of aquatic life use in a
given waterbody if levels of selenium in tissue are below the criteria even if levels of selenium in water
exceed the water-based criteria.

Ambiguity on these issues in the criteria document will likely result in inconsistent application of the
media/criteria combinations between EPA regions and may result in unnecessary negotiations regarding
selenium criteria between EPA regions and their respective states.

EPA needs to provide implementation methodologies and examples on how to translate the fish tissue
criteria to a wasteload allocation and NPDES permit limits. EPA should clarify if States can use only water
column criteria to derive wasteload allocations and NPDES permit limits. In addition, EPA needs to
describe how the intermittent exposure criteria should be implemented, since it is not an acute
criterion.

EPA-HQ-OW-2004-0019-0398-A1; Kansas Department of Health and Environmental
Watershed Planning, Monitoring & Assessment Section; Posted 10/15/15

KDHE supports EPA developing criteria using whole-body fish tissue to conduct impairment assessments
for selenium. The criteria using whole-body fish tissue more closely links the pollutant to the aquatic life
designated use we are trying to protect and because the predominant pathway for harmful selenium
accumulations is through the food chain rather than exposure in the ambient water column. We also
appreciate the elimination of the acute criterion, which the current value for Kansas is 20 ng/l. We rarely
see those levels in our monitoring network and, when we have, there has been no incident of mortality
of aquatic life under those elevated concentrations.

For previous TMDL and NPDES purposes, KDHE has contracted the Kansas Biological Survey to conduct a
few assessments of site specific selenium data collected from the water column, sediment and fish
tissue at monitoring stations on the Arkansas, Pawnee and Solomon Rivers. In doing so we found weak
or non-existent linkages between ambient water concentrations and corresponding selenium
accumulations in fish tissue. Some instances of elevated fish tissue concentrations above the proposed
criterion of 8.0 or 11.3 mg/kg were observed, but not at a frequency one would expect to occur with
consistent water column concentrations well over our current chronic criterion of 5 ng/l.

This causes us to worry that adoption of the proposed water column criterion will present an overly
conservative prospectus of impairment that corresponding fish tissue levels contradict. Our preference
would be to use the water column values as a non-regulatory trigger threshold to conduct more direct
assessment of fish tissue accumulations to ascertain if an impaired condition is present. Alternatively, as
more fish tissue data become available, we may suggest a Kansas-specific relationship between
observed water column concentrations and corresponding fish tissue levels as the basis for deriving
appropriate NPDES permit limits and Section 303(d) listing thresholds for selenium in the future.

-------
EPA-HQ-OW-2004-0019-0397-A1; Peabody Energy; Posted 10/15/15

Peabody disagrees with a single water column concentration applied across the nation. EPA used site
specific data to derive the proposed water column concentration. Using the same method site specific
data from other areas would generate a different limit. It would provide increased accuracy to allow site
specific generation of water column concentrations.

Peabody supports the use of fish tissue based standards as a method of determining appropriate Se
standards. Concentrations of Se in the water column are often not predictive of concentrations of Se in
fish tissue. Directly testing fish tissue appears to be the most practical approach at this time. Fish tissue
limits and egg tissue limits should be included directly in NPDES permits using water column
concentrations as a trigger only.

EPA-HQ-OW-2004-0019-0395-A1; CONSOL Energy; Posted 10/15/15

CONSOL is in support of EPA's decision to develop aquatic life water quality criterion for selenium,
including the tissue-based criteria approach, and feels this is an improvement over the 2004 draft
criteria document. These criteria are more consistent with the latest science regarding selenium toxicity.
Upon reviewing the Draft document, we would like to submit the following General Comments:

• We would like to acknowledge the extensive effort that EPA and others put into development of
these updated chronic selenium criteria and realize the efforts made to create a scientifically
sound criteria document.

• We support the use of a fish-tissue based chronic criterion as the superseding criterion, as this is
the more ecologically relevant measure of toxicity for selenium.

• We are generally in support of the following sections of the 2015 draft criteria document:

o Tissue-based standards
o Use of ECioS

o Timing of tissue data collection

• Exclusion of juvenile survival data

CONSOL does not agree with the approach used to develop national water-column criteria. A single
nationwide standard is not appropriate. The use of site-specific water-column standards, calculated
using EPA"s Equation 18 from the criteria document, would be the more scientifically justifiable
approach.

EPA-HQ-OW-2004-0019-0390-A1; West Virginia Coal Association (WVCA); Posted 10/14/15

As set forth above, WVCA is pleased that the Draft Selenium Criterion are based upon fish tissue
concentrations of selenium, with water column numbers considered secondary for determining the
health of the aquatic system. As EPA has acknowledged, certain selenium species are more
bioaccumulative than others, and site-specific factors may affect selenium toxicity. Therefore the
hierarchy set for the application of selenium criteria, where egg-ovary numbers are considered over

-------
whole body numbers, which in turn are considered over water column numbers, is appropriate and
meaningful.

WVCA has concerns with a number of components of the Draft Selenium Criterion. While WVCA
believes that the document contains far more political rhetoric than appropriate for a scientific criterion,
we will not belabor our comments with this matter and instead will focus on the technical components
of the Draft Selenium Criterion. Specifically, the methodology utilized in calculating the Draft Selenium
Criterion is substantially different than typically utilized in preparing aquatic life water quality criteria.
While WVCA agrees that the selenium criteria should be based upon selenium body burdens for fish, we
do not agree with EPA's data decisions for inclusion and interpretation of certain studies which are
critical to the calculated criteria. WVCA is also concerned with EPA's water column elements of the
chronic criterion, which are back-calculated from the body burden elements using an unconventional
methodology that is at odds with the procedure for preparing water quality criteria set forth in EPA's
1985 Guidelines for Deriving Numeric National Water Quality Criteria for the Protection of Aquatic
Organisms and Their Uses (1985 Guidelines). Finally, EPA places illogical conditions on the applicability
of the selenium criterion, particularly regarding "fishless waters."

EPA-HQ-OW-2004-0019-0393; North Dakota Department of Health; Posted 10/14/2015

The NDDoH does not support a chronic water column criterion. The NDDoH believes that the
assumptions linking levels of selenium in fish tissue to water column concentrations do not account for
the numerous naturally occurring uncertainties.

The principle avenue for selenium exposure by fish and other aquatic organisms is through ingestions
and not through the water column. Since selenium concentrations in North Dakota are predominately
natural and increases in water column concentrations are usually linked to precipitation and runoff
events, the NDDoH does not believe that a chronic water column criterion (unless triggered by a known
tissue concentration) is defensible. To assign a chronic water column criterion to waterbody without
adequate science is disingenuous to the state's citizens.

In summary, NDDoH's view is that the draft selenium chronic water column criteria are premature. The
linkage between the tissue concentrations and the water column tenuous, and that the process for
implementing the intermittent criterion impractical due to naturally occurring background
concentrations. North Dakota also supports the comments provided by the state of Kentucky's Division
of water.

Although NDDoH recognizes that the criterion are not self implementing, and that each state may adopt,
revise or disregard these recommended criterion based on the applicable science for their waters,
setting a water quality criterion well below naturally occurring background levels places an undo burden
on the states to address that value during their triennial standards review. It would be better for EPA to
remain silent, until the sound scientific basis is established, and let those states that desire to be more
stringent proceed on their own. Also, it should be noted specifically that states may adopt criteria less
stringent than the promulgated value if the science specific to their aquatic ecosystems supports it.

17

-------
EPA-HQ-OW-2004-0019-0406-A2; Kentucky Chamber of Commerce; Posted 10/15/2015

While the Chamber supports U.S. EPA's efforts to update the recommended aquatic life chronic criterion
for selenium to recognize the preference of using fish tissue over a water-column derived standard, the
Chamber is concerned about aspects of the proposed criterion related to circumstances where water
column concentration criteria would be deemed controlling. The preference for a fish tissue criterion
over a water column criterion is a strength of the proposal because as noted in relevant scientific
research, the toxic effects of selenium in fish are due to the bioaccumulation of selenium through diet
and not the water column concentration level.

The Chamber appreciates the opportunity to provide these comments and urges U.S. EPA to narrow the
recommended criterion to fish tissue criteria, which are deemed to take precedence of any water
column indicators.

The two qualifiers used in footnote 3 of Table 1 of the draft criterion would trigger the controlling
applicability of a water column concentration criterion and create implementation concerns due to a
lack of adequate definition. First, the condition for "primacy" of a water column concentration criterion
for "new or increased inputs of selenium until equilibrium is reached" is not defined by specific factors
regarding the nature of the discharge, including the duration, magnitude, or frequency of such inputs.
Therefore any new inputs could potentially trigger the use of the water column criterion instead of the
fish tissue criterion. Second, the condition does not differentiate discharge scenarios where similar
discharges of selenium have been occurring to the water body, which is frequently the case in coal
mining regions of Kentucky. Third, the term "equilibrium" is especially troubling and is also not
adequately defined. The lack of clarity from a technical standpoint would create much uncertainty with
respect to how equilibrium conditions would be written into permits.

More fundamentally, the record does not provide a scientific basis to support the need for a stringent
water column criterion to supplement the fish tissue standard with respect to all new "inputs" of
selenium. Specifically, the fish tissue and egg/ovary criteria would still adequately define whether the
aquatic life use is protected, and NPDES permits could be written with monitoring conditions for the
water column and fish flesh to ensure the aquatic life use is protected. Therefore, the Chamber
recommends deleting the conditions in footnote 3 that remove the preference for fish tissue values over
water column values for new or expanded discharges of selenium.

EPA-HQ-OW-2004-0019-0379-A1; Pennsylvania Department of Environmental Protection;
Posted 09/30/2015

Pennsylvania's current chronic selenium water quality criterion is 4.6 ug/L for the protection of aquatic
life. EPA's proposed selenium criterion for lotic waters is 3.1 ug/L. Industries that may be affected by the
adoption of this lower proposed selenium criterion could include mining activities, coal mining and
processing facilities, coal-fired power plants, and coal refuse facilities. As a result, this criterion must be
carefully evaluated and considered before final adoption.

-------
EPA-HQ-OW-2004-0019-0380-A1; California Association of Sanitation Agencies (CASA); Posted
10/01/2015

Over the years, CASA members have worked with other stakeholder groups to provide comments on
previous versions of the proposed selenium criteria. As such, we are very supportive of the approach
taken in this most recent revision. This approach incorporates multiple tiers of criteria to establish
selenium thresholds protective of aquatic life uses. As such, use of fish organ (ovary/eggs) and tissue
selenium data are used exclusively when such data are available. In the absence of these direct
measurements, water column selenium criteria are being proposed. By incorporating overall fish organ
or fish tissue thresholds, the proposed criteria is directly assessing the aquatic life beneficial use and
only relies on the less direct and less reliable water column criteria when direct organism measurements
are unavailable or the water body does not currently support fish populations. Use of this approach will
allow for a more direct assessment of the beneficial use impacts associated with selenium exposure and
should significantly minimize unnecessary and costly selenium treatment controls in systems currently
achieving these direct beneficial uses. In addition to commending the EPA on this proposed approach,
CASA strongly encourages EPA to utilize such a multiple tiered approach in future criteria development
and/or reevaluation whenever appropriate.

EPA-HQ-0W-2004-0019-0416-A1; Wyoming Mining Association (WMA); Posted 11/02/2015

In conclusion, WMA supports EPAs efforts to develop selenium standards that protect aquatic life.
However, the final standards must be derived and applied in an appropriate, site specific and attainable
manner tailored to the aquatic life being protected. WMA additionally supports the selenium comments
from the National Mining Association and their consultant, GEI.

WMA can support the use of fish tissue standards as a method of determining appropriate selenium
standards. Because concentrations of selenium in the water column are not always predictive of
concentrations of selenium in fish tissue, directly testing fish tissue appears to be the most practical
approach at this time. Fish tissue limits should be included directly in National Pollutant Discharge
Elimination System (NPDES) permits using water column concentrations as a trigger only where fish are
present on the specific site.

Need for Site Specific Standards and Guidance for Areas with Naturally Elevated Selenium: WMA
believes it is inappropriate to specify national water column-based standards. The great variety of
geologic environments, climates, and landscapes lead to tremendous variability in the aquatic life that is
potentially affected. National standards cannot address this level of variability.

As the Environmental Protection Agency (EPA) acknowledges in the draft criteria document, weathering
and leaching of seleniferous rocks and soils can lead to naturally elevated concentrations of selenium in
ground and surface waters. Many healthy streams, ponds, and wetlands exist in Wyoming with waters
that may periodically or permanently exceed the published water column standards. Often these waters
provide the only source of water for wildlife and waterfowl in the local environment. For these reasons a
standard, and especially a national standard, is unacceptable.

-------
EPA needs to develop and explain clear and practical procedures and approaches for states to use in
developing achievable site specific criteria for sites with naturally elevated selenium. The general and
vague descriptions presented thus far do not provide adequate guidance and in fact, provide the EPA
with ample opportunity to simply deny state efforts without explanation or justification.

EPA-HQ-OW-2004-0019-0384-A2; Cameco Resources; Posted 10/13/2015

Finally, we fully support the comments of the North American Metal Council-Selenium Working Group
(NAMC-SWG) on both the 2014 and 2015 draft selenium criterion documents. Cameco encourages the
US EPA to work with the NAMC-SWG given this group's sound, science-based perspective on issues
pertaining to selenium and many years of expertise.

EPA-HQ-0W-2004-0019-0409-A2; OC Public Works; Posted 10/20/2015

The effort to update the Criterion is appropriate given that the science has progressed so much in recent
years and a significant update of the California Toxics Rule may be warranted as a follow-up action.
USEPA's effort to use tissue-based criteria that directly link to the true protection of designated uses is
also appropriate and a significant advance over traditional water column-based criteria.

We have reviewed the comment letter from Gel Consultants, Inc. and support their recommendations,
especially those on the chronic tissue criterion concentrations based on revisions to the treatment of
three species. Our general comments below are primarily focused on the water column portion of the
Criterion.

For the comment 1 above, USEPA should either delay the finalization of this Draft Se Criterion or
incorporate the latest scientific consensus on the development of aquatic life criterion to ensure that
the Draft Se Criterion is forward-looking and consistent with the latest science.

The pioneering work by Samual Luoma and Teresa Presser (extensively cited by the Draft Se Criterion),
including work in the Newport Bay Watershed, recognizes the data gaps, uncertainties and the
inappropriateness of the 'one size fits all' approach for water column criterion and proposed water
column guideline for the watershed. In fact, due to heterogeneity, more than one water column
guideline was proposed for different subareas of the Newport Bay watershed. This is a more appropriate
approach.

While the tissue-based criteria appear to be based on sound and best available science, the water
column portions of the Criterion have significant issues. The Draft Se Criterion includes 4 numerical
criteria, all based on fish egg/ovary numbers, so egg/ovary, and to some extent whole fish, should be
the primary numbers and all others secondary. Water column numbers, due to significant uncertainties
and large variability in different waterbodies and environments (as elaborated below), are not
equivalent with the tissue criteria and should not be treated thus.

In the past, the water column-based water quality criteria have traditionally been derived from toxicity
tests directly. In fact, the 1985 Guidelines were specifically prepared for this purpose. Without a peer-

-------
reviewed guideline that specifies a standard method to derive the water column-based criterion
indirectly, as being proposed by the Draft Se Criterion, the water-column portion of the Criterion is
highly questionable. Instead of being an enforceable criterion, water column numbers should be used as
part of the implementation measures to gauge the progress toward attainment of tissue (including
egg/ovary) criteria.

Large Uncertainties in Dataset Underlying Water Column Criteria: The water-column criteria, as derived
by the bioaccumulation model, failed to acknowledge the excessively large uncertainties associated with
a wide range of parameters that the model uses to translate tissue criteria to water column criteria. The
trophic transfer factors (TTFs) can often vary 50% or more (Table B8), with many data unusable due to
poor or even inverse correlations between the concentrations in the pray and predator (e.g. graphs on
B6, B18, B63, and B67). Enrichment factors (Efs), more commonly called particle-water partitioning
coefficients (Kd), can often vary by an order of magnitude or more (Appendix H, where EF varies from
less than 0.1 to more than 10). With Kds and each of several TTFs bringing considerable uncertainties,
the derived water column criteria suffer from exorbitant errors and uncertainties that render their
correlation with tissue-based criteria extremely tenuous.

EPA-HQ-OW-2004-0019-0378; Public comment submitted by C. Lish; Posted 09/25/2015

I am pleased to see that the EPA has made improvements in the 2015 draft, compared to what was
present in the 2014 draft. I approve of the EPA for lowering the criterion for lotic (flowing) water
systems from 4.8 g/L to 3.1 g/L. I also approve of the clarification that water column values have
primacy over fish tissue values under two circumstances: 1) "Fishless water" (waters where fish have
been extirpated or where physical habitat and/or flow regime cannot sustain fish populations); and 2)
New or increased inputs of selenium from a specific source until equilibrium is reached.

I strongly encourage the Environmental Protection Agency (EPA) to strengthen the chronic selenium
criterion. While I understand that measuring the level of selenium in fish tissue is a scientifically
defensible method for determining the current impacts of selenium in a waterway, it is wholly
inadequate for enforcement of permit limits for selenium released from industry point source
discharges.

EPA-HQ-0W-2004-0019-0405-A2; Virginia Coal & Energy Alliance (VCEA); Posted 10/15/2015

We support EPA's decision to base its recommended criteria on fish tissue concentrations of selenium,
and to provide a hierarchy where egg-ovary numbers are considered over whole body numbers, and
both, in turn, are considered over water column numbers. This approach is appropriate, given the fact
that certain selenium species are more bioaccumulative than others, and site-specific factors may affect
selenium toxicity.

The Virginia Coal and Energy Alliancel ("VCEA") writes to comment on EPA's Draft Aquatic Life Ambient
Water Quality Criterion for Selenium-Freshwater 2015, which was published in the Federal Register on
July 27, 2015. We support the use of a fish-tissue based chronic criteria as the overriding criteria, as this

21

-------
is the most ecologically relevant measure of selenium toxicity. However, as noted below, we have
concerns about the manner in which the criteria were derived and will be implemented.

In offering these comments, we endorse and incorporate the questions and concerns raised by West
Virginia Coal Association ("WVCA") and National Mining Association ("NMA") in their written submittals.
In particular, we question EPA's justification and rationale for deviating from its long-standing
procedures and practices for developing water quality criteria, as set forth in the agency's 1985
Guidelines for Deriving Numeric National Water Quality Criteria for the Protection of Aquatic Organisms
and Their Uses ("1985 Guidelines"). Although we generally agree that selenium criteria should be based
upon selenium body burdens for fish, we do not agree with EPA's data decisions for inclusion and
interpretation of certain studies that are critical to EPA's proposed criteria. We also share WVCA's
concern with EPA's water column elements of the chronic criterion, which are back-calculated from the
body burden elements using an unconventional methodology that is at odds with EPA's normal
procedure.

Anticipating that EPA's national recommendations may one day be considered for adoption in Virginia,
we are deeply concerned that many of the fish species evaluated by EPA are simply not present in our
state waters. In other words, the "representative aquatic systems" considered by EPA are not, in fact,
representative of Virginia. In particular, of the ten species evaluated by EPA, seven are not present or
widely occurring in the state (Dolly Varden, Northern Pike, Desert Pupfish, Cutthroat Trout, Rainbow
Trout, Brown Trout and White Sturgeon). We respectfully submit that before Virginia proceeds to adopt
any of EPA's final criteria recommendations, the underlying analysis would have to be redone in a
manner consistent with the actual "representative aquatic systems" in Virginia.

EPA-HQ-OW-2004-0019-0394-A2; Appalachian Mountain Advocates and Sierra Club
Environmental Law Program; Posted 10/14/2015

Commenters are greatly concerned about EPA's Draft Recommended Aquatic Life Ambient Water
Quality Chronic Criterion for Selenium, notice of which was provided in the Federal Register on July 27,
2015. These groups are very familiar with the dangers posed by selenium pollution, specifically selenium
pollution from coal mines and related facilities in the central Appalachian region. Our organizations have
been the primary drivers of enforcement of the existing selenium standards in the region, repeatedly
overcoming efforts by industry and compliant state regulators to avoid and delay addressing the
problem of selenium pollution. If EPA's Draft Criterion—which gives primacy to fish-tissue
concentrations1 in determining compliance—is finalized and adopted by states in the region, we believe
that citizen enforcement will be significantly undermined and that selenium pollution in the region will
not be adequately addressed by the underfunded and industry-friendly state regulators.

Commenters are concerned not only that the Criterion is effectively unenforceable, but also that it
contains serious scientific flaws that render it unprotective of sensitive aquatic life, aquatic-dependent
wildlife, and endangered species. In order to develop criteria that are both practically applicable and
fully protective of sensitive species, EPA must revise its fish tissue elements downward before
translating them to enforceable water column criteria.

22

-------
The Criterion Should Be Expressed As Practically Enforceable Water Column Elements

In passing the CWA, Congress recognized the fact that water quality standards - which existed prior to
1972 - would not, of themselves, protect and improve water quality. Accordingly, Congress established
the National Pollutant Discharge Elimination System (NPDES), providing a mechanism for clear
application and enforcement of water quality standards. Further frustrated with a lack of progress in
realizing the promise of narrative water quality goals, Congress again amended the Act in 1987, at that
time requiring the development and application of numeric criteria for waterways affected by toxic
pollutants. These revisions clearly illustrate Congress' intent to assure that water quality standards and
goals are specific and translated into enforceable limitations on pollution sources.

Water quality criteria thus not only measure whether water bodies are meeting the uses mandated by
the CWA, but also form the basis for establishing effective controls on water pollution to further the
CWA's goal of "restoring] and maintaining] the chemical, physical and biological integrity of the
nation's waters." See 33 U.S.C. § 1251(a). As EPA has recognized, water quality criteria must "serve the
dual function of establishing water quality goals for a specific waterbody and providing the basis for
regulatory controls." EPA Water Quality Standards Handbook at 4.6 (emphasis added). See also 40 C.F.R.
§ 130.3 (noting that water quality standards "serve the dual purposes of establishing the water quality
goals for a specific water body and serving as the regulatory basis for establishment of water quality-
based treatment controls and strategies"). Although a fish tissue-based criterion may be an accurate
way to measure the threat posed by selenium in a waterbody (if the criterion is set at the appropriate
level), it fails to provide the basis for effective regulatory action.

Based on the Commenters' extensive experience, the adoption by any central Appalachian state of a
criterion that gives precedence to fish tissue elements will present obstacles to enforcement that
undermine the dual function of water quality standards and will result in a lack of protection of the
aquatic life in Appalachian streams, rivers, lakes and reservoirs. Because of a history of lax or non-
existent enforcement by regulatory agencies in central Appalachia—particularly with regard to selenium
pollution from coal mining operations—we strongly oppose the adoption of fish tissue criteria that are
not translated to independently enforceable water column criteria. Indeed, the only reason that the
serious problem of selenium pollution from coal mines is being addressed in the region at all is that
citizens have been able to compel compliance with the existing water column criteria in permitting and
enforcement actions. The state agencies have opposed and undermined those efforts at every step,
consistently choosing delay and appeasement over meaningful regulation. Our experience enforcing the
selenium standard in central Appalachia has taught us that the only way that coal operators and others
in the region will be compelled to comply with selenium standards is if there is an enforceable water
column number. We expect that this is true in many other areas of the country as well.

Indeed, in 2005, the USEPA/U.S. Department of Interior Tissue-based Criteria Subcommittee issued a
draft report summarizing its opinions on aquatic life water quality standard guidelines. The report
cautioned that fish tissue criteria alone would be insufficient to address "both scientific and regulatory
needs concerning the relationship between chemical loadings and accumulated chemical residues in the
tissues (i.e. bioaccumulation)." Science Advisory Board Consultation Document, Proposed Revisions to
Aquatic Life Guidelines, Tissue-Based Criteria for "Bioaccumulative" Chemicals at 10. In the

-------
Subcommittee's opinion, there was a "need to develop guidelines for translating tissue-based aquatic
life...criteria into corresponding concentrations in environmental media (e.g. water)..." Id At 13. The
Subcommittee subsequently listed "implementability" as a reason to develop fish-tissue-to-water-
column translations, noting that "monitoring and enforcing pollutant discharge limits on the basis of
measured chemical concentrations in tissues of organisms may not be practical or desirable..." Id. The
central Appalachian states' inability or unwillingness to enforce the existing, simple selenium water
column criteria demonstrates the imprudence of adopting fish-tissue criteria that are significantly more
difficult and costly to implement.

1 For ease of use, these comments refer to the whole body, muscle, and egg/ovary elements of EPA's Draft
Criterion as "fish tissue" elements.

EPA-HQ-0W-2004-0019-0414-A2; Colorado Mining Association (CMA); Posted 10/28/2015
CM A supports with the fish tissue based standard approach

CMA agrees that fish tissue standards are an appropriate method for determining selenium standards.
Water column concentrations are not always predictive of concentrations in fish tissue due to many site-
specific factors. Therefore, the water column concentrations should be used as a trigger only. In the
event a water column concentration is exceeded, fish tissue testing should be used to determine
compliance.

In conclusion, CMA supports EPA's efforts to derive selenium standards protective of aquatic life.
However, such standards should be applied practically in order to prevent undue cost and burden on
industry. CMA also supports comments from the National Mining Association and Gel Consultants.

CMA believes the water column concentrations are overprotective

The water column concentrations for lentic and lotic environments were derived using conservative
assumptions that result in water column concentrations that are overly protective for many locations in
the country. It appears that the water column concentrations were derived using the 20th percentile of
the distribution of water column concentrations from selected sites. However, each of the water column
concentrations shown in the distribution was based on site-specific parameters and is actually
protective of location from which it was derived. The distribution shows that the appropriate criterion
actually ranges from 0.23 ng/L to 50.4 ng/L for lentic sites and 1.2 ng/L to 40.6 ng/L for lotic sites.
Arbitrarily using a metric such as the 20th percentile is not a valid method for determining a nationwide
criterion.

Using these overly-restrictive water column concentrations as a trigger for fish tissue sampling will result
in widespread fish tissue sampling that is extremely costly and difficult to measure and may not be
necessary at a number of locations. Nationwide criteria may not be appropriate considering the large
range in water column concentrations that is seen. Instead, it may be appropriate to address the water
column concentrations regionally.

-------
EPA-HQ-0W-2004-0019-0404-A2; Kentucky Coal Association (KCA); Posted 10/15/2015

As you are aware, Kentucky was the first state to include selenium criteria based on fish tissue levels in
its Water Quality Standards (WQSs). KCA firmly believes water quality standards must be based on
sound scientific rationale and appreciates EPA's recognition of this important tenet of water quality
standards.

KCA conditionally supports EPA's approach to developing the chronic criterion for selenium. KCA
supports EPA's draft criterion to the extent it establishes a preference for the use of a fish tissue
criterion over a water column criterion. Existing science strongly supports a fish-tissue approach, and
KCA appreciates that EPA is taking steps to update its criterion to reflect the scientific consensus on the
use of a fish tissue based standard. KCA is concerned, however, that EPA has inadvertently developed a
standard that invalidates this fish tissue preference for new or expanded discharges and in fishless
waters, two scenarios of particular importance to KCA's membership.

The Water Column Values Derived by EPA are Overly Conservative and Inappropriate.

The purpose of the selenium criterion is the protection of aquatic life. In the context of selenium, the
most scientifically appropriate method of determining whether aquatic life is impacted by any particular
discharge is fish tissue sampling because numerous factors beyond the mere concentration of selenium
in the water column influence selenium's impact on aquatic life in each stream. Without measuring and
studying, redox potential, adsorption of iron and manganese, and the oxidation state of selenate or
selenite in each and every water column, it is impossible to determine whether or not selenium will
have unacceptable impacts on aquatic life in each and every water column. Simply put, different
amounts of selenium have different effects on aquatic life in different waters. As such, a uniform
nationwide water column value for selenium is not useful to the ultimate determination of whether the
use of a given water will be protected, and therefore, adoption of a uniform numeric water column
standard runs contrary to the intent of the rule. By contrast, a fish tissue standard provides a clear and
concise standard resulting in a value with excellent repeatability nationwide.

The draft criterion includes qualifiers in footnote 3 of Table 1 that are not based on appropriate
scientifically derived numeric values, and conditions for implementation that are not appropriate in a
water quality standard. As currently drafted, the criterion would present significant challenges to states
that would adopt and implement the criterion.

EPA-HQ-OW-2004-0019-0385-A2; Kentucky Division of Water (KDOW); Posted 10/13/2015
Deference to Tissue Based Criterion.

The toxic effects of selenium in fish are not a response to water column concentrations, but result from
accumulation of selenium in fish tissue from dietary uptake (USEPA 1998). The base of the food web
(plants, bacteria and invertebrates) is relatively insensitive to selenium; however, fish are exposed to
potentially chronically toxic concentrations through dietary uptake (Chapman et al. 2010). KDOW
believes a water-column threshold which prompts a collection of fish tissue is an appropriate and

-------
protective action with regard to aquatic life. However, KDOW does not believe it is necessary to use a
water column number independent of available fish tissue data unless fish tissue data are unavailable.

KDOW conditionally supports EPA's approach to developing the chronic criterion for selenium with
respect to establishing a preference to the use of a fish tissue criterion over a water column criterion.
The science is clearly leading in this direction. However, KDOW is concerned that EPA has not followed
its own guidance (Stephan, et al. 1985) in the development of the criterion, because EPA has relied on at
least one study (Linville, 2006) that does not meet the appropriate standard of rigor. KDOW is also
concerned that EPA's methodology for calculating the water column criterion from fish tissue is subject
to significant uncertainties and is unprecedented. In addition, the criterion includes qualifiers in
footnote 3 that are not based on scientifically derived numeric values, include conditions for
implementation that are not appropriate to a water quality standard, and present significant challenges
to states that would adopt and implement the criterion.

EPA proposes egg/ovary data take precedence over whole-body, muscle or water column data when
available. KDOW believes this is an appropriate application of the criterion since selenium toxicity is
particularly manifested in fish reproduction and embryo development. However, the reliance on whole-
body or muscle tissue residue concentration is equally sound and provides a protective, real-time
implementation of a reliable tissue-based alternative when egg/ovary tissue data are unavailable. This
approach parallels that of Kentucky's in recognizing that tissue bioaccumulation is an indicator of
potential toxicity in the aquatic environment and provides the most reliable medium for monitoring
selenium and protecting aquatic habitat.

1. Adopt the statement that fish tissue data should take precedence over water column data in
assessing selenium levels (July 27, 2015 80 FR No. 143 at 44352).

2. Delete the following phrase from footnotes I and 2 (July 27, 2015 80 FR No. 143 at 44352):

"except in certain situations. See footnote 3."

3. Delete the following phrase from footnote 3 (July 27, 2015 80 FR No. 143 at 44352): "Water
column values have primacy over fish tissue values under two circumstances: I) "Fishless
Waters" (waters where fish have been extirpated, or where physical habitat or flow regime
cannot sustain fish); and 2) New or increased inputs of selenium until equilibrium is reached."

EPA-HQ-OW-2004-0019-0415-A1; Idaho Department of Environmental Quality (DEQ); Posted
10/28/2015

DEQ appreciates the updates and revisions in the most recent draft version of the criteria and urges EPA
to consider expounding on the document in consideration of DEQ comments. Attached please find our
general and specific comments for EPA's consideration in finalizing this document and promulgating a
new national recommendation for selenium criteria to protect aquatic life.

The document remains impressive in its depth and complexity. We commend EPA in their efforts to pull
together the extensive scientific information regarding selenium toxicity to aquatic life in the updated
draft selenium criteria.

-------
Specific Questions & Comments

Footnote 3 in the table on page XIV summarizing EPA's 4 part criterion says "Water column values are
based on dissolved total selenium in water" [Emphasis added], while the final statement on page 3 says
"These water quality criterion elements apply to the total of all oxidation states (selenite, selenate,
organic selenium, and any other forms)." [Emphasis added]. The latter is vaguer, possibly read to be
inclusive of particulate matter. Please resolve this apparent difference, be absolutely clear and
consistent on this matter.

EPA-HQ-OW-2004-0019-0412-A2; Utility Water Act Group (UWAG); Posted 10/28/2015

The agency has proposed a criterion that has four elements, two fish tissue-based and two water
column-based. EPA recommends that fish tissue elements be given precedence over the water column
elements when both types of data are available, based on an understanding of the latest science that
indicates fish tissue selenium concentrations are a more direct measure of potential selenium toxicity to
aquatic life than water column concentrations. The Draft Criterion contains a recommendation that
states and tribes adopt a similar approach into their water quality standards.

Many UWAG members own and operate coal-fired electric generating facilities. As selenium is a natural
trace element in coal, the management of selenium streams originating from coal combustion residues
is necessary for the operation of these facilities. Therefore, any proposed changes to the national
recommended water quality criterion intended to influence regulatory requirements associated with
selenium is of interest to UWAG. While the final revised aquatic life criterion for selenium should be
adequately protective, it should not be overly conservative such that unnecessary regulatory burdens
are imposed on economic activities important to the states and the nation.

UWAG commends the agency for its long-term commitment to ensuring the revised criterion is based on
a strong, updated scientific foundation. We support EPA's conclusion that, based on the latest scientific
information, "selenium toxicity to aquatic life is primarily driven by organisms consuming selenium-
contaminated food rather than by being directly exposed to selenium dissolved in water." 80 Fed. Reg.
at 44,351. UWAG also supports the agency's effort to provide detailed site-specific criteria
methodologies. We have serious concerns, however, with the science underlying the draft chronic water
column values and believe these issues must be addressed before EPA releases any final criterion.

UWAG appreciates the opportunity to submit comments and requests the agency to consider these
comments based on relevance and merit.

Alternatively, UWAG believes EPA's existing nationally-recommended chronic aquatic life criterion (5
Hg/L) is protective of all waterbody types. The implementation of this criterion in National Pollutant
Discharge Elimination System (NPDES) permits has resulted in the restoration of selenium-sensitive
species at many locations. EPA has provided no evidence to suggest that the current water column
criterion is not protective for waters where the fish tissue criteria are not applicable (e.g., no fish tissue
data are available).

-------
UWAG appreciates the opportunity to provide these comments. We believe that the agency has
correctly prioritized the proposed fish tissue criterion over the water column criterion.

UWAG's primary concern with the translation methods is that EPA does not adequately address the
inherent variability in the underlying data used to back calculate a water column concentration despite
acknowledging it. The methods result in a single concentration derived using average or median values
for factors that can be highly variable. There are potentially serious consequences if the derived water-
column concentration has considerable uncertainty, which is likely if the inherent variability of the
factors from which it is derived is not considered. Applying that single concentration to a waterbody
without taking into account that variability could lead to erroneous impairment or non-compliance
determinations or overly stringent NPDES limits resulting in significant costs to both states and
dischargers with little to no discernable environmental benefit. It also could result in criterion that is
under protective of aquatic life. To address sites where one or more of the factors is highly variable, EPA
should revise the Draft Criterion to allow the use of a range of site-specific water-column criteria for that
particular site. The bottom line is that any translated water-column concentration must reflect the
minimization (or elimination) of all possible sources of error.

Additionally, while information on the advantages and disadvantages of the two translation methods is
provided, EPA does not provide any information comparing the variability of the results from the two
methods. A side-by-side comparison of the variability of water-column concentrations derived with the
two methods for the same waterbody would be useful to stakeholders in determining the translation
method to use for a specific waterbody. UWAG requests that EPA perform the analysis and include the
results, including relevant statistical parameters (e.g., range, average, coefficient of variation), in the
final criterion.

EPA-HQ-OW-2004-0019-0411-A2; American Petroleum Institute (API); Posted 10/28/2015

First and foremost, we strongly support EPA's decision to develop Se criteria that are both
toxicologically- and ecologically-relevant.

The document is clearly laid out, the derivation method for fish tissue-based criteria follows standard
protocols for water quality criteria development, and thorough descriptions of the steps used in the
development of the criteria are provided. We are in support of the following core components of the
2015 draft criteria document (see Sections 1.1.1 through 1.1.4 below):

¦ Tissue-based (vs. water concentration-based) criteria;

¦ Use of ECioS (i.e., 10% effect concentrations);

¦ Timing of tissue data collection; and,

¦ Exclusion of juvenile survival data.

1.1.1 Tissue-based Criteria

We strongly support the approach of employing a fish tissue-based Se chronic criterion approach (i.e.,
egg/ovary and whole-body) in the final document as the primary criterion. Fish tissue-based Se criteria

-------
are consistent with the latest scientific information regarding the toxicology of Se to aquatic life. Tissue-
based criteria are the most ecologically-relevant for Se, as they are based on the chronic toxicity
pathway, which includes the bioaccumulation of Se through dietary exposure and incorporates such
variables as chemical reaction rates and exchange rates between sediment, water, and organisms (Brix
and DeForest 2008, Chapman et al. 2010). Further, the critical exposure route for fish is dietary organic
selenium, which is the basis for all egg and ovary studies linked to toxicity in offspring (Janz et al. 2010).

EPA will need to consider very carefully how to implement tissue-based selenium criteria in determining
water quality impairments and in setting technically sound water-quality-based permit limits. Permit
limits must account for site-specific conditions. Quantitative links between effluent
concentrations/loadings and fish tissue concentrations must account for bioavailability and chemical
speciation, and not be speculative but supported by the preponderance of measured, site-specific data.
We recommend EPA prepare and release for public comment a comprehensive selenium criteria
implementation guide.

EPA-HQ-OW-2004-0019-0391-A1; North American Metals Council (NAMC); Posted
10/14/2015

FISH TISSUE CRITERIA areas of Agreement: NAMC agrees that the focus on chronic effects, as opposed to
acute effects, is appropriate, technically defensible, and that it is supported by extensive scientific
evidence. Similarly, we agree fully with EPA that reproductive effects, linked to the magnitude of fish
egg/ovary selenium concentrations, are of greater ecological concern and provide a more reliable basis
for the criterion than non-reproductive endpoints (e.g., survivorship, growth).

The tissue criteria concentrations are not technically defensible and are demonstrably overly
conservative. These inappropriately low tissue criteria concentrations and those for lentic waters will
result in a very high incidence of false alarms and thus a serious misallocation of resources, thereby
reducing rather than enhancing the nation's ability to address environmental problems. They will also
unduly penalize industry with consequent negative economic ramifications, without an increased level
of environmental protection.

Finally, we support fully the option for the development of site-specific criteria based on appropriate
scientific studies. Cases of naturally elevated selenium concentrations in water have been documented,
as have elevated selenium concentrations in fish above the proposed criteria, without adverse effects
(Chapman et al., 2010; 2015 EPA Selenium Draft). In these cases, ambient-based tissue standards may
be appropriate.

EPA-HQ-0W-2004-0019-0401-A1; Colorado Wastewater Utility Council; Posted 10/15/2015

As stated in our prior review of the 2014 draft document, we commend EPA's effort in developing Se
criteria that are consistent with the latest science regarding Se toxicity. We support the use of a fish-
tissue based chronic criterion as the overriding criterion, as this is the most ecologically relevant
measure of toxicity for Se. We also agree with the use of ECi0s rather than some other endpoint, as this

-------
is the more conservative approach. We also support EPA's conclusions regarding the timing of tissue
data collection that samples collected within one year of each other are reasonable for acceptability.

The EPA approach in the 2015 draft Se criteria document is more in line with standard water quality
criteria development methodology (Stephan et al. 1985) than previous attempts at revising and
updating the Se criteria, and includes a critical evaluation of 16 studies on various fish species and
results in Se tissue thresholds for twelve fish species in ten genera. Criteria calculations follow
recommendations by Stephan et al. (1985) and use the 5th percentile calculation accounting for the
relative sensitivities of all species in the data set. This approach results in more scientifically defensible
criteria than the previous draft tissue criterion based on a single study.

As we stated in our review of the 2014 draft document, this document and the tissue-based criteria
approach is a significant improvement over the 2004 draft criteria document (EPA 2004). We strongly
support EPA's decision to develop Se criteria that are toxicologically and ecologically relevant.

The document is clearly laid out, follows standard protocols for water quality criteria development, and
includes thorough descriptions of the steps used in development of the criteria. We are in support of the
following core components of the 2015 draft criteria document (see Sections 2.1 through 2.4 below):

• Tissue-based standards,

• Use of EClOs, and

• Timing of tissue data collection

• Reliance on maternal-transfer related data over juvenile survival data

In addition, in Sections 3 and 4 of this review, we have provided several recommended revisions and
considerations that we believe will make the 2015 draft criteria document even more scientifically
sound.

We do not agree with the approach used to develop national water-column criteria. Our review provides
discussion on why single nationwide standards are not appropriate and why site-specific water-column
standards, calculated using EPA's Equation 18 from the criteria document, are a more scientifically
justifiable approach. In addition, we provide an example of how limited the data set used by EPA to
develop the national water-column criteria is, and how inclusion of additional data can make a
substantial difference in the calculated water column criteria. Lastly, we would strongly recommend EPA
consider changing the paradigm for National Pollutant Discharge Elimination System permitting such
that direct use of tissue criteria would be possible. To our understanding, there is nothing in the Clean
Water Act that would preclude this. By writing permits to the most defensible endpoint of tissues, EPA
would eliminate the need for backing into water concentrations by methods that invariably add
significant error and uncertainty.

1.1 Tissue-based Standards

-------
the chronic toxicity pathway which includes bioaccumulation of Se through dietary exposure and
incorporates such variables as chemical reaction rates and exchange rates between sediment, water,
and organism (Brix and DeForest 2008, Chapman et al. 2009). In addition, we strongly urge EPA to
consider how such tissue-based criteria can be directly incorporated into National Pollutant Discharge
Elimination System (NPDES) permitting, as discussed further later in this review.

Based on the results of our analysis, we recommend updating the proposed egg/ovary, whole-body, and
muscle chronic criteria to include our suggested changes (Table 8). In addition, because nationwide
water column-based criteria should not be derived, we advise EPA to only recommend site-specific
water column-based criteria on a site-by-site basis (Table 9).

EPA-HQ-0W-2004-0019-0400-A1; The National Mining Association (NMA); Posted 10/15/2015

As stated in our prior review of the 2014 draft document, we commend EPA's effort in developing Se
criteria that are generally consistent with the latest science regarding Se toxicity. We support the use of
a fish-tissue based chronic criterion as the overriding criterion, as this is the most ecologically relevant
measure of toxicity for Se. We also agree with the use of ECi0s rather than some other endpoint, as this
is the more conservative approach. We also support EPA's conclusions regarding the timing of tissue
data collection that samples collected within one year of each other are reasonable for acceptability.

• Tissue-based standards,

• Use of EClOs, and

• Timing of tissue data collection

• Exclusion of juvenile survival data

-------
Our comments on the 2015 EPA Selenium Draft focus primarily on the fish tissue criteria derivations.
While we agree that criteria derived from fish will protect the aquatic community, the criteria must be
technically defensible and not overly conservative. Neither is now the case. The present proposed
criteria are based on technically flawed data and incorrect assumptions. We provide evidence for higher
but still environmentally protective concentrations (i.e., at least 20 mg/kg egg/ovary selenium), which
was also noted in our comments to the 2014 EPA draft. We are concerned that the present overly
conservative fish tissue criteria, in addition to the overly conservative lentic water criterion and other
inappropriate, unrealistic components of the 2015 document, will result in a high rate of false alarms
that will further result in a serious misallocation of resources, thereby reducing rather than enhancing
the nation's ability to address environmental problems and unnecessarily penalizing industrial activities
with concomitant economic costs but without concomitant environmental benefits.

We do not agree with the approach used to develop national water-column criteria. Our review provides
discussion on why single nationwide standards are not appropriate and why site-specific water-column
standards, calculated using EPA's Equation 18 from the criteria document, are a more scientifically
justifiable approach. In addition, we provide an example of how limited the data set is, and how
inclusion of additional data can make a substantial difference in the calculated water column criteria.
Lastly, we would strongly recommend EPA consider changing the paradigm for National Pollution
Discharge Elimination System (NPDES) permitting such that direct use of tissue criteria would be
possible. To our understanding, there is nothing in the Clean Water Act that would preclude this. By
writing permits to the most defensible endpoint of tissues, EPA would eliminate the need for backing
into water concentrations by methods that invariably add significant error and uncertainty.

1.1 Tissue-based Standards

As stated in our previous review, we strongly support the approach of a fish tissue-based Se chronic
criterion approach (egg/ovary and whole-body) in the final document as the primary criterion. Fish
tissue-based Se criteria are consistent with the latest scientific information regarding the toxicology of
Se to aquatic life. Tissue-based criteria are the most ecologically relevant for Se, as they are based on
the chronic toxicity pathway which includes bioaccumulation of Se through dietary exposure and
incorporates such variables as chemical reaction rates and exchange rates between sediment, water,
and organism (Brix and DeForest 2008, Chapman et al. 2009). In addition, we strongly urge EPA to
consider how such tissue-based criteria can be directly incorporated into NPDES permitting, as discussed
further later in this review.

EPA-HQ-OW-2004-0019-0388-A2; J. R. Simplot Company; Posted 10/14/2015

The J.R. Simplot Company (Simplot) has the following comments on the 2015 Draft Criterion for
consideration to aid in developing a protective criterion that is not unduly conservative.

The 2015 Draft Criterion, overall, is based on a scientifically defensible approach and rationale. Simplot
supports the choice of a fish egg/over tissue value as the criterion. There are though, several aspects of
the document where EPA chose to be overly conservative. In the comments that follow, Simplot
identifies several issues where conservatism unnecessarily increases uncertainty in the approach taken

-------
and the resulting criterion value derived. In particular, our comments focus on the brown trout study
(Formation Environmental 2011). The unnecessary conservatism is also evident in the selection of the
final values derived to makeup the various elements for the National Criterion, particularly the
derivation of the final lotic and lentic values for water. Simplot also notes that EPA failed to address
issues identified in the review of the 2014 Draft Criterion in respect to the field data associated with the
brown trout data.

A3. Derivation of Water Column Values and Uncertainty in Enrichment Factors

[Pages 90-92, Section 4]

This four part criterion advocates the primacy of the egg/ovary element over other elements of the
criterion except under the conditions listed in Table 4.1, footnote 3 (page 91). But the statement, " ...
and inclusion of the water column elements into the selenium criterion ensures protection when neither
fish egg-ovary nor fish whole-body or muscle tissue measurements are available, and provides
consistent coverage for all waters" appears to provide for an optional use of a water criterion when fish
tissue data are not available instead of generating the necessary data. In other words, States may simply
default to the water criterion if no fish tissue data were available. In fact this appears to be amplified by
the following, "Where states use the selenium water column concentration criterion element values
only (as opposed to using both the water column and fish tissue elements) for conducting reasonable
potential (RP) determinations and establishing water quality-based effluent limitations (WQBELS) per 40
CFR 122.44(d), existing implementation procedures used for other acute and chronic aquatic life
protection criteria would be appropriate."

The language above imp lies that only the water criterion element maybe used as opposed to using the
fish tissue element for deriving WQBELs for NPDES permits which is likely to result in unnecessarily low
permit limits.

Allowing a water concentration to take precedence in any circumstance, where it is not specifically
derived from bioaccumulation information available for a state or region, is not consistent with the
recommendations in Appendix K, nor the scientific information available indicating that numerous
factors that can affect selenium bioaccumulation.

Simplot understands EPA's desire to ease the burden for State's and regulated entities to assess
compliance relative to a tissue standard by deriving an aqueous criterion equivalent. However, selection
of a single value for a selenium water criterion on a national scale when the prevailing information
indicates site or regional specificity in the bioaccumulation of selenium simply is not consistent with the
science nor adequately representative of all sites. Multiple factors affect development of a water value
from a tissue threshold that creates uncertainty. That uncertainty is multiplied by developing and
applying a single water criterion across the entire US with the only distinction being lotic or lentic
habitats. This compounding of uncertainty stems not only from species specific bioaccumulation and
effect for fish, but also from site specificity in the uptake of selenium into the base layer of the food
chain (i.e., enrichment factors) and the high level of uncertainty associated with enrichment factors
(EFs). Reviewers of the 2014 Draft Criterion (i.e., Deforest and Cutter, ERG 21 04) have pointed out that
the variability in derivation of the EF (i.e., algae, algae and sediment, sediment alone as the particulate
fraction) and subsequent use in the Presser and Luoma (2006) model (i.e., kinetic/rate approach, and

33

-------
not an equilibrium thermodynamic) which pose additional uncertainties that are not accounted for in
the model.

Furthermore, because the difference in EFs is wide ranging, based on the type of system being evaluated
(low selenium background or reference sites vs high selenium impact sites); the selection of streams that
are used to derive a single composite national criterion for water is critical if not impossible to be
representative. There is typically an inverse relationship between the exposure concentration and the EF
such that low aqueous selenium will result in relatively high EFs and high aqueous selenium will result in
low EFs.

Much like the existing revised national copper criterion has been developed to take into account site
specific modifying factors that affect toxicity using the Biotic Ligand Model, the selenium criterion should
similarly also provide a similar path. That framework is already present in the existing document.

Because EPA has derived a national aqueous criterion value, many states may simply opt to use that
value rather than develop the necessary information to provide a state or regional specific criterion for
selenium. As noted in Simplot's prior comments on the 2014 Draft Criterion, a reasonable approach is
that the aqueous value should ideally serve as a "trigger or screening value" to set in motion the
potential need for additional monitoring and data collection rather than it being a set criterion value.
For fishless waters, monitoring would entail collecting the necessary information to conduct trophic
modeling and with the assumption that the species to be protected is the fish species downstream of
the fish less water as described in Appendix K of the 2015 Draft Criterion.

EPA-HQ-OW-2004-0019-0374; Anonymous public comment; Posted 09/23/2015

The 2015 draft Se fish tissue duration and frequency are incompatible with EPA's 2000 fish tissue
monitoring guidance, which specifically names selenium as one of the pollutants it applies to:

Guidance for Assessing Chemical Contaminant Data for Use in Fish Advisories Volume 1 Fish Sampling
and Analysis Third Edition.

The fish tissue monitoring guidance offers far greater scientific rigor than EPA presented in its 2015 draft
fish tissue criterion duration and frequency discussion. For the sake of consistency of state approaches
for assessing fish tissue, please either incorporate the concepts of that guidance, or explain why you
believe that the protection of human individuals allows a far more lenient approach than the protection
of aquatic populations.

The relevant part of the fish tissue monitoring guidance, cited by the mercury implementation guidance,
is presented below:

6.1.2.7.2 Comparison of Target Analyte Concentrations with Screening Values for Issuing Fish
Advisories - Using the statistical model described in Section 6.1.2.7.1, target analyte concentrations
from replicate composite samples at a particular site can be compared to screening values using a t-test.
Assume that zi is the contaminant concentration of the ith replicate composite sample at the site of
interest where i=l,2,3,...,n and, furthermore, that each replicate composite sample comprises m
individual fish fillets of equal mass. To test the null hypothesis that the mean target analyte

34

-------
concentration across the n replicate composite samples is equal to the SV versus the alternative
hypothesis that the mean target analyte concentration is greater than the SV, perform the following
steps:

1. Calculate ~z, the mean target analyte concentration of observed replicate composite samples at a
site: ~z = [Summation] zi / n

where the summation occurs over the n composite samples.

Calculate the estimate of the Var(~z), sA2 :

sA2 = [[Summation](zi - ~zA2] / [n(n - 1)]

where the summation occurs over the n composite samples.

Calculate the test statistic: tc = (~z - SV) / s

The null hypothesis of no difference is rejected in favor of the alternative hypothesis of
exceedance if tc > t alpha, n-1

where t alpha, n-1 is the tabulated value of the Student-t distribution corresponding to level of
significance and n-1 degrees of freedom. Note that the inequality is in one direction (>) since it is
exceedance of the SV that is of interest. When several sites are sampled and/or fish of different
size ranges are collected, it is important to conduct the test separately at each site and for each
size range. Combining sites or size ranges introduces variance components that are not
accounted for in this procedure. The variance estimate may be larger with the additional
sources of variability, and more replicate samples may be needed to detect a significant overall
exceedance of the SV.

EPA-HQ-OW-2004-0019-0371-A2; Wyoming Department of Environmental Quality (WDEQ);
Posted 09/15/2015

Sample Representativeness and Requirements

WDEQ/WQD requests that EPA include a discussion on the variability in fish tissue concentrations,
including the intra-fish, intra-species and inter-species variability within the revised criteria document.
WDEQ/WQD is concerned that fish tissue variability may cause fish samples to fluctuate between
attainment and non-attainment of standards for each successive assessment cycle. Further, the criteria
document does not include recommendations on a sufficient sample size that can account for this fish
tissue variability. Accordingly, WDEQ/WQD requests EPA to include recommendations on sample sizes
for the various fish tissue elements to minimize mischaracterization of a waterbody.

Water Column Data

EPA's draft selenium criteria includes four elements: a fish egg and/or ovary concentration, a fish whole-
body or muscle concentration, a 30-day average concentration for lotic and lentic systems, and an
intermittent exposure water column concentration. The draft criteria document recommends that
states adopt all four elements of the draft criterion into water quality standards as a single criterion

35

-------
composed of multiple parts. Under "steady state" conditions (i.e., food-web equilibrium), EPA explicitly
affirms the primacy of the whole-body or muscle elements over the water column element, and the egg-
ovary element over any other element. The draft criteria document also states that freshwater aquatic
life would be protected from toxic effects of selenium by applying all four elements of the criteria.

Though the draft criteria provides adequate protection from selenium toxicity through all four elements,
WDEO/WQD is concerned that only water column measurements will be available for the majority of
Wyoming waters. The use of water column elements may not accurately reflect concentrations that are
protective of aquatic life, especially since toxicity to aquatic life is primarily driven by dietary exposure
(i.e., consumption of selenium-contaminated prey) rather than direct exposure to dissolved selenium in
water. Further, water column elements are derived from fish tissue concentrations by way of food-web
modeling, hence water column elements are the least accurate of the criteria elements. WDEQ/WQD
requests that the criteria document outline that states have flexibility to not identify waters as impaired
if data show an exceedance of the water column value and no fish tissue are available. The criteria
document could recommend that states conduct tissue monitoring in these instances to confirm
impairment. These provisions may prevent states from placing waters on the 303(d) List when aquatic
uses are actually being supported.

EPA-HQ-OW-2004-0019-0376-A2 [Comment 0381-A3 is a duplicate of 0376-A2]; Texas
Commission on Environmental Quality (TCEQ); Posted 09/24/2015

Comments on Proposed Draft Criterion

EPA's statement that the criterion is comprised of four elements is misleading. The draft criterion
document states "the draft criterion has four elements" (Table 1), consisting of two fish tissue-based
and two water column-based elements." However, EPA is requesting states and authorized tribes to
protect for a total of six different numeric aspects, as identified above. EPA should clearly state the
various components of the criterion and provide sufficient guidance to incorporate these components
into water quality management programs.

EPA-HQ-OW-2004-0019-0413-A2; Federal Water Quality Coalition (FWQC); Posted
10/28/2015

We commend the Agency for revisiting the selenium criteria, and for developing new criteria that better
reflect the state of the science than do the current criteria. In particular, these are aspects of the Criteria
Document that we support:

• The Draft Criteria Document appropriately focuses on chronic effects, and does not recommend
an acute criterion.

• In assessing chronic issues, the Draft Criteria Document appropriately focuses on reproductive
effects, which provide a more reliable basis for the criteria than non-reproductive effects such as
survivorship and growth endpoints.

-------
• Studies concerning reproductive effects are used appropriately. Studies on juvenile survival,
including overwinter survival, are not used directly.

• The Draft Criteria Document clearly states as a general matter (with exceptions discussed below)
that fish tissue data should take primacy over water column data in assessing whether the criteria
are met.

• The Draft Criteria Document allows for development of site-specific criteria where that is
appropriate for particular situations.

While we support those aspects of the Draft Criteria Document, there are a number of specific issues on
which we have concerns about the scientific basis for the recommended criteria. Also, there are several
issues on which we think that the EPA approach needs to be clarified. These issues are discussed in
detail below.

Converting Data to Fish Tissue Levels: As noted above, the FWQC supports the use of fish tissue levels in
whole body, muscle, or egg-ovary concentrations as the primary form for selenium criteria to be
expressed. However, prior to establishment of final fish-tissue and organ concentrations, several issues
need to be addressed with respect to uncertainties involved in converting data to fish tissue levels in
whole body, muscle or organ. Methods to address uncertainty, such as log/log plot regression, linear
plot regression, and central tendency of the ratios need to be considered. Further, sound scientific
methods exist to control for natural variability through sampling design, ecoregion classification, and
regression model application.

II. PRIMACY OF FISH TISSUE DATA

As noted above, the FWQC supports EPA's structuring of the new criterion so that fish tissue data has
"primacy" over water column data. The peer reviewers generally supported that approach as well. In
fact, some of the reviewers recommended that the water column levels be considered only as
"triggers," instead of as part of the criterion. In those comments, the reviewers indicated that the
Agency should clarify exactly what "primacy" means - in other words, how fish tissue and water column
data should be considered in determining whether the criterion has been attained. However, despite
these recommendations, the Agency has not provided that clarity, making several statements about the
primacy issue that are not consistent with each other. EPA should state clearly that the fish tissue
elements of the criterion do have primacy - in other words, if the fish tissue data show that the fish
tissue level in the criterion is being met, then the waterbody is meeting standards. The Agency should
also consider making the water column levels into "triggers" rather than elements of the criterion.

Here are some examples of what the peer reviewers had to say about primacy:

"Although I agree that the primacy of each criterion element is logical, it is not clearly stated whether a
water Se criterion could be adopted into a permit limit. For example¦, if compliance with the lotic or lentic
Se criterion is demonstrated, is measurement offish tissue Se concentrations necessary? If a water body
meets a fish tissue-based Se criterion, but not a surface water criterion, would the water body be
considered in compliance? I believe the answer to the latter is "yes", but this does not seem to be clearly
stated in the draft AWQC document." (p. 18)

-------
"The use of water column concentrations of selenium as environmental assessment tools or as triggers
for additional assessment is fraught with uncertainty from several sources, which are discussed in
subsequent sections of this review." (p. 25)

"However, to use water column [Se] as a criterion in of itself in the absence of tissue [Se] data is a recipe
for inappropriate conclusions, which may penalize industry (i.e., false positives) or cause harm to certain
fish populations (i.e., false negatives). I strongly believe that water column [Se] should be used more as a
"trigger" to initiate further monitoring that includes collection offish for tissue [Se] determinations." (pp.
30-31)

"However, based on comments provided above (in 1 and especially 2a(iii)), relying on water column
dissolved [Se] has a high likelihood of generating both false positive and false negative results with
respect to regulatory action. I think the proposed water column criteria (a) should be used as triggers to
initiate further monitoring offish tissue [Se], (b) should be made more conservative (reduced) by
application of a safety factor to avoid false negatives, and (c) that the simple classification of a water
body as lentic or lotic should be modified to include more quantitative measures of flow such as water
residence time and/or mean annual water velocity. Given that many impounded riverine systems in the
USA are essentially lentic systems for much of their river-miles, perhaps a water column trigger [Se] could
be set at 1 ug/L (same as the current Canadian [CCME] water quality guideline for Se). If exceeded, this
trigger value would result in further action in terms offish collections for tissue [Se]." (pp. 97-98)

In responding to these concerns, EPA could have made a simple statement that if the fish tissue level in
the criterion is met, then the criterion has been attained, regardless of the water column level. At one
point (cited below), the Agency does say that tissue data "will always supersede" water column data.
However, at other points (also cited below), EPA indicates that if the fish tissue levels are being attained
but the water column levels are not, then more inquiry is called for. We believe that such an "inquiry" is
simply inconsistent with the concept that the fish tissue data have "primacy." To ensure that the fish
tissue data are treated as the primary element of the criterion, we believe that EPA should consider
treating the water column levels as "triggers" for further assessment, rather than as components of the
criterion.

Here are some of the EPA statements on "primacy" that give rise to confusion:

"Regarding the purpose of the water criterion elements, one of the purposes of the water-based
criterion element(s) was to facilitate having an easily implemented value in permits. If a
waterbody meets the fish tissue limits yet not a surface water criterion, for an existing input
source, it is likely that the site specific dynamics of selenium bioaccumulation may be different
than the basis of the national criterion, and this waterbody may be a candidate for a site-specific
water value." (p. 15)

"The hierarchy of the criteria is such that, for an existing input, a tissue sample will always
supersede the result of the water element, because the tissue sample(s) are most closely
associated with the endpoint of concern: reproductive impacts on aquatic species. In the
situation the peer reviewer presented, tissue sample(s) could be collected to confirm or refute
the result of the water sample(s)." (p. 21)

38

-------
"Regarding uncertainty associated with the water column criterion elements, EPA recognizes
this uncertainty, but notes that the water column element is useful for implementing in permits,
and in certain situations water will be the preferred sample in the case of water bodies with new
discharges of selenium or where adequate fish samples are not available. In situations where
both fish tissue and water samples are available, the water column values are backstopped by
the fish tissue criteria elements in the hierarchal construction of the criterion and thus provide a
means for reducing the uncertainty." (p. 25)

"Regarding the tiered approach, the water column criteria is backstopped with the hierarchal
construction of the tissue criteria; exceedance of a water column value, can be confirmed via
sampling offish tissue." (p. 30)

"The water column criterion element values were derived using the egg-ovary criterion element
as the basis of the calculation. EPA developed water column criterion values because routine
water concentration monitoring of selenium is more practically feasible than routine tissue
monitoring, and used a hierarchal (tiered) approach to insure that water column values
indicating a selenium problem could be vetted using a fish tissue sample, because the fish tissue
selenium concentrations are indicative of potential effects. Water column values are necessary
because they may be the only media data available - i.e. streams where fish are absent, or
temporally necessary due to a new discharge where selenium has not accumulated in the
system, and so there is uncertainty in the relationship between the discharge and the potential
for bioaccumulation in fish at the site." (pp. 97-98)

Each of those EPA statements gives a different rationale for including water column levels in the
criterion, and each expresses the relationship between the fish tissue levels and the water column levels
in a different way - from fish tissue "superseding" water column, to fish tissue "backstopping" water
column. The Agency needs to provide a clear, consistent explanation of how the criterion will work

EPA-HQ-0W-2004-0019-0389-A1; Thunder Basin Coal Company, LLC (TBCC); Posted 10/14/15

Thunder Basin Coal Company, LLC (TBCC) operates surface coal mines located in the Wyoming Powder
River Basin and holds valid permits to mine, which include provisions to protect water quality and
compliance with all applicable water quality regulations. We support National Mining Association's
(NMA) comments and Wyoming Mining Association's (WMA) comments as provided and emphasize the
following major points:

Incorporation of Background Conditions into Site-specific Criteria: It is inappropriate to specify national
water column-based standards for selenium or any other criterion. The great variety of geologic
environments, climates, and landscapes lead to tremendous variability in the aquatic life that is
potentially affected. National standards cannot address this level of variability. As EPA acknowledges in
the draft criteria document, weathering and leaching of seleniferous rocks and soils can lead to naturally
elevated concentrations of selenium in ground and surface waters. Many healthy streams, ponds, and
wetlands exist in Wyoming with waters that may periodically or permanently exceed the published
water column standards in a natural environment. Often, these waters provide the only source of water
for wildlife and waterfowl in the local environment. EPA should develop methods to use natural

39

-------
background Se concentrations to develop ambient based site-specific criteria (either tissue or water-
based) where elevated concentrations are present unrelated to human-induced sources. EPA's
discussion of site-specific standard development includes nothing on this issue and needs further
clarification. For these reasons, a standard, and especially a national standard, is unacceptable.

EPA needs to develop and explain clear and practical procedures and approaches for states to use in
developing achievable criteria for their respective states with consideration of the items mentioned
above. The general and vague descriptions presented thus far do not provide adequate guidance and in
fact, provide the EPA with ample opportunity to simply deny state efforts without explanation or
justification.

Fish Tissue Based Standards: We support the use of fish tissue standards as a method of determining
appropriate selenium standards. However, each state must take on this responsibility and develop state-
specific standards that apply to their regions. Because concentrations of selenium in the water column
are not always predictive of concentrations of selenium in fish tissue, directly testing fish tissue appears
to be the most practical approach at this time. Fish tissue limits should be included directly in NPDES
permits using water column concentrations as a trigger only where fish are present on the specific site.

EPA-HQ-OW-2004-0019-0402-A2; The Stakeholders Implementing TMDLs in the Calleguas
Creek Watershed; Posted 10/15/2015

1. A Universal, Nationally Applicable Water Column Number Is Inappropriate Due to the Site-Specific,
Bioaccumulative Nature of Selenium. As such, the 2015 Draft Selenium Criterion Should Only Be Based
on Fish Tissue Elements, with Water Column Concentrations Used as a Tool for Implementation of the
Criterion.

As stated in Comment #4, the Stakeholders strongly support the approach recommended by USEPA in
the 2015 Draft Selenium Criterion pertaining to the tissue-based elements as it provides for the direct
assessment and protection of beneficial uses. Notwithstanding this support, the inclusion of water
column elements within the 2015 Draft Selenium Criterion is inappropriate as these elements will be
either over- or under-protective of the aquatic life present in most water bodies. The type of aquatic
environment (e.g. lotic/lentic, marsh/riparian, etc.) and food webs present in a waterbody effect
selenium bioavailability and toxicity.10, n-12,13 In addition, the species of selenium, particulate selenium
concentrations, and the resultant biogeochemical transformations and accumulation in the food web
can differ substantially even at similar dissolved concentrations.

For instance, in the Calleguas Creek watershed, located in Ventura County in Southern California, co-
located fish tissue and water column selenium concentration data were collected from 2008 to 2013 in
Revolon Slough, a freshwater tributary to Calleguas Creek. In this lotic system, the water column
concentrations are significantly greater than the water column element of the 2015 Draft Selenium
Criterion (3.1 ng/L) for over 84% of results and by as much as a factor of eleven (Figure 1). Despite the
water column concentrations, which would indicate an impairment to aquatic life according to the 2015
Draft Selenium Criterion, the muscle tissue concentrations are almost all below (with two exceptions)
the muscle tissue element of the 2015 Draft Selenium Criterion (11.3 mg/kg). As such, if the water

-------
column element of the 2015 Draft Selenium Criterion were applied in this water body, unnecessary
management actions would need to be taken to meet a concentration in the water column that clearly
has no relation to the water column concentration necessary to protect aquatic life.

Further, in the Newport Bay watershed, a relatively small watershed located in Southern California,
initial model runs of the Luoma Presser model for two tributaries, Peters Canyon Wash and Big Canyon
Wash, demonstrate a wide range of water column concentrations needed to protect fish: approximately
16 - 27 ng/L in Peters Canyon Wash and 2-3 ng/L in Big Canyon Wash. The difference in the predicted
water column concentrations is due to the relatively high proportions of selenite and the high median Kd
values (i.e. EF) present in Big Canyon Wash. Even within this small watershed, the use of a universally
applicable water column concentration of 3.1 ng/L may be both significantly over-protective and under-
protective.

Table 3.13 and Figure 3.9 of the 2015 Draft Selenium Criterion also clearly demonstrate why the water
column elements will be either over- or under-protective of the aquatic life present in most water
bodies. The data in the table and figure show that for 53 various lentic and lotic systems, the calculated
range of protective water column concentrations varies by orders of magnitude (0.23 - 50.44 ng/L for
lentic systems and 1.19 - 40.60 ng/L for lotic systems). Given the wide range of values calculated to be
protective, it is clearly inappropriate to establish water column elements outside of a site-specific
setting. Further, the significant variability demonstrates that water column concentrations are an
unreliable measure for the protection for beneficial uses.

Given the site-specific nature of the water column concentrations that protect aquatic life, a more
appropriate alternative would be to utilize water column concentrations as an implementation tool for
the criterion, rather than as part of the criterion itself. For example, the criterion could require that
where tissue values are exceeded, the water column elements would be used for implementation
purposes to help determine the extent of BMPs necessary to attain a water column concentration that
would attain the fish-tissue values.

Precedent for utilizing very specific implementation tools as part of an objective has been established in
the State of California's Water Quality Control Plan for Enclosed Bays and Estuaries - Part 1 Sediment
Quality16 (Phase I SQOs), approved by USEPA on August 25, 2009. For example, if the template set by the
Phase I SQOs is followed, each of the 69 aquatic sites evaluated in Section 3.2.5 of the 2015 Draft
Selenium Criterion would use the model to calculate the water column concentration appropriate for
each individual site. As a result, all 69 aquatic sites would target a water column concentration that
would most appropriately result in protection of beneficial uses (e.g., neither under- nor over-
protective). This evaluation would guide implementation of the tissue-based criterion, rather than
establishing separate water column concentrations as part of the criterion itself. Under the current
proposed approach in the 2015 Draft Selenium Criterion, only two (2) of the 69 sites (one lentic and one
lotic) are assigned water column concentrations which are appropriate (neither over- nor under-
protective).

Given the lack of readily available treatment technologies for selenium, and the additional difficulties for
urban environments in Southern California (land availability, space requirements, etc.), there is
significant impact from establishing water column concentrations that are not directly linked to site-

41

-------
specific conditions. In addition to not being directly linked to beneficial use protection, utilizing water
column concentrations may unnecessarily require implementation of significant and costly BMPs with
no net environmental benefit. Based upon evaluations conducted over the last 10 years in Southern
California, these additional unnecessary costs can range in the hundreds of millions of dollars.

In addition, utilizing water column elements outside of a site-specific setting may cause the unintended
consequence of establishing effluent limits in National Pollutant Discharge Elimination System (NPDES)
permits which are inappropriately low, but may not be able to be raised to the appropriate value once
an SSO has been adopted due to anti-backsliding concerns (Section 402(o) of the Clean Water Act). As
such, dischargers may be put into a position where Minimum Mandatory Penalties are levied against the
discharger, even though the discharger is discharging at a concentration that would meet an effluent
limit which is fully protective of aquatic life. Utilizing the water column concentrations elements of the
criterion as an implementation tool would avoid this unnecessary outcome.

Requested Actions:

Remove the water column elements from the 2015 Draft Selenium Criterion as elements of the criterion
and instead utilize water column concentrations as an implementation tool (similar to the approach
utilized by the State of California in the Phase I SQOs).

States Should Clearly Be Allowed to Adopt SSOs that not Only Modify Each of the Four Elements of the
Criterion, But that Also Allows States to Opt to Eliminate Aspects of the Criterion (e.g., Water Column
Concentrations)

The Stakeholders strongly support the approach in the 2015 Draft Selenium Criterion that states "all
four elements of the selenium criterion can be modified to reflect site-specific conditions." This
approach explicitly provides States the opportunity to account for site-specific factors that are
fundamental to developing and adopting appropriate regulations for selenium. As noted in Comments
#4 and #5, site-specific factors can vary even within a small watershed, let alone when extrapolated to
the national scale.

However, to develop fully appropriate objectives at the State or local level, States also need the
flexibility to not only modify the elements, but also to eliminate certain elements of the 2015 Draft
Selenium Criterion. The need for this flexibility applies mostly to the water column elements, though
certain States in the arid West may also wish to focus on only whole-body fish tissue due to the practical
limitations of obtaining egg/ovary samples. As noted in Comment #5, the water column elements could
be moved from the criterion to an implementation tool. If USEPA determines that approach is
unworkable at the national level, States could implement such an approach if the criterion specifically
allows for the elimination of the water column elements via SSOs.

Page K-2 of Appendix K of the 2015 Draft Selenium Criterion cites the following rationale for
recommending the water column elements of the 2015 Draft Selenium Criterion:

"Although the selenium concentration in eggs or ovaries is the most sensitive and reliable basis for a
criterion, implementation can be challenging because most state and tribal Clean Water Act programs
require the expression of water quality criteria as an ambient concentration in the water-column. "

42

-------
However, not all States require the expression of water quality criteria as an ambient concentration in
the water-column. For example, the State of California has been developing SSOs that do not include
water column concentrations as part of the objective, but rather uses water column concentrations as
an implementation aspect of the objective to guide management actions and permit development.
Providing the flexibility to the States to opt to eliminate aspects of the criterion is critical and must be
maintained and explicitly provided for in the 2015 Draft Selenium Criterion.

Further, in Southern California, it is not practical to rely on the collection of eggs and/or ovaries because
fish eggs and ovaries may be insufficiently present or too unreliable to obtain. In this instance, utilizing
an SSO which utilizes whole-body fish and/or muscle tissue is significantly more practical. Therefore,
States should also retain the flexibility to focus on whole-body fish and/or muscle tissue and not be
required to include egg/ovary elements in SSOs.

Requested Actions:

Revise the language on Page 95, Section 5 of the 2015 Draft Selenium Criterion to read as follows:

"All four elements of the selenium criterion can be modified to reflect site-specific conditions where the
scientific evidence indicates that different values will be protective of aquatic life and provide for the
attainment of designated uses. Furthermore, the egg/ovary element and the water column elements
can be eliminated through site-specific objectives, consistent with Appendix K."

Revise Table 4.1 on Page 91 of the 2015 Draft Selenium Criterion to read as follows:

Add footnote 618 (place after "Fish Tissue" and "Water Column"): Each of the four elements can
be modified through site-specific objectives, consistent with Section 5 and Appendix K.

Add footnote 718 (place after "Egg/Ovary," "Monthly Average Exposure," and "Intermittent
Exposure"): Through site-specific objectives, consistent with Section 5 and Appendix K, these
elements can be eliminated as part of the criterion.

Revise Appendix K to provide guidance on the rationale, procedures, and pathways to eliminate certain
elements of the 2015 Draft Selenium Criterion through SSOs.

Draft Selenium Criterion, Page 100, Section 6. Note that prior comments request additional footnotes
to this table. The actual numbering depends upon the approach (es) selected by USEPA in subsequently
revised criterion. Numbering is provided for illustrative purposes to note that the requested footnotes
are additions to the proposed criterion.

-------
COMMENTS PREVIOUSLY MADE ON 2014 EXTERNAL PEER REVIEW DRAFT CRITERION AND NOT
ADDRESSED IN 2015 DRAFT SELENIUM CRITERION

The Stakeholders strongly request the incorporation of the following previously submitted comments on
the 2014 External Peer Review Draft Criterion. The comments provided below will support effective
implementation of the criteria under the wide range of conditions present in waterbodies in California.
Clarification Is Needed Regarding the Applicability of the Four Elements of the Criterion to Ensure that
Fish-Tissue Elements Supersede the Water Column Elements.

The Stakeholders strongly support the approach recommended by USEPA on Page 91 of the 2015 Draft
Selenium Criterion of establishing "the primacy of the whole-body or muscle elements over the water
column element, and the egg-ovary element over any other element." As selenium is primarily
accumulated in organisms through diet, water column concentration-based criteria are viewed by many
as inappropriate, especially for predicting chronic effects.5,6 Tissue-based approaches ensure a direct
assessment of the beneficial uses that are being protected. The benefit offish-tissue assessment is
reflected on Page 2, Section 1 of the 2015 Draft Selenium Criterion:

"... fish-tissue values more directly represent chronic adverse effects of selenium than the conventional
water concentration approach ... because chronic selenium toxicity is primarily based on the food-chain
bioaccumulation route, not a water route of exposure."

Selenium bioaccumulation and toxicity cannot be predicted based solely on selenium concentrations in
water.7 Ecological risks from selenium are governed by uptake that occurs at the base of the food web
(primarily via primary producers and microorganisms), dietary exposure and toxicity, the timing of
exposure (e.g., during gestation in fish and birds), and transfer through the food web. Selenium uptake
within a food web is both species- and environment-specific.8 Assimilation and retention of selenium in
organisms differs between species and environments at all levels of the food web, making it difficult to
predict concentrations and toxic exposure at different trophic levels. The poor linkage between
dissolved selenium and selenium exposure and toxicity in the food web makes it particularly difficult to
determine impairment in a watershed based on water concentrations alone.

It appears that the 2015 Draft Selenium Criterion intends to capture this critical aspect of selenium risk
and exposure by structuring the criterion such that the fish-tissue elements supersede the water-column
elements. However, this critical aspect of the criterion lacks clarity throughout the document. One
example is on Page 90 where the four-part criterion is identified, but language is absent to note that the
fish tissue elements supersede the water column elements. It appears as if footnotes which refer to the
table footnotes on the following page intend to establish the supremacy of the tissue-based elements,
but it is unclear if that is the intent of the 2015 Draft Selenium Criterion. The supremacy of the tissue-
based elements of the criterion is so fundamental that it needs to be abundantly clear and explicitly
included in all aspects of the criterion, including the Fact Sheet, Executive Summary, and in all instances
where elements of the criterion are discussed or described.

In addition, the language currently used in the 2015 Draft Selenium Criterion to describe the relationship
between fish tissue and water column elements is "primacy" or "precedence." This language leaves
room for interpretation and discretion whereby the water column elements may still be applied even

44

-------
where fish tissue data are available. As noted above and as supported by the 2015 Draft Selenium
Criterion, tissue-based approaches are superior to water column approaches for the protection of
beneficial uses. A more direct and clear term would be to state that fish tissue elements supersede
water column elements and/or attainment of fish-tissue elements is deemed equivalent to attaining any
water column elements. This approach is being employed in site-specific objectives currently under
development in certain areas in California.

Lastly, the current language provides for fish tissue elements to have "primacy" where tissue data are
available. Page 92 of the 2015 Draft Selenium Criterion states (emphasis added):

"Inclusion of the fish whole-body or fish muscle element into the selenium criterion ensures the
protection of aquatic life when fish egg or ovary tissue measurements are not available, and inclusion of
the water column elements into the selenium criterion ensures protections when nether fish egg-ovary
nor fish whole-body or muscle tissue measurement are available, and provides consistent coverage for
all waters ".

And, footnote 1 and 2, respectively, in Table 4.1 on Page 91 of the 2015 Draft Selenium Criterion state
(emphasis added):

"Overrides any whole-body, muscle, or water column elements when fish egg/ovary concentrations are
measured except in certain situations."

"Overrides any water column elements when both fish tissue and water concentrations are measured
except in certain situations."

However, further clarity is necessary regarding data availability and applicability of the four elements of
the criterion. For example, in many monitoring programs in California, water bodies are monitored more
frequently for water column than for tissue as tissue samples for both fish and birds are collected during
the nesting season (typically, late spring). This approach ensures that the most critical condition, the
breeding season, is captured in the monitoring data (as fish are also a dietary item for shorebirds, both
fish and birds are collected concurrently). Throughout the rest of the year, water column samples are
obtained and the data are used to gauge implementation actions. However, the language in the current
2015 Draft Selenium Criterion could imply that even in a monitoring program that is specifically
designed on tissue-based approaches, such as the one described here, the water column elements of
the criterion would apply throughout the year when tissue samples are not being collected concurrently.
If the 2015 Draft Selenium Criterion is interpreted in this way, water bodies could be identified as not
meeting the Criterion based upon water column concentrations, despite tissue data demonstrating that
the water body meets the Criterion. Therefore, clarifying and explanatory language needs to be included
to ensure that the water column elements are not inappropriately applied over tissue-based elements.
One potential approach is to limit the applicability of water concentration elements to instances where
tissue is not collected within the same calendar year.

Requested Actions:

Revise the 2015 Draft Selenium Criterion to clearly establish that the fish tissue elements supersede the
water column elements:

45

-------
Modify the terminology throughout the document from "primary" and "precedence" to
"supersedes"

Modify Table 4.19 as follows:

Modify Footnote 1: "Overrides any Supersedes all whole-body, muscle, or water column
elements when fish egg/ovary concentrations are measured except in certain situations. See
footnote 3."

Modify Footnote 2: '"Overrides any Supersedes all water column elements when both fish tissue
and water concentrations are measured except in certain situations. See footnote 3."

Add Footnote 6 (place after "Fish Tissue"): Fish Tissue elements supersede all water column
elements when tissue data are available within the same calendar year.

Add Footnote 7 (place after "Water Column"): Water column elements (both the Monthly
Average Exposure and Intermittent Exposure) only apply where fish tissue data are not available
within the same calendar year.

Provide more direct and explanatory language throughout the document that clearly limits the
applicability and use of water column elements, including but not limited to Page 90 of Section 4
(National Criterion for Selenium in Fresh Waters). Example modified language for Page 92, first
paragraph:

"Inclusion of the fish whole-body or fish muscle element into the selenium criterion ensures the
protection of aquatic life when fish egg or ovary tissue measurements are not available, and inclusion of
the water column elements into the selenium criterion ensures protections when neither fish egg-ovary,
nor fish whole-body or muscle tissue measurement are available, and provides consistent coverage for
all waters. Therefore, when fish egg or ovary tissue measurements are available, the fish egg or ovary
tissue measurements should be the sole measurements used to determine whether or not the selenium
criterion is being attained, regardless of the presence or absence of any other measurements. Similarly,
when fish egg or ovary measurements are not available, but fish whole-body or fish muscle tissue
measurements are available, the fish whole-body or fish muscle tissue measurements should be the sole
measurements used to determine whether or not the selenium criterion is being met regardless of the
presence or absence of water column measurements. Water column measurements should only be used
to determine whether or not the selenium criterion is being met if fish egg, fish ovary, fish whole-body,
and fish muscle tissue measurements are all not available. Further, as water column data may be
collected more frequently than tissue data, the water column elements do not apply unless tissue data
has not been collected within the same calendar year."

Corresponding changes would need to be made throughout the documentation, including but not
limited to the table on Page 4 of the Fact Sheet.

46

-------
EPA-HQ-0W-2004-0019-0419-A2; Lima Refining Company; Posted 11/03/2015
COMMENT#!

The modeling approach used to translate from an egg - ovary tissue based criterion to a water column
concentration subsumes within it a number of factors that are poorly quantified and which introduce
uncertainty into the results. As such, the accuracy of the water column criteria values and the level of
protection they provide are uncertain and potentially problematic for dischargers.

The strength of the approach that serves as the basis for the Se criterion proposed by EPA is that it is
based on accumulation of tissue Se at the proximate site of action of toxicity (for reproductive effects),
the egg or ovary, or in the associated whole body or muscle tissues. The available data to support the
reproductive effect levels of Se in tissues are reasonably strong. From a practical perspective, the need
to prescribe water column concentrations of Se for use in permitting is also recognized. However, the
approach used for this purpose introduces considerable uncertainty into the process. The model that is
applied was originally proposed more than 30 years ago (Thomann et al., 1981) and has been widely
applied since that time. The fact that it represents many of the features that are important
determinants of the level of selenium accumulation by aquatic organisms is important, but it does not
ensure the accuracy of the results produced by its use.

EPA employs several simplifications to the time variable version of the bioaccumulation model to make
the translation from the egg — ovary tissue criterion concentration to a water column concentration,
one that should result in a tissue Se concentration that equals the egg — ovary benchmark. First, it does
not consider (i.e., it neglects) the time variable nature of Se accumulation and effects. That is, the steady
state ("SS") form of the solution is used to interpret the inherently time-variable process of Se
accumulation. Although this SS solution (Equation 1 from USEPA, 2015, page 55) includes the kinetic
coefficients that control a time variable response to changing environmental exposure levels, it fails to
represent what is inherently a time variable response in the presence of typically non-constant
environmental exposure levels. Second, EPA also assumes, on the basis of a questionable and qualitative
rationale, that several of the important terms in this equation (terms related to selenium uptake from
water and biomass dilution due to growth) can be neglected. As a result, the calculation used to derive
EPA's proposed water quality criterion of 3.1 ng/l does not accurately reflect all relevant scientific data,
and should be reconsidered in the Final Rule.

COMMENT ttlA. Effect of Growth Rate

EPA, in keeping with the kinetic modeling approach used by Luoma and Rainbow (2005), assumed that
for selenium, the growth rate is a relatively inconsequential parameter under most circumstances.
Although the publication does make the simplifying assumption that g can be neglected, it is not the
case that this assumption was made specifically in regard to selenium. Luoma and Rainbow note that
"Growth was not included in most model forecasts reported here, and did not seem necessary. That
does not mean that growth considerations are never important."

Part of EPA's rationale for neglecting the growth rate coefficient was that "high consumption rates of
selenium-contaminated food may counteract selenium dilution that occurs with the addition of body
tissue during periods of fast growth". This statement should only be considered in a qualitative sense.

-------
Further quantitative justification is needed to demonstrate the appropriateness of this approximation in
the course of making such an important simplification to a quantitative modeling framework.

While Luoma and Rainbow (2005) do include some steady state results for Se, the predicted and
measured values were limited in several respects. Specifically, the results for Se were limited to marine
organisms (mussels, clams and zooplankton) from saltwater settings (all of the data were from San
Francisco Bay, other than one data point from the Mediterranean Sea). The comparisons were also
limited to median concentrations for both modeled and measured results. Importantly, there were no
results for fish.

EPA also completed a sensitivity analysis (Appendix J, USEPA, 2015) related to the importance of growth
rate. However, this analysis does not provide enough detail to assess the validity of what was done.
Based on the description provided, the analysis was based upon Equation 1 from USEPA, 2015, page 55,
as a SS evaluation. Equation 1 from USEPA, 2015, page 55, includes the sum of elimination and growth
rate coefficients (Ke + g) in the denominator. However, the value of Ke that was used for the comparison
is not mentioned. Thus, the degree that this parameter value may have influenced the results of their
sensitivity analysis cannot be assessed from the description provided. At the very least, EPA must
provide the values used for these coefficients before finalizing this review, so that states and
stakeholders can provide meaningful input on this modeling.

The sensitivity analysis completed by EPA was limited in that it was not performed for all trophic levels.
Appendix J in USEPA, 2015, states that the sensitivity analysis for growth was only performed for trophic
levels 2 and 3 (TL2 and TL3). Thus, it would appear that the analysis did not consider the influence of g
on Se accumulation by trophic level 1 (TL1, consisting of algae, periphyton, detritus), where most Se
enters the food chain (USEPA, 2015, page 12) and where uptake is entirely from water. This is a
potentially significant oversight, because the same rationale for the effect of growth rate can be applied
to both algae and periphyton. In this case, it cannot be argued that food consumption [by plants] is high
when growth rates are high, yet changes in growth rate may have a direct influence on the
concentrations that result. Thus, "biomass dilution" may be important to consider at TL1.

There is relevant scientific data that supports EPA following such an approach. The effect of growth rate
on metal accumulation by plants was clearly demonstrated in a study by Hill and Larsen (2005). The
authors measured metal concentrations in periphyton grown on tiles placed in a shallow creek for four
weeks. The periphyton that grew on shaded tiles had a relatively low g (0.078/day) in comparison to
unshaded tiles (g = 0.189/day) and UV-shaded tiles (g = 0.204/day). The ratios of the higher metal
concentrations in the low growth rate biofilm to the lower metal concentrations in the high growth
biofilms are summarized on Figure 1 (below) for each of the metals considered. As indicated by these
ratios, the low growth rate biofilm typically had about 3 to 4.5 times higher metal concentrations than
either of the two high growth rate biofilms. While metal uptake rates were found to be relatively
independent of growth rate, the higher growth rate coefficient apparently resulted in biomass dilution
of accumulated metal. Such results highlight why it may be important to consider (rather than neglect)
growth rate in data and modeling analyses. Although the study by Hill and Larsen did not include results
for selenium, it is reasonable to expect that a similar finding would apply to selenium as well.

-------
Considering that this process occurs at the base of the food web, where most of the transfer of Se to the
food web occurs, this is an important factor that should not be ignored.

Therefore, EPA should provide an explanation of each parameter that is used in calculating the proposed
water quality standard.

COMMENT ft2. Uncertain water column criteria cause undue burden on dischargers.

The modeling approach used to translate from an egg - ovary tissue-based criterion to a water column
concentration subsumes several, poorly quantified factors. In doing so, it introduces uncertainty into the
results. As such, the accuracy of the water column criteria values and the actual level of protection they
provide are uncertain. This may be problematic for dischargers who must achieve water quality based
permit limits. As a result, the discharger may have an unduly burdensome need to implement treatment
that may not be required to protect the resident biota.

Given the potential uncertainties discussed above, we respectfully submit that it is not appropriate at
this time and until the above uncertainties are addressed, to recommend to the states the use of the
water quality criterion of 1.2 ng/L for lentic systems and 3.1 ng/L for lotic systems. Given the complex
nature of the tiered selenium criterion, and that typical monitoring tends to be at the water column
level, states may opt to use the water column criterion to set permit limits. The above discussion
indicates that the water column criteria are set at conservative concentrations as they are derived from
a series of factors to translate from measured egg ovary effects levels to estimated water column effects
concentrations. While egg-ovary and fish tissue criteria are based directly on actual effects data, water
column criteria are not (they are calculated based on the egg — ovary criterion). The outcome of having
to meet the water column criterion would be permit limits that are unduly burdensome on dischargers.

The preceding point is demonstrated by the large range in water borne concentrations levels that are
shown in Figure 3.9 (USEPA, 2015, page 84). The graph shows that environmental protection is still in
place for most of the sites (likely including LRC) at water borne concentration levels well beyond the
conservative 3.1 ug/liter (20% probability) value. For LRC, imposing the uncertain proposed 2015 USEPA
selenium criterion would cause an unnecessary reduction to the plants SeT effluent limit that the plant
could not meet without incurring financial hardship, if it could be met at all.

EPA-HQ-OW-2004-0019-0387-A1; Colorado Department of Public Health & Environment;
Posted 10/14/2015

The WQCD has concerns with EPA's proposed lentic and lotic water column elements. While the WQCD
understands the need for a water column element to simplify implementation of the proposed selenium
criterion, the approach taken seems inappropriate.

Our first concern relates to the overall concept of recommending a single water column number for all
waters nationwide. While EPA cited a few of the key points from the Pellston Workshop on "Ecological
Assessment of Selenium in the Aquatic Environment" (a 2009 gathering of 46 selenium scientists) in the
2015 draft criteria document, the following was not included:

-------
"A single, universal, dissolved water quality value is inappropriate for predicting toxicity. The dissolved
Se concentration benchmark that is necessary to protect one site may be either insufficiently protective
or unnecessarily protective at another site." (Chapman et al., 2009)

This is an important consideration, especially considering how low the proposed water column elements
are compared to most states' current criteria. The variability in protective water column values for
individual sites is evident in the data used by EPA to develop the water column element (Figure 3. 9,
pasted below). Every single water concentration shown in Figure 3. 9 is protective at a particular site.

Original letter contains Figure 3.9 - Probability distribution of the water column concentrations
translated from the egg-ovary criterion element at 20 lentic and 22 lotic aquatic sites. See original letter.

In addition, the proposed water column elements are not actual toxicity thresholds like the other
elements of the criterion or criteria in general. The 20th percentile of water column concentrations
back-calculated from the egg-ovary criterion at 20 lentic and 33 lotic sites has no actual toxicological
relevance as an effect level. Use of the 20th percentile of site-specific data only means that 20% of the
sites in the distribution are "underprotected" and 80% are "overprotected". Only at sites with the exact
food web/bioaccumulation conditions as those sites falling on the 20th percentile line will the aquatic
community appropriately protected.

The WQCD would appreciate EPA's continued evaluation of options for the water column element.
While we understand site-specific standards could be developed where necessary and appropriate, we
have serious concerns about the blanket application of such a stringent criterion with potentially little
ecological relevance.

EPA-HQ-OW-2004-0019-0410-A2; West Virginia Department of Environmental Protection
(DEP); Posted 10/28/2015

EPA's 2015 Draft proposes two water column elements of 1.2 ng/L for lentic waters, and 3.1 ng/L for
lotic waters. EPA's current national recommended chronic criterion for selenium is 5 ng/L, which was
developed by EPA to protect bluegills, a recreationally important species. The chronic water column
elements in EPA's proposed criteria are not developed based on the current EPA guidelines for
developing water quality criteria, but instead are back-calculated from the body burden criteria using
trophic transfer functions. EPA's new approach in the 2015 Draft constitutes a full departure from EPA's
approach in its 2014 pre-draft selenium criteria. Furthermore, DEP is finishing up a two-year
reproductive success study in waters with elevated selenium compared to control water. Preliminary
results of this study show the 5 ug/L water column concentration is protective of West Virginia aquatic
life. Based on these concerns, West Virginia believes the appropriate approach is to retain the nationally
recommended water quality criterion of 5 ng/L for water column concentrations.

-------
Comment Category 2.2 - Comments of General Nature Concerning

« « , | ih[, ,\r. i Hii| i. i I indeii I k
-------
when comparing deformities from control and Se-diets. Linville (2006) also found that effects at Stage
45 for control and selenium-spiked diets were significantly different using Tukey's Honest Difference
Test for the maternal transfer study. For more detail please see Section 3, pg 37 of the selenium
criterion document, and Appendix C.

The bluegill (Lepomis macrochirus) data (Hermanutz 1992, 1996, as corrected byTao, 1999) reanalysis
resulted in a statistically superior fit to the previous TRAP-derived ECi0. As a result, the new ECio
increased from 11.36 mg/kg dw to 14.7 mg/kg EO dw. For more detail please see Section 3, pg 41 of the
selenium criterion document.

2.2 Responses concerning averaging period for the water column criterion element

EPA has described the selection of the 30-day averaging period in detail in the final 2016 selenium
criterion document.

2.2 Responses concerning selection of measure of effect

Many commenters supported the measure of effect (ECio) used by EPA for this bioaccumulative
pollutant; a few commenters disagreed. Typically EPA's national 304(a) criteria have been based on an
EC2o for pollutants whose toxicity is dominantly expressed through water exposure. Because selenium is
bioaccumulative in biota and is persistent in the ecosystem, EPA selected the ECio to protect populations
of all sensitive (fish) species. In addition, a protective measure of effect is critical for selenium due to the
steepness of the dose-response curve, compared to other toxicants. Effects of selenium exposure move
from beneficial, to low adverse effects, to major adverse reproductive effects over a very narrow dose
range (e.g., 2 mg/Kg bw dw). The selection of the ECio as the measure of reproductive effect for this
chemical was identified in the problem formulation for the selenium criterion in the 2014 draft criteria
document, and did not change in the problem formulation for the 2015 draft criterion document, or in
the 2016 final criterion document. Further, EPA does not agree with the comment that an effects
endpoint of 25% (EC25) is negligible, particularly for a bioaccumulative pollutant.

2.2 Responses to general comments

The nature of selenium bioaccumulation may vary on a site-by-site basis in a manner sufficient to justify
development of site-specific criteria. Reviewers identified several important issues that will be
addressed in the technical information to be developed to guide the implementation of the criteria.

2.2 Responses to comments on the sensitivity of bluegill and catfish

EPA analyzed all available, reliable data for bluegill and catfish and included these data in distributions
used to derive the chronic selenium criterion. EPA identified three studies available for these species:
Coyle (1993), Doroshov et al., 1992, and Hermanutz et al., 1992. The Doroshov et al. catfish study was

-------
not used by EPA, because fish were exposed via injection, not maternal transfer, which was the route of
exposure in the other studies used in the derivation of the criterion. These data were not included in
derivation of the criterion because the injection route of exposure is not an acceptable experimental
protocol for studies used in criterion derivation due to its difference from exposure routes in the
environment (for selenium, diet). While EPA's analysis indicated that the Hermanutz ECio was the
lowest of the 3 bluegill studies (Hermanutz, Coyle, and Doroshov) at 14.7 mg/kg dw (vs 9.0 mg/kg dw
from Lemly's analysis), the geometric mean of the studies (bluegill SMCV) was 20.6 mg/kg dw. Based on
the available data and application of our Guidelines Methodology, we conclude that the criterion is
protective of bluegill.

One comment cited field data from Belews Lake and concluded that catfish are "very sensitive" to
selenium poisoning and that the final chronic value needs to be lower than 10 mg/kg dw to be
protective of bluegill and catfish. In the effects characterization, EPA describes field evidence from Hyco
Reservoir that found catfish representing multiple year classes present even after most other fish
species were reproductively extirpated from the lake. This indicates that, at a minimum, catfish are no
more sensitive than other species for which we have reliable egg-ovary data for (i.e., centrarchids like
bass and bluegill), and that they are likely less sensitive, due to their presence in these studies after
other species disappeared. Thus, EPA concluded that the egg-ovary criterion is expected to be protective
for catfish (i.e., ictalurids), despite the absence of valid egg-ovary test data.

2.2 Response to comments on the phenomenon of "winter stress syndrome"

Due to the lack of demonstration of effects due to winter stress in the field, (e.g., Hermanutz et al.,
1992, and others), there is great uncertainty as to the actual field occurrence of these effects observed
in the lab studies (Lemly 1993, Mclntyre, 2010). In addition to Hermanutz et al., 1992, investigation of
the potential for winter stress effect (or winter stress syndrome) has been done in field studies of many
types of juvenile fish species (northern pike, fathead minnows, creek chubs, white suckers, and slimy
sculpin) inhabiting areas receiving complex metal mine effluents containing elevated selenium
concentrations. None of these studies found evidence of winter stress, thus casting doubt on the actual
occurrence of this effect in natural aquatic ecosystems. Therefore, the EPA no longer considers "winter
stress syndrome" as a viable effect that should be assessed separately for selenium. Furthermore, the
effect concentration and proposed whole body tissue criterion based on observations in the Lemly and
Mclntyre lab studies is similar to the final whole body criterion value of 8.5 mg/kg dw, therefore the
level of protection afforded by the whole body criterion should be protective for winter stress
syndrome, should it occur, despite the lack of a separate assessment.

2.2 Responses to comments on translation offish tissue to water

EPA was made aware of an overweighting of certain waters in the 2014 comments provided by the
public. The issue stemmed from EPA's use of all fish species monitored at a site. Because the number of
fish species were not equal at all sites, this led to the possibility that those sites with several species (up
to 6) were more heavily represented in the database than those sites with a single fish species. In lentic

-------
waters, this occurred at several sites with low selenium in the water column (leading to a high EF, and
fish tissue values that were protective and low). The opposite scenario occurred in lotic waters, where
waters with higher selenium (and lower EFs) had a larger number offish species available. This led to
those waters providing several fish tissue values that were protective at higher selenium concentrations.
Because the goal is to provide a national criterion protective of aquatic life designated uses in all lotic or
lentic waters, a decision was made to use the most bioaccumulative fish species available at a site. This
insured that protection would be provided to those fish that may be sensitive to selenium and have a
diet that reflects a bioaccumulative food web. The resulting water column criterion elements ensure
that aquatic life uses for selenium-sensitive fish species are protected on a national basis for an
estimated approximately 80% of waterbodies.

If the characteristics of a site are such that data demonstrate that a different, perhaps higher, water
column criterion element would be equally protective, EPA notes that site specific criteria can be
developed. EPA has provided information on how a site-specific selenium criterion could be developed
in Appendix K.

An alternative approach for deriving protective national criteria was provided in a 2014 report by
DeForest et al. 2014, provided to EPA in 2015 comments by NAMC-SWG. EPA has reviewed the detailed
and comprehensive information and analyses that the report presents. This report used many of the
same studies that EPA used in the derivation of the national chronic criterion, but excluded a key study,
conducted on white sturgeon. EPA notes the recommendation in the report that the egg/ovary criterion
be set at 20 mg/kg (higher than EPA's recommendation of 15.1 mg/kg) is primarily due to the exclusion
of the white sturgeon study. When adjusted to include this study, the DeForest water column
recommendations become 4 ng/L and 1.5 ng/L for lotic and lentic systems respectively. This Deforest -
approach based analysis and resultant water column values would represent protection of 75% of
waterbodies, while EPA has made a policy decision to select the 80 centile of water values in derivation
with the national water column value (slightly more protective). The water column values calculated
using this are strikingly similar to EPA's water column recommendations of 3.1 ng/L and 1.5 ng/Lfor
lotic and lentic systems. This result further substantiates and supports EPA's water column values,
despite the use of different analytic methods in the two approaches.

2.2 Specific comments

EPA-HQ-OW-2004-0019-0384-A2; Cameco Resources; Posted 10/13/2015

Based on our review, we were disappointed to see that there appears to be no substantive changes to
the 2015 draft despite the comments we submitted on the previous published External Peer Review
Draft Aquatic Life Ambient Water Quality Criterion for Selenium- Freshwater 2014 (EPA 822-P-14-001;
2014 draft) on June 13, 2014 (J. Leftwich, Cameco, to K. Gallagher, US EPA). As an example, the
egg/ovary selenium criterion (15.8 mg/kg) is only slightly changed from the 2014 draft criterion (15.2
mg/kg) and is still overly conservative because, as previously detailed there is scientific evidence to
suggest an environmentally protective criterion of at least 20 mg/kg egg/ovary selenium.

-------
EPA-HQ-OW-2004-0019-0394-A2; Appalachian Mountain Advocates and Sierra Club
Environmental Law Program; Posted 10/14/2015

IV. The Concentrations of the Fish Tissue Elements Are Too High to Protect Sensitive Species

EPA's Guidelines for Deriving Numerical National Water Quality Criteria for the Protection of Aquatic
Organisms and Their Uses (1985) explain that water quality criteria should fully protect sensitive species
that are "commercially or recreationally important." EPA derived its egg/ovary element, which forms the
basis for its other fish tissue elements, by averaging the genus mean chronic values for what it claims
are the four most sensitive genera for which adequate data exist. Draft Criterion at 27. The resulting fish
tissue elements are not adequate to protect certain sensitive species that are commercially and
recreationally important, such as species of bluegill and catfish.

In a letter to EPA expressing concern over the egg/ovary criterion in EPA's 2010 draft proposal, selenium
expert Dr. Dennis Lemly of the USDA Forest Service concluded that EPA's inclusion of more tolerant
species in the criterion evaluation and development resulted in a proposed criterion that would have
allowed mortality to exceed allowable limits in more sensitive species. Dr. Lemly stated that scientific
studies show quite clearly that a criterion of 17.07 mg/kg for fish eggs/ovaries will jeopardize two of the
most important freshwater fish families in North America: Centrarchidae and Ictaluridae. For example,
(1) An EPA field study published in the peer reviewed journal Environmental Toxicology and Chemistry
(Hermanutz et al 1992) found that ovary selenium concentrations of 9 mg/kg dw or greater resulted in
40% higher mortality and 80% more edema in larval bluegill sunfish that controls for an EC40-80
(converted from wet weight using 80% moisture, based on mean wet weight +- one standard deviation).
The results of this study are not included in EPA's draft criterion calculation, and (2) A laboratory study
at the University of California (Doroshov et al. 1992) found that the EC50 for larval mortality of channel
catfish and bluegill sunfish occurred at egg selenium concentrations of 7.2 and 15.0 mg/kg dw
respectively (lower limit of 95% confidence intervals). These mortality data were not included in the
data used to derive the FCV.

Extensive field data from the Belews Lake case example, which includes reproductive analysis from
young-of-the-year stock assessment, clearly show that catfish are very sensitive to selenium poisoning in
a real-world setting. . .equal to or greater than sunfish (Cumbie 1978, Cumbie and Van Haron 1978,
Holland 1979, Garrett and Inman 1984, Lemly 1985). . . .

The FCV needs to be lower than 10 mg/kg dw in order to protect sunfish and catfish at an EC10 level,
which is the level of protection afforded to trout by the 17.07 draft criterion value.

Letter to Mr. Joseph Beaman, Chief, USEPA, Office of Water, Ecological Risk Assessment Branch,
Washington, DC from A. Dennis Lemly, Ph.D., Research Fish Biologist, USDA Forest Service, Southern
Research Station, Piedmont Aquatic Research Laboratory, July 6, 2010 at 1-3 (emphasis added). Clearly,
EPA's proposed egg/ovary element of 15.8 mg/kg would not protect those species at the EC10 level that
EPA has used to derive its current proposed criterion.

In addition to improperly averaging values across genera, EPA failed to adequately account for "winter
stress" in sensitive bluegill species. As EPA recognized in its Draft Criterion document, a study by Dr.
Lemly found the protective chronic selenium whole body concentration for juvenile bluegill to be 5.85

-------
mg/kg prior to winter stress. Instead of using this protective value for the bluegill's genus mean chronic
value, EPA averaged it with the values from Mclntyre et al.'s 2008 study, which also purported to
account for winter stress, but arrived at a much less protective concentration of over 9 mg/kg. Draft
Criterion at 122-23. Reliance on the Mclntyre study to account for selenium is misplaced, however,
because that study failed to actually induce winter stress, in part, because it did not control photoperiod
or discuss the impacts that the lack of photoperiod controls may have on the interpretation of study
results. Indeed, EPA explicitly acknowledges that the "effects of photoperiod on fish and other
ectotherms are well-documented." Draft Criterion at C-158. EPA must fully account for winter stress,
using studies that actually induce such stress by recreating realistic winter conditions including reduced
photoperiod, when revising its fish tissue concentrations to ensure protection of sensitive aquatic
species.

For the foregoing reasons, EPA must significantly reduce the concentrations allowed under its fish tissue
elements to ensure they are protective of sensitive species, aquatic-dependent wildlife, and threatened
and endangered species. EPA must then translate those revised tissue concentrations to enforceable
water column criteria that can be practically implemented to achieve the regulatory requirements of the
Clean Water Act.

EPA-HQ-OW-2004-0019-0386-A2; Asarco LLC; Posted 10/13/2015

Asarco supports the comments submitted on the draft criteria revision by the North American Metals
Counsel ("NAMC"), particularly those portions of the NAMC comments arguing that: (1) the tissue
criteria are skewed downwards by reliance on some flawed studies and higher criteria would in fact be
protective; (2) "never to be exceeded" tissue criteria are unduly restrictive and should be replaced by
seasonal averages; (3) the methodology for translating tissue criteria to water quality criteria is flawed
and will result in overly low water quality criteria.

EPA-HQ-OW-2004-0019-0371-A2; Wyoming Department of Environmental Quality (WDEQ);
Posted 09/15/2015

Limitations of Fish Reproductive Tissue Endpoint

From 1971 to 2000, Wyoming was the third driest state in the country; only Nevada and Utah were
drier. Mean annual precipitation was 13 inches; 30% of the state received less than 12 inches of
precipitation per year and 67% of the state received less than 16 inches of precipitation per year1. As
such, between 70 and 80% of Wyoming's 11 Streams" are ephemeral and/or intermittent without the
natural hydrologic potential to support fish populations2.

WDEQ/WQD understands that EPA developed the draft selenium criteria to protect the most sensitive
species and life stages, but the resulting criteria are likely over-protective for most of Wyoming's
streams that lack any fish species. WDEO/WQD appreciates EPA's discussion that states should adopt
criteria to protect downstream fish species for waters that lack fish. Unfortunately, due to limited staff
and resources, it is not practical for Wyoming to develop and adopt water quality standards for

-------
approximately 70% of our streams based on downstream fish assemblages. Further, many of these
streams have no hydrologic connectivity or are distant from fish -bearing streams. It is also unrealistic
for the State to apply overly stringent criteria to most of our waters since this may result in unnecessary
impairment decisions and treatment for point source dischargers.

WDEQ/WQD requests that EPA develop a recalculated criteria without fish reproductive endpoints for
waters that lack fish. Currently, EPA has derived the draft selenium criteria as a function of fish egg-
ovary values; this approach precludes the use of the recalculation procedure for waters that lack fish
since recalculations will be inherently fish -based. Development of a recalculated criteria without fish
endpoints would require more toxicological data on additional taxa (e.g., invertebrates including
crustaceans) in order to accurately protect aquatic life that may reside in fish less waters. EPA should
make it clear within the draft criteria document that there is no expectation for the recommended
selenium criteria to be applied to waters without fish.

EPA-HQ-OW-2004-0019-0399; Clean Water Team; Posted 10/15/15

Thank you for the opportunity to comment. The selenium criterion proposed is overprotective for the
desert southwest ecoregion. The proposed regulation:

does not recognize that that over millennia, naturally-occurring levels of selenium has resulted in
ecoregion-wide biologic adaptation; does not recognize that the predominant form of selenium in the
Las Vegas Wash, selenate, is a micronutrient required by endemic life forms ranging from bacteria to
mammals. This form of selenium demonstrates the least toxicity in aquatic environments, and
widespread selenium toxicity has not been noted in birds or fish in the Las Vegas Wash or Lake Mead;
proposes that one tissue test result above the limit stated for ovary or muscle tissue indicates non-
compliance. This requirement is unique to this regulation and does not allow for error. No other
regulation imposes such a rigorous requirement. For lotic environments, cannot be met in most of the
Colorado River system due to naturally-occurring selenium.

EPA-HQ-OW-2004-0019-0391-A1; North American Metals Council (NAMC); Posted
10/14/2015

2.2 Improving the Scientific Foundation for the Fish Selenium Criterion

The proposed tissue criterion in the 2015 EPA Selenium Draft is overly conservative and not technically
defensible. A higher criterion would provide environmental protection without unduly penalizing human
industrial or other activities with concomitant negative economic impacts, and without unnecessarily
expending limited regulatory resources to the detriment of genuine environmental issues. Our
comments below focus on three studies that have been misrepresented and thus inappropriately lower
the egg/ovary selenium criterion. We also comment on the misuse of technically indefensible
overwinter survival data to limit inappropriately the reproduction-based whole body criterion, which
impacts the egg/ovary selenium criterion. Finally, we provide evidence that egg/ovary selenium
concentrations above the proposed criterion occur in fish from reference areas and provide additional

-------
data supporting a higher fish tissue selenium criterion than provided in the 2015 EPA Selenium Draft,
specifically at least 20 mg/kg dw for egg/ovary.

2.2.1 Brown Trout EC10

The EPA brown trout egg/ovary EC10 of 18.09 mg/kg dw was derived only considering survival, not
deformities, and without considering the data from full hatch through 15 days post swim-up test. EPA
should have used the combined endpoint, which integrates lethal and sublethal endpoints. The
uncertainties identified by EPA as justification for selective data exclusion are not technically defensible.
Examination of the hatchery data from the Yellowstone cutthroat trout studies (e.g., the wild run of fish
from Henry's Lake) clearly indicates that slight abnormalities are consistent with a natural rate of
deformities6 and better explains the observed higher incidences of slight deformities for hatchery
brown trout. Accordingly, the combined data should have been used, combining fish ranked as 0 (no
deformities) and 1 (slight deformities).

As noted in the NAMC comments on the 2014 EPA draft (Attachment 1), there is no sound basis for
using either the survival endpoint alone or the deformities endpoint alone when the combined survival
and deformities endpoint is available. EPA's first application of regression analysis for estimating ECx
values was in the 1999 ammonia criteria document. Whenever possible, that document used the
combined survival and growth endpoint, which it termed "biomass." Given EPA's past preference for
combining the important endpoints, it is unclear why, in this case, EPA did not combine the survival and
deformity endpoints for the brown trout study. There is no technical justification for this unusual and
arbitrary decision.

The Toxicity Relationship Analysis Program (TRAP) calculation for the combined endpoint grouping the 0
and 1 deformity data, including the hatchery data, results in an EC10 of 20.5 mg/kg dw as derived by the
authors of the brown trout study. (See Figure 1, below, as presented to EPA as part of Simplot's
comments on the 2014 EPA draft.) The EPA-proposed EC10 of 18.09 mg/kg dw for brown trout based
only on the partial data set should be replaced by the conservative EC10 of 20.5 mg/kg dw, which
considers the full data set. As noted in the NAMC 2014 comments (Attachment 1), there is no evidence
that selenium is having effects on the combined survival and deformities endpoint at concentrations
below 20.5 mg/kg dw. There is less than a 10% effect (relative to the background response) below this
concentration. The EPA's TRAP calculation undershoots this data point, thus yielding an overly
conservative EC10.

Original letter contains Figure 1 - TRAP tolerance distribution graphical output for the model fit yielding
an EC10 of 20.5 mg/kg dw for the combined survival and deformities endpoint. See original letter.

2.2.2 Bluegill EC10

As noted in the NAMC 2014 comments (Attachment 1), EPA obtained its bluegill reproductive genus
mean chronic value (GMCV) as the geometric mean of EClOs from three studies: 20.75 mg/kg dw egg
from Doroshov et al. (1992),7 24.55 mg/kg dw egg from Coyle et al. (1993),8 and 12.68 mg/kg dw ovary
from the combined Hermanutz et al. (1992, 1996)9, 10 studies.

-------
The EC10 obtained from the two Hermanutz et al. studies is an outlier (about half the EC10 of the other
two studies) and does not reflect the fact that the correct endpoint reported by Hermanutz et al. (1992)
was 30 mg/kg. As detailed in the NAMC comments on the 2014 EPA draft (Attachment 1), use of these
two studies is not scientifically defensible. They were not designed to obtain an egg/ovary selenium
benchmarkll and there is a clear disconnect in these studies between the well-characterized water
selenium concentrations and the supplementary information on selenium in the fish.

Accordingly, the two Hermanutz et al. studies should not be used to determine the combined species
mean chronic values (SMCV) and GMCV. Instead, these values should be derived solely from the other
two studies - Doroshov et al. (1992) and Coyle et al. (1993). DeForest et al. (2012) reported a species
mean of 21.5 mg/kg dw for these two studies in their derivation of a selenium benchmark.12

2.2.3 White Sturgeon EC10

The 2015 EPA Selenium Draft references the Linville 2006 white sturgeon study, which was not included
in the 2014 EPA draft. That study reported the lowest EC10 and thus the lowest GMCV, which drives the
current egg/ovary criterion. NAMC notes that this study is very weak, with relatively little effect
determined at the highest tested concentration. Effects were noted for both edema and weight of
larvae, but not for survival, length, or weight. The combined effects of edema and deformities were
27.8% at 20.5 mg/kg dw egg selenium and 13.3% at 11 mg/kg dw egg selenium.

The Linville study data do not allow calculation of a technically defensible EC10 because a concentration-
response curve was not established. There simply are not sufficient data spanning a no effect to effect
range to derive a reasonable dose-response curve. Thus, there were insufficient data points to provide
an unambiguous EC10, as required by EPA's computer program, TRAP.14, 15 EPA, however, ignored the
TRAP requirements for a technically defensible EC10 and calculated a highly uncertain EC10 based on a
partial response at the highest selenium concentration. As the 2015 EPA Selenium Draft states, "[a]n
EC10 based on only one partial response would not ordinarily be included in the chronic data set...."16

Problems with determining any EC10 from Linville are apparent upon review of Appendix C of the 2015
EPA Selenium Draft. The EPA EC10 is not the only value that could have been derived and was
determined subjectively as the most conservative value. The EC10, therefore, is highly uncertain and
most likely overly conservative as the slope was determined based on the single highest effect value.

The technically indefensible partial response used to derive a white sturgeon EC10 was explicitly
included for two technically inappropriate reasons - (1) because sturgeon are a threatened or
endangered species and (2) because of a non-reproductive study. The non-reproductive study resulted
in a whole body EC10 for green sturgeon of 16.36 mg/kg dw and for white sturgeon of 23.94 mg/kg dw.
Both EClOs were based on percent body weight increase. There were no mortalities of white sturgeon
during the study, while for green sturgeon survival data resulted in a whole body EC10 of 28.93 mg/kg
dw. De Riu et al. (2014) noted that the lower body weight increase could have been due to insufficient
food and thus "low energy reserves in the green sturgeon."

As previously noted, NAMC agrees fully with EPA that, based on extensive scientific evidence,
reproductive effects in conjunction with the magnitude offish egg/ovary selenium concentrations are of
greater ecological concern and provide a more reliable basis for the criterion than non-reproductive

-------
endpoints (e.g., survivorship, growth). Accordingly, EPA's reliance on a non-reproductive endpointto
support a highly questionable reproductive endpoint is not technically defensible.

EPA is proposing a national selenium fish egg/ovary criterion driven by two studies with sturgeon: a
partial response for white sturgeon that is only slightly above a negligible effects endpoint of 25% and a
slight reduction in body weight for green sturgeon, which are typically found in marine or estuarine
waters, not freshwaters. Neither of these responses could reasonably be related to effects that might
threaten populations. NAMC reminds EPA that its national water quality criteria do not attempt to
protect all species at all times, "[b]ecause aquatic ecosystems can tolerate some stress and occasional
adverse effects, protection of all species at all times and places is not deemed necessary."

NAMC notes the coincidence resulting from the use of the Linville study in that it limits the fish tissue
criterion relative to the Lemly flawed winter stress syndrome without creating a problem of defending
the criterion's protectiveness for juvenile overwinter survival.

2.2.4 Overwinter Survival

The Lemly bluegill overwinter survival value,when averaged with the Mclntyre et al. overwinter value,25
appears to limit how high the reproduction-based whole-body criterion can be without creating a
problem of defending the criterion's protectiveness for juvenile overwinter survival.26 As highlighted in
the NAMC comments on the 2014 EPA draft (Attachment 1), juvenile overwinter survival is not a
technically defensible issue based on the Lemly (1993) study.

The juvenile bluegill overwinter survival testing conducted by Lemly (1993) to demonstrate increased
overwinter mortality due to selenium is technically flawed. There was only one treatment, and thus no
dose-response relationship was established. The relatively high selenium toxicity in Lemly's (1993) study
was not found by Mclntyre et al. Although the Mclntyre et al. study has been criticized for a possible
difference in light-dark test conditions, this relatively minor difference from the Lemly (1993) study
should not have obviated replicating that Lemly findings of relatively high selenium toxicity if, in fact, the
Lemly study results were robust and repeatable. The Mclntyre et al. study should be viewed as the more
reliable result because it has the greater number of concentrations tested; the more realistic exposure
pathway (i.e., yeast to oligochaetes to fish), and the initial exposure period at a warm temperature.

Lemly's winter stress syndrome is based on the hypothesis that fish rely on stored energy to survive
winter months due to low food availability that results in increased metabolism and energy deficiency.
In fact, fish do not generally rely on stored energy over the winter months as active feeding occurs.

NAMC notes that the seven studies on active feeding referenced in the previous sentence above were
provided in the NAMC comments on the 2014 EPA draft (Attachment 1) and, thus, were known to and
available to EPA for its consideration. Five of the seven references, however, were neither cited nor
considered in the 2015 EPA Selenium Draft. Two of the references (Bennett and Janz, 2007(a) and
Bennet Janz, 2007(b)) were cited but not considered in the context of active winter feeding. It is unclear
why EPA ignored these technically valid comments on the previous draft document, which calls into
question the integrity of the public review process and the technical validity of this revised draft
document.

-------
Field studies have provided no support for Lemly's hypothesis of a winter stress syndrome related to
selenium exposures. Hermanutz et al. exposed bluegills to selenium in outdoor experimental streams
overwinter but did not report increased overwinter mortality. Aspects of the winter stress syndrome
hypothesis have been investigated in field studies of juvenile fish inhabiting areas receiving complex
metal mine effluents containing elevated selenium concentrations. Support for the winter stress
syndrome hypothesis would have come from decreased growth and energy storage over winter, more
so with elevated selenium concentrations than in reference areas. The opposite was found to occur in
juvenile northern pike (Esox lucius), burbot (Lota lota), fathead minnows (Pimephales promelas), creek
chubs (Semotilus atromaculatus), and white suckers (Catostomus commersoni). Slimy sculpins (Cottus
cognatus) exhibited changes in whole body triglycerides consistent with the winter stress syndrome
hypothesis, but this occurred at both sites with elevated selenium and reference sites, thus also
providing no support for the winter stress hypothesis related to elevated selenium concentrations.

It is misleading for the draft criteria document to average together the Lemly (1993) and Mclntyre et al.
(2008) results to create a Lepomis non-reproductive, cold season GMCV, against which the document
compares its reproduction-based criterion.A Lepomis non-reproductive, cold season GMCV should not
be provided; rather, it should be acknowledged that uncertainty remains regarding juvenile overwinter
sensitivity while noting that the weight of field and laboratory evidence indicates that the reproductive
endpoint is the critical endpoint. The NAMC notes that EPA states "The observations of Lemly (1993b)
are evidence that larval deformity, not juvenile mortality, is the most sensitive endpoint."

2.3.3 Translation from Tissue to Water and High Incidence of False Alarms

The 2015 EPA Selenium Draft method to translate tissue selenium concentrations to water involves
using data from the fish that accumulated the highest concentration at each site that had an EF
(enrichment factor) rather than the geometric mean. This approach results in an inappropriately low
water selenium criteria that will inevitably generate false alarms.

We note that the 2015 EPA Selenium Draft no longer contains the binary statistical analysis that was
present in the 2014 document. That analysis used a much larger data set that was not confined to sites
where algae/detritus/sediment were all measured. It used all available measurements of water and
whole body fish selenium (i.e., fish bioaccumulation factors). The water concentrations in the 2014
binary statistical analysis were adjusted to estimate a site's 95th percentile high concentration based on
previous estimates, which were based on time series simulations, of what a 30-day once-in-three year
concentration represented. As detailed in the NAMC comments on the 2014 draft, this information
provided the basis to demonstrate the expected high incidence of false alarms. EPA should explain why
it opted to remove that data set from consideration.

NAMC does not support the 30-day averaging period and continues to believe (Attachment 1) that a 60-
day averaging period would be highly protective.

In its comments on the 2014 draft (Attachment 1), NAMC urged EPA to consider a report that had been
recently completed by the NAMC-SWG, Selenium Partitioning between Water and Fish Tissue in
Freshwater Systems: Development of Water-based Selenium Screening Guidelines. That report, the
result of a three-year research effort, contained new information that filled existing data gaps on the

-------
translation of tissue-based criteria to water-based selenium screening levels. Specifically, the report
derived and recommended technically- and statistically-defensible lentic and lotic water selenium
thresholds, based on regression relationships using tissue thresholds previously determined using a
species sensitivity distribution (SSD) from literature values for reproductive effects on fish, due to
dietary exposure to selenium. That report is neither cited nor considered in the 2015 EPA Selenium Draft
and no rationale is given for this surprising omission of relevant, technically defensible, useful
information.

-------
Comment Category 2.3 - Comments Concerning Fish Tissue Criterion
Elements

2.3 Summary

Comments concerning the fish tissue criterion element as it appears in Table 1 of the July 2015 draft are
included in this section. Some commenters felt that that the fish tissue egg ovary criterion was not
legitimate claiming that it was based on the worst case assumption of the most sensitive species. Others
commented that the egg-ovary criterion was derived using too few data. The "never to be exceeded"
frequency generated much comment. Many interpreted the phrase to mean that a single fish tissue
sample measurement above the criterion value would result in non-attainment.

2.3 Response to comments

2.3 Response to comments on the methodology for deriving fish tissue criteria values

While the Guidelines EPA uses to derive aquatic life criteria do not specifically address bioaccumulative
pollutants like selenium, EPA followed the Guidelines and used the conceptual approach from EPA's
Ecological Risk Assessment Paradigm in deriving the selenium criterion.

1. First EPA gathered all data on the toxic effects of selenium on aquatic organisms, as criteria are
meant to protect the entire aquatic community.

2. Second, EPA recognized that the mode of action of selenium and its duality as a nutrient and a
toxicant would require special consideration in the criteria derivation process. Given the
available data, and deliberations from the 2009 Pellston Conference on selenium, EPA focused
on toxicity tests that used maternal transfer of selenium via diet, and subsequent larval toxicity.

3. Using the sensitivity distribution (SD) methodology of the Guidelines, EPA assembled a fish egg-
ovary species sensitivity distribution (SSD) based on ECi0s for all available reliable studies. We
evaluated the available invertebrate tests, and considered them for meeting the 1985
Guidelines requirements for meeting the 8 minimum data requirements, but we did not use
them quantitatively in the calculation of the egg-ovary criteria, since the available toxicity tests
for invertebrates only use whole body selenium residues. The available tests also indicate that
invertebrates are not as sensitive as fish. Amphibian data was also evaluated, but the data
quality of the available studies was not sufficient for quantitative inclusion in the analysis.

These studies are discussed in sections 3 and 6 in the criteria derivation.

4. We also used principles developed in an EPA White Paper reviewed by the SAB that provided for
considering mode of action and other characteristics of pollutants like Contaminants of
Emerging Concern (CECs) (and to some extent selenium). (U.S. EPA 2008). We were able to
apply these principles because we have specific information about the mode of action and
critical effect of selenium in aquatic ecosystems established via peer-reviewed publications in
the open scientific literature and reviewed at a SETAC Pellston Workshop (Chapman et al. 2009),
specifically, that egg-laying vertebrates are the most sensitive to selenium, and that
reproductive effects are the most sensitive measurement endpoint.

-------
Regarding comments on the tissue derivation procedures, EPA first incorporated the commenters'
suggestion to use empirical data on muscle or whole body concentrations where such data were
available, rather than an estimate derived using a median Conversion Factor (CF), into the criterion
element derivation. For the remainder of tissue values that had to be estimated using a CF, EPA
evaluated the approaches proposed by commenters. EPA concluded that using the median or using total
least squares (TLS) regression are appropriate ways of setting CFs, while using ordinary least squares
(OLS) is not supportable for this purpose. The proposed OLS method suffers from added uncertainty
since it only considers vertical error (y-axis). However, the TLS method evaluated by EPA considers both
horizontal and vertical error, and therefore does not have the uncertainty associated with OLS predicted
CFs that depend on which tissue is assigned to which axis. The similarity in CF outcomes between TLS
and the use of medians (original EPA method) provided support to retain the simpler median-based
method, particularly since the TLS method is not amenable to calculation using readily available
spreadsheet software. A more comprehensive description of the evaluation and method used in the
2016 final criterion document is located in Appendix N.

2.3 Response to comments on the normalization of data to dry weight

Regarding the wet weight to dry weight conversion factors, this is a very common conversion when
assessing toxicity or concentrations in tissues; data are available in the open literature on wet weight to
dry weight ratios. For the purposes of EPA's analyses, the wet to dry weight conversion factors used in
the Agency's analysis for 38 fish species are documented in Appendix B, pages B2- B-4.

2.3 Response to comments on the duration of the fish tissue criterion element

Many commenters did not understand the specification of instantaneous duration (not instantaneous
frequency, as some commenters misunderstood). As noted in the 2015 draft table footnotes,
instantaneous refers to the fact that collecting multiple rounds of samples over a long period of time
(duration) is not considered essential when proper sample collection is planned and conducted, and
when the selenium source is relatively stable. Fish in an ecosystem integrate selenium exposure
occurring through the food web/diet over time, and only change gradually. EPA recognizes that there
could be some variability in concentrations in fish tissue, due to residence time of the pollutant in the
fish habitat or depuration of selenium in females via reproductive loss of selenium to eggs. There are
some data comparing eggs and ovary concentrations post spawn to suggest that body burdens may be
lower in females for a period of time post spawn based on lower concentrations in the ovaries. There
are no data for comparisons in whole body or muscle.

EPA is developing technical support materials regarding how to sample for fish tissue, and is not
recommending that a single fish having selenium concentrations above the criterion be considered an
exceedance of the criterion. These materials will be made available for public comment prior to
finalization.

-------
2.3 Response to comments on the frequency of the fish tissue criterion element

The "never to exceed" frequency was developed to reflect the fact that selenium, as a bioaccumulative
pollutant and persistent, stable element, will accumulate and remain in a food web over a long period of
time. EPA's aquatic life criteria are traditionally based on aqueous concentrations as toxicity for the
chemicals for which criteria have been developed is generally due to aqueous exposure. These
concentrations can vary significantly both short-term and long term, allowing organisms in the aquatic
community to recover. In contrast, selenium is a bioaccumulative pollutant, and fish tissue
concentrations have been directly correlated with the adverse effect. Observed reductions in fish tissue
concentrations occur slowly, even after removal of a selenium source. Data from Belew's Lake, NC,
indicate that once selenium releases into the lake were terminated, it took fish populations 5-10 years
to recover. Thus, because selenium is bioaccumulative and persistent, clearly an exceedance frequency
of once-in-3-years would not be protective of the aquatic life use as it would permit selenium to
accumulate in a system above the criterion where it would remain in the food web and fish tissues for
an extended period of time. Therefore, EPA determined that the fish tissue element of the criterion
cannot be exceeded in order to be protective of the aquatic community. EPA is developing technical
support materials regarding how to sample for fish tissue, and is not recommending that a single fish
having selenium concentrations above the criterion be considered an exceedance of the criterion. These
materials will be made available for public comment prior to finalization. EPA has clarified that the
selenium criterion is focused on the protection of populations, not individuals. For example, from the
criterion table, Footnote 6 states that "Fish tissue data provide instantaneous point measurements that
reflect integrative accumulation of selenium over time and space in fish population(s) at a given site."

However the criterion document does note, "the assessment endpoint for selenium is the protection of
fish populations. In some waters where ESA-listed fish species occur, a protection goal oriented to
protection of individuals may be more appropriate. This should be reflected using site-specific data to
derive an SSC for the site."

Regarding comments that EPA has deviated from its 1985 Guidelines in using a "never to exceeded"
frequency, the 1985 Guidelines also encourage the use of "good science" and best professional
judgement to develop protective criteria for the pollutant at hand. EPA notes that for the most part the
1985 Guidelines reflect practices and assumptions that the major exposure to the chemical under
analysis is through the water column. This is not the case for selenium, a bioaccumulative pollutant that
is persistent in ecosystems. Recommending frequent excursions above the criterion concentration
would not be protective of aquatic life.

2.3 Response to comments on definition of fishless waters and new and increased discharges

The 2016 criterion now clarifies that water column values are derived from fish tissue via
bioaccumulation modeling under the assumption of steady state conditions. EPA agrees with the
commenter regarding increased inputs, as it is possible for a discharger to calculate a new limit based on
the existing water and fish tissue (assuming steady state) to insure the increase would not cause an
exceedance to the fish tissue criterion. Since there is uncertainty in the relationship between water and

-------
fish tissue under a new discharge scenario, use of the water column value will insure that the fish in the
system is protected until steady state is achieved in the fish.

2.3 Responses to footnote 5 of the 2015 criterion table, regarding changes in selenium
concentration in fish tissue related to new discharges

Regarding the KDOW comment referring to footnote 5 of the 2015 draft document, this footnote has
been edited.

Section 6.1 of the 2016 selenium criterion document notes that researchers often report concentrations
of selenium in fish eggs or ovaries Osmundson et al. (2007) found reduced levels of selenium in ovaries
after spawning, presumably due to the loss of selenium through spawning and release of eggs with
relatively high concentrations of selenium. Of the 14 chronic values determined from the maternal
transfer reproductive studies, 11 values represent selenium measured in eggs. Three values represent
selenium measured in the ovaries: Schultz and Hermanutz (1990), Hermanutz et al. (1992, 1996) and
Carolina Power & Light (1997). However, information in two of these studies indicates that the
concentrations of selenium in the ovaries were similar to concentrations in eggs. Schultz and Hermanutz
(1990) measured selenium in fathead minnow ovaries at the end of the study from fish that presumably
had spawned. The authors found the concentrations of selenium in the ovaries and embryos of the
fathead minnows exposed to the same treatments to be similar. Hermanutz et al. (1992, 1996) sampled
adult female bluegill just prior to spawning and at the end of the test (post spawning) and found no
decreases in the concentration of selenium in the post-spawned fish. In the Carolina Power & Light
(1997) study, the third study, selenium in ovaries of largemouth bass (Carolina Power & Light 1997) was
measured from fish sampled just after spawning. No comparison to pre-spawning fish or selenium in
eggs can be made for the largemouth bass study, however, the ECio of 26.3 mg Se/kg ovary dw was mid-
range of sensitivity indicating this test was not overly conservative due to lower selenium measurements
in post spawning ovaries. Based on the indications that the selenium concentrations in the ovaries were
similar to that in eggs in the Schultz and Hermanutz (1990) and Hermanutz et al (1992, 1996) study
observations stated above, egg selenium and ovary selenium were considered equal for the toxicity data
set. Any potential error resulting from this assumption would be conservative since the effect of
spawning only lowers the selenium concentration in the ovary. EPA recognizes selenium ovary
concentrations may vary in field collected samples due to fish reproductive cycles and will address such
concerns in the implementation information.

The Conley (2014) reference has been reviewed and considered. The agency is aware that spawning
strategy likely plays a role in maternal transfer and deposition of selenium into the eggs. The issue raised
in Conley 2014 will be addressed in the technical support materials under development regarding the
timing and type of sampling (egg-ovary vs whole body vs muscle tissue) based on species spawning
strategy (single batch versus multiple smaller batches) and seasonality (spring [e.g., sunfish, some
salmonids] or fall [e.g. brown trout] versus sporadic spring/summer [e.g., cyprinids]), while also
providing flexibility to states to adapt their monitoring programs to address the additional requirements
for fish tissue monitoring for selenium.

-------
2.3 Specific comments

EPA-HQ-OW-2004-0019-0403-A2; Iowa Department of Natural Resources; Posted 10/15/15

EPA states that all the criteria elements except for the 30-day average should be implemented as never
to exceed values: only one exceedance will impair the waterbody. The basis for this approach is not well-
justified, deviates from all previous criteria recommendations, and is not well-suited to the
implementation of a tissue-based criterion. For example, due to between-year variability in levels of
toxic contaminants in fish, the fish consumption advisory protocol used by Iowa DNR specifies that two
consecutive samplings are needed that show contaminant levels above the advisory threshold before a
fish consumption advisory is issued. This confirmation-type approach is used to ensure, to a reasonable
degree, that levels of a toxic contaminant are, in fact, above the consumption advisory threshold.
Without such an approach, advisories could easily be based on sampling error associated with year-to-
year variability in fish tissue contaminant concentrations. EPA should consider allowing states to adopt a
similar approach for implementing tissue-based aquatic life criteria for selenium where impairment
would be identified if the tissue-based criteria were exceeded more than once in three years.

Iowa DNR questions the expression of the tissue-based criteria elements as mg/kg drj/ weight.
Historically and currently, all of Iowa's fish tissue data have been reported from laboratories as mg/kg
wet weight. Most of Iowa's historical fish tissue data came from the U.S. EPA Region 7 lab and were
reported as wet weight values. If EPA plans to recommend adoption of dry weight concentrations of
selenium, then EPA should provide a cross-walk of dry-weight to wet-weight tissue concentrations of
selenium.

EPA-HQ-OW-2004-0019-0397-A1; Peabody Energy; Posted 10/15/15

Due to natural variability fish tissue samples should be averaged as opposed to being limited by
individual samples

EPA studies and other research indicate that there will be natural variability in tissue samples. Single
samples will not represent the environmental condition with precision. Using the "seasonal average"
instead of "instantaneous" limits will provide a more accurate tissue criterion requirement.

Peabody disagrees with applying only the proposed water column concentration where a new source
exists

Applying the water column concentration only until a new source has been active for several months
may result in unwarranted violations. For NPDES applications, a report only requirement along with a
reasonable potential analysis would be appropriate for a limited period of time after discharge is
initiated. Once equalization is determined to have occurred the fish tissue and egg tissue should apply if
the water column limit is exceeded. In fact, EPA leadership recently stated in a stakeholder meeting that
the fish tissue standard should be applied as opposed to the guidance language.

-------
Peabody disagrees with applying only the proposed water column concentration to fish less streams

It is unreasonable to apply a water column concentration designed to protect fish in streams that do not
have fish. We suggest that a better approach is to test fish tissue downstream, where fish are first
present, to determine if a stream segment is in attainment. Macroinvertebrates are less sensitive to Se
as compared to fish. If there is a concern for macroinvertebrates, site specific limits should be developed
based on the aquatic life that is present. Recent statements by EPA leadership in a stakeholder meeting
indicate that states will be given the necessary latitude to test fish tissue downstream or to develop
standards based on protecting macroinvertebrates. These options should be clearly reflected in the
guidance document.

EPA-HQ-OW-2004-0019-0395-A1; CONSOL Energy; Posted 10/15/15

For tissue criteria frequency, the use of term "never to be exceeded" is neither appropriate nor in-line
with standard criteria attainment requirements. Clarification of tissue sampling requirements and use of
an alternative approach such as the geometric mean of samples collected, with an allowable exceedance
frequency of no more than once every three years on average is needed. This would be in-line with
other criteria.

CONSOL believes that the establishment of criteria for streams with no existing fish populations needs
to be clarified. We disagree with the default approach to use water-column criteria. A more scientifically
rigorous alternative would be the use of the chronic invertebrate data provided by EPA to develop
invertebrate tissue-based site-specific criterion and associated protective water-column criteria.

EPA contends that the water column elements should have primacy in "fishless waters" and in waters
with new or increased selenium inputs until equilibrium is reached. These decisions should be handled
on a case by case basis with final authority given to the individual state authority.

EPA-HQ-OW-2004-0019-0390-A1; West Virginia Coal Association (WVCA); Posted 10/14/15

The expression of the fish tissue numbers in the Draft Selenium Criterion as concentrations that are
"never to be exceeded" is not plausible for implementation. This implies that a single fish with a
concentration above the criteria would require 303(d) listing of the receiving stream. The criteria should
be revised to include better language to clarify EPA's intent for the frequency.

WVCA does not agree with EPA's footnotes setting primacy of the water column elements in certain
circumstances. EPA contends that the water column elements should have primacy in "fishless waters"
(waters where fish have been extirpated, or where physical habitat and/or flow regime cannot sustain
fish populations), and in waters with new or increased selenium inputs until equilibrium is reached. First
and foremost, these are not actual elements of the criteria, but instead are efforts to incorporate
implementation determinations directly into the chronic criterion. These decisions cannot be made
hoiistically and must be evaluated on a case by case basis. States are given the authority to make
permitting and stream listing decisions, which consider the specific facts of each circumstance to make
the best decisions for the waters of the State. EPA cannot circumvent this authority and rational
decision-making with footnotes to a recommended criterion.

-------
For fishless waters, particularly those where the flow or other elements of the physical habitat is
insufficient for fish populations, EPA's recommended approach is nonsensical. Certainly, fish are
protected where selenium concentrations in the headwaters do not exceed the concentrations required
in the segments where fish are present. An assumption that segments without fish must have selenium
concentrations lower than segments with fish defies logic.

EPA attempts to reach too far with its "guidance" on implementation of the selenium criteria. These
decisions must be made based on each specific circumstance. If EPA disagrees with a permitting or
stream listing decision, we have no doubt that EPA will exercise its authority to express its concerns to
the State.

EPA-HQ-0W-2004-0019-0414-A2; Colorado Mining Association (CMA); Posted 10/28/2015

CMA disagrees with the use of the instantaneous measurement for fish tissue compliance

Fish tissue concentrations show variability due to a number of factors including, but not limited to,
analytic methods, laboratory procedures, sampling procedures, fish species, seasonal variability, and
other environmental characteristics. Because of this variability, CMA believes a more appropriate
comparison to the fish tissue criteria would use the average of the fish tissue measurements, or some
other statistically derived value, rather than the instantaneous measurement.

CMA disagrees with EPA's application of water column concentrations

EPA proposes to apply the water columns concentrations under two scenarios: 1) fishless streams, and
2) new discharges. On fishless streams, it would be inappropriate to apply the water column
concentrations. The water column concentrations were derived to protect fish from reproductive effects
of selenium. If there are no fish in the stream, the water column concentrations are overprotective for
the resident aquatic life which likely consists predominantly of macroinvertebrates. Rather, EPA should
allow the option of 1) fish tissue testing at a point further downstream where fish can be found, or 2)
development of site-specific criteria protective of macroinvertebrates, to determine compliance.

For new discharges, applying the water column concentrations for an initial period of time while the
stream biota assimilates to the new source discharge can result in needless violations. Instead, the EPA
should allow application of report only limits to the new discharge until a reasonable potential
assessment can be completed. If selenium limits are determined to be appropriate, then a water column
concentration trigger and fish tissue based limit can be applied.

EPA-HQ-OW-2004-0019-0385-A2; Kentucky Division of Water (KDOW); Posted 10/13/2015

KDOW believes that footnote 5 is not entirely accurate and this issue warrants more thorough
explanation in the document. The statement: "selenium concentrations in fish tissue are expected to
change only gradually over time in response to environmental fluctuations" does not reflect the
variations of selenium concentrations in fish tissue resulting from reproduction and depuration.

-------
Delete footnote 5 (July 27, 2015 80 FR No. 143 at 44352) as this is not entirely accurate and should
otherwise be addressed more appropriately elsewhere in the document.

Exceptions to Tissue-Based Criterion

KDOW has several serious concerns regarding footnote 3 in Table 1. EPA does not define, by magnitude
of concentration, frequency, or by duration, the discharges that should be considered "new or increased
inputs of selenium." There are countless scenarios that create implementation challenges for states. EPA
also does not define, by aquatic conditions or by duration, the term "equilibrium." Attempting to define
and defend when a stream is in equilibrium or steady-state in order to determine when and what
criterion should apply creates a legally and technically unworkable, and unnecessary, requirement for
states. Without such terms being clearly defined, states cannot determine in a scientifically defensible
manner when such "new or increased inputs" should cause water column values to have primacy over
fish tissue values. A water quality standard, or a permit condition based on a water quality standard,
that relied upon such "new or increased inputs" and "equilibrium" in order to determine when water
column values would have primacy over fish tissue values would create legal jeopardy for states
implement this requirement. As such, a standard or permit relying on this approach as proposed would
routinely be found as arbitrary and capricious and thus would likely be invalidated.

Essentially these same site-specific concerns were identified during external peer review of the draft
criterion. Dr. Kevin V. Brix commented: "[t]he issue of generating pulse loads of [selenium] that may
ultimately result in [selenium] accumulation in sensitive downstream systems (e.g., pulse loads in a river
that discharges to a wetland) is a legitimate concern. However, in my opinion, this is a site-specific issue
and it is not reasonable to establish national WQC that ensure protection of these sites... "

KDOW believes that the "Fishless Waters" language in footnote 3 is unnecessary given that footnotes 1
and 2 address the order of criterion element primacy. Furthermore, use of "(waters where fish have
been extirpated, or where physical habitat and/or flow regime cannot sustain fish)" is unnecessary and
inadvisably seems to establish and thereby limit the conditions by which there would be a lack of
available fish tissue, which KDOW does not believe is EPA's intention.

EPA-HQ-OW-2004-0019-0413-A2; Federal Water Quality Coalition (FWQC); Posted
10/28/2015

I. NEW ISSUES IN 2015 NOTICE

In the Selenium Notice, there are two issues, not raised by the 2014 draft document, on which EPA
seeks comment. Since those issues are new, we will address them first, before proceeding to other
substantive issues.

As noted above, EPA has applied a general principle, in the Draft Criteria Document, that fish tissue
levels of selenium take precedence over water column levels in determining whether the new criteria
(and, by extension, the aquatic life designated use) are being attained. We agree strongly with that
principle.1 For that reason, we are concerned with the new exception to that principle that EPA suggests
in the Notice. The Agency indicates that in situations where there are "new or increased selenium
inputs," EPA recommends that fish tissue levels should not take precedence over water column levels,
until "sufficient time has passed" to allow equilibrium to be attained in the food web. We disagree with

-------
the broad adoption of this new policy, for several reasons. First, there is already a regulatory/permitting
mechanism to address potential effects of new and increased discharges: the antidegradation program.
Second, some new or increased inputs could be very small (i.e., de minimis), giving rise to no concern
over potential effects on the aquatic system. Moreover, as a scientific matter, while there are some
situations in which there may be lag time between new discharges and their effects, there are also
situations in which fish tissue levels have been seen to respond rapidly to increases (or decreases) in
selenium loadings. If there is a concern in a particular situation, which cannot be addressed through
antidegradation review, then development of a site-specific criterion could be considered. But that is
not a reason to apply a broad principle to all cases where there are new or increased inputs.2 EPA's "new
or increased" exception to "fish tissue precedence" should not be adopted.

1 As discussed below, we believe that EPA needs to explain, in more detail, what it means when it states that the
fish tissue levels "take precedence" over water column levels.

2 Also, EPA's suggested policy, in which water column levels would take precedence for "a few years," until
equilibrium is reached, creates substantial permitting problems. A scenario could readily occur in which initial
permit limits are imposed back on a "lack of equilibrium." Then, several years later, when equilibrium has been
established, the discharger would be entitled to ask that the limits be revised, to reflect precedence of fish tissue
data over the water column levels. However, if any compliance schedule that was included in the permit has, by
then, expired, then anti-backsliding restrictions could apply, and the permittee could be unable to obtain the
revised permit limits that would be justified by the updated scientific evidence. That result should be avoided.

1. Format of Fish Tissue Criteria as "Never to be Exceeded" Numbers

As part of the criteria process, EPA has defined the frequency element by stating that the new fish tissue
criteria are "never to be exceeded." We believe that this frequency specification is fundamentally
inconsistent with the nature of the effects being addressed, as well as EPA precedents in similar
situations. In the Draft Criteria Document, EPA recognizes that uptake and depuration of selenium are
usually gradual. But that would justify a long-term average form for the criteria, so EPA's use of an
instantaneous form ("never to be exceeded") contradicts the Agency's own scientific conclusions. It also
does not consider basic facts about fish: the pollutant content in individual fish, whether for selenium,
mercury or another parameter, can vary enormously. Therefore, it is critical that fish tissue data be
looked at across a set offish, representative of the species collected in a waterbody. To apply a "never
to be exceeded" test to a single fish, collected at one site, and to extrapolate that data to make
attainment decisions as to a water body, would be arbitrary and capricious in the extreme. It would also
be inconsistent with the Agency's own guidance for other fish tissue criteria- specifically, for mercury.
There, EPA guidance makes clear that review of average tissue levels, looked at across several species
representing at least two trophic levels, is the appropriate method of determining whether the criteria
are met in the waterbody. A similar policy should be followed for the new selenium criteria.

V. USE OF A "NEVER TO BE EXCEEDED" FISH TISSUE VALUE

While there is general agreement among the peer reviewers that EPA acted properly in establishing fish
tissue data as the primary element of the new criterion, there were concerns voiced over the Agency's
statement of that element as a "never to be exceeded" value. EPA provides no clear explanation of the

-------
rationale for that decision; while EPA then states that it will provide further detail in guidance, this is an
important issue that needs to be addressed now, before the criterion is issued in final form.

Here are the issues with the "never to be exceeded" value that the peer reviewers expressed:

"It is unclear however, how the USEPA will interpret the "never to be exceeded" criteria.

Biological variability, coupled with uncertainty regarding the residence of mobile fish species, will make it
likely that some fish in a given collection may exceed the guidelines. It is unclear if a result from one fish
(i.e. a single exceedance) will render a given management area in non-compliance, or if some average
value is intended as the trigger." (p. 19)

"Although the EPA's AWQC, including the draft water Se criteria, are not to be exceeded more than once
in three years, the fish tissue-based Se criteria are "never to be exceeded." To my knowledge, the
"frequency" component of AWQC is rarely incorporated into permit limitations, so the implications offish
tissue-based Se criteria "never to be exceeded" are not entirely clear to me. The "frequency" component
was initially incorporated into AWQC based on the premise that ecosystems will not be harmed if the
number of criterion excursions is limited and/or there are compensating periods of time below the
criterion over which the ecosystem can recover. As far as I can tell, the draft AWQC document for Se does
not explain the basis for the "never to be exceeded" frequency decision for fish tissue. It seems that there
should be some level of consistency between the allowable "frequency" for fish tissue-based and water-
based Se criteria." (p. 44)

EPA addresses these concerns in two ways. First, it indicates that guidance will be issued:

"Regarding the interpretation of the "never to be exceeded" criteria, this is an issue that will be
addressed in the technical support documents." (p. 18)

On a more substantive level, here is EPA's response:

"Regarding the criterion frequency, selenium is persistent, however, it may be sequestered in an
environmental compartment that precludes it from uptake in biota (e.g., burial in sediment). EPA's
AWQC are traditionally based on aqueous concentrations since toxicity for the chemicals for which
criteria have been developed is generally due to aqueous exposure. These concentrations can vary
significantly both short-term and long term, allowing organisms in the aquatic community to recover.
In contrast, selenium is a bioaccumulative pollutant, and fish tissue concentrations have been directly
correlated with the adverse effect. Also reductions in fish tissue concentrations occur slowly, even
after removal of a selenium source. Since exceedance of these concentrations is correlated with
adverse effect occurs EPA determined that these levels cannot be exceeded in order to be protective
of the aquatic community. Technical support information regarding fish tissue sampling issue is being
developed by EPA."

Unfortunately, this response does not answer the question at issue. Exceedance of the concentration
specified in a criterion is always connected with an adverse effect-that says nothing about what
frequency should be specified in the criterion. Is EPA saying that if one fish from one site is found to
possess a tissue level of selenium higher than the criterion, the entire waterbody must be classified as in
nonattainment? And if that is so, does that mean that if, on the next sampling event, a fish is found that

-------
has less than the criterion value of selenium, the waterbody is no longer in nonattainment? Surely that
is not what EPA means - but by specifying that the criterion is "never to be exceeded/' that is a reading
of the EPA decision that some will take. And even guidance can only have a limited scope in this
circumstance - the guidance will have to explain what "never" means. If the Agency truly wants to
examine these issues in a fuller way, and to do so through issuance of guidance, then it should eliminate
the "never to be exceeded" language from the criterion structure, and leave the frequency issue to be
worked out in the technical guidance to come.

EPA-HQ-OW-2004-0019-0415-A1; Idaho Department of Environmental Quality (DEQ); Posted
10/28/2015

Does "never to be exceeded in fish tissue" really mean not even in one fish sample, ever? Or is some
averaging allowed, intended, possible? Could there be compositing of samples for a species from a site
for example, as a way to reduce costs as well as smooth out some variability? If just one sample
exceedance will bust the criterion, then how are analytical outliers, or a wandering individual to be
taken into account? A better discussion on the relation of individual effect thresholds to population
effects, including statistical foundation, is needed to understand the monitoring and assessment
implications of the "never to be exceeded in fish tissue" statement. EPA should address directly and
clearly in revisions to this criterion proposal issues of measurement variability as well as environmental
variability, to arrive at an allowable frequency of exceedance; the never to be exceeded requirement
should be dropped.

EPA-HQ-0W-2004-0019-0401-A1; Colorado Wastewater Utility Council; Posted 10/15/2015
3.5 Recommendations for "Never to be Exceeded" Frequency

Although it is inherent in the way this document was written that the burden to determine how a tissue-
based standard will be implemented will be left to the States, we would like to comment on the
"instantaneous" and "never to be exceeded" language that is presented in relation to the proposed
tissue standard.

Generally, the term "never to be exceeded" requires substantive clarification to make clear what is
expected - an issue that cannot wait for some future undetermined "implementation guidance"
document. The way the language is written currently suggests that a single fish tissue sample with a
concentration above the criteria (egg/ovary, whole-body, or muscle) would result in non-attainment - a
result that is directly contradictory to the data used to develop the criterion. However, other
alternatives could be employed (as discussed in our previous review) to ensure that a single fish would
not be used to determine attainment. We also support the very detailed analysis of this issue contained
in last year's comments by NAMC-SWG (2014).

-------
3.6 Recommendation for Fish less Waters

In cases where fish populations are not present due to flow limitations the aquatic life to be protected
are the macrobenthos. In EPA's analysis of invertebrate data, the mayfly, Centroptilum triangulifer, was
found to be the most sensitive, with a GMCV of 24.2 mg Se/kg dw wb. Effect concentrations for the
other invertebrates for which chronic data were available were substantially higher than 24.2 mg/kg.
Therefore, based on these data, an invertebrate Se tissue concentration of 24.2 mg/kg dw would be
protective of the attainable uses of those fishless waters, i.e., invertebrates.

The key issue is to ensure protection of downstream waters through evaluation of fish tissue attainment
measurements. If downstream receiving waters are in attainment with water quality standards (based
on tissue-based criteria) it is not necessary to protect non-existent fish in the upstream waters.

However, it may be appropriate to develop site-specific standards to protect other aquatic life in these
fishless streams.

5.1 Other Considerations

As stated in our review of the 2014 criteria document (GEI 2014b), we agree that any fish tissue
collected is representative of accumulation over time and could be considered an "instantaneous"
measurement of the current conditions. However, we believe the use of the term "instantaneous" leads
to substantial confusion regarding the tissue sampling requirements. We support the thorough
discussion of this topic in the NAMC-SWG 2014 review and agree that there will be natural variability in
the tissue samples collected, and a single sample may be over- or under-representative of site
conditions. The best way to represent environmental conditions is to average the tissue samples in some
manner. We recommend replacing "instantaneous" with "seasonal average" in the tissue criterion
requirements.

Regarding implementation of the tissue-based criteria, the use of "never to be exceeded" frequency is
inappropriate and not in line with standard criteria attainment requirements. We recommend
clarification of tissue sampling requirements and use of an alternative approach such as the geometric
mean of samples collected, with an allowable exceedance frequency of no more than once every 3 years
on average.

There are several other components of the criteria that should be reevaluated by the EPA. The use of
"never to be exceeded" frequency for the tissue criteria is inappropriate and not in line with standard
criteria attainment requirements. We recommend clarification of tissue sampling requirements and use
of an alternative approach such as the geometric mean of samples collected, with an allowable
exceedance frequency of no more than once every 3 years on average - consistent with other criteria.
Another issue that requires reevaluation is how to establish criteria for streams with no existing fish
populations. The default by EPA is to use water-column criteria; however, we provide discussion of an
alternative involving use of the chronic invertebrate data provided in the EPA document.

Overall, this document is a substantial improvement over pre-2014 Se criteria documents, and we look
forward to the final draft document, with the hope our recommendations are taken into consideration.

We also have general concerns with the language in Footnote 3 of the criteria table. This footnote
states:

-------
Water column values are based on dissolved total selenium (includes all oxidation states, i.e.,
selenite, selenite, organic selenium and any other forms) in water. Water column values have
primacy over fish tissue values under two circumstances: (1) "Fishless waters" (waters where
fish have been extirpated, or where physical habitat and/or flow regime cannot sustain fish);
and (2) New or increased inputs of selenium until equilibrium is reached.

Our concerns are how this language will potentially be used by state agencies and their (NPDES permit
writers. In the case of new or expanded discharges, the permit writer would be required to put a water
quality-based effluent limit (WQBEL) in the permit based on the proposed water quality criterion during
the time period it takes for the stream to reach "equilibrium", which may take months or even years.
When equilibrium is reached, and a fish tissue concentration is deemed appropriate, the limit derived
from the fish tissue concentration-based effluent limit may likely be higher than the existing limit based
on the water column value. As a result of EPA's anti-backsliding and antidegradation provisions, state
agencies and permit writers generally do not incorporate WQBELs in permit renewals that are less
stringent that those that were previously issued. Therefore, it is likely that the fish tissue-based effluent
limit will not be permissible after a water column limit is already in place.

This same concern could also apply to existing dischargers, which currently have water column criteria in
place with limited or no existing fish tissue data. During the permit renewal process implementation of
fish tissue-based limits could be complicated due to the anti-backsliding antidegradation provisions
unless an exception in included in any selenium implementation guidance that EPA issues. Another
approach to eliminating this issue would be to implement a tiered approach such as that approved in
Kentucky, in which the water column values are considered "thresholds" rather than "criteria".

Another issue that requires reevaluation and clarification is how to establish criteria for streams with no
existing fish populations. The default approach from EPA is to use water-column criteria; however, an
alternative would be the use of the chronic invertebrate data provided by EPA to develop invertebrate
tissue-based site-specific criterion.

Another topic that needs further consideration is the use of natural background Se concentrations to
develop ambient based site-specific criteria. We have provided discussion of how this has been
successfully done in Colorado in our previous review (GEI 2014a), and how it should be considered on a
case-by-case basis nationwide. EPA's discussion of site-specific standard development is lacking any
discussion of naturally elevated selenium in water and fish tissues and needs further clarification.

EPA-HQ-0W-2004-0019-0400-A1; The National Mining Association (NMA); Posted 10/15/2015
2.3.2 Protection Goal and High Incidence of False Alarms

The duration and frequency for the tissue criteria could be interpreted to mean that no individual fish in
a water body is ever allowed to exceed the tissue criterion. Given that the goal is to protect populations
of fish, not individual fish, we request that EPA clarify that the protection goal is, in fact, fish
populations, not individual fish. Therefore, the criteria should be applied as geometric means for a
sampled population, not individual fish.

-------
We are also concerned that the tissue criterion, inappropriately expressed as "never to be exceeded" on
an instantaneous basis, will result in a high incidence of false alarms that will lower national and state
abilities to address genuine environmental problems. Such a specification is unprecedented for EPA
aquatic life criteria and EPA has presented no rationale for this recommendation. Nor has EPA explained
how it should be interpreted or implemented. Because EPA's permit program incorporates assumptions
of lognormal concentration distributions in its permit derivations, we are unable to understand how
either the permit writer or the discharger can design for a criterion that is "never to be exceeded," since
lognormal distributions have no concentration that is never exceeded.

We point out that an EPA requirement to "never exceed a tissue criterion" has significant implications in
implementation and engineering design of water treatment facilities. A "never to exceed" criterion
means that industrial and municipal water treatment facilities must over-design either in treatment
capability or water holding capacity to avoid never having a tissue exceedance. This effectively lowers
the selenium tissue criterion to a much lower concentration than proposed in the current document to
ensure there is never an exceedance. We do not have the data at present to estimate the impact, but it
would be significant. A "never to exceed" is unrealistic, technically indefensible, and unworkable.

Accordingly, we recommend that, in accord with EPA Guidelines, "instantaneous" be replaced with
"seasonal average," and "never to be exceeded" be replaced with "not to be exceeded more than once
in three years on average" applicable to the seasonal average concentration.

As noted, NAMC is extremely concerned by an expected high rate of false alarms. State pollution control
agency budgets are fixed. Each time there is a false alarm, state government resources that had been
allocated to solving genuine environmental problems must be reallocated to resolve the false alarm,
even if that only means overseeing a tissue monitoring study performed by the discharger and arriving
at a conclusion after examining the results. Because we are supportive of the mission of the state
pollution control agencies and of EPA, we are concerned by the fact that setting the criteria
inappropriately low will result in a serious misallocation of resources, thereby reducing rather than
enhancing the nation's ability to address environmental problems.

3.5 Recommendations for "Never to be Exceeded" Frequency

Generally, the term "never to be exceeded" requires substantive clarification to make clear what is
expected - an issue that cannot wait for some future undetermined "implementation guidance"
document. The way the language is written currently suggests that a single fish tissue sample with a
concentration above the criteria (egg/ovary, whole-body, or muscle) would result in non-attainment - a
result that is in direct contradiction to the data used to develop the criterion, which are based on ECi0s
calculated from means of treatments. However, other alternatives could be employed (as discussed in
our previous review) to ensure that a single fish would not be used to determine attainment. We also
support the very detailed analysis of this issue contained in last year's comments by NAMC-SWG (2014).

-------
3.6 Concerns with Primacy of Water Column Values

We also have serious concerns with Footnote 3 of the proposed criterion. Footnote 3 outlines two
scenarios in which water column values have primacy over fish tissue values - "fishless waters" and
waters with "new or increased inputs of selenium until equilibrium is reached." Specifically, Footnote 3
states:

Water column values are based on dissolved total selenium (includes all oxidation states, i.e.,
selenite, selenate, organic selenium and any other forms) in water. Water column values have
primacy over fish tissue values under two circumstances: (1) "Fishless waters" (waters where
fish have been extirpated, or where physical habitat and/or flow regime cannot sustain fish);
and (2) New or increased inputs of selenium until equilibrium is reached.

While we understand the concerns EPA is seeking to address with respect to these two scenarios, we
think that the proposed language, when taken in conjunction with the anti-backsliding provisions of the
CWA, the requirements of the National Pollutant Discharge Elimination System (NPDES) permitting
process for establishing water quality-based effluent limits (WQBELs), and state antidegradation
regulations could in some instances effectively nullify the fish tissue approach all together. While EPA
considers this an "implementation" issue, it is in fact the language of the criterion itself that will cause
the issue and it cannot be rectified with implementation guidance. However, we think that there are
ways to address EPA's intent in the criterion document that will allow state programs to maintain the
primacy of the fish tissue values in permits without compromising the quality of downstream waters or
waters with new or expanded selenium discharges.

The CWA's anti-backsliding provision provides that "a permit may not be renewed, reissued, or
modified...to contain effluent limitations which are less stringent than the comparable effluent
limitations in the previous permit." CWA Sec. 402(o). While there are listed exceptions to the anti-
backsliding requirements, many state agencies and their NPDES permit writers construe these
exceptions narrowly and generally interpret anti-backsliding to mean that permits cannot be modified or
renewed with less stringent WQBELs than those contained in the previous version. Additionally, certain
anti-backsliding exceptions for WQBELs are tied to both attainment and antidegradation regulations,
which may cause additional issues with their application in some state programs.

As written, with respect to new or expanded discharges, the criterion would require a permit writer to
put WQBELs based on the proposed water column criteria into NPDES permits during the months or
even years that it takes for the receiving water to reach "equilibrium."1

Due to the conservative estimates used to derive the water column concentrations, it is possible that in
many instances a post-equilibrium discharge limit derived from a fish tissue concentration-based
effluent limit will be higher than one based on the water column value. However, because a discharger's
permit will already contain the more stringent limitation, even after equilibrium is reached a discharger
may only be allowed to have a limitation based on a fish tissue concentration put into their permit if
that limitation would be more stringent than the one based on the water column value. In other words,
there will likely be instances in which a fish tissue-based effluent limitation will not be permissible if
states adopt the criterion as proposed despite the fact that such a limitation would represent the best
available science and reflect the EPA's intent as we understand it. This same analysis applies to existing
discharges where states may choose to first apply a WQBEL based on the proposed water column

-------
criterion where data regarding fish tissue and fish egg ovary selenium concentrations may be scarce, not
collected during permit renewal, or otherwise unavailable at the time of permit renewal.

While for some states programs this may not pose an issue (and we support such states' interpretation
of anti-backsliding), and for others implementation guidance may be able to address this issue by fitting
the criterion into one of the anti-backsliding exceptions, it is unclear if that will be legally possible, and
regardless it would not matter because as a practical matter dischargers would have already had to
install control technologies to treat to the more stringent water column-based limit despite the fact that
such controls are not necessary to protect aquatic life. In more extreme cases, dischargers may not be
able to get a permit at all based on Footnote 3 due to elevated natural background levels of selenium,
despite the fact that certain levels of selenium could be discharged safely as demonstrated by the fish
tissue concentrations, because permit writers would be inappropriately basing their calculations on the
water column number. Because of this, we make two recommendations. The first is deleting Footnote 3
as written, and instead including a provision for new or expanded discharges whereby limitations are
derived from fish tissue concentrations but compliance cannot be tested until the appropriate amount
of time has elapsed for the receiving water to reach equilibrium (i.e., a bioaccumulation study special
condition requirement). The second is renaming the water column "criteria" as water column
"thresholds," as was done in Kentucky's recently approved selenium water quality criteria revision. This
would clarify EPA's intent to allow fish tissue concentrations to have primacy over the water column
values.

With respect to "fishless waters," Footnote 3 will likewise lead to situations in which dischargers must
install expensive control technologies or not be able to discharge selenium at all despite the fact that
such requirements are not necessary to protect aquatic life. We understand that EPA is required under
the CWA to protect downstream waters when deriving water quality criteria. However, a scientifically
defensible approach to "fishless waters" that EPA should provide to states as an option in the criterion is
the derivation of a site-specific criterion that takes into account the sensitivity of the macrobenthos in
the receiving water to protect the receiving water's attainable uses (i.e., to protect the receiving water's
macrobenthic communities), while protecting downstream waters through determination of the flow
rate of the receiving water into the nearest fish-containing water and ensuring that the fish tissue
standard is being met in the downstream fish populations. Such a criterion would ensure that both in-
stream and downstream aquatic life are protected without arbitrarily assigning overly-protective water
column-based limitations to permittees. We therefore again recommend deleting Footnote 3 as written,
and instead including a provision for "fishless waters" that allows for the development of site-specific
criteria that protect the attainable use of in-stream macrobenthos as well as downstream fish
populations through application of the tissue-based criterion (see additional discussion in Section 3.6.1).

Concerning a related issue, we also note that EPA guidance limits the terms of compliance schedules to
five years, as do many state permitting programs based on EPA's guidance. Because in some situations
where selenium levels are being decreased based on the new criterion it may take longer than 5 years
for corresponding fish tissue concentrations to decrease, NMA recommends including a provision in the
criterion stating that longer compliance schedules may be needed with respect to bioaccumulants.
Although again this could be considered an "implementation" issue, because it is a facet of the criterion
that could directly conflict with existing state water quality programs - thereby limiting states' ability to
adopt it - it is necessary to adopt language in the criterion itself that addresses this concern.

-------
5.1 Other Considerations

As stated in our review of the 2014 criteria document (GEI 2014b), we agree that any fish tissue
collected is representative of accumulation over time and could be considered an "instantaneous"
measurement of the current conditions. However, we believe the use of the term "instantaneous" leads
to some confusion regarding the tissue sampling requirements. We support the thorough discussion of
this topic in the NAMC-SWG 2014 review and agree that there will be natural variability in the tissue
samples collected, and a single sample may be over- or under-representative of site conditions. The best
way to represent environmental conditions is to average the tissue samples in some manner. We
recommend replacing "instantaneous" with "seasonal average" in the tissue criterion requirements.

Based on the results of our analysis, we recommend updating the proposed egg/ovary, whole-body, and
muscle chronic criteria to include our suggested changes (Table 8). In addition, because nationwide
water column-based criteria cannot not be derived reliably, we advise EPA to only recommend water
column-based criteria be developed on a site/state/region-specific basis (Table 8).

There are several other components of the criteria that should be reevaluated by the EPA. The use of
"never to be exceeded" frequency for the tissue criteria is inappropriate and not in line with standard
criteria attainment requirements. We recommend clarification of tissue sampling requirements and use
of an alternative approach such as the geometric mean of samples collected, with an allowable
exceedance frequency of no more than once every three years on average - consistent with other
criteria. Another issue that requires reevaluation is how to establish criteria for streams with no existing
fish populations. The default by EPA is to use water-column criteria; however, we provide discussion of
an alternative involving use of the chronic invertebrate data provided in the EPA document.

EPA-HQ-0W-2004-0019-0389-A1; Thunder Basin Coal Company, LLC (TBCC); Posted 10/14/15
Frequencies of Exceedance

The "Never to be Exceeded" and "Not more than Once (Exceedance) in Three Years on Average" could
prove very difficult for regulators and permittees to work with. EPA needs to explain how single sample
exceedances, outliers, and false positives will be taken into account given the great variety of species
and environments in which the new standards will be applied. In addition, if averaging is allowed, then
EPA needs to better explain their rationale and procedures.

-------
EPA-HQ-OW-2004-0019-0372; Anonymous public comment; Posted 09/23/2015

Please provide scientific evidence that justifies not following the 1985 Guidelines in setting the 2015
draft Se tissue criterion duration at "instantaneous" and its frequency at "never".

The 1985 Guidelines assert: "A statement of a criterion as a number that is not to be exceeded any time
or place is not acceptable". In addition, "rather than try to reinterpret a [never-to-exceed] criterion that
is neither useful nor valid, it is better to develop a more appropriate way of stating criteria."

The Guidelines also say, "the concentrations of a pollutant in a body of water can be above the CCC
without causing an unacceptable affect if (a) the magnitudes and durations of the excursions above the
CCC are appropriately limited and (b) there are compensating periods of time during which the
concentration is below the CCC."

EPA's 2015 draft selenium document rationalizes its choice of instantaneous on the grounds that "fish
tissue data provide point, or instantaneous, measurements". It is not apparent that EPA understands
that tissue concentrations vary from fish to fish, and that a fish tissue sample collected at one instant
would be expected to differ from a nearby fish tissue sample collected at a different instant.

Given that the criterion statement implies that the occurrence of a single fish with tissue concentration
marginally above the criterion at any time, no matter how infrequent, can prevent the water body from
attaining its designated use as a fishery, please provide scientific evidence to support your duration and
frequency statements.

The "never" frequency implies that a system experiencing even a marginal exceedance can never
recover. In fact, the scientific evidence in Chapman 2010 (cited in the EPA's draft) indicates that systems
that once experienced long periods of exceedances several fold above the criterion have all since
recovered.

Furthermore, while it is understandable that there is a range of scientifically defensible approaches,
there appears to be an extreme difference between the 2015 draft fish tissue duration (instantaneous)
and frequency (never) versus the Utah selenium bird egg criterion, which EPA approved as the
geometric mean of samples, thus to be exceeded around 50% of the time. Please explain how this great
difference in acceptable approaches is rationalized.

Standards for Fishless Waters and Use of Nationwide Water Column Based Standards

Approximately 70 to 80 percent of Wyoming streams are ephemeral or intermittent without the natural
hydrologic potential to support fish populations (National Hydrologic Database). It is unreasonable to
apply a water column concentration designed to protect fish in streams that do not have fish. It is also
unreasonable to develop national standards to protect downstream fish species for all fishless waters;
many Wyoming streams lack hydrologic connection or are very distant from streams with fish. For
waters such as these, it is unnecessarily overprotective to apply such a national standard.

Waters with New or Increased Selenium Inputs

Applying the water column concentration only until a new source has been active for several months
may result in unwarranted violations. For NPDES applications, we recommend reporting and a

-------
reasonable site specific potential analysis for impacts on local aquatic life for a specified period of time
after discharge has begun.

EPA-HQ-0W-2004-0019-0416-A1; Wyoming Mining Association (WMA); Posted 11/02/2015
Frequencies of Exceedance

Standards for Fishless Waters and Use of Nationwide Water Column Based Standards

Approximately 70 to 80 percent of Wyoming streams are ephemeral or intermittent without the natural
hydrologic potential to support fish populations (National Hydrologic Database). It is unreasonable to
apply a water column concentration designed to protect fish in streams that do not have fish. It is also
unreasonable to develop national standards to protect downstream fish species for all fishless waters
because many Wyoming streams lack hydrologic connection or are very distant from streams with fish.
For waters such as these, it is unnecessarily overprotective to apply such a national standard. National
standards could result in costly and unnecessary water quality exceedances and require treatment as
well as unnecessary reductions in discharges if standards are not or cannot be met.

WMA believes that an acceptable option for some fishless waters would be to test macroinvertebrates
to develop site specific limits. Macroinvertebrates are less sensitive to selenium compared to fish, but as
other reviewers have commented, more data should be gathered for these species prior to
implementation of the standards.

Waters with New or Increased Selenium Inputs

Applying the water column concentration only until a new source has been active for several months
may result in unwarranted violations. For NPDES applications, a better approach would be to require
reporting and a reasonable site specific potential analysis for impacts on local aquatic life for a specified
period of time after discharge has begun.

EPA-HQ-OW-2004-0019-0402-A2; The Stakeholders Implementing TMDLs in the Calleguas
Creek Watershed; Posted 10/15/2015

1. Justification and Explanation for the Selection of Execeedance Frequencies is Absent and Must be
Provided in Order to Comment on the Appropriateness of Such Frequencies. However, Even Without
Any Provided Explanation or Justification, the Exceedance Frequency for the Fish Tissue Elements of
the Criterion of "Never to Be Exceeded" is Inappropriate, Impractical and Contrary to USEPA Guidance.

As previously noted in the Stakeholder's comments on the 2014 External Peer Review Draft Criterion, an
exceedance frequency of "not more than once in three years on average" is identified for the water-

-------
column elements of the criterion, yet no explanation or justification is provided to support this proposed
frequency. Therefore, it is not possible to comment on the appropriateness of this selection. Given that
this is such an important aspect of the criterion, it is unclear why the 2015 Draft Selenium Criterion was
not updated to provide the requested justification. Additional supporting documentation needs to be
provided with an additional opportunity to comment on this issue.

However, even without additional supporting explanation or justification, the exceedance frequency for
the fish tissue elements of the criterion of "never to be exceeded" is impractical, inappropriate, and
contrary to the guidance provided by the USEPA in the Guidelines for Deriving Numerical National Water
Quality Criteria for the Protection of Aquatic Organisms and Their Uses19 (1985 Guidance Document) and
the Technical Support Document for Water Quality-based Taxies Control20 (1991 TSD). In the 1991TSD,
USEPA explicitly states (emphasis added):

"To predict or ascertain the attainment of criteria it is necessary to specify the allowable
frequency for exceeding the criteria. This is because it is statistically impossible to project that
criteria will never be exceeded. As ecological communities are naturally subjected to a series of
stresses, the allowable frequency of pollutant stress may be set at a value that does not
significantly increase the frequency or severity of all stress combined. "

An exceedance frequency of "never to be exceeded" cannot reasonably or foreseeably be implemented
in the impairment assessment of waterbodies required under Section 303(d) of the Clean Water Act. By
the definition of "never to be exceeded," once a fish tissue exceedance is documented, there is no way
to make a valid statistical determination that the waterbody has improved or meets the criterion. The
2015 Draft Selenium Criterion creates an unreasonable and unnecessary requirement that can be
avoided by applying existing federal technical criteria guidance.

To address the shortcomings of the exceedance frequency of "never to be exceeded," one potential
alternative is the development of a statistical threshold value (STV), a concept described in USEPA's
nationally recommended Recreational Water Quality Criteria21 (RWQC) released by USEPA in 2012. The
STV approximates the 90th percentile of the water quality distribution and is intended to be a value that
should not be exceeded by more than 10 percent of the samples taken. This concept can be applied as a
frequency by stating that there should not be greater than a ten percent excursion frequency of the
selected STV magnitude.

Requested Actions:

• Revise the frequency for the fish tissue elements of the criterion from "never to be exceeded"
to an alternative frequency that allows a certain percentage of exceedances over a unit of
time.

• Provide the rationale for the exceedance frequency of "not more than once in three years on
average" for the water column elements and any other exceedance frequencies included in a
revised criterion (e.g., frequencies for the tissue-based elements).

SUMMARY OF KEY TECHNICAL ISSUES

-------
New Comments on the 2015 Draft Selenium (Addressing Modifications to the 2014 External Peer
Review Draft Criterion)

• For evaluating "fishless waters," surrogate biological parameters should be used in place of
water column values.

• Considerations for new or increased inputs should be included as implementation guidance and
not included directly as part of the criterion.

The Stakeholders have the following specific technical comments related to the summary of key issues
above:

NEW COMMENTS ON THE 2015 DRAFT SELENIUM (ADDRESSING MODIFICATIONS TO THE 2014
EXTERNAL PEER REVIEW DRAFT CRITERION)

1. Clarifying Language Added to Footnote 3 of Table 4.1 Is Appreciated.

Notwithstanding Comment #10 the Stakeholders appreciate the USEPA's effort to clarify the basis for
water column values in footnote 3 of Table 4.1 of the 2015 Draft Selenium Criterion by adding "(includes
all oxidation states, i.e., selenite, selenate, organic selenium and any other forms)."

2. Considerations for New or Increased Inputs Should Be Included as Implementation Guidance and
Not Included Directly as Part of the Criterion

As noted in Comment #2, the 2015 Draft Selenium Criterion includes two circumstances where water
column values have primacy over fish tissue values:

"1) 'Fishless waters' (waters where fish have been extirpated or where physical habitat and/or
flow regime cannot sustain fish populations); and 2) New (see glossary) or increased inputs of
selenium from a specific source, until equilibrium is reached."

The Stakeholders reaffirm that water column values should not have primacy over fish tissue values
under any circumstance, including both of these circumstances. This comment addresses the aspect of
establishing primacy of water column values over fish tissue values for new or increased inputs of
selenium from a specific source.

USEPA's intent behind including a mechanism to prevent new or increased inputs of selenium from a
specific source from causing increased bioaccumulation in fish over a period of time is understood.
However, this aspect addresses implementation of the criterion itself, and not whether or not
designated uses are being protected.

Therefore, it is not appropriate to include this mechanism as part of the criterion as it is more directly
related to implementation of the criterion (e.g., NPDES permits). Given that addressing these site-
specific situations seems to be more appropriately done through the NPDES permitting process for
individual dischargers, it seems that this mechanism would be much more appropriately incorporated as
an implementation tool for the criterion. Precedent for utilizing very specific implementation tools as
part of an objective has been established in the State of California's Water Quality Control Plan for
Enclosed Bays and Estuaries- Part 1 Sediment Quality4 (Phase I SQOs), approved by USEPA on August 25,
2009. Incorporating this mechanism, as well as the edits to the water column element as a whole as
described in Comment #5, as an implementation tool would give permit writers more flexibility to
address new or increased inputs and identify the water column concentration that is most appropriate
for a dischargers given situation. Reserving the water column concentrations as an implementation tool

-------
for the criterion would allow NPDES permits to individualize the water column concentrations needed to
protect or obtain the specific tissue concentrations, negating the need to attempt to incorporate such
implementation concerns in the criterion itself. By continuing to include the water column concentration
directly into the criterion itself, EPA is limiting the ability for States to address these site-specific issues
(such as an increased input or a new discharge).

Further, the concept of "increased inputs" is very vague and broad. While at the national level broad can
have benefits in providing flexibility to States for implementation, the term should be limited to
instances where such inputs have the expectation to negatively impact tissue concentrations such that
the designated use would not be protected.

Requested Actions:

• Revise the language in Table 4.1 of the 2015 Draft Selenium Criterion to read as follows:

o Modify Footnote 3: "Water column values are based on dissolved total selenium
(includes all oxidation states, i.e., selenite, selenate, organic selenium and any other
forms) in water. Water column values have primacy overfish tissue values under two
circumstances: 1) "Fishlcss waters" (waters where fish have been extirpated or where
physical habitat and/or flow regime cannot sustain fish populations); and 2) New (see
glossary) or increased inputs of selenium from a specific source7 should not cause or have
the reasonable potential to cause the fish tissue elements to be exceeded in any
downstream waterbody until when equilibrium is reached. This aspect shall be
implemented as described in [Insert Reference to section titled 'Implementation for
Assessing Protection of Fish Tissue When New or Increased Inputs of Selenium from a
Specific Source Are Present'l."

• Add a section to the 2015 Draft Selenium Criterion titled "Implementation for Assessing
Protection of Fish Tissue When New or Increased Inputs of Selenium from a Specific Source Are
Present". Move the following language from Page 93, Section 4 into this section with the
underlined modification:

New inputs are defined as new activities (see glossary) resulting in selenium being released into a lentic
or lotic waterbody. Increased input is defined as an increased discharge of selenium from a current
activity released into a lentic or lotic waterbody that has the potential or reasonable potential to result in
exceedances offish tissue concentrations in any downstream waterbody. New or increased inputs witt
likely have the potential to result in increased selenium in the food web, likely potentially resulting in
increased bioaccumulation of selenium in fish over a period of time until the new or increased selenium
release achieves a quasi-"steady state" balance within the food web. EPA estimates that concentrations
of selenium fish tissue will not represent a "steady state" for several months in lotic systems, and longer
time periods (e.g., 2 to 3 years) in lentic systems, dependent upon the hydrodynamics of a given system;
the location of the Se input related to the shape and internal circulation of the waterbody, particularly in
reservoirs with multiple riverine inputs; and the particular food web. Estimates of steady state under new
or increased selenium input situations are expected to be site dependent, so local information should be
used to better refine these estimates for a particular waterbody. Thus, EPA recommends that fish tissue

-------
concentration not override water column concentration until those periods of time have passed in lotic
and lentic systems, respectively when incorporating the criterion from a specific source into appropriate
regulatory mechanisms (e.g., NPDES permits), new or increased inputs that have the potential or
reasonable potential to cause exceedances offish tissue concentrations in any downstream water body,
water column concentrations that will result in attainment of the fish tissue concentrations should be
developed (or the site-specific situation and included in the applicable regulatory mechanism.

1. For Evaluating "Fishless Waters," Surrogate Biological Parameters Should Be Used In Place of Water
Column Values.

The 2015 Draft Selenium Criterion includes two circumstances where water column values have primacy
over fish tissue values:

However, water column values should not have primacy over fish tissue values under any circumstance,
including both of these circumstances. This comment addresses the aspect of establishing primacy of
water column values over fish tissue values for "fishless waters."

Page 92, Section 4 of the 2015 Draft Selenium Criterion states (emphasis added):

"The magnitudes of the water column elements are derived from the egg-ovary element
coupled with biogccumulgtion consideration."

As shown in Figure 2.4 of the 2015 Draft Selenium Criterion (presented on the following page for ease of
reference), the pathway for translating the concentration of selenium in eggs and ovaries to a
concentration of selenium in the water-column involves a number of steps. To proceed from one step
along the pathway to another, relational parameters (i.e., a species-specific Trophic Transfer Function
[TTF], an egg-ovary to whole-body conversion factor [CF], or a site-specific enrichment factor [EF]) must
be used. Given that these parameters are species-specific or site-specific, there are a range of values
which may be appropriate for each parameter. As a result, with each step that is taken from the egg-
ovary element to the water-column element, a greater degree of uncertainty is introduced. For this
reason, it is much more accurate to use a surrogate parameter, which represents a step along the
pathway closer to the egg-ovary element upon which all other elements are derived, as opposed to the
water column-element, which represents the end of the pathway, when trying to determine the health
of "fishless waters". Additionally, as stated on Page K-13, Appendix K, it is appropriate to use surrogate
parameters when evaluating the health of a waterbody (emphasis added):

"In addition, states and tribes may evaluate upstream waters without fish by measuring the
selenium concentration in water, biotic and/or abiotic particulate material, and/or the tissues
of invertebrate aquatic organisms that reside there."

USEPA's peer reviewer, Dr. Gregory A. Cutter, reminded USEPA of the Chapman et al (2009 and 2010)
recommendations from the SETAC Pellston workshop which specifically states, "Understanding Se
speciation is critical to understanding its mobility, transformation, partitioning in the environment, and
potential risk to aquatic ecosystems".2 The Stakeholders agree with Dr. Cutter's critique of USEPA's
translation from fish tissue to water column. Incorporating the very straightforward suggestions below

-------
to enable use of biological surrogates would go a long way towards addressing the Stakeholder's
concerns in this matter.

Original letter contains Figure 2.4 - Conceptual model for translating the selenium egg-ovary
concentration to a water column concentration. See original letter.

As a result, rather than defaulting to the water column element derived to protect fish when evaluating
"fishless waters", a surrogate parameter (e.g., tissue of invertebrate aquatic organisms, particulate
material) should be used when evaluating "fishless waters".

Requested Actions:

• Revise the 2015 Draft Selenium Criterion as follows to clearly establish that, if the water column
element continues to be a part of the 2015 Draft Selenium Criterion as more than an
implementation tool, surrogate parameters should be used in place of the water column element
when evaluating "fishless waters":
o Modify Table 4.1s as follows:

¦ Modify Footnote 1: "Overrides any whole-body, muscle, or water column
elements-when fish egg/ovary concentrations are measured except in certain in
situations. See footnote 3. For evaluating "fishless waters" (waters where fish
have been extirpated or where physical habitat and/or flow regime cannot
sustain fish populations), a surrogate parameter (e.g., tissue of invertebrate
aguatic organisms, particulate material) derived from the egg-ovary element
using a species-specific Trophic Transfer Factor fTTFl or a site-specific
enrichment factor [EF1 should be used."

¦ Modify Footnote 2: "Overrides any water column elements when both fish tissue
and water concentrations are measured except in certain in situations. See
footnote 3. For evaluating "fishless waters" (waters where fish have been
extirpated or where physical habitat and/or flow regime cannot sustain fish
populations), a surrogate parameter (e.g., tissue odnvertebrate aguatic
organisms, particulate material) derived from the egg-ovary element using a
species-specific Trophic Transfer Factor fTTFl or a site-specific enrichment factor
[EFj should be used."

Modify Footnote 3: "Water column values are based on dissolved total selenium (includes all oxidation
states, i.e., selenite, selenate, organic selenium and any other forms) in water. Water column values have
primacy overfish tissue values under the following circumstances: 1) "Fishless waters" (waters
where fish have been extirpated or where physical habitat and/or flow regime canncl sustain fish
populations); and 2) ... "

EPA-HQ-OW-2004-0019-0388-A2; J. R. Simplot Company; Posted 10/14/2015

A. Implementation Processes

A.l. Fishless Waters and "Increased" Inputs

[Page xiv, Executive Summary, Table 1]

-------
EPA implements the tissue criterion as an instantaneous measurement never to be exceeded.
Maintaining the fish tissue criterion as an instantaneous measurement implies that a single fish with a
concentration of selenium in its t issue exceeding the tissue threshold would be a violation of the
criterion. Much of the field data collected by numerous investigators has indicated that individual fish
will have varying levels of selenium as part of their body burden. While fish tissue may represent an
integrated exposure, each fish may integrate that exposure differently depending upon its history at a
site. While the criterion as it is written may imply the instantaneous measurement is meant to be
temporally instantaneous as alluded to in the footnote,1 it should be more specific about how tissue
selenium is measured, including the number of fish needed for a representative sample for a site. In the
ecological risk assessment process (EPA 1997, 1998) EPA expects that exposure point concentrations are
based on the average of multiple samples, represented by the 95th upper confidence limit (UCL) of the
arithmetic mean as a "high end" estimate of the average. It would seem practical that decisions about
criterion exceedance not rely on a single measurement, but on a representative exposure estimate.

There are two characteristics of the criterion that will have far reaching and likely unintended
conservative consequences. As noted in Table 1 of the 2015 Draft Criterion (page xiv), tissue data take
precedence over water concentrations, except when the following conditions are present:

1. "Fishless waters" (waters where fish have been extirpated or where physical habitat and/or flow
regime cannot sustain fish populations); and/or

2. New or increased inputs of selenium from a specific source until equilibrium is reached.

The 2015 EPA Selenium Draft defines fishless waters as waters with insufficient instream habitat and/or
flow to support a population of any fish species on a continuing basis, or waters that once supported
populations of one or more fish species but no longer support fish (i.e., extirpation) due to temporary or
permanent changes in water quality (i.e., due to selenium contamination), flow, or instream habitat.
Because of the inability to collect sufficient fish tissue to measure selenium concentrations in fish in such
waters, water column concentrations are prescribed as best representing selenium levels required to
protect aquatic communities and downstream waters in such areas.2

The above definition of the "process" for fishless waters is inconsistent with Appendix K of the External
Peer Review Draft (Section 1.2.1.2), where EPA states that "when fish are naturally absent from a
waterbody, states and tribes should target the most sensitive fish species inhabiting downstream
waters."3 Recommendations from Appendix K are consistent with current risk assessment methods in
that consideration of exposure is taken into account. Default application of an aqueous criterion based
on fish tissue to "fishless waters" will result in values at or below background for many stream systems,
and thus will certainly result in a large number of false positives with consequent dilution of efforts to
address real environmental problems.

The 2015 EPA Selenium Draft defines new inputs or increasing inputs as: "new activities resulting in
selenium being released into a lentic or lotic waterbody. Increased input is defined as an increased
discharge of selenium from a current activity released into a lentic or lotic waterbody. New or increased
inputs will result in increased selenium in the food web, likely resulting in increased bioaccumulation of
selenium in fish over a period of time until the new or increased selenium release achieves a quasi
"steady state" balance within the food web. EPA estimates that the concentration of selenium in fish

-------
tissue will not represent a "steady state" for several months in lotic systems ... EPA recommends that
fish tissue concentration not override water column concentration until these periods of time have
passed in lotic and lentic systems, respectively."4

This issue has a number of nuances and complications, such as establishing an initial permit limit for
selenium. Most likely, this initial permit limit will be "conservative" (i.e., more stringent than might be
using a fish tissue measurements to compare to the criterion). Once fish tissue monitoring and
comparison to the criterion can occur, an increase in the selenium limit may not be possible due to the
"antibacksliding" provision in the Clean Water Act. Also, the frequency of monitoring that a discharger
will have to perform will be higher than otherwise would be required.5 Finally, the discharger will have
had to invest in the control equipment necessary to meet the initial permit limit. Thus, considerable
resources will likely have been expended that were not necessary to protect aquatic life.

This new addition to the selenium criterion will be problematic for many regulated entities. By using the
language "new or increased" rather than "new and increased" EPA has automatically assumed that any
new inputs into a system will automatically increase selenium concentrations. How then does this affect
for example a treatment plant designed to reduce selenium? By the definition above, it is a new input.
Specific language needs to be included that excludes capture and reintroduction discharges that treat
and reduce selenium concentrations.

As defined above, increased input is an increased discharge of selenium from a current activity released
into a lentic or lotic waterbody. What is the time frame for "current activity"? More importantly, what is
the definition of "activity"? Would a spring discharge that is increasing be considered a current activity,
even if it has been present and discharging for hundreds of years? What is the time frame for
"increasing" for the determination of whether or not the "activity" is increasing or simply fluctuating
naturally?

As noted in comments below, EPA's action to implement a traditional aqueous criterion for selenium will
create unnecessary burdens on regulated entities due to an overly conservative "National Criterion" in
water that is inherently uncertain. Much like the methyl mercury criterion, the selenium criterion
document needs to identify a tissue threshold that is the basis for the criterion and provide methods for
regional application and derivation of an aqueous value.

We recommend that existing Footnote 3 in Table 1 be deleted and replaced with the following:
"Limitations are derived from fish tissue concentrations, but compliance cannot be fully tested until the
appropriate amount of time has elapsed for equilibrium in the receiving water for new and increased
inputs of selenium.6 In the interim, both tissue and water column concentrations in the receiving water
can be monitored. For fishless waters, the most sensitive fish species inhabiting downstream waters
should be monitored."

Implementation Approach Needs Changes to Address Fishless and Increased Inputs Waters

The summary of the draft criterion (Table 1, page xiv) presents a number of issues associated with
implementation of the criterion. Specifically, for certain situations the deviations from the use of the fish
tissue value result in a very conservative water column value being the criterion. Such an approach
defeats the rationale for establishment a criterion based on fish tissue and will likely result in significant

-------
negative consequences for dischargers. Simplot recommends several changes to Table 1 to address
these consequences including the use of a tiered and combined monitoring approach to specific
situations.

Regarding the draft proposal, the TCEQ would like to emphasize the need for the development of
implementation guidance due to the complexity of the various aspects incorporated in the draft
criterion.

EPA-HQ-OW-2004-0019-0412-A2; Utility Water Act Group (UWAG); Posted 10/28/2015

In addition, UWAG has the following concerns with the technical basis and implementation of the
"never to be exceeded" averaging period:

• EPA's assumption that tissue selenium concentrations change only gradually over time (and thus
a measured concentration in the tissue represents a relatively unvarying level for at least a
portion of the fish's longevity) is not universally accurate. In some cases the selenium content in
fish tissue changes relatively rapidly. Figures 1 to 5 of these comments indicate the
concentration of selenium measured in otolith (ear bone) cross sections for four fish species
(channel catfish, sauger, flathead catfish, bluegill) collected in the Ohio River (reference site) or
in the mixing zone of an FGD and/or fly ash wastewater discharge during October 2013.

Selenium was measured in the otoliths using laser ablation mass spectrometry inductively
coupled plasma. Otoliths are ideal structures for reconstructing the exposure history of a given
fish to trace elements. Unlike tissue samples, which reflect contaminant concentrations only in
the recent past offish, otoliths illustrate a lifetime exposure. In each graph the black line
represents a running average concentration value and the word "distance" is the radius distance
of the otolith, from fish hatching (centrum) to the time of fish collection (right outer edge). As
"distance" increases, fish age increases. Fish length, age, and measured selenium concentration
in fillet (muscle) samples are also indicated.

The figures, collectively, indicate varying exposure to selenium for each fish. In other words, the
exposure to selenium can be gradual (resulting in a shallow slope of the mean otolith selenium
concentration) during some periods, but temporally variable (resulting in a steep slope of
selenium concentrations) during other periods. Most studies report a significant positive
correlation between tissue selenium and otolith selenium content within the past 6 to 12
months of the fish's life. Depending on when the fish are collected and analyzed for selenium
tissue content, compliance with regulatory tissue criterion could either be verified, or not
verified. Thus, to assess compliance with tissue criterion several individuals of a given species
need to be analyzed because the variability of selenium content in individual fish may be
marked.

• EPA provides no clarification on whether the proposed averaging period applies to a single fish
(of several analyzed from various species) whose tissue concentration exceeds the tissue
criterion, or whether it applies to an average (or geometric mean) value for: 1) a particular
selenium-sensitive species; or 2) for all species evaluated at a site (i.e., a grand average value for

-------
all species pooled). The exact manner in how fish tissue data are evaluated for criterion
compliance at a given site is extremely important. Clarification on this issue is needed in the
final criterion documents.

• Assuming EPA intends that no single fish sample exceed a tissue criterion, then this policy is
unprecedented. In 2001, EPA issued an ambient water quality criterion recommendation for
methylmercury for the protection of people who eat fish (66 Fed. Reg. 1,344). EPA later issued
implementation guidance in response to stakeholder requests. (US EPA, 2010a). In the guidance,
EPA states clearly that - for assessment of criterion compliance - stakeholders are to evaluate
levels of methylmercury in several species of fish (representing at least two trophic levels), and
that average methylmercury tissue concentrations are to be calculated:

Statistical tests for comparing the average concentration from multiple replicate
composite samples to the criterion may be conducted where a sufficient number of
replicates have been collected. [EPA] recommends using the t-test to determine whether
the mean concentration of mercury in composite fish tissue samples exceeds the
screening value. This test involves a statistical comparison of the mean of all fish tissue
data to the criterion.

US EPA, 2010a at 59.

• Establishing that the proposed tissue criterion should be attained at all times is erroneous
because all of the toxicity (and other underlying) data have, themselves, variability. A particular
chronic toxicity test uses average values of parameters: 1) toxicant concentrations in control
and treatment groups (e.g., all water quality variables); and 2) fish testing endpoints (e.g.,
tissue concentration, survival, growth). Indeed, the reason for the testing of replicates is to
understand the magnitude of variability. EPA's 1985 Guidelines support that a criterion should
be implemented consistent with how it was developed. 1985 Guidelines at 4.

• Setting a water quality criterion that is "never to be exceeded" is very problematic from a
permitting standpoint. Regarding the proposed revised chronic selenium fish tissue criterion, a
permitting authority may need to derive a numeric effluent limitation to ensure that tissue
levels outside of the mixing zone do not exceed - at any time - the tissue criterion. Calculating
an effluent limitation that never causes, or contributes to, an exceedance of an applicable
water quality criterion will be extremely stringent because the factor of effluent variability is
removed. The concentration (or loading) of all pollutants regulated by the CWA exhibit some
level of variability. A permitting authority must consider effluent variability when determining
the need for a water quality-based effluent limit (WQBEL):

When determining whether a discharge causes, has the reasonable potential to cause, or
contributes to an in-stream excursion above a narrative or numeric criteria within a
State water quality standard, the permitting authority shall use procedures which
account for existing controls on point and nonpoint sources of pollution, the variability of

-------
the pollutant or pollutant parameters in the effluent, the sensitivity of the species to
toxicity testing (when evaluating whole effluent toxicity), and where appropriate, the
dilution of the effluent in the receiving water.

(40 CFR 122.44(d)(l)(ii))

When determining the need for a WQBEL, a permitting authority also must consider the dilution
and fate of a particular pollutant in the receiving water:

[EJffluent data alone, showing toxicity at the RWC [receiving water concentration], may
be adequate to demonstrate the need for a limit for toxicity or for individual
toxicants...The regulatory authority will need to prioritize, on a case-by-case basis, the
importance of all data and information used in making a determination.

(US EPA, 1991. p. 49)

• Federal regulations require that permit writers calculate two types of effluent
limitations: an average monthly limitation and a maximum daily limitation. EPA's
"NPDES Permit Writers Manual" (US EPA, 2010b) clarifies this requirement:

The NPDES regulations at § 122.45(d) require that all effluent limitations be expressed,
unless impracticable, as both AMLs [average monthly limitation] and MDLs [maximum daily
limitation] for all discharges other than POTWs and as both AMLs and average weekly
limitations (A WLs) for POTWs. The AML is the highest allowable value for the average of
daily discharges over a calendar month. The MDL is the highest allowable daily discharge
measured during a calendar day or 24-hour period representing a calendar day.

(p. 6-34)

Even when a daily maximum WQBEL is established by a permitting authority, multiple samples
may be collected within a 24-hour period as there is variability in the effluent concentration
during this period.

In summary, there will be pragmatic problems with how a criterion that is expressed as "never to be
exceeded" is implemented in wastewater permits. It does not appear EPA has considered any of these
real-world consequences that will be imposed on states to decide during implementation.

Exceedance frequency of tissue criterion "never to be exceeded" must be clarified.

EPA must clarify that it is the protection of fish populations and not individual fish that is intended by
the term "never to be exceeded." As written, the language suggests that a single fish tissue sample with
a concentration that exceeds the proposed tissue criterion would result in non-attainment. UWAG does

-------
not believe that is EPA's intent. We agree with the comments of expert peer reviewer Vince Palace,
Ph.D.:

Biological variability, coupled with uncertainty regarding the residence of mobile fish species,
will make it likely that some fish in a given collection may exceed the guidelines. It is unclear if a
result from one fish (i.e. a single exceedance) will render a given management area in non-
compliance, or if some average value is intended as the trigger.

2014 Peer Review at 86.

Assuming EPA intends such a literal reading, however, UWAG finds several flaws with the proposed
exceedance frequency. First, EPA's Derivation Guidelines very clearly oppose adoption of a "never to be
exceeded" approach. 1985 Guidelines at 4. ("A statement of a criterion as a number that is not to be
exceeded any time or place is not acceptable because few, if any, people who use criteria would take it
literally and few, if any, toxicologists would defend a literal interpretation."). Second, even if EPA made a
policy decision to depart from the 1985 Guidelines prohibition on a never to be exceeded frequency, the
agency provides no technical justification to support such a drastic departure. EPA has not even
provided a sound policy rational for adopting such a conservative approach. Finally, the following
statements in the Draft Criterion appear in conflict with EPA's proposed exceedance frequency:

• In Appendix J, Section 3.0 at J-12, EPA states:

It can be concluded that the kinetics of selenium accumulation and depuration are
sufficiently slow that applying a 30-day averaging period to the water criterion
concentration affords protection even under unrealistic worst case conditions.

• In Table 4.1, footnote 5, EPA references the "never to be exceeded" exceedance frequency with
the following:

Instantaneous measurement. Fish tissue data provide point measurements that reflect
integrative accumulation of selenium over time and space in the fish at a given site.
Selenium concentrations in fish tissue are expected to change only gradually over time in
response to environmental fluctuations.

• In Appendix J, Section 3.0, EPA presents the results of modeling analyses which illustrate the
uptake and depuration rates of selenium in various aquatic biota, under varying exposure
regimes (steady, uniform spikes, and smooth variable). Regarding the relationship between
averaging period and kinetics speed (rate of uptake or depuration), EPA states at J-5:

Because short averaging periods are more environmentally conservative than long
averaging periods, selecting parameter values for fast kinetics is more environmentally
conservative. Figure J1 reflects environmentally conservative choices for k values.

Thus, it appears that EPA chose to ignore its own conclusions that the kinetics of selenium uptake are
gradual. UWAG finds these positions contradictory and, like EPA, we believe the "never to be exceeded"
averaging period is overly conservative. EPA should abandon or clarify that implementation of a literal

-------
reading of the proposed "never to be exceeded" is inconsistent with the science on selenium kinetics
and depuration and in violation of the agency's 1985 Guidelines.

Primacy of Water Column Criterion

Footnote 3 of Table 4.1 indicates that the proposed numeric lentic and lotic water column elements
have primacy over the proposed fish tissue elements in two circumstances: 1) waterbodies with no
sustaining fish populations due to extirpation or habitat limitations; and 2) new or increased inputs of
selenium until equilibrium is reached. UWAG recommends that footnote 3 be revised to delete the
second sentence or to indicate that a state may determine that the water column values have primacy
in those two circumstances. We believe that states should have flexibility to decide whether
implementation of EPA's policy decisions as set forth in the footnotes to Table 4.1 is appropriate.
Moreover, deleting this language from footnote 3 is consistent with EPA's statement that the timing for
when a waterbody will reach "steady state" or equilibrium is dependent on site specific conditions and
local information. Draft Criterion at 93.

EPA's rationale for this policy is as follows:

New inputs are defined as new activities resulting in selenium being released into a lentic or lotic
waterbody. Increased input is defined as an increased discharge of selenium from a current
activity released into a lentic or lotic waterbody. New or increased inputs will likely result in
increased selenium in the food web, likely resulting in increased bioaccumulation of selenium in
fish over a period of time until the new or increased selenium release achieves a quasi-"steady
state" balance with the food web. EPA estimates that the concentration of selenium in fish tissue
will not represent a "steady state" for several months in lotic systems, and longer time periods
(e.g., 2 to 3 years) in lentic systems, dependent upon the hydrodynamics of a given
system;...Estimates of time to achieve steady state under new or increased selenium input
situations are expected to be site dependent, so local information should be used to better refine
these estimates for a particular waterbody.

(Draft Criterion at xv)

UWAG agrees that site-specific attributes should be recognized and assessed to determine when the
new or increased loading of selenium becomes fully assimilated in the water and biota of the receiving
stream. We believe, however, that the technical basis of the proposed policy is speculative. And, the
policy could result in the unnecessary and costly treatment of selenium at facilities.

While UWAG understands the agency's position that a tissue criterion should not apply in settings where
compliance with a tissue criterion cannot be assessed, UWAG disagrees with the agency's position that
the proposed water column criterion should always take precedence over tissue criterion when a new or
expanded discharge of selenium occurs. The latter is unnecessary because other requirements, such as
anti-degradation, may apply to such discharges ensuring adequate water quality protection. For
example, states typically address new (and some address expanded) discharges as part of their anti-
degradation review, which will afford them the opportunity to determine the appropriate criterion for
the discharge. UWAG believes EPA has not provided evidence to suggest the agency has fully assessed
the implications of footnote 3 in the context of existing regulatory programs.

-------
EPA's assumption that there will be some period during which a waterbody will assimilate new or
expanded discharges before reaching "steady state" or equilibrium (e.g., between months to possibly
three years) is reasonable. There are clearly some settings, however, where the response of tissue
selenium to either new inputs, or to eliminated inputs, is relatively rapid. The kinetics of selenium
biomagnification is highly site-specific as the efficiency of the transfer of dissolved selenium to water
column particulate matter is variable across locations. This is observed in Figure 6, which shows average
muscle selenium concentrations in bluegill collected from Hyco Reservoir during 1987 - 2014.

Original letter contains Figure 6 - Graph of average selenium concentrations in bluegill muscle samples
collected in Hyco Reservoir; North Carolina, 1987-2014. See original letter.

• In 1989, Roxboro Steam Electric Station completed the installation of a dry fly ash disposal
system. Reductions of fly-ash derived selenium loading to the lake first began in July 1985. The
concentration of selenium in bluegill tissue samples collected just after the dry handling system
was installed (1990) was markedly lowered. Though selenium levels in tissue samples collected
in 1991 increased somewhat, steadily decreasing concentrations in following years (up until
about 1995) were observed.

• In 2008, Roxboro Steam Electric Station began operation of a flue gas desulfurization. A new
source of selenium, present in FGD wastewater, was introduced to the lake. Annual fish tissue
monitoring continued following this operational change. During 2011 to mid-2012 selenium
levels in bluegill muscle tissue samples began to increase, though the concentrations never
approached the markedly high tissue concentrations observed during 1976 - 1987. The
relatively rapid response in bluegill tissue samples to the (initial) elimination of selenium, and
the (subsequent) addition of selenium after FGD operation, is somewhat unexpected due to the
long hydraulic residence time of the reservoir (4.6 years). One possible explanation for this rapid
tissue response is the accelerated and efficient transfer of aqueous selenium to suspended
particulate matter (which may include plankton and zooplankton) in the lake.

• Reash et al. (2006) reported concentrations of selenium and other trace elements in bluegill
gonad samples collected just before and just after the installation of an FGD system at American
Electric Power's James M. Gavin Plant. The FGD system became operational in December 1994.
Bluegills were collected in October 1994 and May 1995. The following table indicates the mean
concentration of selenium in ovary and teste tissue samples during the two periods.

Original letter contains Table 2 - Mean Selenium Concentration in Bluegill Gonad Tissue Samples. See
original letter.

The reductions in tissue selenium concentration mirrored the change in water selenium levels; the mean
selenium concentration in October 1994 was 15 ng/L whereas the mean concentration in May 1995 was
2-3 ng/L.

If, after considering these comments, EPA does not decide to remove the "new or expanded discharge"
footnote from Table 4.1, we request that the agency limit the applicability of the policy. UWAG
recommends that EPA, in its final guidance, allow states the flexibility of determining whether primacy
of the water column elements of the criterion is absolutely needed for maintenance of applicable water
quality standards. There are certainly some settings in which a new or expanded discharge will clearly
not result in tissue selenium levels in excess of the criterion values. These include large impounded
rivers and lakes with relatively short hydraulic retention times. UWAG recommends that EPA consider a

-------
threshold attribute that, if met, may justify primacy of the water column criterion. For example, if the
calculated instream waste concentration (IWC) of selenium - accounting for the new or expanded
discharge - meets or exceeds 50% using effluent design flow and a receiving stream flow equal to the
7Q10 value, the applicability of the water column criterion may take primacy. Low risk settings should
be identified by EPA and the agency should allow states to use all reliable, relevant data for a specific
location.

EPA-HQ-0W-2004-0019-0419-A2; Lima Refining Company; Posted 11/03/2015

COMMENT #4. The document does not provide guidance as to when a waterbody is impaired.

EPA's proposal sets the fish-tissue concentration limit as a "never to be exceeded" limit, however, there
is no discussion in the proposed criterion about the practicality of enforcing such a limit. States must use
the criteria published by EPA to create or revise water quality standards for each state's waters. See, 33
U.S.C. § 1313(c)(2)(B). Following implementation of those water quality standards, states must provide a
list of those waters in their state where effluent limitations are not sufficient to meet the applicable
water quality standard. See, 33. U.S.C. § 1313(d). In this proposed criterion, EPA does not acknowledge
the practical impact of setting a water quality standard at a "never to be exceeded" limit. Is a water
body impaired at the water column, fish tissue or egg ovary level? What happens if one fish out of 10
exceeds the FCV (based on tissue), which typically is based on an EC10? If its reproductive success is
reduced by 10%, it could be reasoned that this equates to a 1% impact on the reproductive success of
the population (a 10% decrease for 10% of the fish). Is the tissue level in a sensitive species, or an
insensitive species? The implication of the "never to be exceeded" fish tissue limit is discussed further.

There is a subtle difference with respect to implications of exposure when the "never to be exceeded"
fish tissue limit is applied. A water quality criterion (WQC) based on a measured water concentration is
relevant to all of the aquatic organisms that are present and exposed to selenium in the water. In the
case of a fish tissue concentration, which may vary from one individual to the next, at any point in time,
an exceedance by a single fish defines the exposure of that individual to Se, not all of the individuals
present at the site. Therefore the application of a "never to be exceeded" fish tissue limit does not
correlate to broad spectrum impact, as does a WQC based on measured water concentrations.

LRC believes that the "never to exceed" limit is overly burdensome to states and dischargers and creates
a situation where a single fish tissue sample could lead to an impairment listing under CWA § 303(d). We
believe the states would look to USEPA to provide guidance as to how to apply the proposed AWQC. The
2015 USEPA Se Draft AWQC does not provide such guidance

COMMENT ft5. EPA's proposal to implement an interim Water Column Criteria for the time period
before steady state is achieved is unduly burdensome, and not based upon the best available scientific
data.

As noted above, LRC acknowledges that the strength of EPA's proposal is the use of fish tissue data to
establish the applicable standard that is protective of aquatic life, but that the calculation used to derive
a water quality standard from that fish tissue criterion is flawed. LRC believes that this flawed
calculation will result in water quality criteria that are overly restrictive and unduly burdensome to

-------
dischargers. Furthermore, while EPA states that fish tissue concentrations can override water column
concentrations, EPA is recommending to states that in implementing a selenium water quality standard
to protect aquatic life, "fish tissue concentrations should not override water column concentrations until
sufficient time has passed to allow equilibrium to be attained in the food web of lentic and lotic
systems" See, Notice of Availability for Request for Scientific Views, 80 FR 44350, at 44353 (July 27,
2015). This approach will result in the application of the most inaccurate and restrictive portions of
EPA's proposal to become effective immediately, with no relief for dischargers in the interim period.
Therefore, LRC requests that EPA include in their fish tissue to water column translation calculation
various measures (i.e., growth considerations) that will provide dischargers with a more accurate
representation of the fish tissue concentration baselines, until those baselines can be confirmed at
steady state. In the alternative, LRC requests that EPA include in the final criterion a recommendation
that the current State selenium standard, which, in Ohio, is 5.0 ug/L for total recoverable selenium, be
maintained as an interim water column standard to meet until steady state conditions are achieved.

EPA-HQ-OW-2004-0019-0406-A2; Kentucky Chamber of Commerce; Posted 10/15/2015

In addition, the "fishless" waters language in footnote 3 of Table 1 could also be interpreted in an overly
restrictive manner. Many discharges are to "fishless" ephemeral streams that lead to and merge into
larger water bodies that support fish within a short distance. The footnote and supporting language in
the draft criterion document could be construed to always require a water column standard to be
imposed in such cases even though fish tissue is readily available in relatively close proximity to the
outfall at a downstream perennial water body. This concern is heightened due to adoption of the Clean
Water Rule, which further expands the number of fishless waters that are deemed a water of the United
States and thus a receiving water under Section 402 of the Act.

Indeed, the water column concentration criteria were developed using a complex formula designed to
model trophic transfer of selenium through the food chain "resulting in the fish tissue concentration
that yields the chronic reproductive effects of concern." Draft Aquatic Life Ambient Water Quality
Criterion for Selenium - Freshwater (2015) (hereinafter Criterion Document), p. 23. The proposed water
column concentrations of selenium in ephemeral "fishless" streams cannot, by definition, cause any
such impacts in those waters, and the potential for such impacts in downstream perennial waters
supporting fish can still be monitored by fish tissue. At most, the determination of when fish are not
available for testing to determine any potential impact of a discharge to an aquatic life use should be
made by the permit issuing authority on a case-by-case basis and should take into account the proximity
of fish in downstream waters that would reflect any impacts from the discharge. Where fish are
available for testing downstream, they can be used to determine the aquatic life use is fully protected in
receiving streams and fish tissue should still be given precedence. Therefore, the fishless stream
restriction on the precedence offish tissue data should be removed or, at a minimum, fishless streams
should not be defined in the criterion document and U.S. EPA should recognize that the issue should be
determined by the NPDES permitting authority.

-------
EPA-HQ-OW-2004-0019-0378; Public comment submitted by C. Lish; Posted 09/25/2015

The language for new or increased inputs of selenium should be modified so as to lengthen the time
period where the fish tissue standard does not override the water column standard. Currently, the
language states a period of "several months," but surface mines in particular change regularly over the
lifespan of the mine, so the characteristics of wastewater outflows also change. Because mine discharge
will often not allow a stream to reach a state of equilibrium, the selenium water quality standard should
retain primacy until the mine permit has been fully bond released.

Enforcement of a selenium standard based on fish tissue for point source discharges will be difficult,
especially in watersheds with multiple selenium discharges. In Central Appalachia, many rivers and
streams have multiple wastewater discharges from multiple coal mines. Because fish move within the
stream, it may be impossible to prove which mine is responsible for the fish tissue selenium exceedance.
Additionally, enforcement of water quality standards in Central Appalachia is often dependent on citizen
enforcement. Testing for selenium in fish tissue will make citizen enforcement much more cumbersome
and less likely to occur.

For these reasons, the EPA needs to implement a protective, enforceable, water column-based selenium
standard that is enforceable by citizens.

EPA-HQ-0W-2004-0019-0405-A2; Virginia Coal & Energy Alliance (VCEA); Posted 10/15/2015

That said, we are deeply troubled by EPA's suggestion that in "fishless waters" (e.g., those where flow or
other aspects of the physical habitat are insufficient to sustain fish), EPA's water column criteria must be
applied. This is fundamentally inconsistent with the federal and state water quality standards program.
Criteria must be protective of uses. Where a use does not exist, the criteria simply do not apply (since
there is nothing to protect). EPA's proposal would make the criteria independent of the use, which is
both illogical and inappropriate. In any event, where and how the criteria are applied are decisions
relegated to the states with primary authority for implementation of the Clean Water Act. These kinds
of implementation details do not belong in EPA's national recommended criteria document.

EPA-HQ-0W-2004-0019-0404-A2; Kentucky Coal Association (KCA); Posted 10/15/2015

Footnote 3 of Table 1 Creates Unnecessary Implementation Problems and Undercuts EPA's Stated
Preference for a Fish-Tissue Based Standard. KCA has serious concerns with Footnote 3 of Table 1 of the
proposed criterion. Footnote 3 outlines two scenarios in which water column values have primacy over
fish tissue values - specifically, in "fishless waters" and waters with "new or increased inputs of selenium
until equilibrium is reached." While we understand that EPA is seeking to address with respect to these
two scenarios, we think that the proposed language, when taken in conjunction with the anti-
backsliding provisions of the Clean Water Act (CWA) and the requirements of the National Pollutant
Discharge Elimination System (NPDES) permitting process for establishing Water Quality-Based Effluent
limits (WQBELs), effectively nullifies the fish tissue approach all together. While EPA considers this an
"implementation" issue, it is not clear that the implementation difficulties posed by the language of the
criterion can, in fact, be rectified through implementation guidance. EPA's intent in the criterion

-------
document to avoid compromising downstream waters or waters with new or expanded discharges can
be accomplished under a standard that does not create unnecessary confusion in future implementation
and maintains the primacy of the fish tissue values in permits.

The "New or Expanded Discharges" Qualifier Will Lead to Unnecessarily Stringent Permit Terms and
Preclude Future Permit Limits Based on Fish Tissue. By requiring a water column based WQBEL for
expanded discharges, the criterion effectively vitiates the ability of permit writers to incorporate more
scientifically justifiable fish-tissue based limits into permits. The CWA's anti-backsliding provision
provides that "a permit may not be renewed, reissued, or modified ... to contain effluent limitations
which are less stringent than the comparable effluent limitations in the previous permit." CWA Sec.
402(o). While there are listed exceptions, the general interpretation of state agencies and their NPDES
permit writers is that permits cannot be modified or renewed with less stringent WQBELs than those
contained in the previous version.

As written, with respect to NPDES permits for new or expanded discharges, the criterion would require a
permit writer to establish WQBELs based on the proposed water column criteria, and that water
column-based WQBEL would apply during the months or even years that it takes for the receiving water
to reach "equilibrium." Due to the overly conservative estimates used to derive the water column
concentrations, it is anticipated that in many instances a post-equilibrium discharge limit derived from a
fish tissue concentration-based effluent limit will be higher than any initial limit based on the water
column value. Because a discharger's permit will already contain the more stringent limitation, even
after equilibrium is reached a discharger will only be able to obtain a more appropriate permit limitation
based on a fish tissue concentration if that limitation would be more stringent than the one based on
the water column value. In other words, under the criterion as it is currently written, there will likely be
many instances in which a fish tissue-based effluent limitation will not be permissible in actual permits,
despite the fact that such a limitation would represent the best available science and reflect the EPNs
intent as we understand it. This same analysis could also potentially be applied by permit writers even to
existing discharges, because states may choose to first apply a WQBEL based on the proposed water
column criterion in cases where data regarding fish tissue and fish egg ovary selenium concentrations is
scarce, was not collected during permit renewal, or is otherwise unavailable at the time of permit
renewal. In essence, the criterion as drafted establishes a sound rule- that fish tissue is the best
indicator of protection of the designated use - then establishes an ambiguous and potentially vast
exception that threatens to swallow the rule whole. Water quality standards must be drafted not only in
a manner that is sufficiently protective of designated uses, but also in a manner that provides clear
guidance to permit writers who must implement the standards.

While implementation guidance may be able to address this issue in some cases where the criterion fits
into one of the anti-backsliding exceptions, it is unclear if that will be legally possible, and regardless it
would provide no relief as a practical matter to dischargers who are forced to install control
technologies to meet the more stringent water column-based limit, despite the fact that such controls
are not necessary to protect aquatic life. In more extreme cases, footnote 3 may prevent dischargers
from obtaining any permit at all, even when their discharges present no risk to aquatic life, due to
elevated natural background levels of selenium in the receiving water, because permit writers would be
inappropriately basing their calculations on the water column number.

-------
Because of this significant and likely unintended consequence posed by Footnote 3 to Table 1, KCA
makes two recommendations. First, EPA could delete Footnote 3 as written, and instead include a
provision for new or expanded discharges whereby limitations are derived from fish tissue
concentrations but compliance cannot be tested until the appropriate amount of time has elapsed for
the receiving water to reach equilibrium (i.e., a compliance schedule). The second is renaming the water
column "criteria" as water column "thresholds," as was done in Kentucky's recently approved selenium
water quality criteria revision. See 401 KAR 10:031 Section 6(1). This would clarify EPA's intent to allow
fish tissue concentrations to have primacy over the water column values.

Concerning a related issue, we also note that EPA guidance limits the terms of compliance schedules to
five years, as do many state permitting programs based on EPA's guidance. Because in some situations
where selenium levels are being decreased based on the new criterion it may take longer than five years
for corresponding fish tissue concentrations to decrease, KCA recommends including a provision in the
criterion stating that longer compliance schedules may be needed with respect to bioaccumulants.
Although again this could be considered an "implementation" issue, because it is a facet of the criterion
that could directly conflict with existing state water quality programs- thereby limiting states' ability to
adopt it- it is necessary to adopt language in the criterion itself that addresses this concern.

The qualifier related to "new or expanded discharges" is also problematic because it is ambiguous. The
criterion does not define, by magnitude of concentration, frequency, or by duration, the discharges that
should be considered "new or increased inputs of selenium." Similarly, the criterion does not define, by
aquatic conditions or by duration, the term "equilibrium." Attempting to define and defend whether and
when a stream is in "equilibrium" or "steady-state" in order to determine when and what criterion
should apply creates a legally and technically unworkable, and unnecessary, requirement for states.
Without such terms being clearly defined, permittees cannot determine in a scientifically defensible
manner when such "new or increased inputs" should cause water column values to have primacy over
fish tissue values.

The "Fishless Waters" Qualifier Also Creates Unnecessary Implementation Problems. With respect to
"fish less waters," footnote 3 will likewise lead to situations in which dischargers must install expensive
control technologies or not be able to discharge selenium at all, despite the fact that such requirements
are not necessary to protect aquatic life. The concept of a "fishless" stream could be used to impose
overly restrictive limits in many cases where discharges occur to "fishless" ephemeral streams, even
where those streams ultimately flow into and merge with larger bodies of water that contain fish only a
short distance away. This is of particular concern to KCA, because mining discharges frequently occur to
these smaller streams, where selenium poses no risk to aquatic life at the outfall or downstream. The
recent adoption of the Clean Water Rule, which significantly expands the number of so-called "fishless"
streams subject to classification as receiving water, only heightens this concern.

We understand that EPA is required under the CWA to protect downstream waters when deriving water
quality criteria. But KCA does not believe that the water column number proposed for "fishless waters"
is necessary to protect downstream waters. KCA believes that more scientifically defensible approaches
to addressing the issue of "fishless waters" may exist. However, it does not appear based on the current
record for development of the criterion that these potential alternative approaches have been
adequately investigated and analyzed.

-------
EPA-HQ-OW-2004-0019-0410-A2; West Virginia Department of Environmental Protection
(DEP); Posted 10/28/2015

Bioaccumulation modeling shows that selenium accumulates in fish tissue based on dietary exposure,
and toxicity occurs primarily through maternal-egg transfer. This research indicates that an appropriate
approach to a selenium criterion is to use fish tissue and/or egg/ovary concentration to determine
selenium toxicity in water. With EPA's proposed criterion, when the chronic water column limit is
exceeded, fish tissue and/or egg/ovary tissue concentrations may be assessed to make a final
determination of exceedance. While DEP agrees with much of the data interpretations that EPA has
provided in this revised draft, DEP believes some changes should be made to preliminary
implementation strategies put forth in the EPA 2015 Draft.

Regarding EPA 2015 Draft consideration of chronic selenium criteria to address waters with new or
increased selenium inputs, while DEP agrees with EPA's approach to new discharges of selenium in
waters previously unimpacted by selenium (new inputs), DEP believes additional inputs of selenium on
already impacted waters (increased inputs) should be handled differently. Specifically, when additional
inputs are proposed on previously selenium-impacted waters, fish tissue concentrations should continue
to take precedence over water column concentrations, as they reflect the bioaccumulation of selenium
in those waters.

In regards to the EPA 2015 Draft descriptive element for fishless waters, EPA defines fishless waters as:

Waters with insufficient instream habitat and/or flow to support a population of any fish species

on a continuing basis, or waters that once supported populations of one or more fish species but

no longer support fish (i.e., extirpation) due to temporary or permanent changes in water

quality (e.g., due to selenium pollution), flow or instream habitat

(EPA 2015 Draft pg. xv).

EPA 2015 Draft recommends that a water column concentration take precedence over fish tissue
concentrations in fishless waters, but this approach is unreasonable for waters which are fishless due to
insufficient flow, or which have been extirpated for reasons unrelated to selenium. In these examples of
fishless waters, basing selenium outputs on receiving water column selenium concentration could not
result in the re-establishment of fish, because the cause of fishless water would not have been resolved.
Rather, in the situations of insufficient flow or fish extirpation due to non-selenium related pollution,
precedence should remain with fish tissue analysis downstream, where conditions do allow for fish
populations. Because the "fishless waters" element better relates to implementation of the selenium
criteria, DEP believes these decisions are best made on a case-by-case basis in National Pollution
Discharge Elimination System (NPDES) permitting, rather than attempting to prescribe implementation
aspects of the criterion in water quality standards.

DEP understands EPA's goal of ensuring that fish tissue concentrations do not increase above the
chronic selenium criteria. However, a more detailed approach is appropriate in considering protective
effluent limits for increased discharges and for fishless waters. The amount of water discharged, the
location with respect to the nearest fish population, and the existing tissue and water column
concentrations in the watershed may all influence the decisions made with regard to effluent limits for

100

-------
new or expanded selenium discharges. The goal is to ensure that water column concentrations do not
increase above the level required to protect the segments where fish are or could be located.

EPA-HQ-OW-2004-0019-0391-A1; North American Metals Council (NAMC); Posted
10/14/2015

2.3.5 Implementation: Fishless Waters and Increasing Inputs

There are two aspects of the proposed criteria that will have far reaching and unintended, inappropriate
consequences. As noted in the 2015 EPA Selenium Draft,53 tissue data take precedence over water
concentrations, except when the following conditions are present:

1. "Fishless waters" (waters where fish have been extirpated or where physical habitat and/or flow
regime cannot sustain fish populations); and/or,

2. New or increased inputs of selenium from a specific source until equilibrium is reached.

The 2015 EPA Selenium Draft defines fishless waters54 as waters with insufficient instream habitat
and/or flow to support a population of any fish species on a continuing basis, or waters that once
supported populations of one or more fish species but no longer support fish (i.e., extirpation) due to
temporary or permanent changes in water quality (e.g., due to selenium contamination), flow, or
instream habitat. Because of the inability to collect sufficient fish tissue to measure selenium
concentrations in fish in such waters, water column concentrations are prescribed as best representing
selenium levels required to protect aquatic communities and downstream waters in such areas.

The above definition of the "process" for fishless waters is inconsistent with Appendix K of the External
Peer Review Draft (Section 1.2.1.2), where EPA states that "[w]hen fish are naturally absent from a
waterbody, states and tribes should target the most sensitive fish species inhabiting downstream
waters."55 Recommendations from Appendix K are consistent with current risk assessment methods in
that consideration of exposure is taken into account. Default application of an aqueous criterion based
on fish tissue to "fishless waters" will result in values at or below background for many stream systems,
and thus will certainly result in a large number of false positives with consequent dilution of efforts to
address real environmental problems.

The 2015 EPA Selenium Draft defines new inputs or increasing inputs as:56 "new activities . . . resulting in
selenium being released into a lentic or lotic waterbody. Increased input is defined as an increased
discharge of selenium from a current activity released into a lentic or lotic waterbody. New or increased
inputs will result in increased selenium in the food web, likely resulting in increased bioaccumulation of
selenium in fish over a period of time until the new or increased selenium release achieves a 'steady
state' balance within the food web. EPA estimates that the concentration of selenium in fish tissue will
not represent a 'steady state' for several months in lotic systems... EPA recommends that fish tissue
concentration not override water column concentration until these periods of time have passed in lotic
and lentic systems, respectively."57

101

-------
tissue measurements to compare to the criterion). Once fish tissue monitoring and comparison to the
criterion can occur, an increase in the selenium limit may not be possible due to the "antibacksliding"
provision in the Clean Water Act. Also, the frequency of monitoring that a discharger will have to do may
be larger.58 And, finally, the discharger will have had to invest in the control equipment necessary to
meet the initial permit limit. Thus, considerable resources will likely have been expended that were not
necessary to protect aquatic life.

We recommend that existing Footnote 3 in Table l59 be deleted and replaced with the following:
"Limitations are derived from fish tissue concentrations, but compliance cannot be fully tested until the
appropriate amount of time has elapsed for equilibrium in the receiving water for new and increased
inputs of selenium. In the interim, both tissue and water column concentrations in the receiving water
should be monitored. For fishless waters, the most sensitive fish species inhabiting downstream waters
should be monitored."

EPA-HQ-OW-2004-0019-0387-A1; Colorado Department of Public Health & Environment;
Posted 10/14/2015

• The proposed criterion also applies to fishless waters, with the goal of protecting downstream
waters. However, because the water column element is derived from the fish egg/ovary
element, application of the water column element to fishless waters may not always be
appropriate.

Many waters in the west have naturally very low flows or are headwater streams, and therefore
only support invertebrates. Water chemistry conditions in such waters can be quite different
from those further downstream. For instance, in areas where fish are absent due to low flow, it
is likely that water inputs further downstream would affect water chemistry (increasing or
decreasing selenium concentrations). Where fish are not part of the expected aquatic
community, and there is evidence of downstream protection, an alternative criterion may be
more appropriate. For example, it may be possible to derive a criterion protective of
invertebrates and/ or amphibians.

Additionally, while states and authorized tribes are required to consider and ensure the
protection of downstream water quality, they are allowed to choose their own preferred
approach in their standards. EPA has provided guidance and suggestions on how states can
include narrative downstream protection criteria in their standards (EPA Publication No. 820-F-
14-002), and provided several different template options for the states to work from and use
based on their individual circumstances. There are many different ways (e.g., permit limits
based on downstream waters) to ensure the protection of downstream water quality, and not
all of them require assigning the most stringent standards to all waters.

For these reasons, the WQCD recommends EPA acknowledge the potential need for and
appropriateness of alternative approaches for selenium criteria in fishless waters.

102

-------
EPA-HQ-OW-2004-0019-0411-A2; American Petroleum Institute (API); Posted 10/28/2015

2.3.4 Back-calculated Particulate Se "Thresholds" Relative to Background

A key implementation issue is how to address background Se concentrations. As discussed in Section
2.3.2, particulate Se concentrations of 2.3 to 4.6 mg/kg-dw, based on the food web models compiled by
the EPA, are predicted to result in the draft fish egg Se criterion of 15.8 mg/kg-dw. For comparison,
Skorupa (1998) noted that background Se concentrations in sediment, freshwater algae, and freshwater
macrophytes range up to 2.0, 1.5, and 2.0 mg/kg-dw, respectively. This means that the most
conservative "safe" particulate Se concentration of 2.3 mg/kg-dw approaches these values. As also
discussed above, there are also two clearly identified reference lakes in the EPA (2015) dataset that have
particulate Se concentrations within the 2.3 to 4.6 mg/kg-dw range. These are particulate Se
concentrations of 3.34 and 3.99 mg/kg-dw for High Rock Lake and Badin Lake, respectively, which were
reported as reference lakes in Lemly (1985). This suggests that the jjFcomP°slte values and/or CFs are
conservative for at least some sites. Consideration of regression-based TTFs could reduce the likelihood
of back-calculating "safe" particulate Se concentrations that fall with the range of reference area
concentrations.

EPA-HQ-0W-2004-0019-0409-A2; OC Public Works; Posted 10/20/2015
'New or Increased Selenium Inputs'

The Draft Se Criterion allows water column portion of the criterion to be the primary one when there
are 'new or increased selenium inputs' to a water body. The scientific basis of this provision appears
tenuous. In addition to the issues presented in the comments 2-6 above, new or increased input to a
waterbody is clearly a site-specific issue and should be treated as such. Since the bioaccumulation model
for any given site is essentially a linear regression model, for any given site where both fish tissue and
water column concentrations have been established, the upper end of water column concentration
where fish species could potentially be impacted can be established through a simple calculation, and
that calculation should be the basis for the limit of new or increased input. If the 'new' input does not
increase the existing water column selenium concentration, or is well below the calculated water
column concentration that could cause fish impairment this provision will impose an overly stringent
and unnecessary requirement.

To justify this 'new or increased input' provision, USEPA states that these inputs will cause excursion of
'steady state' or 'equilibrium' thus their impacts on fish species cannot manifest themselves in time to
be captured by monitoring. This is reasonable but it should only apply when a short term, new or
increased, and high concentration input cause an excursion above the water column concentration that
could potentially cause fish impairment under equilibrium condition. If the new or increased input is
below this concentration, it would be scientifically impossible that fish t issue can be impacted when
steady state is reached.

The 'equilibrium' and 'steady state' language in the Draft Se Criterion overlooked the fact that these
terms are relative and represent an ideal condition. In nature, few water bodies are in steady state and
many exhibit considerable excursions. The language does not acknowledge the considerable natural
variations (seasonal, interannual, or even dry/wet weather) in most waters.

103

-------
'Fishless Water' Paradox

For water bodies with fish, the issues associated with water column portion of the Criterion (discussed in
comments 2-6 above) will not pose a problem due to the primacy of fish tissue numbers. This is
reasonable because fish tissue level is directly linked to designated uses. If fish are absent, however, and
the primary criterion becomes water column based, a paradox is created. A water body with aqueous
selenium concentrations exceeding the Draft Se Criterion could be fish less due to one of three reasons:
1) selenium is naturally-occurring but other natural factors preclude fish from establishment, such as
limitations of habitat, including intermittent flows; physical and chemical constraints, predatory
pressure, lack of food items, etc. Fish would have been extirpated even if selenium levels were reduced
to levels below the Draft Se Criterion; 2) anthropogenic factors that, by altering physical, chemical (other
than selenium), biological factors and flow regime, result in the extirpation of resident fish species even
if selenium levels were reduced to a level below the Draft Se Criterion; 3) anthropogenic input of
selenium that result, entirely or in combination with other factors, in the extirpation of resident fish
species. The intention of the Draft Se Criterion is clearly to protect the third scenario, but the 'fishless
waters' issue will likely result in expenditure of considerable resources to manage scenarios 1 and 2,
which is essentially protecting designated uses that do not exist or protecting designated uses that are
not impacted by selenium. This issue becomes even more acute considering the significant issues with
the development of water column portion of the Draft Se Criterion discussed above.

For comment 8, a fish habitat survey should be conducted, either independently or as part of a use
attainability analysis, to determine the existence of designated use and viability offish species in the
subject waters before selenium water column guidelines are applied. Water column guidelines or
screening levels can be used as part of the habitat and use assessment. If water column selenium
concentration is determined not to be a contributing factor to fish extirpation, no selenium reduction
effort should be implemented or prioritized. If the monitoring suggests that selenium is causing or
contributing to cause fish extirpation, an adaptive approach should be designed to improve all of the
factors to help fish population to be reestablished. With fish tissue concentration to determine
attainment, water column guidelines can be used to assess the progress of selenium reduction efforts.

104

-------
1 II ii ii II ii ii! II III 1 i1 II ! 1 II III II II III * 1 II |1 ' IIII III I \ I' II 1 III III III I 1 II' II i« IIII' 'in

Elements

2.4 Summary

Comments concerning the water column criterion element as it appears in Table 1 of the July 2015 draft
are included in this section. Several commenters provided comments on magnitude, duration, and
frequency. However, the bulk of the comments were focused on the intermittent equation and how it
should be implemented. Some commenters noted that the criterion values should include uncertainty
data.

2.4 Response to comments

2.4 Responses concerning the intermittent water column criteria element

Regarding the intermittent exposure criterion element, EPA recognizes that not all exposures are
continuous and developed the intermittent criterion element due to concern that intermittent discharge
sources may not be accounted for with the national 30-day average chronic water column criterion
element, yet intermittent discharges of sufficient magnitude and frequency could accumulate through
the food web and ultimately result in chronic impacts on aquatic life.

EPA believes, and many comments support EPA's conclusion, that it is unnecessary to have an additional
acute criterion element which addresses acute, water column-only events, because selenium is
bioaccumulative and toxicity primarily occurs through dietary exposure. Although selenium may cause
acute toxicity at high concentrations (which would be captured by the intermittent criterion element)
the most deleterious effect on aquatic organisms is due to selenium's bioaccumulative properties; these
effects occur at lower concentrations than acute effects. Chapman et al. (2009) noted that selenium
acute toxicity has rarely been reported in the aquatic environment and that traditional methods for
predicting effects based on direct exposure to dissolved concentrations do not work well for selenium.

Regarding the intermittent criterion element, as described in Section 2.7.9 of the 2015 draft, this
criterion element fills a need to account for intermittent discharges that may be elevated but
infrequent, and to account for associated potential loadings and resulting chronic risk (not risk from
intermittent or variable exposures per se) to downstream waters. Modifications of the criterion can be
made on a site-specific basis, with data from the site. Biokinetic modeling, if used, should be adapted to
a site using appropriate data - the values EPA presented represent default values that will be generally
protective.

FWPC commented that they "agree with the concept that intermittent exposure to selenium, as with
other pollutants, is less toxic than continual exposure." EPA is not articulating such a concept in the
document. Repeated exposures of selenium are expected to result in accumulation through an
ecosystem's food web leading to sustained exposure. Because the effect related to the intermittent
criterion element is the chronic effect not effects due to variable exposure, additional toxicity testing is
not needed, as suggested in some comments.

105

-------
The citation (Hoang and Klaine 2008)1 provided by FWPC, as evidence to support FWPC's assertion that
"intermittent exposure to selenium, as with other pollutants, is less toxic than continual exposure" is for
a study on an invertebrate species with a water column only exposure, which is not representative of
the dietary exposure considered important for selenium and discussed in the criterion document.

Further the study authors, Hoang and Klaine 2008, specifically identify an important issue not described
in the public comment: because of latent mortality observed in their study they indicate that "standard
toxicity tests using continuous exposures would underestimate Se toxicity." This comports with EPA's
practice of developing criteria durations typically less than the toxicity test experimental duration. These
authors did find recovery of the aquatic invertebrate tested after water column only exposures, most
likely because the pulsed selenium diminished in water column, unlike dietary exposures where
selenium bioaccumulates. Further, all exposures in this study were for 12 hours of water column
exposure, less than typical acute test exposure durations. Dietary exposures in an ecosystem are
unlikely to be of such short exposure durations.

The intermittent criterion element is meant to protect receiving and downstream waters from
bioaccumulative impacts by limiting the amount of selenium that is available to be taken up by biota and
bioaccumulated to levels of concern in sensitive species. Because (a) the derivation of the 30-day
averaging period, and the subsequent derivation of the intermittent criterion element, and (b) the
considerations upon which EPA based its decision to use the ECio both involve the kinetics of
bioaccumulation, EPA understands the connection the comment is making between the two. However,
EPA does not agree that it is addressing the same issue twice. Rather EPA is addressing two distinct
issues that are both affected by bioaccumulation and kinetics.

EPA notes that several reviewers viewed the intermittent criterion element as a reasonable surrogate
for an acute criterion for protecting aquatic ecosystems, particularly downstream lentic waterbodies,
from the effects of intermittent discharges of selenium. EPA has also further examined the lentic/lotic
classification issue, and has evaluated each site used in the criterion development individually to ensure
it was not mischaracterized.

2.4 Specific comments

EPA-HQ-OW-2004-0019-0403-A2; Iowa Department of Natural Resources; Posted 10/15/15

(4) EPA states that the "Water Intermittent" criterion is not an acute criterion. Also, it is not an
independent criterion but a re-expression of the 30-day criterion. States should not be required to adopt
both the 30-day average criterion and the intermittent criterion because the chronic effect through
dietary consumption of selenium (not short-term) is the cause for selenium toxicity to aquatic life based

1 Hoang, T.C., Klaine, S.J. 2008. Characterizing the toxicity of pulsed exposure to Daphina magna. Chemosphere
71(2008) 429-438.

106

-------
on the available toxicity data. Meeting the 30-day chronic criterion will ensure that acute toxicity can be
prevented.

(5) States should have the option to use "water intermittent" criterion as an alternative to the 30-day
average criterion for intermittent discharges of selenium, but the intermittent and 30-day average
criteria should not be applied together in the implementation of Clean Water Act programs.

EPA-HQ-0W-2004-0019-0389-A1; Thunder Basin Coal Company, LLC (TBCC); Posted 10/14/15
Intermittent Water Column Exposure Criteria

Due to the wide variation in site specific selenium concentrations even in small areas and since selenium
concentrations can change rapidly due to flow and geochemical conditions, the intermittent exposure
criteria will be very difficult and costly to correctly determine for regulators and permittees alike. The
proposed criteria and procedures should be better explained by EPA or dropped from the proposed
standards and replaced by a simpler chronic standard.

EPA-HQ-OW-2004-0019-0393; North Dakota Department of Health; Posted 10/14/2015

Additionally while EPA's attempt to address the intermittent nature of the selenium concentrations in
the water column is commendable, the implementation process is not practical, and again places an
unsupported burden on the state. The translation between intermittent high concentrations and tissue
concentration does not withstand scientific rigor. Until a broader dataset is established, a link between
intermittent concentrations and tissue concentrations should not be established.

EPA-HQ-0W-2004-0019-0416-A1; Wyoming Mining Association (WMA); Posted 11/02/2015
Intermittent Water Column Exposure Criteria

EPA-HQ-OW-2004-0019-0394-A2; Appalachian Mountain Advocates and Sierra Club
Environmental Law Program; Posted 10/14/2015

II. The Chronic Water Column Element Should Be Expressed as a Four-Day Average

As explained above, an inviolable water column criterion is necessary to achieve the dual purposes of
setting water quality goals and providing the basis for effective regulatory controls. While we believe
that EPA must give precedence to the water column elements of its Draft Criterion, those elements must

107

-------
be revised to ensure that they can be practically enforced and implemented. The Draft Criterion
includes, as one of its four elements, a "Monthly Average" water column element that is based on the
"30-day average water concentration." Draft Criterion at xiv, 22. This 30-day average replaces the
existing criterion for chronic exposure, which is expressed as a "four-day average." See 64 Fed. Reg
61,182 at 61,194-61,195. By shifting from a four-day to a 30-day average, EPA has removed important
protections for aquatic life. EPA has not explained the basis for its shift from a four-day to a 30-day
average. This shift and the lack of explanation are particularly problematic because multiple EPA
guidance documents explicitly state that four-day averaging periods are preferred while 30-day
averaging periods should be discouraged. The documents that support a four-day averaging period
include EPA's "Guidelines for Deriving Numerical National Water Quality Criteria for the Protection of
Aquatic Organisms and Their Uses (1985)"8 and "Technical Support Document For Water Quality-based
Toxics Control (1991)."9 In its final selenium criterion, EPA should return to the use of a four-day
average.

A. EPA's Use of a 30-day Average is Inconsistent with EPA Guidance

One of the documents that supports a four-day average over a 30-day average was directly relied on by
EPA in preparing the proposed selenium criterion. EPA's public notice states that the proposed selenium
criterion is "based on the latest scientific information and current EPA policies and methods, including
EPA's Guidelines for Deriving Numerical National Water Quality Criteria for the Protection of Aquatic
Organisms and Their Uses (1985) (EPA/ R-85-100)." 80 Fed. Reg at 44,351. Despite this apparent
reliance, EPA has ignored the Guidelines' express statement that a four-day average is more protective
than, and therefore preferable to, a thirty-day average.

The Guidelines first observes that the averaging period should be shorter than the test used to derive
the criteria, stating that "Life-cycle tests with species such as mysids and daphnids and early life-stage
tests with warmwater fishes usually last for 20 to 30 days. An averaging period that is equal to the
length of the test will obviously allow the worst possible fluctuations and would very likely allow
increased adverse effects." Guidelines for Deriving Numerical National Water Quality Criteria for the
Protection of Aquatic Organisms and Their Uses (1985) at 5.

The Guidance then expands on the benefits of a four-day averaging period:

An averaging period of four days seems appropriate for use with the CCC [criterion continuous
concentration] for two reasons. First, it is substantially shorter than the 20 to 30 days that is obviously
unacceptable. Second, for some species it appears that the results of chronic tests are due to the
existence of a sensitive life stage at some time during the test, rather than being caused by either long-
term stress or long-term accumulation of the test material in the organism. The existence of a sensitive
life stage is probably the cause of acute-chronic ratios that are not much greater than 1, and is also
possible when the ratio is substantially greater than 1. In addition, some experimentally determined
acute-chronic ratios are somewhat less than 1, possibly because prior exposure during the chronic test
increased the resistance of the sensitive life stage. A four-day averaging period will probably prevent
increased adverse effects on sensitive life stages by limiting the durations and magnitudes of
exceedences of the CCC.

108

-------
Guidelines for Deriving Numerical National Water Quality Criteria for the Protection of Aquatic
Organisms and Their Uses (1985) at 5.

The Guidance upon which the proposed selenium criterion is supposedly "based on" thus recognizes
that an averaging period of 30-days "is obviously unacceptable" and recommends instead an averaging
period of four-days.

Other EPA publications also recommend the use of a four-day average. The "Technical Support
Document For Water Quality-based Toxics Control (1991)" states that "a 4-day averaging period is
recommended for application of the CCC in aquatic-life criteria for both individual pollutants and Whole
Effluents." Technical Support Document at Appendix D-2. Just as the Guidelines did, the Technical
Support Document makes clear that the "averaging period should be substantially less than the lengths
of the tests" on which it is based. Id. at Appendix D-3. The chronic exposure tests cited by EPA in its
Draft Criterion document include studies of 30 or fewer days. See e.g. Draft Criterion at 102 (describing a
28-day study of "fry surviving at swim-up"); 36 ("the 30-day larval survival test"). Because these tests
were as little as 30 days, the averaging period should be substantially shorter.

The Technical Support Document provides several additional reasons why a four-day averaging period is
recommended:

• It is substantially shorter than the 20- to 30-day duration of most chronic tests and is somewhat
shorter than the 7-day duration of the Ceriodaphnia life-cycle test.

• For both endrin and fenvalerate, Jarvinen et al. found that a 72-hour exposure caused about the
same amount of effect on the growth of fathead minnows in early life-stage tests as did a 30-
day exposure to the same concentration.

• In some life-cycle tests on effluents with Ceriodaphnids, concentrations of effluents that were a
factor of 1.8 greater than the CCC caused unacceptable effects in 4 or 5 days.

• It is not so short as to effectively defeat the purpose of the concept of the averaging period.

Id. at Appendix D-2 (internal citations omitted).

B. EPA's Use of a 30-day Average Will Render the Water-Column Element Unenforceable

Of equal or greater importance is the fact that the use of a 30-day average will render the water-column
based chronic element unenforceable. As EPA has provided no explanation as to how the 30-day
average is to be implemented, courts and state regulators are free to interpret the element as they see
fit. At least one federal judge has interpreted a four-day average as requiring four consecutive days of
sampling. Such an interpretation applied to a 30-day standard would make it impossible for citizen
groups to monitor compliance with a selenium water quality standard and would greatly increase the
cost to states of determining compliance with the standard.

In a Clean Water Act citizen suit enforcing a permit condition that forbade violations of water quality
standards, a West Virginia federal district court acknowledged that the citizen plaintiffs had presented
selenium monitoring data in which "some months have two days of measurements per location, [and]
other months have only one day." Ohio Valley Environmental Coalition, Inc. v. Consol of Kentucky, Inc.,

109

-------
2014 WL 1761938 at *16 (S.D.W.Va., 2014). Although the court acknowledged that "every measurement
reported exceeds [the existing chronic criterion of] 5 ug/l/' it concluded that "it is not clear that any of
these measurements are actually chronic measurements, that is, four-day average concentrations." Id.
The implication of the district court's decision is that four consecutive days of sampling data are
required to prove a violation of a standard expressed as a four-day average. The further implication is
that a thirty-day average would require thirty consecutive days of sampling data. Such a requirement
would be logistically complicated to the point of impracticality, and would be prohibitively expensive
given the costs of the sampler's time and the laboratory fees for each sample. This would thwart
enforcement in the not-uncommon situation where regulators have not imposed end of pipe numerical
effluent limitations but rather rely on general permit conditions that prohibit violation of water quality
standards.

C. EPA Must At Least Clarify How the 30-day Average is to be Implemented and Enforced

As explained in the Guidelines and Technical Support Document, the water-column based element
should be expressed as a four-day average. If EPA does not intend to utilize a four-day average, but does
intend to allow the 30-day average to be implemented and enforced based on less than 30 days of data,
EPA must at least clarify that. For example, if EPA intended that the 30-day average be implemented as a
monthly average, EPA should state that directly, and should reference 40 C.F.R. § 122.2, which defines
"average monthly discharge limitation" as "the highest allowable average of 'daily discharges' over a
calendar month, calculated as the sum of all 'daily discharges' measured during a calendar month
divided by the number of 'daily discharges' measured during that month." In other words, EPA should
clarify that compliance can be determined based on less than 30 samples taken within a given month.

III. The Intermittent Exposure Water Column Element Needs Clarification

In order to address the cumulative, chronic effects of shorter-term pulses of selenium pollution, EPA
included an intermittent exposure water concentration element in its Draft Criterion. Draft Criterion at
86. Compliance with this element is determined using an equation that involves the concentration of
selenium during pulse events as well as the "average background selenium concentration" during the
rest of the 30-day measurement period. Although the Commenters generally agree with such an
approach at this time, the element as proposed suffers from similar implementation problems as the
monthly average element. In particular, the data necessary to determine the "background
concentration" term of the intermittent exposure element equation will be lacking in most
circumstances. As explained above, thirty consecutive days of water column data are rarely available to
either citizens or regulators, such that, even where data exists to show high pulses of selenium,
compliance with the intermittent exposure element cannot be easily determined. EPA should make clear
that citizens and regulators may extrapolate from more limited data to determine the "background
exposure" occurring during the non-pulse days of the 30-day period.

EPA-HQ-OW-2004-0019-0385-A2; Kentucky Division of Water (KDOW); Posted 10/13/2015
Intermittent Exposure

110

-------
The element of the criterion which addresses intermittent exposure in the water column is presumably
intended to mitigate exposure to high concentration pulses or "spikes" in selenium entering a
waterbody and accumulating in food particles at a concentration that may cause chronic toxicity to the
aquatic habitat. The calculated intermittent criterion is to be determined site-specifically and depends
on the frequency and magnitude of selenium "spikes" over a 30-day period. To account for variability in
the selenium concentrations, the mathematical average is used as the input data. The calculated
intermittent criterion would be applied to the mathematical average of the spike concentrations. While
the formula EPA provided for "Intermittent Exposure" is sufficiently understandable, the
implementation of this criterion is not practical.

On its face, the Intermittent Exposure criterion element appears to have been developed and is
intended to apply to a narrow range of discharge scenarios or types. In so doing the criterion
unnecessarily complicates implementation as a "national" criterion. There are a wide variety of
discharge in which the intermittent exposure criterion could otherwise potentially be interpreted to be
applied if adopted and implemented by states. Consequently, this proposed EPA approach is
cumbersome and impractical to implement. EPA's contract peer review also concludes that "the
intermittent approach proposed is unnecessarily complicated if the simpler approach continues to
achieve the same objective."

KDOW requests that EPA defer to its contractor peer review which indicates that while the proposed
method for addressing intermittent and time-varying discharges may be reasonable if adequate data are
available, given that limited bio-kinetic data are available, it is premature to implement such as
approach for setting water quality criteria. As such, KDOW urges EPA to reconsider this approach and to
rely on what KDOW and EPA agree is the appropriate endpoint: fish tissue data. As with Kentucky's
chronic criterion, requiring tissue sampling in the event the water column concentration exceeds a given
threshold (i.e. 5.0 ng/L) protects the critical aquatic endpoint, which are monitored to detect any
potential toxicity concerns. EPA has clearly articulated and developed tissue based chronic criteria for
the purpose of protecting against chronic endpoints that the Intermittent Exposure criterion is also
intended to protect. EPA must ultimately succinctly and definitively answer the question: Are technically
sound tissue-based criterion protective or not, to which the science supports an affirmation in response.

Delete footnote 4 (July 27, 2015 80 FR No. 143 at 44352) and the "Intermittent Exposure" criterion as this
criterion is scientifically premature and is impractical to implement.

EPA-HQ-OW-2004-0019-0387-A1; Colorado Department of Public Health & Environment;
Posted 10/14/2015

The Intermittent Exposure criterion element seems complicated to implement. Would calculation of this
element require daily water collection to determine when and how long elevated selenium
concentrations are observed? If selenium concentrations are constant, can this element of the criterion
be disregarded? Does EPA expect that intermittent pulses of selenium occur in all types of waters, or are
there particular situations EPA has in mind where an intermittent criterion would be most applicable?

Ill

-------
EPA-HQ-OW-2004-0019-0413-A2; Federal Water Quality Coalition (FWQC); Posted
10/28/2015

1. Adjustment for Intermittent Exposures

In the Criteria Document, EPA has included a provision for adjusting the applicable water column criteria
to less stringent levels where the discharge of selenium is "intermittent." We agree with the concept
that intermittent exposure to selenium, as with other pollutants, is less toxic than continual exposure,
and we support the Agency's incorporation of this concept into the criteria. However, we have several
questions and concerns about the new "intermittent" criteria. For one thing, it is not clear when the
adjusted criteria would apply; EPA indicates that they are triggered when selenium is present at an
"elevated concentration." But that term is never defined. In addition, it is not clear how the criterion is
applied when the background level is high. It appears that the equation would lead to a zero limit in that
circumstance, which would obviously not be attainable.

EPA should also clarify the scope of the "intermittent" process - in particular, as to what types of
discharges it is meant to cover. EPA indicates that it would apply to "short-term exposures that
contribute to chronic effects through selenium bioaccumulation (e.g., stormwater overflows from
storage ponds or other intermittent releases." However, that discussion conflates two very different
types of releases: intermittent discharges of process wastewater, and releases of stormwater during and
after wet-weather events. In contrast to process discharges, stormwater discharges tend to peak quickly
(in terms of concentration of pollutants) and then decline rapidly as well. In these circumstances,
bioaccumulation is less likely, and there is also evidence that aquatic toxicity effects are reduced as
compared to constant-discharge situations. Until EPA has reviewed this issue, and determined the
nature of the selenium toxicity effects caused by stormwater discharges, it should not apply the
selenium criteria to those discharges.

I. INTERMITTENT CRITERION

One of the aspects of the criterion that most troubled the peer reviewers was the new criterion level for
intermittent exposures. They raised a number of serious concerns with that aspect of the criterion, and
suggested other options for EPA to consider. We believe that those concerns are well-founded, and that
the Agency should seriously consider adopting an alternative approach to intermittent exposures.

Here are some of the statements by the peer reviewers on the intermittent exposure issue:

"I also do not agree with the intermittent exposure criterion; it is unclear why it was developed, how it
could be implemented consistently and reliably, and in general I think it just adds too much complexity to
an already complex (indeedperhaps the most complex) water quality criterion." (p. 10)

"[tjhe discussion of background science, data and methods used in the intermittent exposure tier of the
present criterion needs significant improvement" (p. 12)

"The draft intermittent exposure Se criteria represent a mathematical manipulation of the monthly
average criteria in order to derive values that would still result in 30-day average concentrations of 4.8
and 1.3 jig/L for lotic and lentic waters, even if those were exceeded for x number of days. A limitation of
this approach is that it does not consider the uptake and elimination kinetics ofSe in aquatic food chains

112

-------
and the influence of exposure duration and magnitude on these biokinetic parameters. In my opinion, a
biokinetic modeling-based approach would be more appropriate for deriving intermittent, or acute,
criteria that are protective against exceeding fish tissue-based Se criteria." (p. 43)

"I am not sure that the criterion equation for intermittent dischargers is meaningful, as it is basically a
mathematical manipulation and does not in any way account for selenium uptake and elimination
kinetics. An alternative approach that the EPA may want to consider is based on biokinetic modeling,
such as that described in Brix and DeForest (2008). The method they described was based on modeling of
a food chain comprised of periphyton, an invertebrate (mayfly), and a fish (fathead minnow). Inputs to
the model include the background water Se concentration, the magnitude of an intermittent Se pulse,
and the duration of the Se pulse. This provides a tool for evaluating whether a Se pulse of a given
magnitude and duration could result in exceedance of a whole-body fish Se criterion, or short-term Se
criteria could be derived for given short-term durations.

For a comparison of the biokinetic-based approach to the intermittent criterion equation in the draft
AWQC document, I assumed that the background water Se concentration is 1 /jg/L, the lotic criterion is
4.8 /ug/L, and the number of days elevated is 4. The intermittent criterion would be 29.5 /jg/L. Just as an
example, if a lotic food chain consisting of periphyton^mayflies^fathead minnows were assumed, a 4-d
pulse of 29.5 /jg Se/L would not be nearly sufficient to reach a whole body Se concentration of 8.1 mg/kg
dw (Fig. 2). There is a rapid increase in predicted Se concentrations in periphyton and mayflies and then a
rapid elimination, but uptake is slower in fathead minnows.

In my opinion, a biokinetic-based modeling approach would be more appropriate for deriving acute or
intermittent water Se criteria." (p. 120)

"It is not clear how the intermittent criterion outlined in the Draft Document will be applied. The
mathematical expression of the criteria on page 93 is clear but the terms surrounding the application of
the criterion are not. For example, the criterion is not intended to apply to "ordinary smoothly varying
concentrations" (Page 94). However, what specifically will constitute a discharge curve that is not
"smooth" has not been defined. It is also not clear what magnitude of selenium concentration spikes
would designate a discharge as having to be regulated as an intermittent discharge. Finally, designation
of an intermittent criterion appears to contradict the data in Appendix G and the statement on page 94
that "kinetics of selenium accumulation and depuration are sufficiently slow that attainment of the
water criterion concentration element by ambient 30 day averages will protect sensitive aquatic life
species even where concentrations exhibit a high degree of variability. While outside the area of our
expertise it is noted that several comments in the public registry suggest that a biokinetic model may be
more appropriate than the application of an expansion of the 30-day average calculation for determining
intermittent criterion." (pp. 123-124)

In response to these concerns, here is EPA's primary response:

"EPA notes that several reviewers viewed the intermittent criterion element as a reasonable surrogate
for an acute criterion for protecting aquatic ecosystems, particularly downstream lentic waterbodies,
from the effects of intermittent discharges of selenium."

113

-------
We believe that this summary response is not adequate to address the significant problems with the
intermittent exposure element of the criterion that were raised by the peer reviewers. Before finalizing
the criterion, EPA needs to consider each of these issues and assess whether changes can be made in
the new criterion, or alternative approaches can be adopted, that would address these problems.

EPA-HQ-OW-2004-0019-0415-A1; Idaho Department of Environmental Quality (DEQ); Posted
10/28/2015

The intermittent water column exposure criterion does not appear to be very practicable. It is highly
unlikely Idaho, or any regulatory entity, will have sufficient daily selenium data to make use of it. EPA
should better explain the utility and application of the intermittent exposure criteria, or better yet,
simplify their proposal by dropping the intermittent criteria as impractical and unneeded.

The intermittent water column criterion is an interesting way to deal with variable exposure. It appears
this criterion is intended to deal with a waterbody that varies in concentration over time. It is not clear
whether EPA envisions or would allow application as well to variation in exposure as fish move among
waters of varying selenium concentrations.

EPA-HQ-OW-2004-0019-0412-A2; Utility Water Act Group (UWAG); Posted 10/28/2015

3.3 Intermittent-Exposure Water Criterion Element: Derivation From the Chronic Water Criterion
Element

EPA has proposed a criterion equation that would adjust the applicable lotic and lentic water criterion to
a less stringent criterion in settings where the discharge of selenium is intermittent. UWAG agrees with
the agency that the proposed fish tissue criterion would override the water-based intermittent criterion
in instances where fish tissue data are available. We also agree that there is merit to the proposed
intermittent exposure value based on the understanding that intermittent exposures to trace elements
and other pollutants is generally less toxic to aquatic life relative to continual exposure. However,

UWAG strongly disagrees with the suggestion that the intermittent value should be applied to
stormwater overflows as suggested by EPA in the Executive Summary of the Draft Criterion (at xiii):

EPA is recommending the intermittent value to address short-term exposures that contribute to
chronic effects through selenium bioaccumulation (e.g., stormwater overflows from storage
ponds or other intermittent releases).

Stormwater discharges are chemically and physically distinct from traditional process waste-stream
discharges. During storm events, concentrations of pollutants increase drastically in a short period of
time, a peak concentration is observed, and then the concentration of pollutants decreases rapidly. For
example, Riscassi et al. (2010) reported that, during a storm event, the cycle of increasing total mercury
concentration, peak concentration, and declining concentration lasted no more than 12 hours. The rapid
peak and declining concentration of pollutants during a storm event is fundamentally distinct relative to
batch or intermittent discharges of process waste-streams. Thus, the potential for bioaccumulation of

114

-------
selenium in the proximal food web is significantly reduced during storm events. Lastly, EPA does not
require states to implement numeric water quality-based effluent limitations for stormwater discharges
(memo from Andrew Sawyers and Benita Best-Wong, US EPA Office of Wastewater Management, to US
EPA Region Water Directors dated November 26, 2014). UWAG requests EPA delete language from the
final criterion that suggests the intermittent criterion should apply to stormwater discharges.

Additionally, UWAG believes the agency should have made an effort to gather information on the
kinetics of intermittent selenium exposure, although UWAG acknowledges that there are likely few
published papers on chronic tests of aquatic life exposed to intermittent pulses of selenium. In one such
study, Hoang and Klaine (2008) reported results of a 21-day chronic test using Daphnia magna exposed
to intermittent pulses of selenium. The study indicated that:

• Survival of D. magna was higher following a second pulsed exposure compared to the first,
suggesting some kind of tolerance induction following the first pulse.

• Growth of D. magna was not affected by the duration or magnitude of the pulses. Likewise,
reproduction was not affected by pulse duration or magnitude, indicating that D. magna fully
recovered (or tolerance was induced) after the pulses.

It is certainly possible that intermittent chronic exposure to selenium could induce tolerance in aquatic
organisms other than D. magna. As such, UWAG requests that EPA evaluate this issue prior to finalizing
the freshwater selenium chronic criterion. Because the most sensitive endpoint to assess selenium
toxicity is survival and deformity incidence in fish, EPA should seek to conduct studies to evaluate
whether intermittent exposure causes the same magnitude of reproductive effects that continual
exposure induces.

EPA-HQ-OW-2004-0019-0406-A2; Kentucky Chamber of Commerce; Posted 10/15/2015

The Chamber also questions the need for an intermittent water column exposure criterion. U.S. EPA's
intent is to mitigate exposure to occasional high concentration spikes in selenium in a discharge.

Because the proposed formula is highly reliant on the availability of adequate data on individual
streams, it would create significant implementation concerns. Additionally, since only chronic impacts
are at issue and the fish tissue standard adequately and directly determines the existence of any
"chronic" impact endpoint, an intermittent standard (even if it were shown to be a reliable indirect
indictor of a chronic endpoint) is not necessary. Finally, the intermittent standard is based upon "further
modeling" beyond that used for determining the water column concentration criteria for lotic and lentic
aquatic systems, and therefore it is also unreliable and unproven. (Criterion Document at p. 26). The
Chamber requests that it not be included in the final recommended criterion.

EPA-HQ-OW-2004-0019-0391-A1; North American Metals Council (NAMC); Posted
10/14/2015

The key problem with the manner in which EPA has approached the development of waterborne
selenium concentrations is that the criteria derived from medians do not apply to medians but rather to

115

-------
the 30-day once-in-three years, upper-tail of the distribution value. There is a clear disconnect between
criteria development and application that will provide ample work and fees for consultants, will have
negative economic consequences, and will not notably advance environmental protection.

116

-------
Comment Category 2.5 - Comments Concerning Items Considered out
of Scope

2.5 Summary

This section includes comments that EPA considered "out-of-scope" given the purpose and objectives of
EPA's 2016 Selenium Aquatic Life Water Quality Criteria recommendations. A range of topics are
commented on such as acute toxicity of selenium, selenium impacts on aquatic-dependent wildlife, the
role sulfate might play in toxicity to aquatic organisms, the applicability of the criterion to marine
environments, economic impacts associated with criterion implementation, and questions about
previous (2002 and 2004) selenium criteria drafts.

2.5 Response to comments

2.5 Response to comments that are out of scope

As shown in Table 2.2 of the 2015 draft, EPA has indicated that acute toxicity is not included in the
assessment. EPA notes that several peer reviewers viewed the intermittent criterion element as a
reasonable surrogate for an acute criterion for protecting aquatic ecosystems, particularly downstream
lentic waterbodies, from the effects of intermittent discharges of selenium. Many public comments
support the development of intermittent criterion element in lieu of an acute criterion.

Regarding risk to aquatic dependent wildlife, EPA understands the potential for risk to birds from
selenium exposure and has begun work on a national criteria for selenium that would protect aquatic-
dependent wildlife.

EPA decided not to include a sulfate correction factor in the 2016 selenium criterion due to uncertainties
in the science. The Deforest et al 2014 report referred to in the NAMC and API 2015 public comments
notes that a sulfate-dependent selenium criteria would apply only to selenate-dominated, well-
oxygenated streams, which is a small subclass of waters in the US. The publication discussed
experiments to assess influence of sulfate on selenate uptake on only one species of macrophyte
(Lemna minor) and one algal species (Pseudokirchnella subcapitata), a very limited data set of primary
producers. The authors themselves note that "It does need to be emphasized here, however, the analysis
currently does not include Se data for periphyton and benthic diatoms, as these data are not available."
The authors also note that "due to methodological challenges and high costs, it is difficult to
comprehensively evaluate the influence of sulfate on bioconcentration and transfer up the food chain."
Similarly, EPA describes effects of mercury on aquatic life, but did not adjust the chronic criterion.

Regarding the need to clarify that the criterion applies to freshwater, the criterion document cover page
indicates this is a freshwater criterion. EPA has added additional text in the body of the criterion
document to ensure that this is clear. It is important to note, however, that the approach used in the
development of the Freshwater Selenium criterion could clearly be applied to other systems, if data for
those systems (e.g., estuarine/ marine food web data, surrogates) were available.

117

-------
Finally, EPA has reviewed all available reliable data and used that as a basis for deriving the 2016 chronic
selenium criterion. This criterion document replaces the 2002 and 2004 draft criteria documents, and
EPA is not responding to comments on those older documents at this time.

2.5 Specific comments

EPA-HQ-OW-2004-0019-0399; Clean Water Team; Posted 10/15/15

The economic burden arid western states will incur complying with the proposed regulation is enormous
and essentially an unfunded mandate to the local community's ratepayers. The proposal for site specific
criteria can help address ecoregion concerns; however, this is an additional cost. We urge EPA to revise
the regulation to include ecoregion specific stipulations especially for criteria that can never be
exceeded in the present draft.

EPA-HQ-OW-2004-0019-0385-A2; Kentucky Division of Water (KDOW); Posted 10/13/2015

The proposed national criterion and its elements appropriately do not include a criterion for acute
toxicity since the mode of action for toxicity in the aquatic habitat is dietary driven; the pronounced
toxic effects of concern are lethality and embryonic teratogenicity driven by the cellular incorporation of
seleno-amino acids. Thus, the mode of toxicity is unlike oral ingestion or dermal sorption that may lead
to system lethality in developed organisms. Therefore, short-lived, high water-column concentration
exposure is not indicative of a toxicity problem.

Establish that the chronic selenium criterion is sufficient to protect streams and aquatic habitats, and
forego any recommendation for state adoption of an acute criterion.

EPA-HQ-OW-2004-0019-0390-A1; West Virginia Coal Association (WVCA); Posted 10/14/15

The Draft Selenium Criterion do not offer adequate detail regarding site-specific factors that inhibit or
enhance selenium toxicity, such as sulfate, and how these factors should be incorporated into
implementation of the criteria.

EPA-HQ-OW-2004-0019-0410-A2; West Virginia Department of Environmental Protection
(DEP); Posted 10/28/2015

EPA's 2015 Draft, like the 2014 Peer Review Draft, removes the acute water quality criterion of 20 ng/L.
DEP agrees with EPA that acute toxicity to selenium occurs at higher concentrations than chronic effect
concentrations, negating the need for an acute selenium criterion in the presence of a protective
chronic criterion.

118

-------
EPA-HQ-OW-2004-0019-0387-A1; Colorado Department of Public Health & Environment;
Posted 10/14/2015

In Appendix A, page A-2, EPA discusses evidence of the relationship between selenate and sulfate (i.e.,
sulfate can reduce the accumulation of selenate). Specifically, EPA stated "A significant inverse
relationship was shown to exist between acute selenate toxicity to aquatic organisms and ambient
sulfate concentrations (Brix et al. 2001 a). Competition with selenate has also been observed for
phosphate in green algae (Riedel and Sanders 1996), and with chromate and tungstate in anaerobic
bacteria (Oremland et al. 1989)". However, it does not appear that the effects of sulfate were
considered in development of the draft criterion. If it is possible for sulfate to affect bioaccumulation of
selenium in the food chain, thus resulting in less selenium in fish tissues than would be predicted by
food web modeling, it seems this would be an important consideration. The WQCD requests that EPA
consider adding more information about why sulfate was not considered in the current draft criterion,
especially considering its importance in previous selenium criteria documents. A section similar to the
"Interactions with Mercury" section would be useful.

EPA-HQ-OW-2004-0019-0411-A2; American Petroleum Institute (API); Posted 10/28/2015
2.3.5 Alternative Sulfate-dependent Selenium Guideline Approach

Sulfate (S04) competes with selenate for uptake by algae and macrophytes (Williams et al. 1994; Lo et
al. 2015) and, therefore, can influence the magnitude of EFs in selenate-dominated systems and the
amount of Se that is ultimately bioavailable to fish. Most recently, Lo et al. (2015) provided a
comprehensive study on how sulfate concentrations influenced the uptake of selenate by an alga and a
macrophyte. DeForest et al. (2014; in revision) provide an example of how these data could be used to
develop sulfate-dependent Se guidelines for selenate-dominated waters. This included the following
steps:

• Compiled paired water selenate, water sulfate, and particulate Se concentrations from
laboratory studies (namely the data recently published in Lo et al. [2015]);

• Applied particulate-to-invertebrate and invertebrate-to-fish eggs regression models to
particulate Se concentrations;

• Conducted multiple quantile regression to predict fish egg Se concentrations (dependent
variable) as a function of water selenate and water sulfate concentrations (independent
variables); and,

• Potential S04-dependent water Se screening guidelines were then validated against colocated
water Se, water sulfate, and fish tissue data from lotic field sites presumed to be dominated by
selenite.

Using the multiple quantile regression analysis offish egg Se concentrations predicted as a function of
waterborne selenate and sulfate concentrations to which particulates were exposed, the following 75th
quantile regression equation was derived:

Original letter contains Equation - not numbered. See original letter.

119

-------
This equation can be rearranged to solve for the waterborne Se concentration which predicts a given
fish egg Se concentration at a given waterborne sulfate concentration:

Original letter contains Equation - not numbered. See original letter.

Assuming the 75th quantile and an egg Se guideline of 20 mg/kg-dw (DeForest et al., 2012), water Se
guidelines were calculated at a large number of lotic field sites from which co-located water Se, water
S04, and fish egg Se data were available. The fish Se ratios (ratio of fish egg Se concentration to the
guideline of 20 mg/kg-dw) corroborated the waterborne Se ratios (ratio of waterborne Se concentration
to the sulfate-dependent guideline) 73% of the time, with 28% of the cases resulting in false positives
and 5% of the cases in false negatives (Figure 5). Therefore, the waterborne Se ratio based on the
sulfate-dependent guideline was protective in 95% of the samples.

Original letter contains Figure 5 - Relationship between mean fish egg Se and waterborne Se ratios
based on the sulfate-dependent screening guideline. See original letter.

Using this equation with the egg/ovary value of 15.8 mg/kg and sulfate concentrations ranging from 50
mg/L to 300 mg/L would result in protective water column criteria ranging from 8.1 ng/L to 97.0 ng/L,
respectively (Table 4).

Original letter contains Table 4 - Protective water column Se concentrations at various sulfate
concentrations. See original letter.

2.4 Applicability to Estuarine Waters

The 2015 draft AWQC are explicitly identified as freshwater criteria. However, it is not clear from the
current draft, whether the EPA would ever intend for the freshwater criteria to be applicable, without
modification, to estuarine or other saltwater environments. This seems counter-intuitive. We
recommend that the EPA explicitly address this issue in the final version of the AWQC document.

EPA-HQ-OW-2004-0019-0391-A1; North American Metals Council (NAMC); Posted
10/14/2015

2.3.4 Selenium and Mercury

The 2015 EPA Selenium Draft does not fully evaluate the interactions of mercury with selenium, which
can reduce selenium toxicity. Although already noted in our comments on the 2014 EPA draft
(Attachment 1), NAMC again recommends that EPA fully evaluate the extensive evidence for
modification of selenium toxicity by mercury relative to the potential for these interactions to comprise
exposure and toxicity modifying factors that should be considered when applying the proposed criteria.

EPA-HQ-OW-2004-0019-0402-A2; The Stakeholders Implementing TMDLs in the Calleguas
Creek Watershed; Posted 10/15/2015

As described in the Stakeholder's comment letter on the 2014 External Peer Review Draft Criterion,
selenium is an issue for many municipalities within Southern California due to the natural geology of the

120

-------
Monterey Formation. This formation is a known source of selenium in California and surface water
concentrations tend to be higher where the geologic formation is present; however, bioaccumulation
and the resulting impacts to beneficial uses varies significantly. Many efforts throughout the State have
focused on tissue-based approaches to more appropriately manage and regulate selenium, with an
emphasis on protection of aquatic life (fish) and aquatic-dependent wildlife (birds).

In certain areas, such as Orange County, local agencies have worked for over a decade with the United
States Environmental Protection Agency (USEPA), United States Fish and Wildlife Service (USFWS),
United States Geological Survey (USGS), the State Water Resources Control Board, and the Santa Ana
Regional Water Quality Control Board, to develop revised objectives for selenium.

In the Calleguas Creek Watershed, Stakeholders have been implementing a Total Maximum Daily Load
(TMDL) for selenium1 since 2006. Implementation efforts have resulted in a greater understanding of
the relationship between water column concentrations and tissue concentrations (birds and fish) in the
watershed. As a result, stakeholders are considering the development of a site-specific objective.

Additionally, there are potentially significant cost implications and the potential need to address natural
sources in the Calleguas Creek Watershed that could be avoided by implementing the changes
requested in the summary of key issues and specific technical comments contained herein. Not
implementing the changes requested in the summary of key issues and specific technical comments
contained herein may unnecessarily require implementation of significant and costly BMPs
(predominately to address natural sources) with no net environmental benefit. Based upon evaluations
conducted over the last 10 years in Southern California, these additional unnecessary costs can range in
the hundreds of millions of dollars. Therefore, the implications of the proposed revisions to the
freshwater selenium criterion are an important issue in Southern California.

EPA-HQ-0W-2004-0019-0409-A2; OC Public Works; Posted 10/20/2015

OC Public Works is concerned that the Draft Se Criterion, in the face of this newly available science, will
rapidly become obsolete and inappropriate. This may have a significant impact on on-going selenium-
related actions in Orange County. For example, a tissue-based site-specific selenium water quality
objective (SSO) is currently under development for the Newport Bay watershed, that will consider
protection of not only aquatic species (mainly fish), but also aquatic dependent wildlife (mainly birds,
including several T&E species). Since selenium is such a unique contaminant, most, if not all, of the new
issues raised at the meeting are relevant to how the aquatic life selenium criterion is derived. Therefore,
EPA should carefully review the information from the Invited Expert Meeting and ensure that the most
up-to-date science is incorporated in the derivation of the Draft Se Criterion.

9. Aquatic-Dependent Wildlife Protection

Protection of wildlife-dependent species is not part of the Draft Se Criterion. Not only will this pose an
issue for the Endangered Species Act Section 7 review process that could complicate the adoption of the
Criterion, future reconsideration of the aquatic-dependent wildlife would require a completely new
analysis of the same dataset and could takes years to complete. This may not be an efficient approach.

121

-------
EPA-HQ-OW-2004-0019-0394-A2; Appalachian Mountain Advocates and Sierra Club
Environmental Law Program; Posted 10/14/2015

V. The Criterion Must Protect Aquatic-Dependent Wildlife

The Clean Water Act mandates that water quality standards protect not only fish, but all aquatic
organisms and other wildlife that depend on healthy streams. Section 303(c) requires that such
standards "shall be established taking into consideration their use and value for.. . propagation of fish
and wildlife." among other things. 33 U.S.C. § 1313(c)(2)(A) (emphasis added); see also 33 U.S.C. §
1252(a) (directing states to develop comprehensive programs for controlling water pollution giving due
regard to improvements necessary to "conserve such waters for the protection and propagation offish
and aquatic life and wildlife"). EPA's regulations require states to develop standards that will "[s]erve
the purposes of the Act/' meaning that they will "provide water quality for the protection and
propagation offish, shellfish and wildlife." among other things. 40 C.F.R. § 130.3 (emphasis added).
Commenters are not aware of any states that have adopted selenium water quality standards
specifically for the protection of aquatic-dependent wildlife and EPA does not have a Recommended
Criteria for selenium to protect aquatic-dependent wildlife. In the absence of any standards that address
wildlife, an approach that focusses solely on aquatic life does not satisfy the requirements of the CWA
because it leaves such wildlife without any protection under the Act from selenium pollution.

Although EPA did not analyze the impacts of its criterion on aquatic-dependent wildlife, existing
evidence makes clear that the concentrations of the proposed fish tissue elements are not protective of
aquatic dependent wildlife. In 2004, EPA proposed but did not adopt recommended criteria that
included a whole-body fish tissue criterion of 7.91 ng/l, which is more protective than EPA's current
proposal. See Notice of Draft Aquatic Life Criteria for Selenium and Request for Scientific Information,
Data, and Views, 69 Fed. Reg. 75, 541 (December 17, 2004). A group of the nation's leading selenium
scientists wrote a white paper vigorously criticizing that criterion as not protective and too high. The
authors explained the history of the EPA's flawed number:

During the past 17 years numerous researchers including those funded by EPA have estimated
that the toxicity threshold for selenium lies below the current chronic aquatic life criterion of 5
Hg/L. Recently, corporate interests have claimed that 5 ng/L is overly restrictive. Because of an
endangered species issue in California, EPA agreed to re-evaluate their CWA criteria guidance
for selenium by 2002. This was problematic because:

• EPA's normal procedure for setting Aquatic Life Criteria does not directly consider
toxicity data for aquatic-dependent wildlife

• EPA has promulgated no separate wildlife criteria for selenium.

• EPA's normal procedure for setting criteria is better suited to non-bioaccumulative
pollutants - selenium is bioaccumulative.

• ESA-listed species every individual of a population "counts" and therefore criteria
guidance would need to be fully protective at an individual-effects level.

122

-------
EPA contracted with the Great Lakes Environmental Center (GLEC) to derive the new selenium
criteria. GLEC was instructed to derive the chronic criterion on a fish-tissue basis rather than on
a water concentration basis. The GLEC derived criterion was released in March 2002. The draft
tissue-based chronic criterion, of 7.9 ng/g, dry weight basis, assumed 20% of the target
population would die. The USFWS asked EPA to not promulgate the criterion because it wasn't
protective of endangered species.

Joseph P. Skorupa, USFWS, Theresa S. Presser, USGS, Steven J. Hamilton, USGS, A. Dennis Lemly, USFS,
Brad E. Sample, CH2M HILL, EPA's Draft Tissue-Based Selenium Criterion: A Technical Review. Spring
2004 at 2-3.

The authors noted significant additional flaws in EPA's proposed criterion that would lead to harm to
wildlife, including threatened and endangered species:

GLEC's assessment of risk to aquatic-dependent wildlife was based on an erroneous draft
wildlife toxicology report. The draft tissue-based chronic criterion for selenium of 7.9 ng/g
would leave a substantive proportion of aquatic- dependent wildlife species unprotected; on the
order of half the species. Aquatic life criteria are considered by EPA to be separate and distinct
from wildlife criteria. Nonetheless, in the absence of promulgated wildlife criteria (as is the case
for selenium), if the aquatic life criteria do not protect wildlife the purposes of the CWA are not
being met. More critically, for waters of the United States supporting ESA-listed aquatic-
dependent wildlife, the criteria would not be approvable for incorporation into state or tribal
water quality standards.

Id Those experts estimated that EPA's previously proposed criterion would have caused reproductive
impairment in, conservatively, 40% and possibly as high as 95% of exposed mallard ducks. See Lemly, A.
Dennis, Assessing the toxic threat of selenium to fish and aquatic birds, Environmental Monitoring and
Assessment 43: 19-35 (1996). Reproductive impairment occurs if ducks are exposed through a
contaminated diet during the development of their chicks. Mallard ducks are ubiquitous, breeding near
and relying on aquatic resources throughout the US. They are primarily vegetarians eating seeds of
grasses and sedges and the leaves, stems and seeds of aquatic plants. They occasionally eat insects,
crustaceans and mollusks, especially when they are young. See

http://www.nhptv.org/natureworks/mallard.htm. While the ducks do not eat fish, "allowing fish tissue
to reach 7.9 ug/g would allow a level of contamination in the other parts of the aquatic ecosystem
sufficient to cause nearly total reproductive failure among mallard ducks." Skorupa et al. at 22. Indeed,
EPA explicitly recognizes that its proposed Criterion may not be protective of certain wildlife species,
particularly those that regularly feed on mollusks. Draft Criterion at 131. The USFWS previously stated
that a protective fish tissue standard for water birds would be 5 ng/g selenium, much lower than EPA's
proposed whole body element of 8.0 mg/kg. See, e.g.. Letter from Virgil Lee Andrews, USFWS Kentucky
Field Office Supervisor to Annie Godfrey, Chief of USEPA Water Quality Standards Section, December 27,
2013. More recently, USFWS stated that a whole body criterion of about 4 mg Se/kg would be necessary
to protect mallards. July 28, 2014 Comments at 18. EPA thus must either revise its fish tissue elements
to ensure that they protect aquatic-dependent wildlife or else issue a concurrent wildlife criterion that
must be adopted along with EPA's recommended aquatic life criterion.

123

-------
Comment Category 2.6- Comments on Various Implementation Topics
2.6 Summary

The purpose of the selenium criterion document is to set forth EPA's basis for and derivation of the
water quality criterion recommendations for protecting aquatic life from the harmful effects of
selenium. Many commenters expressed the need for EPA to develop implementation materials,
provided feedback on how the criterion should be implemented, and some identified specific topics that
EPA should address in upcoming implementation materials. There were a wide range of these types of
comments including: how to measure attainment in monitoring, how to distinguish between lotic and
lentic systems, and how to establish NPDES permit limits and thresholds for 303(d) listings and TMDLs.

2.6 Response to comments

2.6 Response to comments concerning implementation

EPA's criterion provides recommendations for states and authorized tribes to consider in their adoption
of water quality standards under CWA section 303(c). The implementation documents that EPA is
developing are intended to provide assistance to states and authorized tribes that adopt into the water
quality standards a criterion based on or similar to EPA's recommended criterion. The implementation
documents are also intended to provide assistance to other stakeholders and the public. If EPA refers to
implementing the 304(a) criterion, it means implementing a state- or tribally-adopted criterion based
on, or similar to, EPA's recommended criterion EPA is currently working on developing technical
information that will help in the implementation of a selenium criterion in the many areas identified in
Specific comments. EPA recognizes that there are numerous aspects of the recommended criterion that
will benefit from technical support documents to enhance implementation of state and tribal criteria,
and is planning to develop such documents and make them available for public comment. Note,
background selenium is addressed in section 1.1 of the response to comments.

2.6 Responses to comments on implementation materials addressing the hierarchy of elements

The design of the tiered criterion is such that the hierarchy allows for the assessment of samples that
are available (i.e., egg-ovary, muscle, whole body, and water). There are practical and scientific
considerations for the collection and assessment of specific types of samples which EPA plans to address
in a detailed technical support document. For example, whole body or muscle tissue from fish may
provide the best estimate of critical body burdens of selenium in fish (male or female) that are not in the
process of spawning. This and other issues will be discussed in the technical support document on fish
tissue monitoring which is under development by EPA at this time.

2.6 Responses to comments on implementation materials addressing collection offish tissue

Regarding collecting fish tissue samples, EPA agrees that technical support information is needed on
sampling approaches for each element of the tissue criterion, since there are various situations where
one tissue type may be favored over another, or the sample logistics (e.g., fish size) may limit the sample
to a certain type of tissue sample. EPA is developing information for states, tribes and stakeholders to

124

-------
consider when sampling fish tissue for the purposes of implementing the freshwater selenium criterion.
In addition EPA has derived and is recommending that states adopt all of the fish tissue elements to
provide maximum flexibility for a states monitoring and assessment program. Adopting all elements
covers all potential logistical, spatial, temporal, and species- and life history-specific considerations.

2.6 Responses to comments on implementation offish tissue in the GLI

EPA thanks MDEQfor the comment identifying implementation issues for selenium fish tissue elements
in the GLI methods. EPA intends to work with MDEQ and the environmental agencies in the Great Lakes
states to address those concerns.

2.6 Response to comments on Analytical Methods

Several commenters requested more information on analytical methods for selenium, especially for fish
tissue. Such methods will be described in implementation materials.

2.6 Response to comments on use of methyl mercury guidance for implementation of the
selenium criterion.

A few comments suggested EPA utilize an existing guidance on fish tissue monitoring issued in 2001 for
methylmercury in implementing the selenium criterion. One key difference between these two
documents is that they were developed for different purposes. The 2001 guidance was developed to
support monitoring for the purposes of developing fish consumption advisories to protect human health
from potential effects of methylmercury based on human diets from a variety of sources. The selenium
fish tissue criterion element was developed to support the protection of aquatic life with fish diets
based on a single ecosystem (with areal extent dependent on species being assessed). There are a
number of issues relevant for aquatic life that must be considered that are not addressed in the 2001
methylmercury guidance to protect human health.

2.6 Response to comments on sampling window recommendations

Several commenters provided comments supporting a 1-year sampling window recommendation
attributed to EPA. EPA has not made such recommendations in the selenium criterion document.
Discussions of appropriate sampling windows will be addressed in implementation materials currently
under development.

2.6 Specific comments

EPA-HQ-OW-2004-0019-0398-A1; Kansas Department of Health and Environmental
Watershed Planning, Monitoring & Assessment Section; Posted 10/15/15

Our monitoring protocols do not accommodate calculation of a 30-day average of selenium
concentrations to assess against the proposed 3.1 ng/l water column criterion. We would appreciate

125

-------
guidance on how to use sample values collected typically once per calendar quarter to assess against the
proposed criterion. We would also like some clarification on how selenium is intended to be analyzed.
The proposal frames selenium as "dissolved total selenium", whereas our laboratories analyze for "total
recoverable selenium". We are unsure if our analyses produce values that may be directly comparable
to how selenium is expressed in the EPA proposal.

Reliance on the newly proposed ambient water column criterion of 3.1 ng/l for lotic waters will surely
increase the number of streams listed under Section 303(d) for impairment by selenium, given current
monitoring protocols are hampered by a preponderance of instantaneous grab samples that weakly
represent 30-day averages as well as limited availability of laboratory resources to conduct fish tissue
analyses.

KDHE appreciates this opportunity to comment and looks forward to working with EPA to properly
assess impairments to aquatic life in Kansas streams caused by natural and anthropogenic-influenced
selenium.

EPA-HQ-OW-2004-0019-0379-A1; Pennsylvania Department of Environmental Protection;
Posted 09/30/2015

• The concepts and methodologies used by EPA to develop this criterion, as well as the
recommendations on how this multi-medium based criterion is to be implemented, are
relatively new. Therefore, we recommend that the Technical Guidance Document precede the
final aquatic life criterion for Selenium.

• EPA wants states to adopt a criterion that includes two media, fish tissue and water-based. The
Clean Water Act program requires the expression of water quality criteria as ambient
concentrations in water. Other fish tissue criteria developed by EPA, such as the methyl mercury
fish tissue based water quality criterion, have been difficult to implement. We see no evidence
that this criterion will be any different.

• The proposed criterion states, "New or increased inputs will likely result in increased selenium in
the food web, likely resulting in increased bioaccumulation of selenium in fish over a period of
time until the new or increased selenium release achieves a quasi-"steady state" balance within
the food web. Estimates of time to achieve steady state under new or increased selenium input
situations are expected to be site dependent, so local information should be used to better
refine these estimates for a particular waterbody. Thus, EPA recommends that fish tissue
concentration not override water column concentration until these periods of time have passed
in lotic and lentic systems, respectively." It is not clear how to implement this recommendation.
EPA needs to clarify how EPA expects the "steady state" of a waterbody should be determined.
What types of "local information" does EPA consider applicable?

• States may be reluctant to adopt this criterion because its conversion from a fish tissue
concentration to a water concentration is extremely difficult to achieve. The proposed fish
tissue criterion would require states to develop site specific water concentration values based
on one of two models; a mechanistic modeling approach or using a field derived
bioaccumulation factor. Both of these conversion techniques require more than one analytical
determination to calculate a protective limit.

• We recommend EPA obtain early approval on the selenium aquatic life criterion from the U.S.
Fish and Wildlife Service (USFWS). Comments received by EPA from USFWS on the pre-draft

126

-------
selenium document contained a number of concerns relating to incomplete data used in the
criterion determination. USFWS recently blocked EPA approval of the nationally recommended
aquatic life criterion for Nonylphenol because they believed the national recommended
criterion is not protective of endangered mussels. We had adopted this criterion for
Nonylphenol in our most recent triennial review of water quality standards. We would like to
avoid being in the same situation relative to selenium.

EPA-HQ-OW-2004-0019-0407-A2; Michigan Department of Environmental Quality (MDEQ);
Posted 10/15/2015

Thank you for the opportunity to provide comments on the Draft Aquatic Life Ambient Water Quality
Criterion for Selenium - Freshwater 2015. The aquatic life criterion for selenium is of interest to the
Michigan Department of Environmental Quality (MDEQ), Water Resources Division (WRD), since there
are some surface waters in Michigan with elevated selenium levels. We appreciate the United States
Environmental Protection Agency's (USEPA) efforts to develop a scientifically defensible surface water
criterion for selenium. We have the following comments on the draft document:

The Methodology Conflicts with the Great Lakes Initiative (GLI)

Michigan promulgated into state rules the selenium Criterion Continuous Concentration of 5
micrograms per liter derived by the USEPA as part of Title 40 of the Code of Federal

Regulations, Part 132, Water Quality Guidance for the Great Lakes System, commonly known as the GLI.
Michigan also promulgated the methodology that the USEPA developed for deriving surface water
values protective of aquatic life into our state rules. As you know, a methodology for developing a fish
tissue level-based water quality standard (WQS) was not included as part of the GU. It is our opinion
that we are required to retain our current selenium WQS until the USEPA revises the numerical WQS in
the GLI. The USEPA should rescind the current selenium criterion and modify the methodology provided
in the GLI to allow the Great Lakes states to use a fish tissue criterion to protect aquatic life from the
effects of selenium.

Additional Guidance is Needed

The document developed by the USEPA provides much detail on how the criterion was derived.
However, it does not provide adequate guidance on how to apply the criterion. For example, additional
guidance for the following is needed:

1. Species selection for tissue-based criteria attainment measurement is critical. It is also
inherently more complicated than water column-based criteria attainment measurement
because many different fish species might be considered for attainment sampling. Because of
this complexity, the US EPA should provide fish species selection guidance for tissue-based
criteria attainment measurement.

2. Footnote No. 3 in Table 1 states that the surface water criterion for selenium will take
precedence over the fish tissue criterion when there are "new or increased inputs of selenium
until equilibrium is reached." Guidance needs to be provided on how states should define a new
or increased input of selenium and how the state can determine that selenium is at equilibrium
in the receiving water.

127

-------
3. The USEPA derived a criterion for total selenium as part of the GLI. The proposed criterion for
selenium is based on dissolved selenium. Since many states develop wasteload allocations based
on total selenium, the USEPA should develop a translator to convert dissolved selenium to total
selenium.

EPA-HQ-OW-2004-0019-0413-A2; Federal Water Quality Coalition (FWQC); Posted
10/28/2015

I. IMPLEMENTATION ISSUES FOR NEW CRITERIA

Beyond the derivation of the numeric levels in the Criteria Document, there are a number of other
issues with the application of the new criteria that EPA should consider, which are set forth below.
Further, as a general matter, it is important for the Agency to recognize that due to the format of the
new criteria (which we believe is scientifically appropriate), implementation issues will be somewhat
different than for other criteria that the Agency has developed in the past. Therefore, EPA should
develop draft implementation guidance, and should make that guidance available for public comment
before it issues the new criteria in final form.

1. Fish Tissue Data Taking Precedence

EPA has stated that while the new draft criteria have four parts (two based on fish tissue levels and two
based on water column levels), the fish tissue elements "take precedence" over the water column
elements. We believe that this precedence policy is fully consistent with, and even mandated by, the
scientific information available, which shows that effects on aquatic life from selenium are indicated
most directly through analysis of fish tissue selenium levels, rather than review of selenium levels
present in waterways. However the EPA criteria document does not make entirely clear what the "take
precedence" policy will mean in practice. For example, in a situation where fish tissue levels in a water
body are below EPA's specified criteria, but the water column levels are above the EPA criteria, we
believe that the scientific conclusion to draw is that the selenium levels are acceptable, and that the
waterbody is not impaired. Since the new criteria will be used (after adoption by States) to make
attainment decisions, EPA should make it clear that in making those decisions, available fish tissue
should dictate the result.

VI. COMMITMENTS TO ISSUE GUIDANCE

Throughout the Response to Peer Review Comments, EPA frequently responds to questions and
concerns raised by the peer reviewers by indicating that it will address those issues in implementation
guidance. As a general matter, EPA states as follows:

"EPA recognizes that there are numerous aspects of the criteria that will benefit from technical support
documents to enhance its application, and is planning to develop such documents." (p. 11)

The Agency also states that:

"EPA, in a parallel process to the criterion development, is working to develop technical information that
will assist in the implementation of the selenium criteria."

We support these commitments by the Agency. It will be critical, as the guidance is developed, for EPA
to issue draft documents for public review and comment. It will also be important for the guidance to
cover the gamut of issues that will arise as the new criterion is implemented. To assist in compiling that

128

-------
list of issues, we have reviewed the Response to Peer Review Comments to cull out the specific
commitments that EPA has made in that document to develop guidance on particular issues. Here they
are:

"Regarding the interpretation of the "never to be exceeded" criteria, this is an issue that will be
addressed in the technical support documents." (p. 18)

"The design of the tiered criterion is such that the hierarchy allows for the assessment of
samples that are available (i.e., muscle, egg ovary, or whole body). There are practical and
scientific considerations for the collection and assessment of specific types of samples which
EPA plans to address in a detailed technical support document. For example, whole body or
muscle tissue from fish may provide the best estimate of critical body burdens of selenium in
fish (male or female) that are not in the process of spawning. This and other issues will be
discussed in the technical support document on fish tissue monitoring which is under
development by EPA at this time." (p. 20)

"EPA also thanks you for the important information on seasonal considerations of collection of
egg-ovary versus whole body or muscle tissue and will consider this information in the technical
support document on fish tissue sampling for selenium." (p. 22)

"The issue of temporal considerations on collection of a specific tissue type (egg ripeness) for
selected species is relevant and will be discussed (to the extent information is available) in the
information developed for states to aid in the implementation of the criterion." (p. 23)

"However, the agency is aware that spawning strategy likely plays a role in accumulation, and
maternal transfer and deposition of selenium into the eggs. This issue will be addressed in the
technical support materials under development regarding the timing and type of sampling based
on species spawning strategy (single batch versus multiple smaller batches) and seasonality
(spring or fall [some salmonids] versus sporadic spring/summer [e.g., cyprinids]), while also
providing flexibility to states to adapt their monitoring programs to address the additional
requirements for fish tissue monitoring for selenium." (p. 32)

"Technical support information regarding fish tissue sampling issue is being developed by EPA."
(p. 40)

"EPA realizes that the nature of selenium bioaccumulation may be effectively reflected through
the development of site-specific criteria. Other public comments that reflect implementation
issues are beyond the scope of the scientific criterion document and will be addressed in
technical support documents under development by EPA." (p. 127)

"The nature of selenium bioaccumulation may vary on a site-by-site basis in a manner sufficient
to justify development of site-specific criteria. The reviewer identifies several important issues
that will be addressed in the technical information to be developed to guide the implementation
of the criteria." (p. 129)

In addition to those specific statements by EPA, the Agency also makes general commitments to develop
guidance on the following additional issues raised by reviewers, including concerns that were initially
raised by public comments and then assessed (at EPA request) by the peer reviewers:

129

-------
"The EPA must provide guidance on several aspects related to implementation of the tiered
criteria approach, at the very least including (a) when to sample fish so that females are in
vitellogenic or pre-ovulatory stages of oogenesis, (b) what sample size offish to collect for tissue
[Se] determinations (I suggest a minimum of n=10 female fish per site), (c) recommended
analytical procedures for quantification ofSe, (d) guidelines for implementation of the 30-day
average water column criterion element (how, when, where), and (e) guidelines for
implementation of the intermittent water column criterion, if the EPA chooses to keep it in the
tiered criterion." (pp. 130-131)

"Averaging period: Comment reasonably addresses the need for clearer implementation
guidance of the intermittent water column criterion, (p. 132)"

"Implementation: The public comments express thoughtful concerns and practical
implementation questions that can serve as prompts to draft additional guidance." (p. 135)

"Intermittent criterion: Several good points are raised in the public comments. Suggestions to
abandon one model for another do not provide adequate support for the suggestion. Practical
implementation concerns are valid and should be addressed." (p. 135)

"Natural background: The public comments correctly identify concerns of naturally occurring
selenium contamination of waters and impacted aquatic life. The draft criterion should explicitly
address these concerns in regards to implementation of the draft criterion." (p. 136)

"Site-specific criteria: There are numerous public comments that should be addressed in
guidance for implementation, (p. 138)

"Tiered criteria: There are numerous public comments that should be addressed in guidance for
implementation." (p. 138)

"Tissue criterion: There are numerous public comments that should be addressed in guidance for
implementation." (p. 138)

EPA-HQ-OW-2004-0019-0376-A2 [Comment 0381-A3 is a duplicate of 0376-A2]; Texas
Commission on Environmental Quality (TCEQ); Posted 09/24/2015

Background

The EPA published the current nationally-recommended chronic selenium water quality criterion in
1987, which is expressed solely as a water column concentration. Since then, there has been
considerable scientific review on the exposure pathways and effects of selenium on aquatic life. EPA
published initial proposed revisions to the criterion in 2004, which were based upon whole-body fish
tissue concentrations. In 2014, EPA developed the 2014 External Peer Review draft, which expanded the
criterion to include four elements, including concentrations of selenium in fish tissue and the water
column. The latest draft proposal includes four elements comprising six aspects of the criterion: (1) fish
tissue concentration in egg-ovaries, (2) fish tissue concentration in muscle, (3) fish tissue concentration
in the whole-body, (4) monthly water column concentration for lentic (non-flowing) systems such as
reservoirs, (5) monthly water column concentration for lotic (flowing) systems such as streams and

130

-------
rivers, and (6) intermittent concentrations for lotic and lentic aquatic systems to account for pulses of
elevated selenium concentrations. EPA's draft proposed criterion incorporates all six of these targets,
and EPA recommends all aspects to be adopted into state water quality standards. When available, fish-
tissue based aspects would override water column concentrations, since selenium is bioaccumulative.
Considerable uncertainty exists regarding how these elements would be implemented in regulatory
programs under the CWA, particularly due to the complex nature of the criterion and lack of
implementation guidance. The TCEQ offers the comments provided below.

C. EPA is requested to develop suggested monitoring guidance for sample protocols, including target
fish species and fish eggs.

Guidance is needed standardizing the various aspects of fish tissue sample collection with respect to
selenium evaluation. Clarifications for the target fish species or trophic levels, time of year of sampling
and harvesting of eggs, as well as general considerations such as sample collection and handling are
needed. Current research has determined that the bioaccumulation of selenium through aquatic food
webs is the primary mechanism of exposure for aquatic life. However, EPA suggests that exposure and
accumulation occurs significantly at base levels of the food chain, and site-specific factors such as
dietary preferences of consumers can play a large role in determining exposure. Guidance is needed to
specify what trophic levels should be sampled for evaluation and compliance with the criteria.
Additionally, the monitoring of fish eggs is not practical for the state or permittees. The harvesting of
eggs may require additional considerations such as spawning periods, special sample collection or
analysis protocols. As a result, it is anticipated that this particular sample effort may be much more
complex than the simple collection of fish for whole-body or tissue analysis. Without guidance and
standard protocols, states and tribes may develop inconsistent methods for sampling and monitoring.

The TCEQ is concerned that the proposed changes could interfere with the effectiveness of the existing
TCEQ water quality management processes. In particular, the comments address the need for
implementation guidance for incorporation of the criterion in wastewater permitting, Total Maximum
Daily Loads, and water quality assessment activities. The major TCEQ concerns included in the
comments are as follows:

• the misleading nature in which EPA describes the criterion as having "four elements" when in
fact there are six numeric components to the complex criterion;

• the lack of implementation guidance accompanying the criterion. Guidance for implementation
of the criterion in permitting, monitoring and assessment programs is needed;

• EPA's expectations for consideration and adoption of the criterion as part of triennial reviews of
water quality standards, particularly in light of the complexity of the criterion and lack of
implementation procedures;

• lack of guidance for the sampling of target fish species, including the harvesting of fish eggs;
and,

• lack of guidance regarding the development of site-specific criteria to account for regional
differences in naturally-occurring selenium.

II. Lack of Implementation Guidance for Incorporation of the Criterion into Water Quality
Management Programs of the Clean Water Act.

131

-------
A. The proposed criterion lacks guidance for implementation. EPA is requested to coordinate with the
states and tribes to provide guidance for implementation of the draft criterion in permitting, Total
Maximum Daily Loads (TMDL) and water quality assessment. EPA should postpone the adoption of the
criteria until all the necessary information, including the guidance, is available for public review and
comment.

Current implementation procedures to establish effluent limits for National Pollution Discharge
Elimination System permits are based upon water column concentrations. Guidance is needed to
address various aspects of the six-part criterion, particularly since EPA is recommending that fish tissue
elements be given precedence over water column components. The following key areas need to be
addressed in the guidance:

• The development of water quality based effluent limits necessary to ensure attainment of the
fish tissue criterion from whole-body, muscle, eggs or ovaries. Since permit limits are not
written directly from tissue-based criteria, guidance is needed to assist states and tribes with
the translation of fish tissue to water-column concentrations for use in site-specific wastewater
permitting and TMDLs.

• The characterization of intermittent exposure scenarios (including determination of background
concentrations), to account for pulses of selenium occurring during single day, high exposure
events.

• Procedures for determining a "steady-state equilibrium" has been achieved in lentic and lotic
environments as a result of increased inputs of selenium.

• Recommendations for determining attainment of criteria during assessment scenarios,
particularly when data to evaluate all aspects of the criterion are not available. It is unclear
whether all six numeric components must be considered or whether a combination of less than
the six components can be used.

It is requested EPA seek input from the states and tribes to develop adequate guidance for the
implementation of the criterion prior to final publication.

B. The proposal lacks clarity regarding the expectations for inclusion of the draft criterion in water
quality standards. EPA should postpone adoption of this document and develop guidance through
coordination with the states and tribes.

Due to the lack of guidance provided by EPA, additional uncertainties for entities with adopted acute
and chronic selenium criteria remain:

• EPA has provided insufficient discussion regarding expectations of the timeframe for inclusion of
this criterion in the triennial reviews of water quality standards. It is unclear whether states and
tribes are expected to immediately adopt this complex criterion with no considerations to
implementation or existing criteria.

• EPA is requested to provide information regarding acute effects. The draft criterion document
states that these endpoints are of minor concern when compared to chronic exposure and
maternal transfer, but the document does not expressly continue to endorse the current
nationally recommended acute criterion. If an acute criterion is no longer needed, please
directly state this in the final document.

The Federal Register notice states that EPA is developing information material to aid state and tribal
adoption and that these materials will be released when the final criterion is published. Given the

132

-------
complex nature of this criterion and the significant change to the approach, EPA should postpone
adoption of the proposed criterion and coordinate with states and tribes regarding the expectations for
inclusion of this criterion in triennial reviews. EPA is requested to develop guidance documents and
propose them together with the final criterion document. Informational material should be provided for
review prior to finalization of the criterion. Without this additional information, stakeholders cannot
completely evaluate the proposal and will miss the opportunity to provide proper feedback on the
proposal.

EPA-HQ-OW-2004-0019-0387-A1; Colorado Department of Public Health & Environment;
Posted 10/14/2015

General Implementation

• While we appreciate EPA's consideration of the difficulties in collecting tissue data (especially
egg/ovary), suggesting that all four elements should be adopted as the criterion seems
complicated and potentially excessive, especially if data for one tissue type will override all
other elements. Including all four elements suggests that, when all four types of data are
available, they should all be collected (even though egg/ovary would be the only data 'used'). Is
this what EPA intends? Perhaps there is a way to present the criterion in a different way so it
recommends only collection of the highest 'tier' component possible (e.g., egg/ovary) and the
lesser-tiered elements (e.g., whole-body/ muscle, water) can be disregarded.

Implementation in Discharge Permits

• Page xvi states "... states should establish additional procedures to facilitate translation of the
fish tissue criterion concentration elements into water concentration permit limits". This
suggests that the water column element of the criterion would not be used in the permit. Is this
the intent? If so, why can't the water column element (i.e., 1.2 or 3.1 ng/L) be used to
determine the permit limit?

• How will the four-part criterion be applied to short-term discharges (e.g., construction
dewatering), since the water column element is based on long-term bioaccumulation?

How will the four-part criterion be implemented in storm water permits, which often have practice-
based, rather than numeric-based, permits?

• The proposed multi-part criterion is based on 'tiered' values, where egg/ovary data have
precedence over whole-body/muscle and water data, and whole-body/muscle has precedence
over water. If a site exceeds the water column value but attains the egg/ovary value, does this
mean the site is in attainment of the chronic selenium criterion?

Many waters do not currently have tissue data available, but do have water column data. If a
waterbody does not attain the proposed water column element, it is expected that tissues
would be collected as soon as possible, if feasible. These kinds of situations may result in many
new (and potentially only temporary, if tissues are in attainment) impairment listings for
waterbodies that do not yet have tissue data.

133

-------
Does EPA plan to include guidance for handling selenium in the 303(d) listing context to
minimize inappropriate impairment listings? For instance, perhaps streams with water column
exceedances should only be placed in 303(d) Category 3 (insufficient data to make a use support
determination) until tissue data are collected.

• Does EPA still recommend a zero exceedance frequency for the fish tissue element of the
criterion? Colorado's data indicate that there is wide variability among individual fish collected
from the same waterbody at the same time. When a single fish may be used as the basis for
impairment, it discourages collection of larger sample sizes, which may provide more insight
into the actual distribution of concentrations. It is also possible that this approach will result in a
finding of impairment when the vast majority of the fish population is well below the standard.

• The draft criterion document focuses on selenium sources related to anthropogenic activities
and does not discuss how to handle the natural processes that can be sources of selenium. As
mentioned above, Colorado has high amounts of selenium-laden shale in much of its geology.
EPA should add a discussion of natural selenium sources in the draft criteria document and
consider including additional alternatives or considerations for site-specific standards in these
situations. Would EPA consider site-specific tissue-based standards in these cases? Or could site-
specific water column-based standards over-ride the national tissue-based criterion elements?

EPA-HQ-OW-2004-0019-0370-A2; USGS Leetown Science Center; Posted 09/08/2015

We evaluated the sample sizes necessary to assess whole-body fish tissue Se concentrations to detect
mean Se concentrations above candidate water quality standards (Hitt and Smith 2014). We used data
from our collections (as published in Presser 2013) as well as data from the West Virginia Department of
Environmental Protection. We found that natural variation in Se concentrations within samples (i.e.,
among individuals sampled from a given location) increases with increasing mean Se values, and this will
affect the sample sizes needed to assess compliance with the proposed standard. The underlying mean-
variance relationship was expected (for reasons provided in the paper) and was observed for several
taxa: Green Sunfish (Lepomis cyanellus), Creek Chub (Semotilus atromaculatus), and Central Stoneroller
(Campostoma anomalum) (see Figure 2 in Hitt and Smith 2014).

Our power analysis can inform sample designs based on the proposed whole-body criterion of 8.0 mg
Se/kg (Figure 1). For example, given a type-l error rate of 0.10 (i.e., bottom-center panel in Figure 1), a
sample of 2 fish would be unable to detect true-mean Se concentrations below 11.0 mg Se/kg with 80%
power. In contrast, samples of 16 fish would be expected to detect true mean concentrations of 9.0 mg
Se/kg (i.e., 1 mg Se/kg above the criterion) with 80% power. If higher type-l error rates are allowed (i.e.,
alpha = 0.20) then lower sample sizes may be necessary to achieve 80% power. However, even with
relaxed alpha levels, samples of 4 or fewer fish would be unlikely to detect an increase of 2 mg Se/kg
above the criterion with 80% power. Given the rapid transition from essentiality to toxicity for Se,
sampling designs capable of detecting near-threshold concentrations may be warranted.

Original letter contains Figure 1 - Power analysis results for Se assessment in whole-body samples of
stream fish (Figure 3 in Hitt and Smith 2014). See original letter.

134

-------
EPA-HQ-0W-2004-0019-0419-A2; Lima Refining Company; Posted 11/03/2015

COMMENT #3. There is insufficient guidance for State implementation that may lead to application of
the least certain criterion (water column).

The 2015 USEPA Se Draft AWQC does not provide sufficient guidance to the states on implementing the
tiered criteria. The Peer Review Team (External Peer Review of the Draft Aquatic Life Ambient Water
Quality Criterion for Selenium — Freshwater 2014) has noted that implementation is a major concern
from the public and that more guidance is needed in order to implement this very complex water
criteria guidance which ideally should include a continuous program to measure egg ovary, fish tissue
and water column selenium levels. Each of these parameters would require individual programs. It does
not appear that the 2015 USEPA Se Draft AWQC has addressed the public's concerns. The provisions of
the 2015 USEPA Se Draft AWQC gives the states great latitude in how the guidelines are used; or not
even used at all. States can use in increasing order of preference:

a) Water

b) Fish tissue (whole body or muscle) — overrides water when measured

c) Egg ovary — overrides whole body, muscle or water column when measured

This tiered approach presents increased levels of complexity that may discourage the states from using
the best and more reliable criteria, which are for fish tissue and/or egg ovary. The water column
criterion, which is the least certain criterion, may be used by the states simply because water column
measurements are the most straight forward and is what typically has been part of state monitoring
programs. In fact, the 2015 USEPA Se Draft AWQC in Section 2.7 Measures of Effect, Subsection 2.7.2
Water on Page 22 states the "While state monitoring programs may sample ambient waters for
selenium, widespread measurements of selenium in fish tissue are relatively rare." The document as
written offers no protection to the discharger.

EPA-HQ-OW-2004-0019-0415-A1; Idaho Department of Environmental Quality (DEQ); Posted
10/28/2015

The attention to implementation of this new criterion is still lacking and should be detailed furthered.
While clearly representing the state of the science, an egg-ovary criterion is also clearly impracticable for
routine criteria compliance monitoring and assessment. EPA has attempted to address this through
translation to other criteria bases, e.g. whole body and muscle tissues, as well as translation to water. In
the end this creates a 4- layered criterion hierarchy that is very complex.

Although this hierarchy allows traditional water testing, this hierarchy is also likely to create pressure for
acquiring "acid test" egg-ovary data that is the most difficult and expensive to obtain. Furthermore,
some fish species are highly mobile which leaves the door open to raise a number of biological and life
history questions that call into question association of fish selenium concentrations to site water quality
that are not simply answered or addressed in routine monitoring programs. This complexity is likely to
further strain already reduced state monitoring budgets.

135

-------
It is not clear how high natural selenium is to be handled, that is, selenium levels that would exceed the
criterion in the absence of human cause. While there are general schemes in water quality standards
that can address this they are under attack. It would be helpful for a modern criterion document to
address the issue of naturally occurring selenium up-front with clear direction to the states. This is
especially important in Idaho where we likely have water bodies with naturally elevated selenium levels.

What about methods of analysis? Currently there is no EPA approved method for analysis of selenium in
fish tissue. While states/dischargers are clearly given analytical flexibility under the NPDES regulations
where no EPA approved method exists, this situation does invite debate, and places an extra burden as
well on states and dischargers. It should be EPA practice to promulgate an approved analytical
method(s) in advance or at the same time it proposes a criterion that calls for new analytical methods.
In general, EPA needs to do a better job describing how a fish tissue criterion is to be implemented in
NPDES permits and TMDLs.

The document leaves it unclear how depth profiles or area-volume relationships fit into evaluation of
the lentic water criterion. Is there to be area or volume weighted averaging? Or is it intended that one
sample, no matter how little time or space it represents, be sufficient to show violation of the criterion?
If the latter, this is does not seem to be reasonable or statistically justified.

While researchers often report selenium concentrations in eggs or ovaries (bottom page 26), such
sampling is not trivial and is far from routine in environmental monitoring by environmental agencies or
regulated entities. Although we end up with perhaps a better criterion from a science standpoint, we get
a practical nightmare from an implementation and measuring compliance standpoint. We believe EPA
needs to pay even more attention to the implementation and monitoring implications of this new
criterion.

EPA-HQ-OW-2004-0019-0411-A2; American Petroleum Institute (API); Posted 10/28/2015

• EPA should prepare, accept public comment on, and release a comprehensive implementation
guide to accompany the final selenium criteria. EPA will need to consider very carefully how to
implement tissue-based selenium criteria in determining water quality impairments and in
setting technically sound water-quality-based permit limits. API recommends the guidance
afford flexibility, not prescription, in translating between tissue and water column selenium
concentrations.

Although EPA clearly notes that fish tissue-based selenium criteria override water column-based criteria
(with the exception of "fishless" waters and for new discharges), EPA needs to more explicitly define
how the water column criteria could be considered "trigger" values for collection of fish tissue data.

We support EPA's conclusions that Se measurements from water and tissue samples collected at the
same site within one year can, for the purpose of developing a database designed to capture a broad
range of conditions, serve as "matched pairs" of measurements, provided the limitations and constraints
of this assumption are fully understood for the specific site. In local evaluations implementing tissue-
based criteria into NPDES permits and general 303(d) assessments, however, the timing of co-located
water and tissue data collection must consider myriad important site-specific factors and non-steady-
state conditions, including the migration of local fish populations, and seasonal and/or storm-related Se
pulses. Allowance of appropriate sampling windows provides time for the analysis of effluent data and

136

-------
potential follow-up tissue sampling. In addition, this large sampling window could be very important in
ephemeral and intermittent aquatic systems where flowing surface water may only be present during
certain times of the year, resulting in limited time periods to conduct fish sampling.

Criteria Implementation

Fish and Water Se Criteria Hierarchy

Although EPA (2015) clearly notes that the fish tissue-based Se criteria override the water column-based
Se criteria (with the exception of "fishless" waters and for new discharges), we recommend that the EPA
more explicitly define how the water Se criteria could be considered "trigger" values that identify the
need to collect data on Se concentrations in fish tissue. In our experience, this is an issue of on-going
concern within both the regulatory and the regulated community.

EPA-HQ-0W-2004-0019-0417-A1; Utility Water Act Group (UWAG); Posted 11/02/2015
1. Implementation guidance is needed for the selenium criterion.

EPA has indicated that it will provide guidance on implementing the selenium criterion. UWAG agrees
that guidance is necessary, indeed imperative, for the criterion to be used in any meaningful way. We
recommend that the following be addressed in the implementation guidance in addition to the topics
EPA has self-identified:

• Clarification of when fish should be collected that addresses the importance of the egg or ovary
maturity level.

• Selection of fish species to base determination of impairment status of a waterbody.

EPA-HQ-0W-2004-0019-0401-A1; Colorado Wastewater Utility Council; Posted 10/15/2015
1.2 Timing of Tissue Data Collection

We support EPA's conclusions that selenium measurements from samples collected at the same site
within one year can serve as matched pairs of measurements, and feel the timing of tissue data
collection will be an important component of implementing tissue-based criteria into NPDES permits
and general 303(d) assessments. Allowance of appropriate sampling windows provides time for analysis
of effluent data and potential follow-up tissue sampling. In addition, this large sampling window could
be very important in ephemeral and intermittent aquatic systems where water may only be present
during certain times of the year, resulting in limited time periods to conduct fish sampling.

137

-------
EPA-HQ-0W-2004-0019-0400-A1; The National Mining Association (NMA); Posted 10/15/2015

1.2 Timing of Tissue Data Collection

We support EPA's conclusions that selenium measurements from samples collected at the same site
within 1 year can serve as matched pairs of measurements, and feel the timing of tissue data collection
will be an important component of implementing tissue-based criteria into NPDES permits and general
303(d) assessments. Allowance of appropriate sampling windows provides time for analysis of effluent
data and potential follow-up tissue sampling. In addition, this large sampling window could be very
important in ephemeral and intermittent aquatic systems where water may only be present during
certain times of the year, resulting in limited time periods to conduct fish sampling.

Another issue that requires reevaluation and clarification is how to establish criteria for streams with no
existing fish populations. The default approach is to use water-column criteria; however, an alternative
states could be provided is the use of the chronic invertebrate data provided by EPA to develop
invertebrate tissue-based site-specific criterion and associated protective water-column criteria.

EPA-HQ-OW-2004-0019-0394-A2; Appalachian Mountain Advocates and Sierra Club
Environmental Law Program; Posted 10/14/2015

EPA has not explained how it and authorized state agencies would incorporate the proposed fish-tissue
elements into enforceable measures needed for NPDES permit limits, TMDLs, and other pollution
control decisions required by the Clean Water Act. Indeed, EPA explicitly recognizes that, if states
choose to use the tissue criterion elements for NPDES permitting purposes, "additional state WQS
implementation procedures (IPs) will be needed to determine the need for and development of WQBELs
necessary to ensure attainment of the fish tissue criterion element(s)." Draft Criterion at 92. EPA
entirely fails, however, to provide any meaningful guidance on what those procedures should entail.
EPA's proposal thus leaves unanswered fundamental questions about how the fish-tissue elements are
to be used when issuing NPDES permits. For instance, how are regulators to determine the "reasonable
potential" for a proposed new discharge to cause or contribute to violations of the fish tissue elements?
How will appropriate "end of pipe" effluent limits be determined? If there is a "reasonable potential,"
when must treatment start? Without clear guidance from EPA, we fear that states will adopt and EPA
will be forced to approve standards that cannot practically be used to set necessary water quality-based
permit limits.2 Our experience has shown that underfunded and/or industry-friendly state regulators will
only impose enforceable permit limits when they are forced to do so by clear standards and
incontrovertible evidence of reasonable potential. A recommended criterion that does not explicitly
establish when permit limits must be imposed but instead injects considerable uncertainty into the

138

-------
reasonable potential analysis invites regulators to acquiesce to industry pressure to impose no limits or
limits that are effectively meaningless.

Likewise, EPA's proposal lacks necessary information regarding how compliance with the fish-tissue
elements should be determined for the purpose of enforcing NPDES permit limits, evaluating waters for
impairment, and developing and enforcing TMDLs. For instance, if a permittee receives a fish tissue-
based NPDES permit limit, where must sampling of fish occur in relation to the discharge? How many
fish must be collected to provide a representative sample?3 How often and at what stages of life must
sampling take place? What fish taxa will be used to determine compliance? How will regulators account
for variation and individual differences and toxicity within taxa depending on, among other things, age,
individual diet, areas of forage, and duration of stay in polluted waters?4 If adequate numbers of fish are
indeed collected, what impact will this have on fish populations that may already be pressured by
selenium and other pollution? How will regulators ensure that endangered species are protected by
sampling protocols such that illegal "take" is avoided? How will impairment be detected in waters where
sensitive species that rapidly accumulate selenium have already been extirpated?5

EPA has not shown that compliance with the fish tissue elements can accurately be determined in most
circumstances. This is particularly problematic in small headwater streams that directly receive much of
the selenium pollution from coal mines in Appalachia. These streams often lack sufficient fish
populations for a truly representative sample to be collected, and downstream reaches with larger fish
populations often receive discharges from many different sources such that responsibility for violations
of the standard will be extremely difficult to assign. See Hitt and Smith (2014). Moreover, if a "species-
composite" method is used to determine compliance with a fish-tissue element, wherein the tissue of all
fish collected is combined for analysis, it is likely to miss impairment of sensitive species that accumulate
selenium more rapidly. USFWS, July 28, 2014 Comments. In sum, regulators who are under heavy
pressure from industry and whose resources are already stretched far too thin are unlikely to develop
and implement the complex, expensive fish sampling and testing protocols necessary to obtain
representative samples to determine a waterbody's compliance with the fish-tissue elements, to the
extent that such protocols are even possible to develop.

Instead of relying on fish tissue standards that present critical implementation problems, EPA should
adopt clearly enforceable water column criteria. EPA's Draft Criterion document recognizes that the
dietary pathway of selenium accumulation can still be accounted for in water column criteria. Using the
methods developed by the EPA and the United States Geological Survey, protective fish tissue
concentrations can be translated to practically enforceable water column criteria. Draft Criterion at 28.
The model developed by USGS and modified by EPA recognizes that diet is the primary pathway of
exposure for selenium and creates a simple, direct linkage between dissolved selenium in the water
column and selenium toxicity to aquatic life. As EPA explains, its approach "explicitly recognizes the
sequential transfer of selenium between environmental compartments (water, particulate material,
invertebrate tissue, fish tissue, and eggs and/or ovary tissue) by incorporating quantitative expressions
of selenium transfer from one compartment to the other." Draft Criterion at 63. Indeed, EPA's analysis
found that that the expected and measured relationships between egg-ovary concentrations and water
column concentrations are "highly correlated."6 Draft Criterion at 125. An inviolable water column
criterion that is based on fish tissue concentrations is therefore scientifically defensible because it
recognizes and accounts for the fact that diet is the primary pathway for selenium uptake.

139

-------
The Draft Criterion's inclusion of water column based elements in no way corrects this fundamental
flaw. EPA has explicitly stated that the fish tissue elements should be given primacy over the water
column elements. Draft Criterion at xiv. That statement essentially eliminates any practical benefits from
including water column elements. This is true despite EPA's qualification of the primacy of fish tissue
elements in "fishless waters." See Draft Criterion at xiv. Although the inclusion of that qualification
improves the Draft Criterion, it does not go nearly far enough. For example, "fishless waters" should not
be limited to "waters where fish have been extirpated or where physical habitat and/or flow regime
cannot sustain fish populations." As explained above, accurately determining compliance with fish tissue
elements requires significant sample sizes, regular collection of which is likely to have a severe impact
on fish populations, particularly in smaller headwater streams. See supra footnote 3. In order to be
practically protective, the water column elements would need to be given primacy in all waters that lack
adequate numbers of fish to regularly collect large tissue samples without adversely affecting resident
populations, not just in waters that completely lack fish. Moreover, in order to combat the survivor bias
inherent in tissue sampling and be truly protective of aquatic life, see supra footnote 5, the water
column elements would need to be given primacy in all waters where sensitive species could be
supported but have been extirpated. See Hitt and Chambers, Temporal changes in taxonomic and
functional diversity offish assemblages downstream from mountaintop mining, Freshwater Science
33:915-926 (2014) (finding that sensitive species have been largely extirpated from sites downstream of
mountaintop removal mining operations as a result of changes to water quality, including increased
selenium pollution).

Likewise, giving water column elements primacy in waters with new or increased selenium inputs prior
to reaching ecological equilibrium will not adequately protect aquatic life. As EPA acknowledges, the
time needed for waters with new or increased inputs to reach such an equilibrium varies greatly from
site to site depending on "the hydrodynamics of a given system; the location of the selenium input
related to the shape and internal circulation of the waterbody, particularly in reservoirs with multiple
riverine inputs; and the particular food web." Draft Criterion at 93. State agencies that are subject to
extreme resource limitations and industry pressures are unlikely to ensure that the elements are
properly implemented. Indeed, the complexity and uncertainty involved in determining when water
column elements should take primacy over fish tissue elements demonstrates the impracticality of EPA's
multi-element Draft Criterion. The better approach that would be both scientifically defensible and
practically implementable would be to adopt a criterion consisting only of translated water column
elements.

Not only is a translated water column criterion scientifically defensible, it is vastly more useful as a
regulatory tool that places minimal burden on states to implement without significant delay or expense.
Most states have specific, federally-approved procedures for how to convert water column criteria to
enforceable restrictions on wastewater discharges, in addition to the technical guidance, training and
other materials on scientifically valid models, necessary background data, sampling protocols, and
acceptable laboratory techniques for the implementation of traditional water column criteria that EPA
has provided. See Draft Criterion at 92 (explaining that existing state procedures will be adequate for
implementing the water column elements of the criteria, but that new procedures will be needed to
implement the tissue-based elements). Water column criteria also can be more easily enforced by
citizens with limited resources when state regulators fail to uphold their duties. Enforcing the proposed
fish-tissue elements, in contrast, will require a case-by-case analysis of the local ecosystem, including

140

-------
collection, processing, and testing of fish tissue, all of which will require significant resources and inject
considerable uncertainty. Thus, in order to achieve the dual purposes of water quality criteria, EPA
should adopt as its Recommended Criteria a set of water column criteria that are translated from
protective fish-tissue concentrations.

EPA-HQ-OW-2004-0019-0373; Anonymous public comment; Posted 09/23/2015

Certain issues considered in EPA's mercury implementation guidance seem highly relevant to
establishing a selenium tissue criterion duration and frequency. Because the draft Se tissue duration and
frequency recommendations are incompatible with EPA's mercury implementation guidance, please
explain why you are not using the concepts from that guidance:

Guidance for Implementing the January 2001 Methylmercury Water Quality Criterion, Final, EPA-823-R-
10-001.

The relevant portion of the mercury implementation guidance is shown below:

4.3 How should waterbody impairment be assessed for listing decisions?

Section 303(d)(1) of the CWA and EPA's implementing regulations require states and authorized tribes
to identify and establish priority ranking for waters that do not, or are not expected to, achieve or
maintain water quality standards. In accordance with this ranking, a TMDL for such waters must then be
established. For purposes of determining impairment of a waterbody and whether to include it on
section 303(d) lists, or in category 5 of the Integrated Report under sections 303(d) and 305(b)15, states
and authorized tribes must consider all existing and readily available data and information (see 40 CFR
130.7).

States and authorized tribes determine attainment of water quality standards by comparing ambient
concentrations to the numeric and narrative AWQC (40 CFR 130.7 (b)(3)). Where a fish tissue criterion
has been adopted, states and tribes should consider observed concentrations in fish tissue in
comparison to the criterion. Where a water column translation of the fish tissue criterion has been
developed and is adopted as part of the state's or tribe's water quality standards, states and tribes
should consider ambient water concentrations in comparison to the translation.

For assessment of concentrations in fish tissue, resources may typically be unavailable to collect an
adequate number of replicate composite samples to support rigorous statistical testing, especially
where it is desirable to evaluate each individual target species separately. In these situations, states
should make direct comparisons between composite sample concentrations and the criterion, as each
composite effectively represents the average concentration observed in several fish.

Statistical tests for comparing the average concentration from multiple replicate composite samples to
the criterion may be conducted where a sufficient number of replicates have been collected. EPA's
Guidance for Assessing Chemical Contaminant Data for Use in Fish Advisories, volume 1 (USEPA 2000c),
at section 6.1.2.7.2, recommends using the t-test to determine whether the mean concentration of
mercury in composite fish tissue samples exceeds the screening value. This test involves a statistical
comparison of the mean of all fish tissue data to the criterion. States and authorized tribes can evaluate
whether the t-test statistic of the mean exceeds the water quality standards. This procedure could also

141

-------
be used to determine impairment, provided it is consistent with a state's water quality standards. States
and authorized tribes might also want to consider the guidance in appendixes C and D of the
Consolidated Assessment and Listing Methodology: Toward a Compendium of Best Practices (USEPA
2002e). Ultimately, the method that states and authorized tribes choose depends on how they express
their water quality standards and apply their water quality assessment methodology.

142

-------
Comment Category 2.8 - Comments Concerning the Protection of

;cies

2.8 Summary

Some commenters were concerned that the criterion would not protect some ESA-listed species and
urged coordination with U.S. Fish and Wildlife Services and other relevant agencies.

,)onse to comments

2.8 Responses concerning threatened and endangered species

Regarding consideration of endangered species, EPA has provided a summary of available data on the
adverse effects of selenium to endangered species, and taxonomically related surrogates. EPA identified
data for white sturgeon (a species with one population listed as endangered, and a closely related
taxonomic surrogate for other endangered sturgeon), several species in the family Salmonidae,
Cyprinidae as well as a pupfish in the Cyprinodontidae, and several less closely related species with tests
in the order Perciformes (sunfish and bass) as surrogates for endangered darters.

Regarding endangered species, EPA recognizes the concern regarding protection of listed species. EPA
has considered and described available data on impacts of selenium on endangered species in the
criterion document as well as impacts on species that are close phylogenetic surrogates of listed species
dataset. EPA notes that site-specific criteria can be developed if sensitive endangered species not
represented by surrogates in the dataset used in the criterion are present at a specific site.

EPA had significant data quality concerns with the Hamilton et.al. 1990 data, most significantly that the
90-day control survival (67%) was below toxicity test acceptability thresholds, such that we could not
use those effects data. We did consider and use the 60-day time point for this study, and concluded that
the egg-ovary transformed whole body criteria of 8 mg/kg dw would protect against growth effects in
juvenile salmonids. Compromising data analysis to include very high control mortality data despite the
availability of other more acceptable data within the same test does not meet EPA's data quality
expectations. Further, use of poor quality data is not an improvement to or modernization of methods
for criteria development.

Published data indicates that endangered fish species have not been found to be more sensitive to
toxicants than common species (Sappington et al 2001, and other). EPA acknowledges that there may be
locations where a lower criterion could be applicable on a site specific basis due to the site-specific
presence of a particularly sensitive species.

EPA disagrees that before making general recommendations to states regarding future state actions to
adopt selenium criteria, it is helpful or legally necessary to first engage in consultation under the ESA to
ensure that any possible subsequent federal action to approve new or revised state selenium criteria
consistent with the national recommendations would necessarily be protective of listed species. The
issuance of national criterion recommendations for selenium does not impose any legally binding
requirements on states. Nor does it authorize any state or federal action that would otherwise be

143

-------
inconsistent with the ESA, simply on the grounds that such action is consistent with EPA's national
recommendations under the Clean Water Act. Since the distribution of listed species which might affect
the appropriate water quality criteria is location-specific, and the national recommendations for
selenium are intended to be generally informative, allowing the most sensitive location-specific
potential concerns to drive national recommendations would tend to inappropriately distort those
recommendations. This approach would tend to produce national criteria recommendations for
selenium that states would need to modify to make less stringent before incorporating into their own
standards, based on the absence of species-specific concerns. And national consultation on general
recommendations for selenium criteria would still be unlikely to obviate the need for step-down
consultations on subsequent federal actions to approve particular new or revised state water quality
criteria for selenium. Therefore, EPA believes that it is more efficient for states to modify national
criteria recommendations for selenium to make them more stringent, as needed based on the presence
of localized species-specific concerns. EPA intends to engage in ESA consultation about a proposed
approval of a state water quality standard under Clean Water Act Section 303(c) to the extent that it
determines that such approval may affect listed species and to the extent that it has relevant discretion
to consider those effects.

2.8 Specific comments

EPA-HQ-OW-2004-0019-0396-A1; National Marine Fisheries Service Office of Protected
Resources Interagency Cooperation Division; Posted 10/15/15

The National Marine Fisheries Service (NMFS), Office of Protected Resources appreciates the
opportunity to provide input on the Environmental Protection Agency's (EPA) external peer review Draft
Aquatic Life Criteria Document for Selenium. We understand that EPA has been developing this criterion
for several years and we appreciate the hard work that has gone into this effort. Our comments focus on
implications of the selenium guideline for NMFS ESA-listed species and age classes foraging in
freshwater. With the exception of shortnose sturgeon, adult NMFS ESA-listed species feed little or not at
all in freshwater. However, there are some reports of freshwater feeding by adults of other sturgeon
species, particularly in females after spawning. The guideline also has implications for NMFS ESA-listed
juvenile salmonids and sturgeon rearing in freshwater.

Conclusion: We conclude that the guideline document needs to convey an understanding that, due to
the vulnerability of ESA-listed species populations, an ESA-listed species need not be toxicologically
more sensitive to a stressor, selenium in this instance, to require site-specific protective criteria. We
encourage EPA to work with us to determine the impact of an EC10 larval deformity response threshold
to the continued existence of populations of ESA-listed species such as shortnose sturgeon and
determine if location specific criteria are necessary for rivers where susceptible populations occur. The
guideline document should clearly indicate that the presence of ESA-listed species can trigger
development of site specific criteria and reevaluation of existing data. In such cases, derivation of site
specific criteria for waters containing shortnose sturgeon would need to incorporate relevant exposure
parameters. In addition, the evaluation and analysis of dietary data in the guideline document should be
updated. For example Hamilton et al. 1990 data set should be reevaluated in an effort to extract
additional information from the 90 day exposure data.

144

-------
EPA-HQ-OW-2004-0019-0415-A1; Idaho Department of Environmental Quality (DEQ); Posted
10/28/2015

We also believe it is EPA's responsibility to conduct Endangered Species Act consultation with the
services nationally on its national criteria recommendations, and not shift this burden to state adoption
of nationally recommended criteria.

EPA-HQ-OW-2004-0019-0394-A2; Appalachian Mountain Advocates and Sierra Club
Environmental Law Program; Posted 10/14/2015

VI. The Criterion Must Protect All Threatened or Endangered Species

Water quality standards must protect all existing uses in a waterbody, which uses often include
supporting species that are listed as threatened or endangered pursuant to the Endangered Species Act.
See 33 U.S.C. § 1313. Additionally, Section 7 of the Endangered Species Act and its implementing
regulations require each federal agency, in consultation with the appropriate wildlife agency, to insure
that any action authorized, funded, or carried out by the agency is not likely to (1) jeopardize the
continued existence of any threatened or endangered species or (2) result in the destruction or adverse
modification of the critical habitat of such species. 16 U.S.C. § 1536(a)(2); 50 C.F.R. § 402.14(a). EPA thus
must ensure that any criteria that it recommends states to adopt will be fully protective of listed species.

EPA's Draft Criterion document concludes that it will protect threatened and endangered species based
off analysis of only two listed species and two additional species that are closely related to listed
species. USFWS records show that there are currently 163 fish species in the US that are listed as
threatened or endangered pursuant to the Endangered Species Act. See USFWS, Environmental
Conservation Online System, Summary of Listed Species.10 EPA cannot safely assume that the two listed
species and two closely related species are good proxies for every single endangered species in the
country that could be exposed to selenium pollution. Indeed, EPA recognizes that "because other
threatened or endangered species might be more sensitive, if relevant new information becomes
available in the future, it should be considered in state- or site-specific criteria calculations." Draft
Criterion at 127. Instead of putting off protection of sensitive endangered species to later state or site-
specific standard setting, EPA must revise its criterion to ensure protection of al[ endangered species. It
is not sufficient to say that the agency lacks information. Rather, in the absence of additional data
regarding selenium-sensitive listed species, EPA must apply a substantial safety factor to its criterion to
ensure protection of such species.

EPA-HQ-OW-2004-0019-0409-A2; OC Public Works; Posted 10/20/2015

For comment 9 USEPA should coordinate the United States Fish and Wildlife Service to reach consensus
on the approach to the protection of aquatic-dependent wildlife.

145

-------
Comment Category 3.1 - Comments on Use and Calculation of ECio
3.1 Summary

This section focuses on the use and calculation of ECio. Comments in section include those on EPA's use
of the TRAP software, and the selection of ECio as the most appropriate endpoint.

Commenters suggested:

• Use of ECio is not well justified by EPA and should be better explained

• Commenters agree that ECio is appropriate, as compared to EC20, but EPA's methodology using
TRAP, which is consistent yet yields values for ECi0s that are different than those estimated by
study authors; these commenters suggested using the EC20 values as an estimate for the ECi0s

• Use of ECio is not protective of listed species
One commenters agreed with EPA's use of ECio.

3.1 Response to comments

3.1 Response to concerns with the use and calculation of ECio

When considering the use of the ECio versus the EC20, an ECio was determined to be a more appropriate
endpoint for tissue-based criteria given the nature of exposure and effects for this bioaccumulative
chemical. EC20s have historically been used in the derivation of EPA criteria applicable to the water
medium. While water concentrations may vary rapidly over time, tissue concentrations of
bioaccumulative chemicals are expected to vary gradually. Thus, where concentrations of selenium in
fish tissue are used as an effect threshold, there is potential for sustained impacts on aquatic systems,
relative to chemicals that are not as bioaccumulative. This calls for use of a lower level of effect to
attain sufficient protection.

Also, detection in fish tissue above these levels can potentially have longer-term impacts due to the
bioaccumulative nature of selenium. This is further demonstrated by field observations that
reproductive effects in fish can take on the order of a decade for a selenium contaminated ecosystem to
recover from (e.g., Belews Lake, NC). Further, and most importantly, the steepness of the dose-response
curve for selenium argues for a more protective value, not less, because small increases in
concentrations of selenium can have larger impacts than for toxicants with shallower dose-response
curves.

The ECio was also preferred for selenium over the NOEC or LOEC as these measures of effect are
influenced by study design, specifically the concentrations tested, the number of concentrations tested,
the number of replicates for each concentration, and the number of organisms in each replicate. As
noted by Campbell (2011), ECi0s and NOECs are generally of similar magnitude, but ECi0s have the
advantage of being more reproducible than NOECs (Van der Hoeven et al. 1997; Warne and van Dam
2008).

Regarding the comment that it is possible that the ECio might not be sufficiently protective of a
particular endangered aquatic species, the commenter did not recommend any specific studies,
regarding any particular endangered species that provides additional evidence for EPA to consider, to be
addressed more thoroughly. Further, published data indicates that endangered fish species have not

146

-------
been found to be more sensitive to toxicants than common non-endangered species (Sappington et al
2001). EPA acknowledges that there may be locations where a lower criterion could be applicable on a
site specific basis due to the site-specific presence of a particularly sensitive species (including those that
are endangered). However, as discussed in section 2.8 above, EPA disagrees that before making general
recommendations to states regarding future state actions to adopt selenium criteria, it is helpful or
legally necessary to first engage in consultation under the ESA to ensure that any possible subsequent
federal action to approve new or revised state selenium criteria consistent with the national
recommendations would necessarily be protective of listed species.

Regarding the comment that "the degree of variability in measured concentrations would seem to
provide rationale in the choice of statistic to best characterize exposure... rather than in the choice of an
ecologically meaningful effects level," the goal of aquatic life criteria is, roughly speaking, to protect
aquatic life from ecologically meaningful and adverse effects. Thus, EPA selected such a level for the
criteria basis, and does not find the rationale that the criteria should be based on statistical utility
compelling.

Following issuance of the 2015 draft, EPA has reviewed the statistical fits of the using the computer
program TRAP (version 1.30a), Toxicity Relationship Analysis Program (U.S. EPA 2013) for all fish species
that have a numeric impact on the criterion, as suggested by a number of commenters. A full discussion
of the refined statistical approach is provided in the 2016 criterion document, with details provided for
each species analyzed.

3.1 Specific comments

EPA-HQ-OW-2004-0019-0403-A2; Iowa Department of Natural Resources; Posted 10/15/15

(6) The use of EC10 as the chronic toxicity end point to derive the criteria is inconsistent with the EPA's
current procedure, which recommends the use of EC20. The use of the end point EC10 as the end point
is not well-justified.

EPA-HQ-OW-2004-0019-0385-A2; Kentucky Division of Water (KDOW); Posted 10/13/2015
Appropriate Endpoint and Criterion Calculations

KDOW agrees with the EPA's selection of the ECio (Effective Concentration) in preference of the EC2o
(the observed or measured effect concentration at 10 or 20 percent, respectively) to measure a chronic
end-point (e.g., growth, development effects or reproduction). Previous national toxicity criteria were
derived using the EC2o- This approach mirrors that taken by KDOW when developing its chronic criterion
for selenium. Calculations were also made from appropriate studies for the NOEC (No Observed Effect
Concentration) and LOEC (Lowest Observed Effect Concentration) with preference given to the ECio.

EPA-HQ-OW-2004-0019-0415-A1; Idaho Department of Environmental Quality (DEQ); Posted
10/28/2015

147

-------
The rationale provided on page 20-21 for favoring an ECio over an EC2o for assessing selenium toxicity,
i.e. stability of tissue concentrations of selenium relative to aqueous concentrations leads to greater
potential for sustained impacts, should be more fully developed. This appears to us to be more of a
policy decision by EPA, than something that is needed to deal with lower environmental variability. In
our view the degree of variability in measured concentrations would seem to provide rationale in the
choice of statistic to best characterize exposure, e.g. choice of a mean or median of observed
environmental concentrations versus an upper percentile, rather than in the choice of an ecologically
meaningful effects level. Variability in exposure among individuals in a population versus variability in
exposure overtime in relation to expected population level effects deserves greater discussion.

EPA-HQ-OW-2004-0019-0411-A2; American Petroleum Institute (API); Posted 10/28/2015
1.1.1 Use ofEClOs

We support the use of ECio values to develop the tissue-based Se criteria, as data allow. Use of ECi0s is
more conservative and is consistent with other recent approaches (e.g., DeForest and Adams 2011).
However, we note that for some studies it may not be possible to calculate technically defensible ECIO
values and note that EC2o (20% effect concentrations) can realistically be used where ECio, values cannot
be calculated (Chapman 2015). In general, we support the data decisions used by EPA for those studies
where ECio values could not be calculated, with the exception of the suggested modifications for specific
studies and their data noted below.

2.1 Review of Toxicity Studies Used for Criteria Development

The 2015 draft Se criteria document includes reproductive toxicity study data for nine fish genera (see
Table 3.1, page 40). As noted previously, we concur with most of the EPA's decisions regarding which
data to use. However, we would like to provide comments and suggestions on the data used to develop
the chronic egg/ovary criterion (and subsequently, the whole-body and muscle criteria). We believe that
incorporation of these suggested changes would result in a more scientifically-defensible egg/ovary
criterion, which would have greater consistency with EPA's other data-usage decisions used elsewhere
in the document.

2.1.1 General

We examined each of the data points deemed to be acceptable by EPA for use in the calculation of the
egg/ovary criterion (see Table 3.1, page 40); the intention was to determine if we saw any potential
issues with EPA's use of the data.

We have some concerns with the EPA's exclusive use of the Toxicity Relationship Analysis Program
(TRAP) for all data analyses conducted, although we understand the value of standardizing the
calculation of chronic values from each study by using one statistical approach. However, many of the
values calculated by TRAP differ considerably from those calculated by the study authors. We would
again recommend that EPA consider comparing effects calculations using other standard toxicological
statistical programs to better understand the variability among programs used by researchers in this
field, and any implications of their choice to use TRAP for the resulting criteria.

148

-------
EPA-HQ-0W-2004-0019-0401-A1; Colorado Wastewater Utility Council; Posted 10/15/2015

1.3 Use of EClOs

We agree with the use of ECio values to develop the tissue-based Se criteria, as data allow. Use of ECi0s
is more conservative and consistent with other recent approaches (e.g., DeForest and Adams 2011). In
addition, for many of the studies, other endpoints (e.g., EC2o) may not be able to be determined based
on the response curves observed in the data.

We also understand that not all available studies provide sufficient data to reliably calculate ECio values.
In general, we support the data decisions used by EPA for those studies, with the exception of the
suggested modifications for specific studies and their data noted later in our review (Section 3.1).

1.4 Review of Toxicity Studies Used for Criteria Development

The 2015 draft Se criteria document includes reproductive toxicity study data for nine fish genera (Table
3.1, page 40). As noted previously, overall, we concur with most of the data usage decisions made by
EPA. However, we would like to reiterate some of our comments and suggestions on the data that were
used to develop the egg/ovary chronic criterion (and subsequently, the whole-body and muscle criteria),
as well as provide new comments and suggestions. We believe incorporation of these suggested
changes would result in an egg/ovary chronic criterion that is even more scientifically defensible and
have greater consistency with EPA's other data-usage decisions used elsewhere in their document.

1.4.1 General

As with the 2014 review, we examined each of the data points deemed acceptable by EPA for use in the
egg/ovary criterion calculation (Table 3.1, page 40) to determine if we saw any potential issues with
EPA's use of the data.

We continue to have some concerns with the use of TRAP for all data analysis (described in Section 4.2),
although we do understand the value in standardizing the calculation of chronic values from each study
by using one statistical approach. However, many of the values calculated by TRAP differ considerably
from those calculated by the study authors. We would recommend EPA consider comparing effects
calculations using other standard toxicological statistical programs to better understand the variability
among programs and any implications of their choice to use TRAP for the resulting criteria.

1.5 Use of TRAP versus Other Statistical Methods

As mentioned previously, we continue to have some concern with the exclusive use of TRAP for
determination of all ECio values used in criteria development. TRAP may work well with certain data
sets, but may pose problems for analysis of others. We provided additional discussion of the potential
issues in our previous review (GEI 2014a).

In many of the studies provided in the 2015 draft Se criteria document, the ECio values calculated by EPA
differed from the original values determined by the study authors and those calculated by DeForest and
Adams (2011). EPA should provide a list of all parameters and "Initial Guesses" on slopes used for
analyses.

EPA-HQ-0W-2004-0019-0400-A1; The National Mining Association (NMA); Posted 10/15/2015
1.3 Use of EClOs

149

-------
We agree with the use of ECio values to develop the tissue-based Se criteria, as data allow. Use of ECi0s
is more conservative and consistent with other recent approaches (e.g., DeForest and Adams 2011). In
addition, for many of the studies, other endpoints (e.g., EC2o) may not be able to be determined based
on the response curves observed in the data.

1.4 Review of Toxicity Studies Used for Criteria Development

1.4.1 General

We have some concern with the ECio calculations used in the document such as the use of TRAP for all
statistical analyses and we recommend EPA consider other options as well. We have also provided
additional information regarding calculation of egg/ovary to whole-body conversion factors (CFs), which
we used to revise the GMCVs and SMCVs discussed previously.

1.5 Use of TRAP versus Other Statistical Methods

150

-------
EPA-HQ-OW-2004-0019-0394-A2; Appalachian Mountain Advocates and Sierra Club
Environmental Law Program; Posted 10/14/2015

Moreover, as USFWS has noted to EPA, use of the EC10 effect level that EPA has employed here, see
Draft Criterion at 21-22, is inappropriate for water quality criteria that apply to listed species. When
dealing with listed species, every individual is important. An EC10 effects level assumes that one out of
every ten individuals will suffer adverse effects. That is unacceptable for listed species. As the Us Fish
and Wildlife Service explained in their comments on the 2014 Draft Criterion,

it is still unclear how an EC-10 standard for fish-tissue criteria relates to threatened and
endangered species conservation. A large majority (>90%) of all species of freshwater fish listed
under the Endangered Species Act (ESA) have not been tested for sensitivity to selenium
toxicity. Assuming that ESA-listed species exhibit a distribution of sensitivities comparable to
non-listed species (as several EPA-funded studies have indicated), it can be expected that in
waters achieving EPA's newly proposed fish-tissue criteria about 5% of ESA-listed species would
experience a 10% or greater level of reproductive toxicity. Also, it can be expected that some
unknown additional percentage of ESA-listed species would experience a level of reproductive
toxicity greater than 0% but less than 10%. It's possible that a toxic standard more protective
than an EC-10, such as EPA's (2012) benchmark dose approach, might be required for waters
that host untested ESA-listed species offish.

July 28, 2014 Comments at 1. In order to ensure that endangered species are protected, EPA must
initiate and complete consultation with the USFWS pursuant to Section 7 of the ESA prior to finalizing
any recommended aquatic life criteria for selenium.

151

-------
Comment Category 3.2 - Comments on Studies of Fish Reproductive
ll lil< ¦ i i11 < - I nil II iiinii< mi' < mi' ii iii-111 ¦< riv.iiMi

3.2 Summary

This section includes comments concerning studies of the reproductive effects of selenium on fish. The
majority of the comments focus on specific studies of three of the four most egg ovary sensitive genera
(i.e., white sturgeon, bluegill, and brown trout). In general, commenters questioned the validity and
applicability of specific studies. Additionally, one commenter questions the overall methodology used by
EPA and suggests that EPA change to consider the unique dietary toxicant nature of selenium, in
particular with respect to threatened and endangered species.

3.2 Response to comments

3.2 Responses concerning studies on fish reproduction effects

EPA had significant concerns with the 90-day endpoint in Hamilton et.al. 1990, most significantly that
the 90-day control survival (67%) was below toxicity test acceptability thresholds, such that we could
not use those effects data. We did consider and use the 60-day time point for this study, and concluded
that the egg-ovary transformed whole body criteria of 8 mg/kg dw would protect against growth effects
in juvenile salmonids.

Updated descriptions of the key toxicity tests and the statistical analyses used to calculate the ECio are
provided in toxicity Section 3 of the 2016 Criterion document, and are summarized below for reference.

Whenever possible, estimates of selenium concentrations associated with a low level of effect (i.e., ECio)
were calculated for each study using the computer program TRAP (version 1.30a), Toxicity Relationship
Analysis Program (U.S. EPA 2013). The program is based on a regression approach that models the level
of adverse effects as a function of increasing concentrations of the toxic substance. With the fitted
model it is possible to estimate the contaminant concentration associated with a small effect. TRAP is
used when there are sufficient data for ECio estimation. For studies with binary data, the analysis
proceeds by tolerance distribution analyses using the log-triangular distribution, unless there is
substantial extrabinomial variability, in which case regression analysis was used. For regression analysis,
the threshold sigmoidal model is used, exposure variables are log-transformed, and effects variables are
weighted appropriately to address their relative uncertainties.

When there are insufficient data for TRAP to fit an effects/exposure curve (no treatments with clear
effects near the ECio and/or significant background variability), the ECio is based on interpolation. To
ensure that the interpolations were comparable to the TRAP analyses, threshold sigmoidal equation was
used. This equation is fitted to two points, and constrained so that 3 equation parameters can be set.
The first set-point is treated as the EC0 with a second associated set-point being the threshold for
background effects values, based on the highest NOEC (HNOEC) datum and other NOEC data. The final
set-point is the LOEC. If the LOEC is a partial effect, then this point was used to estimate the equation
slope. If the LOEC is a 100% effect, it is specified as the ECioo; with the EC0 specified, then this
relationship dictates the equation slope. It should be noted that despite the superficial resemblance of

152

-------
these analyses to TRAP they are also subject to the uncertainties associated with the interpolation
method.

It should be noted that TRAP involves the assumption that (a) there is a single underlying relationship of
the effects variable to the exposure variable which follows the specified equation and (b) the exposure
variable is known with negligible error, with uncertainty being predominantly in the effects variable.
Some of the reproductive data involved multiple sources of variability that lead to both multiple
relationships across different cohorts of offspring and uncertainty in the exposure variable, so that the
resulting TRAP curves were more approximate, and TRAP error estimates are generally not useful. These
issues can also affect the interpolation protocol. It should also be noted that estimating a concentration
associated with a low effects level, such as an ECio, is especially uncertain when treatments yielding
partial effects values are lacking in the concentration response data produced by a study. These two
issues prevented the use of TRAP for some datasets. When the data are insufficient to provide any
meaningful ECio by the first two approaches, the study should either not be used for criteria
development or a chronic value should be set by other means than an estimated ECio if possible.

Brown Trout (Salmo trutta)

For the (brown trout) Formation Environmental study (2011), EPA continues to use the 2015 draft report
analysis approach, with a refinement of the statistical fit. Thus, EPA recalculated the ECio for larval
survival for the hatch through swim-up portion of the test based on the same data as in the 2015 draft,
but using an improved statistical analysis and fit of the data. The study had two phases, hatch-to-swim
up, and swim up-to-15 days post swim-up. There are two experimental complications that affected the
interpretation of these data: (a) elevated deformity rates among the offspring that were to serve as
hatchery-originated method controls (very low selenium exposure) and among some of the low
exposure field-collected organisms, and (b) the accidental loss of a number of individuals from several
treatments during the 15-day post swim up portion of the test due to overflow of the tank water.
Because of uncertainties as to how to best address the loss of fish during the overflow event during the
second phase of the test, and also because of the preferential selection of healthy fish introduced during
the thinning process prior to the post-swim up potion of the test by the experimenters, where only
those individuals presumed to be healthy were retained, survival during only the first portion of the test
(hatch to swim up) was used because it provides a more reliable chronic value.

The dataset of percent survival from hatch to swim up versus the selenium concentration in the eggs
was an excellent dataset and provided a good foundation for setting benchmark concentrations for
selenium. There is a narrow range between the NOEC (20.5 mg/kg) and an LOEC with severe effects
(26.8 mg/kg, 61% mortality) that leaves little uncertainty in what an appropriate low effects
concentration should be. There were sufficient data for TRAP to estimate a curve, using weighted least-
squares nonlinear regression with the threshold sigmoidal model generating an ECio of 21.0 mg/kg.

Bluegill {Lepomis machrochirus)

In the 2016 selenium criterion document, EPA used three Bluegill studies in the derivation of the tissue-
based criteria: Doroshov et al. (1992), Coyle et al. (1993), and Hermanutz et al. (1992, 1996). The text
below describes an overview of EPA's re-examination of the Hermanutz blueguill data, which EPA
retained 2016 criterion document, and described in full in the 2016 selenium criterion document in
section 3.

153

-------
Hermanutz et al. (1992), and Hermanutz et al. (1996) exposed bluegill sunfish to sodium selenite spiked
into artificial streams which entered the food web, thus providing a simulated field exposure
(waterborne and dietary selenium exposure. These data come from a series of three studies lasting from
8 to 11 months, conducted over a 3-year period. All three studies began with exposure of adult bluegill
sunfish in the fall, and with respective studies ending in the summer of the following year. Temperatures
averaged 4.6, 4.1 and 4.5°C during the winter months and averaged 26.4, 23.9 and 22.4°C during the
spawning months (June-July) for Studies I, II and III, respectively.

The egg-cup data for all streams of Studies I, II, and III of this experiment were combined and analyzed in
response to measured selenium concentration in the maternal ovaries (mg/kg dw) using TRAP; i.e., all 3
experiments were used in the 2016 selenium criterion analysis for this study. Data for streams receiving
water-borne selenium were combined with data for streams recovering from the previous year's
contamination. The absence of effects at high tissue levels (55 mg Se/kg ovary dw) in the recovering
stream of Study II did not affect the ECio estimate because it was outweighed by three other points
showing severe effects at concentrations as low as 16.7 mg Se/kg ovary dw. Several egg-cup endpoints
were analyzed by TRAP independently (% edema, % lordosis, and % hemorrhage) and in combination (%
normal and surviving).

The best fit and most sensitive was the combined percent normal and surviving larvae. Due to
inadequate partial effects for the ovary analysis, a threshold sigmoidal model was used to interpolate an
ECio estimate between the first interpolation point set to the highest no observed effect concentration
(HNOEC) of 14.0 mg/kg and the average background survival/normal of 69.1% and the second point set
to the LOEC of 16.7 mg/kg and a survival/normal of 5.76%. The resulting ECio is 14.7 mg/kg ovary dw.

The SMCV for bluegill reproductive endpoints based on ECio values is 20.6 mg Se/kg dw in egg/ovary,
based on the ECio values of Doroshov et al. (1992), Coyle et al. (1993), and Hermanutz et al. (1992, and
1996 as corrected by Tao et al. 1999).

White sturgeon (Acipenser transmontanus)

The discussion below provides an overview of EPA's statistical re-analysis of the Linville (2006) data for
the dietary/reproduction part of the experiment.

EPA agrees with the comment regarding the egg injection aspect of the Linville 2006 sturgeon study:
"results obtained through short-term exposure of seleno-L-methionine via injection into larval fish yolk
sacs are not environmentally relevant and do not correlate to environmental dietary exposure during
the period of embryonic development."

EPA notes it did not, in any draft nor in the final criterion, use the Linville egg injection data from the
study in the criterion calculation, as clearly explained in the criterion document. There are dietary,
maternal transfer aspects in the Linville study, and these are the data that were used by EPA. Linville
(2006) data EPA considered regarded the effect of elevated dietary selenium on the health and
reproduction of white sturgeon. Adult female white sturgeon (approximately 5 years old) were fed
either a control diet or a diet spiked with selenized yeast for six months in a freshwater flow through
system. At the end of the dietary exposure, females were induced to spawn and fertilized with non-
exposed male milt. Large cohorts of fertilized eggs from individual females were collected and

154

-------
separately hatched and progeny from two control females and three treatment females were examined
for length, weight, edema and deformities. The test fulfilled Guidelines requirements for larval control
survival and the highest test concentration indicated evidence of reproductive effects.

No selenium effects were observed for length or weight of larvae but effects were observed for both
edema and skeletal deformities. EPA evaluated larval survival and abnormality frequency at stage 45.
Because the mortalities for each cohort were recorded up to the sample collection time for
abnormalities, a combined effects variable was derived based on the total proportion of hatched larvae
which were both alive and without any abnormalities at stage 45.

Estimation of the ECio was conducted using weighted nonlinear regression analysis with the threshold
sigmoidal model equation (TRAP version 1.30a).

Since the study yielded only one definite partial effect, TRAP cannot be used to estimate a
concentration-response curve. Instead, TRAP was used to interpolate between the last two points to
estimate the ECio (see Appendix C for detail). The resultant TRAP slope is 2.96 and the interpolated ECio
is 15.6 mg/kg.

The white sturgeon ECio of 15.6 mg/kg egg dw is important because this species is a commercially and
recreationally important fish species in the Pacific Northwest, and also serves as a surrogate for other
sturgeon species in the United States (see Section 6.3, Protection of Threatened and Endangered
Species), and has a population listed as endangered in the Kootenai River in Idaho and Montana.

Other Fish Species

Regarding the use of the Esox data, EPA exhaustively examined the data. EPA does not believe that the
authors' ECio calculation is scientifically defensible. The spacing between exposures is too large to
estimate the ECio either by the authors' linear regression approach, or by EPA's nonlinear regression
approach. But based on its cluster of three values with concentrations near 34 mg/kg dw, and having a
24% effect, EPA estimates that Esox is somewhat sensitive, but not among the four most sensitive
species.

3.2 Specific comments

EPA-HQ-OW-2004-0019-0395-A1; CONSOL Energy; Posted 10/15/15

CONSOL is concerned about the decisions EPA made in calculating the chronic tissue criterion
concentrations. We ask that the EPA adopt the recommendations provided by the National Mining
Association (NMA) on revisions to the bluegill and brown trout Genus Mean Chronic Values (GMCVs) or
Species Mean Chronic Values (SMCVs).

Fish Body Burden Recommendations

CONSOL has concerns with the usage of specific bluegill, brown trout, and white sturgeon data in the
criteria calculations.

• Bluegill - We have major concerns about the data used from the Hermanutz et al. (1992, 1996)
studies. These concerns were previously submitted to the EPA by commenters in response to

155

-------
EPA's 2014 draft criteria. EPAs decision to exclude Study III and several streams from Study II
because they were considered to be "recovering" streams and were no longer receiving aqueous
selenium calls into question the validity of its use. These fish were exposed to selenium and EPA
has not properly explained why these studies were not included in the criteria development.
Also, the use of results from Study I and selected results from Study II over two different years is
not common practice in the development of criteria. We recommend excluding all of the data
from the Hermanutz studies. Hermanutz et al. (1992) had a chronic value of 11.36 mg/kg while
the more reliable studies by Doroshov (1992) and Coyle (1993) had chronic values of 20.05
mg/kg and 24.55 mg/kg respectively. The impacts of the use of this invalid study are significant.

• Brown trout - During the Formation Environmental (2011) study a tank overflowed resulting in
the loss of many study fish. In the 2014 draft, EPA presented two approaches for dealing with
this loss of these study fish:

1. Assume that all fry lost were dead or deformed ("worst case" assumption), or

2. Assume that fry lost had the same rates of mortality and deformities as those not lost
("optimistic" assumption).

In the 2015 draft EPA acknowledged the uncertainty with how to best address the loss of fish,
and decided to calculate an ECio for survival only during the first portion of the test, prior to the
accident. This resulted in calculation of an ECio of 18.09 mg/kg egg dw for larval survival. We
recommend using the brown trout ECio calculated under the "optimistic" /realistic assumption
for dealing with the lab accident that resulted in loss of the study fish. This data would represent
what was observed in the remaining fish population that was not lost to overflow. An
assumption that the fish lost from the study have the same rate of deformity and mortality as
the other fish in the study is much more scientifically defensible.

• White sturgeon - CONSOL believes that the Linville (2006) white sturgeon data should be
excluded from the 2015 draft criteria calculations. We recommend these data be excluded from
the calculations due to lack of sufficient response and a scientifically defensible ECio calculation.
These data were not included in the 2014 draft. White sturgeon is now the most sensitive
species in the database. The EPA calculated ECio for total deformities (edema+ skeletal) of 16.3
mg/kg was calculated based on a partial dose response. Use of an ECio based on a partial
response is not appropriate for national criteria development, especially when it results in a
value becoming the "most sensitive" and thereby driving the final criterion. In fact, on page 33
of the 2015 draft criteria document, EPA states that "an ECio based on only one partial response
would not ordinarily be included in the data set".

EPA-HQ-OW-2004-0019-0390-A1; West Virginia Coal Association (WVCA); Posted 10/14/15

In this case, the four lowest GMCVs are for white sturgeon, bluegill, brown trout, and largemouth bass.
EPA's data decisions for the lowest three species - white sturgeon, bluegill, and brown trout - depart
significantly from acceptable scientific procedure and from the requirements of the 1985 Guidelines.
The following data decisions resulted in body burden elements that are overly stringent and inaccurate.

156

-------
(1) EPA should have excluded Linville (2006) for white sturgeon. EPA correctly determined that
this study was not appropriate for use in criterion calculation in its 2014 peer review draft
criterion. However, due to pressure from US Fish and Wildlife Service, EPA included this
inappropriate study in the 2015 Draft Selenium Criterion. EPA explicitly acknowledges that
Linville (2006) does not qualify for inclusion: "An ECio based only on only one partial
response [27.8% effect at the highest egg selenium concentration] would not ordinarily be
included in the chronic data set." (Draft Selenium Criterion, p. 33). Clearly, EPA seems to
believe that the use of bad science is appropriate when no valid data exists. This is
disturbing on the most basic level.

(2) EPA should have excluded Hermanutz (1992) for bluegills due to serious issues with the
scientific validity of the test. Multiple commenters raised these concerns with EPA in
response to EPA's 2014 draft criteria. The difference caused by this invalid study is
substantial. Whereas Hermanutz (1992) had a chronic value of 11.36 mg/kg, the more
reliable bluegill studies had chronic values of 20.75 mg/kg (Doroshov, 1992) and 24.55
mg/kg (Coyle, 1993). The Hermanutz study should be excluded and the criteria recalculated.

(3) For brown trout, EPA mishandled the interpretation of data where fish that accidentally
were lost due to a laboratory accident causing overflow from the test chambers (Formation,
2011). EPA calculated an ECiofor survival during only the first portion of the test, resulting in
an ECioof 18.09 mg/kg. The more defensible approach is to excluded these fish (324 lost of
11,130 fish) from the results of the toxicity testing, resulting in an assumption that the fish
accidentally lost from the testing had the same rate of deformity and mortality as the other
fish in the study.

These adjustments to the database should be made, and the criteria calculations revised accordingly.

EPA-HQ-OW-2004-0019-0396-A1; National Marine Fisheries Service Office of Protected
Resources Interagency Cooperation Division; Posted 10/15/15

The selenium guidelines are unique because selenium is a dietary toxicant and an appropriate
methodology needed to be developed. Likewise, when evaluating dietary data, EPA needs to include
dietary considerations in its discussion. NMFS agrees that control survival rate is an important
component of study quality, but control survival alone should not be a basis for evaluating such data.
For example, dietary studies, particularly those measuring growth, need to account for any influence of
dietary content in addition to the stressor being evaluated. Specifically proximate analysis (lipid,
carbohydrate, protein, fiber, moisture, and ash) of the diets need to be comparable. Hamilton et al.
(1990) did not report proximate analysis, but the diet composed of reference site mosquitofish spiked
with selenomethionine could be considered a dosed reference diet that would account for any effects of
nutritional quality or other toxicants in the diet made using mosquitofish from the selenium-
contaminated San Luis Drain. The two diets provided similar results indicating a selenium dose response
decline in mass and length throughout the study. All treatments experienced high mortality at 90 days,
negating the possibility that control mortality would be due to deficiency in selenium in the
uncontaminated control diet. The 90 day data were also marked by a slight decline in growth rates.
Mortality and growth rate decline at 90 days was greatest in the higher selenium exposures.

At 90 days a confounding stressor (e.g., disease, parasitism) likely influenced results, but the effect of
that stressor was most severe in treatments that were already compromised by selenium toxicity. Given
EPA's interest in updating the methodology, exploration of ways to extract useful information from

157

-------
datasets such as the 90 day exposures in Hamilton et al 1990 offers an opportunity to expand the
dataset.

Inclusion of larval deformity data for white sturgeon: The only sturgeon data considered in the original
2014 draft selenium guideline were for growth decline in white sturgeon. NMFS appreciates EPA's
consideration of De Riu's 2014 data on the sensitivity of green sturgeon to selenium in deciding whether
to incorporate Linville's 2006 larval deformity data for white sturgeon. EPA's calculated larval deformity
ECio of 16.3 mg/kg egg dw is close to the selenium egg-ovary guideline of 15.8 mg/kg dw (<5% higher)
and EPA concluded that the guideline is protective because it falls below the white sturgeon larval
skeletal deformity ECio. However, confidence interval for the ECio of 16.3 mg/kg egg dw likely overlaps
the guideline and Linville also reported an ECio for either larval edema or skeletal effects that was below
the guideline, at 15.01 mg/kg dry weight. In stating that the guideline is protective when it is roughly
equivalent to an EC10 for larval deformity, EPA needs to take into consideration whether an effect level
of 10% represents an impact that can contribute to the decline in recruitment and may adversely affect
threatened and endangered fish species under the jurisdiction of NOAA Fisheries.

Addressing potential for greater selenium sensitivity in ESA-listed species: While EPA acknowledges De
Riu's data suggests that green sturgeon are more sensitive to selenium than white sturgeon, EPA
concludes "inclusion of the white sturgeon's ECIO in the dataset provides surrogacy for the threatened
and endangered species from this group". Further, EPA recommends that, since ESA listed species may
be more sensitive to selenium, relevant new information should be considered in state- or site-specific
criteria calculations (page 127). This does not convey the greater importance of toxicity to individuals of
ESA-listed species, regardless of whether the species EC10 is similar to or lower than that of non-ESA
listed species. Populations of a listed species may not be able to sustain a 10% decline in reproduction,
while stable populations of non-ESA listed species may not be affected. Further, EPA does not suggest
that existing data should be reassessed for waters where selenium-susceptible ESA-listed species occur
and does not specifically indicate the presence of ESA-listed species among reasons for deriving site-
specific criteria (Section 5). New data should not be the only consideration for state- or site-specific
criteria calculations that are protective of ESA-listed species. Reassessing existing data is the most logical
approach to deriving such criteria. For example, if the generic guideline is based on the geometric mean
of observed EClOs for a particular response, a site-specific criterion intended to be protective of ESA-
listed species may be a lower ECxx calculated from existing data or, if an EC10 is determined to be an
appropriate threshold, the appropriate criterion may be the lower bound or some lower quantile within
the credible interval around the geometric mean of observed EClOs.

Extrapolating to water column and whole body concentrations and calculation of site-specific criteria:

White sturgeon parameters used in the derivation of whole body tissue and water column guidelines
and site specific guidelines do not reflect the environmental conditions under which shortnose sturgeon
would accumulate selenium. Shortnose sturgeon ingest much sediment when foraging, and they
consume mainly mollusks from bottom sediments. Stomach contents of juvenile shortnose sturgeon
have been reported to contain 85-95% non-food matter, mostly mud (Curran and Ries, 1937; Dadswell,
1979). EPA's model does not account for species that consume large amounts of sediment/particulate
bound selenium. In addition, the trophic transfer factor for white sturgeon used in EPA's model indicates
a diet composed predominantly offish, insects, and benthic crustaceans, with a dietary mollusk
composition of only 9%. Mollusks tend to accumulate selenium at higher concentrations than other
trophic level 2 organisms so fish that consume a large portion of their diet as freshwater mollusks, such

158

-------
as shortnose sturgeon, will exhibit greater selenium bioaccumulation than fish in the same waters that
consume primarily insects or crustaceans.

EPA-HQ-OW-2004-0019-0385-A2; Kentucky Division of Water (KDOW); Posted 10/13/2015
Use of Appropriate and Peer Reviewed Science

The EPA appropriately reviewed the body of available scientific literature when it developed its database
for criterion formulation, which resulted in a pertinent subset of studies based on the understanding of
the mode of selenium toxicity. However, the final chronic value of this draft criterion is based in-part on
reliance on a study that was not peer reviewed and for which the results obtained were driven only
partly by dietary uptake. EPA's reliance on the bioassay results from the Linville (2006) study is
concerning from the standpoint of the use of this data in the development of the tissue-based Final
Chronic Value (FCV) from which to recommend 304(a) Se criteria.

The Linville (2006) study results were based in-part on data gleaned from test offspring where the larval
uptake (exposure) of selenium was not solely through dietary uptake; rather, the larval yolk sacs were
injected with seleno-L-methionine directly. This is not the pathway in the aquatic habitat by which fish
are exposed to selenium, but more importantly the study does not account for complex processes in
dietary exposure and subsequent cellular incorporation, including ratios and forms of other chemical
compounds and elements in the natural environment that may affect the incidence of teratogenicity;
therefore this exposure is not considered empirical.

EPA has rejected the use of studies based on this method of exposure (i.e. direct injection); in prior
drafts of selenium criteria documents (e.g. see page 6, Appendix G, EPA 2004 draft selenium criteria
[e.g. Doroshov, 1992]) as a matter of misapplication of procedures to derive water quality criteria
(Stephan, et al. 1985). KDOW recognizes that the long ovarian cycle may make it difficult and/or
expensive to study teratogenicity effects in white sturgeon, however that is not a scientifically valid
reason to go forward with the a study utilizing inappropriate methodology. Therefore, results obtained
through short-term exposure of seleno-L-methionine via injection into larval fish yolk sacs are not
environmentally relevant and do not correlate to environmental dietary exposure during the period of
embryonic development. To adequately control the experiment and shadow exposure to selenium in
the natural habitat, parental female exposure to selenium through dietary uptake followed by
contaminant transfer from the parental female to the eggs is required.

Furthermore, the stress of handling and injecting the selenium on the test fishes was not thoroughly
accounted, and the sample size was inadequate to allow for statistical comparability. Additionally, the
high mortality rate (45 to 70 percent) of the larvae that underwent injection does not meet the rigor and
test acceptability standards outlined in guidance (Stephan, et al. 1985). Therefore, this study should also
be rejected due to insufficient control performance.

Summary

For the reasons provided above, KDOW supports EPA's taking the following approach to developing
selenium criteria and encourages EPA in the formal selenium criteria proposal to:

159

-------
1. Exclude the Linville (2006) study from the EPA selenium criterion calculations for similar reasons
that EPA has excluded other studies.

EPA-HQ-OW-2004-0019-0388-A2; J. R. Simplot Company; Posted 10/14/2015

An Egg/Ovary Tissue Criterion is the Correct Approach, but the Effects Values Need to be Recalculated

The establishment of an egg or ovary fish tissue value as the criterion is the right approach for the
protection of sensitive species. A review of the studies used (Table 3-1, page 40) show that two of the
studies (white sturgeon and brown trout) warrant a reevaluation of the ECio calculated by EPA. For
example, use of all the data from the white sturgeon study would result in an ECio of 19 mg/kg (dw)
instead of the 16.27 mg/kg (dw) value chosen by EPA.

Brown Trout Study Endpoint should be no lower than 20.5 mg/kg (dw)

Formation Environmental has conducted for Simplot a series of studies for over the past decade
examining the toxicity of selenium on brown and cutthroat trout in eastern Idaho. These studies have
included both field and laboratory work. The brown trout study in particular, has been extensively
reviewed, including an external peer review panel assembled by EPA. One of the reviewers, in response
to one of the "charge questions" posed by EPA, stated the following:

"The final point in Note B to Charge Question #5 is "EPA's belief that given its effort to
understand how the study uncertainties affect the resulting ECio, the study contains valuable
information that should not be discarded. EPA does not believe the uncertainties in the brown
trout ECio are much greater than, and may in fact be less than, those for other studies used in
criteria derivation. "I concur with this view, and note that this study has been vetted to a
degree that I have never before encountered." (ERG 2012- emphasis added)

Simplot commented extensively in 2014 that EPA had evaluated the brown trout study in an excessively
conservative method. EPA has revised the calculation of an endpoint from the brown trout study,
however EPA has not fully followed the recommendations of the peer reviewers. Simplot recommends
the use of the study data to calculate a combined endpoint (survival and deformity) ECio value of 20.5
mg/kg (dw) or higher as compared to EPA's calculated value of 18.09 mg/kg (dw).

B. Effects Data

B.l. White Sturgeon Study

(Page 33, Section 3.1.1.1)

EPA included the white sturgeon study from Linville (2006) and derived an ECio of 16.3 mg/kg dw using
only partial response data from the final stage of a multi-staged reproduction test. In doing so, EPA lost
some of the variability that would occur in the dose response estimate simply due to ignoring those
earlier stages. Below is a dose response curve utilizing all three stages of the original data using the
TRAP model for the same combined endpoint EPA used (edema and skeletal deformities).7 The ECio for
this dose response curve is 19 mg/kg dw. As illustrated in the figure below, the TRAP model attempts to
fit the data, but the dose response is incomplete. However, by including all effects data from the final

160

-------
stages of the test as the original study authors did, the TRAP model is not trying to fit a dose response
curve to a single data point as its basis for the slope of the curve. It is not clear why EPA chose to not use
the earlier stages in their analysis, but if the original author was able to identify edema and skeletal
deformities in these earlier stages, then those data would be applicable for use in defining the dose
response just as the final stage was applicable. EPA should provide a transparent justification for not
including all of the data for the three final stages of the test or include all three stages and adjust the
white sturgeon ECioto 19 mg/kg dw.

Original letter contains Figure 1 - TRAP Model output for the Linville (2006) study using the final three
stages from the study. See original letter.

B.2. Brown Trout Study

(Page 34, Section 3.1.1.2.1, Section 6.1.6, and Appendix C summary for brown trout)

Endpoint Selection and the Overflowing Test Chamber

In the 2014 Draft Criterion, of the six different types of endpoints evaluated for the brown trout study,
and EPA selected the most conservative endpoint. One key aspect to EPA's endpoint selection revolved
around how to handle the loss of fish from a test chamber due to overflow from an overflow pipe
getting clogged (due to uneaten food). Simplot (and others) submitted extensive comments on how to
handle the loss of fish due to the test chamber overflow and other aspects of the different endpoint
evaluations.

For the 2015 Draft Criterion, EPA has decided to use the larval survival endpoint from the first portion of
the test rather than other endpoints they derived. EPA cites the rationale for using this endpoint by
stating, "Because of uncertainties as to how to best address the loss of fish during the overflow event
during the second phase of the test, an ECio for survival during only the first portion of the test was
calculated."8 The EC io for this endpoint was 18.09 mg/kg egg dw for larval survival during the first
portion of the test, hatch to swim up.

EPA has once again selected the most conservative endpoint, "Larval survival was the most sensitive
endpoint, followed by the combined survival and deformities endpoint, with deformities being the least
sensitive. For each of the three endpoints, the ECio for the worst case scenario was lower than the ECio
for the optimistic scenario"9

EPA states the following, "In contrast to other ECio calculations, this approach is free from all
assumptions, and even with respect to larvae that fail to reach swim up, the ECio calculation is based on
measured, rather than assumed, values."10 For this conclusion, EPA set non-swim ups to their actual
value rather than assuming them to be dead as they had done previously per comments from USFWS.11
When the non-swim-ups were set to zero for survival, the ECio was 20.62 mg/kg dw, which is very close
to and within the range of the three optimistic scenario ECi0s is defined.

The Formation Environmental brown trout study has been extensively studied by EPA and various
experts retained by EPA. EPA put together an external peer review (Eastern Research Group 2012) to
review the Formation Environmental 2011 study and EPA's analysis of the Formation Environmental
2001 study. One such reviewer commented:

161

-------
"The final point in Note B to Charge Question #5 is "EPA's belief that given its effort to
understand how the study uncertainties affect the resulting ECio, the study contains valuable
information that should not be discarded. EPA does not believe the uncertainties in the brown
trout ECio are much greater than, and may in fact be less than, those for other studies used in
criteria derivation." I concur with this view, and note that this study has been vetted to a degree
that I have never before encountered." [Emphasis added.]

Thus, the Formation Environmental brown trout study has been extensively and comprehensively
reviewed. This decision to use the most conservative endpoint is not consistent with the peer-review
comments and document. Throughout the discussions of the brown trout data and analyses conducted,
EPA overall concludes that the worst case assumption was not a technically sound analysis approach.12

"In summary, the positive correlation between survival during the hatch to swim up portion of the test
and survival during the 15 day post swim up portion of the test, combined with the lack of correlation
between mortality during the hatch to swim up portion of the test and overflow loss during the second
stage of the test, suggests that the overflow loss may represent a random technical error not related to
the health of the individuals lost. The relationship between selenium egg concentrations and overflow
loss was lower for the larvae hatched from hatchery fish compared to the larvae hatched from field
collected fish; however, among field treatments ranging from 6.0-20.5 mg/kg there was no correlations,
further supporting the hypothesis that the overflow event was a random occurrence unrelated to the
health of larval fish." [Emphasis added]

However, it is unclear why EPA would go through the effort to document and illustrate that the overflow
loss was not related to the health of the organisms and then disregard any of these endpoints from the
optimistic scenario.

By a process of elimination, only the optimistic scenario is a logical choice, and the endpoints remaining
for developing viable ECi0s would then include the following:

• survival- hatch through 15 day post swim up (optimistic);

• deformities - hatch through 15 day post swim up (optimistic);

• combined endpoint - hatch through 15 day post swim up (optimistic);

• survival - hatch through swim up

For each of these endpoints, EPA derived EClOs (with the lowest sum of squares) which are as follows:

• ECio survival- hatch through 15 day post swim up (optimistic)= 20.4 mg/kg dw

• ECio deformities- hatch through 15 day post swim up (optimistic) = 21.95 mg/kg dw

• ECio combined endpoint- hatch through 15 day post swim up (optimistic)= 21.16 mg/kg dw

• ECio survival- hatch through swim up= 18 .09 mg/kg dw

The three different endpoints converge on such a narrow range (20.4 to 21.95) which is a difference of
only 1.5 mg/kg dw, but EPA then used a value of 18.09 mg/kg dw as the brown trout ECio which is
separated by a wider margin from the lowest of the other endpoints . It may be the most conservative
value, but it is not the most representative value. Based on these data, the correct value for the brown
trout ECio is best represented by the combined endpoint of 21.16 mg/kg dw.

162

-------
Deformity Endpoint

EPA also stated that a complicating factor for use of the deformity data and the combined endpoint data
was the presence of multiple minima and high variability among low selenium and hatchery treatments.
Simplot commented on these concepts in the 2014 Draft Criterion and provided a revised summary and
derived ECio to provide evidence that the high variability in low and hatchery endpoints is a function of
the ranking system for deformities.

For its deformity assessment and combined endpoint, EPA only used fish that were ranked as "0" (i.e.,
no deformities present). However, a less conservative approach, and likely a more realistic approach,
would be to include fry ranked as 0 or 1 (slight deformities) as insignificant deformities and to include fry
ranked as 2 or 3 as significant deformities. Simplot (Simplot 2013) presented these data to EPA and
Idaho Department of Environmental Quality (IDEQ) in response to the US Fish and Wildlife Service
(USFWS) comments.

The rationale for the approach described above stems from the MacDonald and Chapman (2009) QA
review on a number of larval deformity assessment studies for selenium where they provided some
guidelines to reduce uncertainty in the results of those assessments. In particular, they examined the
frequency of overlap or non-overlap between two observers rating of deformity rankings for larval fish.
They found a lower average difference in observed ranking by two individuals for those fry ranked as 2
and 3 versus those ranked as 0 and 1. In other words, significant deformities were more consistently
identified similarly by two different individuals, whereas insignificant or low levels of deformities were
more often scored differently. The biggest differences occurred for edema for fish ranked in the zero
and one category, but differences were also noted for skeletal and craniofacial deformities in these
same categories. Fewer differences (i.e., more agreement) were found between the two observers for
fish ranked as 2 and 3 level deformities. The findings of this evaluation suggest that enough uncertainty
is present in the identification of no and low levels of deformities that overly conservative estimates of
deformities may be made by only considering fry ranked as zero deformities. From this, the biological
relevance and association of actual effects must be considered for fish that are considered to be only
slightly deformed versus those that have more severe deformities.

When considering deformities based on these two categories for the wild brown trout, the mean
proportion for insignificant deformities (i.e., ranked as 0 or 1) was 92.3 percent (range: 67 to 99%) at an
egg selenium concentration of 20.5 mg/kg dw13 or less. There is a stark contrast to the mean proportion
normal (i.e., ranked as 0) used in the EPA analysis which was 61 percent (range: 31 to 88.7%) at an egg
selenium concentration of 20 .5 mg/kg dw or less. Similarly, the mean proportion for insignificant
deformities (i.e., ranked as 0 or 1) was 17.5 percent (range: 4 to 28%) versus a mean proportion normal
of 5.9 percent (range: 2 to 11.3%) at an egg Se concentration of greater than 20.5 mg/kg/ dw. For the
hatchery fish, the mean proportion normal (i.e., ranked as 0) was 64.4 percent (range: 32.8 to 85.4%).
For that same group of fish, the mean proportion of insignificant deformities (i.e., ranked as 0 or 1) was
88 percent (range: 57.8 to 100%). If only the fish from background areas are considered, (i.e., those fish
from up gradient of Sage Creek including the CC-150 and CC-350 locations), the mean proportion normal
(ranked as 0) is 55 percent (range: 31 to 81 percent). Using these same locations, the mean proportion
of insignificant deformities is 90 percent (range: 67 to 98%).

163

-------
Evidence Supporting a Combined Endpoint

Based on the evaluation of these data, it is apparent that the uncertainty questions EPA raises in the
2014 and 2015 Draft Criterion concerning the deformity rate can be significantly reduced by separating
the data into two categories, insignificant (rank 0 and 1) and significant (rank 2 and 3) deformities. The
organization of these data is such that the analysis can be easily done and there is evidence that
suggests (i.e., the hatchery fish data), an insignificant level of deformities can and does occur even at
low levels of selenium exposure. Consider for example the Yellowstone cutthroat trout (YCT) studies
where "wild fish" treatments were tested simultaneously with Henry's Lake "hatchery" fish. A similar
level of variability is observed with the YCT "hatchery" fish as that found with the brown trout hatchery
fish. However, Henry's Lake "hatchery" fish were not hatchery fish at all, but a wild run of fish from
Henry's Lake where the Idaho Department of Fish and Game (IDFG) has set up a station to retrieve ripe
and running YCT to extract eggs for propagation in other systems through Idaho. These wild fish,
assessed for deformities using the ranking system similar to that used for brown trout (i.e. 0, 1, 2, and 3)
had a similarly variable range of deformities despite low selenium exposure.

For two species, tested in similar manners, low selenium exposures result in a variable range of
deformities (if only fish ranked as zero are considered). For both species, that variability is significantly
reduced by simply using the fish ranked as zero and one and dividing the data into groups of
insignificant and significant deformities. The result of analyses using these data as described above is
increased certainty in the ECio values derived from data with this level of detail.

Using EPA's approach to derive a combined endpoint, Simplot derived an endpoint for surviving fish with
in significant deformities (i.e., 0 and 1 rankings) (see Figure 2). The ECio for this approach is 20.5 (LCL =
19.81, UCL=21.24) mg/kg dw egg selenium.14 It is not surprising that when the data are grouped in a
logical manner, the combined endpoint reflects nearly the exact ECio as the ECio for survival for all of the
data, and the ECio for the survival data from hatch to swim up when the non-swim ups were set to zero.

By including fry ranked as 0 and 1, the analysis appears to be more robust (though no statistical analysis
has been done to confirm this). This illustrates that two different populations of data for brown trout
(i.e., low selenium exposure hatchery fish and higher exposure wild fish) demonstrate a very similar
range of insignificant deformities, similar to observations made in the Yellowstone cutthroat trout
studies that used a wild run of "hatchery fish" and wild fish from the sites adjacent to the Smoky Canyon
Mine. When selenium exceeded 20.5 mg/kg dw in eggs for brown trout as shown, then the response is
immediate.

Original letter contains Figure 2 - Brown trout combined endpoint (deformities ranked as 0 and 1) used.
See original letter.

B.3. Cutthroat Trout Whole Body Conversion Factor

[Page 73, Table 3.12, footnote b]

The footnote states "The reason for the higher variability in the cutthroat trout CF values is due to the
relatively higher CF values in the hatchery fish from the Formation study." It is important to note that
the "hatchery" here is a natural run of Yellowstone cutthroat trout from Henry's Lake and it is not a
hatchery at all, but is a station set up by the IDFG to collect ripe and running males and females for the

164

-------
purpose of collecting eggs and milt. The higher CFs for these fish is simply a function of fish from a
different natural environment.

C. 2. Citation Clarification

[Page 73, Table 3.12, footnote b]

Any references to Newfields should be deleted and replaced with Formation (2011) if specifically
referencing the brown trout study.

EPA-HQ-OW-2004-0019-0387-A1; Colorado Department of Public Health & Environment;
Posted 10/14/2015

The WQCD has concerns about the use of the Hermanutz et al. (1992 and 1996) data for bluegill (pages
37-39). First, the ECio from these studies (11.36 mg/kg) is approximately half the ECi0s from the two
other bluegill studies (20.75 and 24.55 mg/kg) in EPA's database. This discrepancy warrants additional
review of all bluegill studies being considered, and such a review has been conducted by EPA and many
reviewers. As a result of these reviews, several concerns have been raised regarding the Hermanutz et
al. (1992 and 1996) studies. For instance, in Study I, the 10 ng/L treatment resulted in an ovary
concentration of 17.71 mg/kg, while the 30 ng/L treatment resulted in an ovary concentration of only
15.46 mg/kg; these results are not what would be expected under typical dosing conditions, suggesting
there may be issues in the study design or methods. Because there are other bluegill data with
consistent results, the WQCD recommends rejecting this questionable study and using only the
Doroshov et al. (1992) and Coyle et al. (1993) study results for bluegill.

EPA-HQ-OW-2004-0019-0413-A2; Federal Water Quality Coalition (FWQC); Posted
10/28/2015

I. SCIENTIFIC CONCERNS ABOUT STUDIES AND VALUES

In evaluating the data that were used to derive the criteria, the FWQC carne to the conclusion that
several of the Agency's conclusions are problematic and should be reevaluated. One example is the
derivation of a very low EC10 value for white sturgeon, based on incidence of larval edema and
deformities. There were not a clear dose-response relationship for those effects across treatment levels.
Use of the data was justified by EPA based on reported adverse effects of selenium exposure for a
different, federally threatened species (green sturgeon). However, that is not an adequate reason to
apply a value that is not scientifically supported. EPA should reconsider the EC10 value of 16.27 mg/kg
dw, and should apply a higher value based on the white sturgeon data that are available.

Flaws are also evident in how the Agency used studies of selenium effects on bluegill sunfish. Two
studies by Hermanutz et al. (one from 1992 and one from 1996) played critical roles in the Agency's
derivation of a Species Mean Chronic Value (SMCV) of 17.95 mg/kg for bluegill. This value is much lower
than the SMCV's reported in other bluegill studies. The problems in using the low value become evident
when one considers that incorporation of that value results in bluegills appearing to be much more
sensitive to selenium than are largemouth bass. Given that studies looking at both of those species

165

-------
simultaneously showed effects on largemouth bass much earlier than effects on bluegill, it is highly
unlikely that the bluegill are more sensitive than the largemouth bass. Therefore, EPA should remove
the Hermanutz studies from the bluegill database before calculating the final criteria.

EPA-HQ-OW-2004-0019-0410-A2; West Virginia Department of Environmental Protection
(DEP); Posted 10/28/2015

EPA 2015 Draft has improved upon their External peer review draft aquatic life ambient water quality
criterion for selenium—freshwater 2014 (2014 Peer Review Draft) in several ways. Among these, EPA
changed its interpretation of a Formation Environmental brown trout study involving a tank overflow
which killed a portion of study fish. Previously, EPA's interpretation assumed 100% of overflowed fish
were deformed/dead. This new draft interprets the Formation Environmental 2011 study overflow
event closer to DEP's interpretation, changing the GMCV for brown trout from 15.91 mg Se/kg dry
weight Egg-Ovary to 18.09; therefore, DEP supports this change.

EPA-HQ-OW-2004-0019-0415-A1; Idaho Department of Environmental Quality (DEQ); Posted
10/28/2015

The Linville (2006) study on white sturgeon departs from norms for acceptable toxicity studies in its
atypical method of exposure — direct injection of larval yolk sacs — and in its high control mortality.

This study was also not subject to peer review. This is all the more concerning as these questionable
results provide the most sensitive genus mean chronic value. As EPA has done in prior drafts, we
recommend EPA not use this study for criteria derivation due to its departure from EPA's guidelines
(Stephan and others, 1985).

EPA-HQ-OW-2004-0019-0412-A2; Utility Water Act Group (UWAG); Posted 10/28/2015

1.1.1 Acceptable Studies of Fish Reproductive Effects of Four Most Sensitive Genera
White sturgeon

EPA should exclude data from the white sturgeon study from calculations used for setting national
criteria. EPA derived an estimated ECio value of 16.27 mg/kg dw based only on a partial response (27.8%
effect larval edema and deformities) at the highest egg selenium concentration (20.5 mg/kg Se egg)
using the most conservative model across a range of slopes. Although UWAG acknowledges that the
incidence of edema and deformities was elevated at the highest dietary exposure treatment, a clear
dose-response relationship considering all treatment levels was not evident. We believe that this reason
alone should be sufficient to exclude the estimated effect level for this study. But additionally, there was
no observed elevated incidence of edema and deformities in the two lowest treatment levels (2.68 and
7.61 mg/kg, respectively). Moreover, EPA acknowledged that adoption of an ECio based on a partial
response would not normally be included in a data set. Draft Criterion at 33. It is inappropriate to
include a partial response study in the dataset, particularly where, as in this case, use of the study

166

-------
resulted in the white sturgeon being the most sensitive species in the dataset and the cause for lowering
the value significantly.

EPA's basis for including the data is that effects due to selenium exposure have been reported for green
sturgeon (a federally threatened species), and that protection of the white sturgeon provides surrogacy
for other threatened or endangered sturgeon species. Draft Criterion at 126-128. We find this
justification weak. Protection of a specific ESA species is not an appropriate justification for developing a
nationally applicable criterion, especially when the chronic toxicity study is speculative in terms of what
the effect level is. Additionally, a white sturgeon chronic toxicity study conducted by Tashjian et al.
(2006) found no effects on juvenile sturgeon survival within a selenium dietary exposure range of 0.4 -
191.1 mg/kg. Growth rate and swimming ability of sturgeon, however, were significantly affected at
dietary exposure concentrations at or above 41.7 mg/kg.

While there have been numerous studies reporting measured levels of pollutants (organic and
inorganic) in white sturgeon (collected mostly from the Columbia River), to date no specific pollutant
has been unequivocally linked to the species' long-term population stability/decline. Changes in
Columbia River substrate composition, however, have been strongly linked to poor recruitment of the
species in some stretches (Steven and McAdam, 2015). This demonstrates that ecological factors are, at
least, just as important as contaminant levels regarding the long-term sustainability of white sturgeon. If
EPA had compelling evidence that levels of selenium in water or food prey organisms was limiting the
persistence and abundance of white sturgeon, then the relevance of the chronic toxicity study
suggesting an ECio value of 16.27 mg/kg would be acknowledged. Instead, the agency has not made this
demonstration.

Bluegill

EPA should not use the unrealistic, conservative reproductive chronic value of 11.36 mg/kg for bluegill
based on Hermanutz et al. (1992, 1996) in deriving the Species Mean Chronic Value (SMCV). EPA relied
on laboratory studies by Doroshov et al. (1992) and Coyle et al. (1993) along with Hermanutz et al.
(1992, 1996) to derive a SMCV of 17.95 mg/kg for bluegill sunfish. See Table 3.1 (maternal transfer
reproductive studies). Draft Criterion at 40. This SMCV is much lower than would be expected and
results in bluegills appearing to be much more sensitive to selenium toxicity than largemouth bass. This
conclusion is inconsistent with field studies (with both species present) that indicate the exact opposite,
i.e., largemouth bass are more sensitive to chronic selenium exposure than bluegill.

EPA's SMCV calculation was influenced by chronic values reported by Hermanutz et al. (1992) and
Hermanutz et al. (1996). In analyzing the results of these two studies, EPA derived a chronic value of
11.36 mg/kg. UWAG questions the validity of this value as it is markedly lower than the chronic values
reported for the two other bluegill studies: Doroshov et al. (1992) and Coyle et al. (1992). Moreover, the
agency acknowledges the conservatism of the chronic value of 11.36 mg/kg:

The ECio value of 11.36 mg/kg Se dw (combined larval survival and edema in response to
Se concentration in the parental ovaries) is considered an environmentally conservative
chronic value for this bluegill study.

Draft Criterion at 38. UWAG's technical experts believe this chronic value should have been discarded as
an outlier. Failing to discard the extremely low reproductive effects-based value of 11.36 mg/kg for this

167

-------
species results in the bluegill being the second lowest GMCV value for the 11 species tested for
reproductive effects. A related species, the largemouth bass, however, is the fourth most sensitive
species for reproductive effects. This conclusion conflicts with the totality of the science as it relates to
these two species.

The following table indicates chronic values for bluegill and largemouth bass, and the ratios of these
values:

Original letter contains Table 1 - Comparison of Reproductive Chronic Value and Whole Body Converted
Chronic Values for Bluegill and Largemouth Bass. See original letter.

One would not expect bluegill to be 1.7 times more sensitive to chronic selenium exposure relative to
largemouth bass in terms of a whole body effect level, and the science does not support such a
conclusion. In field studies where both species were adversely affected by selenium-laden wastewaters,
UWAG finds no evidence that bluegills were more sensitive (i.e., suffered significant population declines
at an earlier time than largemouth bass). The studies at Belews Lake and Hyco Reservoir clearly showed
that largemouth bass suffered selenium-induced population declines before effects on bluegill. Cumbie
and Van Horn (1980) reported that in Belews Lake the standing crop and density of largemouth bass was
virtually zero during rotenone cove surveys conducted during 1976. During this same year, the standing
crop and density of bluegill were somewhat reduced but a viable population was still maintained.
Crutchfield (2000) reported that in Hyco Reservoir the density of largemouth bass was reduced
significantly starting in 1982, while marked reductions in bluegill density were not observed until 1986.

Studies also show that bluegill are able to accumulate very high tissue levels of selenium without
seemingly adverse population-level effects. In a fly ash-impacted receiving stream (Stingy Run, Ohio)
average whole body and ovary selenium concentrations in bluegill were 17.26 mg/kg and 32.48 mg/kg,
respectively (Reash et al., 2006). These values are significantly higher as compared with EPA's calculated
species mean chronic values for bluegill: 8.41 mg/kg (whole body) and 17.95 mg/kg (egg/ovary).

In summary, UWAG urges EPA to not use the unrealistically conservative reproductive chronic value of
11.36 mg/kg. The effect of using this value is to calculate SMCVs for bluegill that are considerably lower
than those for largemouth bass, which, as discussed above, is not supported by field studies.

Hermanutz Bluegill Study:

"In my opinion, the uncertainty in the moisture content of the bluegill ovaries in the Hermanutz et al.
(1992, 1996) [studies], along with the uncertainties in the ovary Se concentrations in Study I, are
sufficiently great that this study should not be included in the SMCV for bluegill, as there are other
studies (Doroshov et al. [1992] and Coyle et al. [1993]) for which dry weight ovary Se concentrations
were reported and the EClOs from those two studies were very comparable." (2014 Peer Review at 28).

4.0 National Criterion for Selenium in Fresh Waters

Table 4.1 lists the proposed revised fish tissue and water column chronic aquatic life criterion. UWAG
offers comments on the following: (1) the numeric fish tissue criterion values, (2) the fish tissue
exceedance frequency of "never to be exceeded," and (3) the requirement that the lotic or lentic water
column criterion have primacy for new or expanded discharges of selenium.

Numeric fish tissue criterion values must be recalculated.

168

-------
UWAG believes EPA must recalculate the egg/ovary, muscle, and whole-body numeric criterion based
on our comments above that: 1) EPA inappropriately relied on an insufficient, "partial" treatment
response in the white sturgeon study to calculate the SMCV value of 16.27 mg/kg, and thus, this value
should not be used as it is not scientifically sound; and 2) EPA should remove the anomalous, obviously
conservative chronic value of 11.36 mg/kg for bluegill calculated based on the Hermanutz et al. (1992,
1996) studies. With these changes, a more scientifically defensible egg/ovary criterion (about 20 mg/kg)
based on sound data would result.

EPA-HQ-OW-2004-0019-0411-A2; American Petroleum Institute (API); Posted 10/28/2015

• Given the very large uncertainties in the concentration-response data for White Sturgeon, the
EPA should not use these data.

• EPA should exclude from its analysis the technically flawed Bluegill ECio of 11.36 mg/kg-dw.

• EPA should reconsider for Brown Trout an ECio of21.16 mg/kg-dw, based on larval mortality and
deformities.

2.1.2 White Sturgeon (Acipenser transmontanus)

In the 2015 draft criteria document, EPA included data from the Linville (2006) White Sturgeon study.
These data were not included in the 2014 draft, and inclusion of this study now results in White
Sturgeon being the most sensitive species in the database used to derive the criteria. Linville (2006) fed
adult female sturgeon a control diet (i.e., 1.4 mg Se/kg-dw) or Se-enriched diets (i.e., 34 mg Se/kg-dw),
with Se provided as selenized yeast for six months. Females were induced to spawn, and eggs were
fertilized with milt from non-Se-exposed males. Eggs were hatched and larvae reared in the laboratory,
with edema and skeletal deformities observed at different stages of development. Eggs (and hatched
larvae) from each female were kept separate. Only a single dietary Se treatment was tested (vs. a
concentration gradient, as is usually tested in these types of studies). Mean egg Se concentrations from
the different females provided three different egg and muscle Se exposure concentrations (due to
between-individual variability in the internal transfer of Se to different tissues).

The levels of effect cited in Linville (2006) were relatively low, with the highest Se tissue concentration
associated with an effect level of 28%. The lack of a complete concentration-response relationship
imparts uncertainty in the concentration-response models. For example, the EPA derived an egg Se ECio
of 16.27 mg/kg-dw using TRAP, but noted that "the modeled ECio was sensitive to the slope of the
model, and the ECio of 16.3 mg/kg was the most conservative model across a range of slopes with
identical goodness of fit" (EPA 2015). In fact, by varying the initial slope estimate in the model, EPA
could also derive egg Se ECi0s ranging from 17.5 to 19.1 mg/kg-dw (each with an equivalent goodness-
of-fit). The geometric mean of four different ECio estimates derived by EPA is 17.8 mg/kg-dw. In addition
to varying initial slope estimates, the choice of concentration-response model also influences the ECio.
The ECio of 16.3 mg/kg-dw was based on the threshold sigmoid model, while ECi0s of 17.9 and 18.9
mg/kg-dw are estimated from the piecewise linear and logistic models, respectively (both are valid
options in TRAP) (Fig. 1). The geometric mean of these possible ECi0s is 17.7 mg/kg-dw, which is

169

-------
comparable to the geometric mean ECio of 17.8 mg/kg-dw, based on the different initial slope
estimates. Note in Figure 1, however, that the uncertainty in all concentration-response slopes is high;
of 4 data points only one provides a measurable effect. A slope effectively based on a single measurable
effect data point is not technically defensible. Thus the Linville (2006) study is inappropriate and should
not have been used to develop a national criterion.

Original letter contains Figure 1 - Fit of different concentration-response models to the White Sturgeon
maternal transfer effects data from Linville (2006). See original letter.

The use of an ECio based on a partial response is not appropriate for national criteria development,
especially when it results in a value becoming the "most sensitive" and thereby driving final criteria
values. In fact, on page 33 of the 2015 draft criteria document EPA states that "an ECio based on only
one partial response would not ordinarily be included in the data set." A similar partial response was
observed in the Fathead Minnow data from the GEI (2008) study; EPA excluded these data from the
chronic dataset because of an "insufficient response". EPA's rationale for using the White Sturgeon data
that also had only a partial response is that there are data suggesting that the federally-listed Green
Sturgeon is also sensitive to Se. The Green Sturgeon study (De Riu et al. 2014), however, consisted of
effects on survival and percent body weight in juvenile sturgeon and is not comparable to the
reproductive endpoints in all other studies used for criteria development. Although we realize that EPA
is concerned with protecting threatened or endangered species such as sturgeon, technically
indefensible data sets cannot be used for this purpose.

2.1.3 Bluegill (Lepomis macrochirus)

In the 2015 draft Se criteria document, EPA utilized three Bluegill studies in the derivation of the tissue-
based criteria: Doroshov et al. (1992), Coyle et al. (1993), and Hermanutz et al. (1992, 1996). While we
agree with the use of the Doroshov et al. (1992) and Coyle et al. (1993) studies, we have concerns
regarding the use of the Hermanutz et al. (1992, 1996) data.

EPA reported an egg/ovary ECio of 12.68 mg/kg for the Hermanutz et al. (1992, 1996) studies. This value
was derived through a re-analysis of the data using TRAP and combining results from the two separate
studies, Studies I and II (Hermanutz et al. 1992, 1996) — in other words, data were combined from
studies conducted over two different years, a practice rarely employed in criteria development. EPA
elected to exclude Study III and several streams from Study II because they were considered to be
"recovering" streams and were no longer receiving aqueous Se dosing. However, fish in these streams
were exposed to dietary Se throughout the study, which is the same exposure route used in other
studies considered as acceptable by EPA, such as the Linville (2006) White Sturgeon study, the Doroshov
et al. (1992) Bluegill study, and the CP&L (1997) Largemouth Bass study. It is unclear why these dietary
studies were included in the criteria development and the "recovering" streams from Studies II and III, in
which no effects were observed, were not.

In addition, we previously expressed concerns regarding the results of Studies I and II and the resulting
analysis for the following reasons:

¦ The two studies were not conducted concurrently, resulting in significant variation in all of the
measured water quality parameters between the two studies (e.g., Study I waters were 5
degrees warmer in summer; Study I had higher hardness, alkalinity, turbidity, and conductivity;

170

-------
and, dissolved oxygen varied differently between seasons and studies; Hermanutz etal. 1992,
1996).

¦ There were differences in how EPA analyzed and used the data from the two studies in
Appendix C of the draft document (EPA 2015). For Study I, EPA calculated geometric means of
two replicate Se concentrations (and used these geometric means to calculate the ECio),
whereas for Study II, EPA calculated geometric mean concentrations but used individual
replicates to calculate ECi0s.

¦ There were differences in egg and larval survival between Studies I and II (EPA 2015). In the egg
cup observations, egg percent survival to day 3 or 4 in the 10 ng/L exposure was much lower in
Study I (i.e., 28.8%) than in Study II (i.e., 57.1 — 57.7%). In the nest observations, the percent
dead larvae in the 10 ng/L exposure was much higher in Study I (i.e., 17%) than in Study II (i.e.,
0.4 — 0.5%).

¦ The ovary Se concentrations resulting from the same water exposures were different. In Study I,
the control exposure resulted in ovary Se concentrations ranging from 0.29 — 2.21 mg/kg,
whereas the control exposure in Study II resulted in ovary Se concentrations ranging from 3.72
— 3.79 mg/kg. In Study I, the 10 ng/L exposure resulted in ovary Se concentrations ranging from
13.73 — 22.85 mg/kg, whereas the 10 ng/L exposure in Study II resulted in ovary Se
concentrations ranging from 33.75 — 39.02 mg/kg. This variability between studies raises
concerns that differences in Se uptake, bioaccumulation, exposure, or other factors could have
been occurring between Studies I and II.

¦ In Study I, the geometric mean ovary Se concentrations were higher in the 10 ng/L exposure
(i.e., 17.71 mg/kg) than in the 30 ng/L exposure (i.e., 15.46 mg/kg). Related to the concerns
discussed above, this unexpected result indicates that there may have been an issue with study
conditions that resulted in inconsistencies with Se uptake, bioaccumulation, or exposure.
Interestingly, invertebrate Se concentrations in Study I were as expected (i.e., they increased
with increasing Se exposure), so it is odd that the fish did not show the same result.

Based on the above-mentioned concerns, as well as the omission of any data from Study III, we
recommend that the Hermanutz et al. (1992, 1996) data be completely removed from the database
used for criteria derivation. If only the other two chronic values for bluegill (i.e., 20.75 mg/kg [Doroshov
et al. 1992] and 24.55 mg/kg [Coyle et al. 1993]) are used, a Bluegill egg/ovary GMCV of 22.57 mg/kg
would be appropriate, and indicates that the ECio EPA derived from the Hermanutz et al. (1992, 1996)
studies would be an outlier.

The much lower ECio mentioned above is also inconsistent with field data for Bluegill reported in Finley
and Garret (2007). Fish populations in Hyco Reservoir (North Carolina) were decimated following high Se
inputs in the early 1980s. After Se inputs were curtailed, Bluegill populations recovered, and by 2003-
2005 the mean whole-body (WB) Se concentration was approximately 8 mg/kg-dw, respectively.
Assuming the EPA's WB-to-egg conversion factor (CF) of 2.13 for Bluegill results in a predicted egg Se
concentration of 17 mg/kg-dw, which falls above the Bluegill ECio of 11.36 mg/kg-dw derived from
Hermanutz et al. (1992, 1996), but below the ECio values of 20.75 and 24.55 mg/kg-dw derived from
Doroshov et al. (1992) and Coyle et al. (1993). This lends further support to use of only the latter ECio
values in calculating the SMCV for Bluegill.

2.1.4 Brown Trout (Salmo trutta)

171

-------
In the 2015 draft Se criteria document, EPA (2015) utilized Brown Trout data from Formation
Environmental (2011). During this study, a tank overflow accident occurred, resulting in the loss of
several study fish. In the 2014 draft, EPA presented two approaches for dealing with the loss of these
study organisms: (1) assume that all fry lost were dead or deformed (i.e., worst case assumption); and,
(2) assume that fry lost had the same rates of mortality and deformities as those not lost. In the 2014
draft, EPA (2014) chose to assume the worst case scenario and derived an egg/ovary ECio of 15.91
mg/kg, using the results from that scenario and the deformity-alone endpoint.

In the 2015 draft, EPA (2015) acknowledged the uncertainty with how to best address the loss of fish,
and decided to calculate an ECio for survival only during the first portion of the test, prior to the
accident. This resulted in the calculation of an ECio, of 18.09 mg/kg egg dw for larval survival.

We still believe that all appropriate data from this study should be used and that the realistic scenario
should be assumed, namely, that the lost fry had the same rates of mortality and deformity as the fry
not lost. EPA did not provide any justification for using the worst case assumption rather than the
realistic scenario. In the course of conducting standard toxicity tests in the laboratory, based on the
expertise of those who conduct them, it has been observed that dead fish actually collect at the bottom
of the aquarium not at the top, where fish would be more likely to be lost in an overflow event.
Therefore, based on what has been observed in standard toxicity tests, EPA should use the realistic
assumption, and continue to use the entire data set that provides both survival and deformity
information, rather than just using survival data from only one portion of the test.

We also believe there is no valid reason to use the deformity-alone endpoint when the combined
survival and deformities endpoint is available. In fact, this is consistent with the EPA's previous approach
in the 2013 ammonia criteria document, in which they used the combined survival and growth
endpoints, termed "biomass" (EPA 2013). Additionally, there were no difficulties in using TRAP to fit the
combined data, and these data and resulting graphs are provided in Appendix C, Figure 1 of the EPA
2015 draft Se criteria document.

In summary, we believe that: (1) data for the overflow aquaria should not be excluded; (2) there is no
reason to assume that the fish lost to the overflow were dead or deformed (similar mortality and
deformity rates would have been expected in the no-overflow aquaria which had similar egg Se
concentrations); and, (3) the most comprehensive endpoint is a combination of larval mortality and
deformities, not the truncated study data used by EPA. Following these considerations, the ECio for the
realistic scenario, as derived by the EPA, is 21.16 mg/kg for egg/ovary. We recommend this ECio as an
appropriate and protective Brown Trout chronic value.

2.1.5 Toxicity Data for Less Sensitive Species

The draft fish egg Se criterion is based on toxicity data for 10 species and 9 genera. Most of the toxicity
data are expressed as ECi0s, but due to the nature of the toxicity data, an EC24 was used for northern
pike and an LOEC was used for fathead minnow (with a footnote for the latter, noting that additional
data suggest that the LOEC is conservative). There are laboratory and field data for other fish species
that are not sufficiently robust for calculation of a threshold (e.g., ECio), but adequate for demonstrating
that these species are not among the four most sensitive, and therefore could be included to increase
"n". This is not inconsistent with the EPA's current guidelines for ambient water quality criteria (AWQC)
development (Stephan et al. 1985), which states that "If the tests were conducted properly, acute values

172

-------
reported as "greater than " values and those which are above the solubility of the test material should be
used, because rejection of such acute values would unnecessarily lower the Final Acute Value by
eliminating acute values for resistant species."

As the EPA discusses in Appendix E of EPA (2015), there are several cyprinid species with sustainable
reproducing populations that have WB Se concentrations greater than the draft WB Se criterion of 8.0
mg/kg-dw. The information (included in Appendix E) could be used to increase the number of genera in
the species sensitivity distribution, including Central Stonerollers (Campostoma anomalum), Red Shiner
(Cyprinella lutrensis), and Sand Shiner (Notropis stramineus). These species were able to maintain
successfully-reproducing populations (similar to their populations at low Se locations) with WB tissue
concentrations over 40 mg/kg dw in a study conducted near Pueblo, Colorado (EPA 2015; Table E-18 and
GEI 2007). Because these species are insensitive to elevated Se, they would not be in the top four most
sensitive genera; however, they can be included in the total number of species. Inclusion of these fish
species would increase the "n" from 15 to 18 (implications of changing "n" on the criterion is provided in
Section 2.2).

2.2 Derivation of Tissue Criterion Elements
2.2.1 Updates to the Egg/Ovary Criterion

Implementing the data usage modifications discussed above (i.e., Section 2.1) results in changes to the
criteria calculations. The following is a summary of our recommended modifications:

• Eliminate the White Sturgeon data. Although a SMCV of 17.8 mg/kg can be calculated as the
geometric mean of the four ECi0s calculated by EPA, any such calculation is not technically
defensible given the few data points and only one partial response.

• Update the Bluegill GMCV to 22.57 mg/kg

o Result of excluding the data from the two Hermanutz et al. (1992, 1996) studies

• Update the Brown Trout SMCV to 21.16 mg/kg

o Result of using all data (i.e., both survival and deformity) with the realistic assumption
for addressing the issue of the fry lost during the accident that occurred during the
study

• Include three additional fish species to increase "n" based on observed insensitivity to Se, similar
to inclusion by EPA of crustacean "placeholder taxa" to fulfill the 8-family rule and increase "n".

The order and chronic values for the top four most sensitive species change as a result of the above
(Table 1). We recommend that the EPA consider these scientifically-defensible data for relevant species.

Fish Tissue-based Toxicity Data

• Given the very large uncertainties in the concentration-response data for White Sturgeon, the
EPA should not use these data.

• We also strongly recommend that the EPA exclude the Bluegill ECio of 11.36 mg/kg-dw derived
from the data presented in Hermanutz et al. (1992, 1996), which is technically flawed,
inconsistent with the laboratory results of Doroshov et al. (1992) and Coyle et al. (1993), and
inconsistent with the field data presented in Finley and Garrett (2007).

173

-------
• Finally, we recommend that the EPA reconsider the ECio for Brown Trout, given that the most
comprehensive endpoint is combined larval mortality and deformities, which results in an ECio
of 21.16 mg/kg-dw based on the realistic assumption for addressing fry lost during the study.

Original letter contains Table 1 - Modified version of Table 3.2 of EPA (2015) following adjustments to
chronic values based on review and analysis of study data. See original letter.

EPA-HQ-OW-2004-0019-0391-A1; North American Metals Council (NAMC); Posted
10/14/2015

The available, technically defensible data clearly indicate that the egg/ovary criterion should be at least
20 mg/kg dw and the whole body/muscle criterion should be proportional. The 2015 EPA Selenium Draft
is proposing a national selenium fish egg/ovary criterion driven by two studies with sturgeon: a partial
response that is only slightly above a negligible effects endpoint, and a slight reduction in body weight.
Neither of these responses could reasonably be related to effects that might threaten populations.

EPA-HQ-0W-2004-0019-0401-A1; Colorado Wastewater Utility Council; Posted 10/15/2015

We do have some concern with several of the data decisions made by EPA in calculating the chronic
tissue criterion concentrations. We have provided recommendations on revisions to the White
Sturgeon, Bluegill and Brown Trout genus mean chronic values (GMCVs) or species mean chronic values
(SMCVs). Should our recommendations be accepted, the result would be minor, but significant, changes,
with an egg/ovary criterion to 17.9 mg/kg, the whole-body criterion to 9.5 mg/kg, and the muscle
criterion to 12.0 mg/kg. These revisions are based on our reanalysis of the studies, recalculation of
tissue-to-tissue conversion factors, or both.

1.3 Reliance on Maternal-Transfer Related Data over Juvenile Survival

We would like to point out one additional point which we strongly agree with. We support EPA's
decision to rely primarily on data from maternal-transfer studies and measures of larval survival and
deformities, rather than relying on juvenile survival data, including overwinter survival, in developing
the Se criteria. Although winter-stress may be a valid hypothesis, there are no data supporting its
occurrence in the field (Janz 2008). Additionally, if other commenters have concerns with the exclusion
of winter-stress data, we would recommend directing them to the Bluegill studies by Hermanutz et al.
(1992, 1996), which included Bluegill exposed to year-round seasonal conditions in an outdoor test
system, and thus include "winter stress" under natural conditions.

1.3.1 Bluegill

In the 2015 draft Se criteria document, EPA utilized three Bluegill studies in the derivation of the tissue-
based criteria: Doroshov et al. (1992), Coyle et al. (1993), and Hermanutz et al. (1992, 1996). While we
agree with the use of the Doroshov et al. (1992) and Coyle et al. (1993) studies, we have several
concerns regarding the use of the Hermanutz et al. (1992, 1996) data.

174

-------
over 2 different years, a data usage practice rarely used in criteria development. EPA elected to exclude
Study III and several streams from Study II because they were considered to be "recovering" streams
and were no longer receiving aqueous Se dosing. However, fish in these streams were obviously
exposed to dietary Se throughout the study, which is the same exposure route that was used in other
studies considered acceptable by EPA, such as the Linville (2006) White Sturgeon study, the Doroshev et
al. (1992) Bluegill study, and the CP&L (1997) Largemouth Bass study. It is unclear why these dietary
studies were included in the criteria development and the "recovering" streams from Studies II and III,
which showed no effects, were not.

In addition, we previously expressed several concerns about the results of Studies I and II and the
resulting analysis:

• The studies were conducted one year apart, which resulted in significant variation in all of the
measured water quality parameters between the two studies.

• There were differences in how EPA analyzed and used the data from the two studies in
Appendix C (see pages C-107 and C-108 of the 2014 draft Se criteria document).

• There were differences in egg and larvae survival between Studies I and II (see pages C-107 and
C-108 of the 2014 draft Se criteria document).

• The ovary Se concentrations resulting from the same water exposures were quite different.
These variations raise concerns that differences in Se uptake, bioaccumulation, exposure, or
other factors could have been occurring between Studies I and II.

• In Study I, geometric mean ovary Se concentrations were higher in the 10 ng/L exposure (17.71
mg/kg) than the 30 ng/L exposure (15.46 mg/kg). Related to the concerns discussed above, this
unexpected result indicates there may have been an issue with study conditions that caused
inconsistencies with Se uptake, bioaccumulation, or exposure.

Based on these concerns, as well as the omission of any data from Study III, we recommend the
Hermanutz et al. (1992, 1996) data could be completely removed from the database used for criteria
derivation altogether. If only the other two chronic values for Bluegill (20.75 mg/kg [Doroshov et al.
1992] and 24.55 mg/kg [Coyle et al. 1993]) are used, a Bluegill egg/ovary GMCV of 22.57 mg/kg would
be appropriate, and show that the value EPA derived from the Hermanutz et al. studies would be an
outlier.

1.3.1.1 Use of Only Egg Data from Bluegill Studies with Egg and Ovary Data Available

Both Doroshov et al. (1992) and Coyle et al. (1993) present tissue data for both ovaries and eggs. It is
unclear why EPA only used egg data from these studies and did not include both egg and ovary tissues
as an average value. On page 78 of the 2014 draft Se criteria document, EPA suggests that it is
appropriate to use the average of egg and ovary concentrations when both tissues are reported when
deriving conversion factors (CFs). We recommend EPA include data for both ovaries and eggs, or at the
very least provide details about its preference for egg versus ovary tissues, as this information will be
important to make sure future toxicity testing and criteria implementation include measurement of Se
in appropriate tissues.

1.3.2 Brown Trout

175

-------
organisms: 1) assume that all fry lost were dead or deformed ("worst case" assumption) and 2) assume
that fry lost had the same rates of mortality and deformities as those not lost ("optimistic" assumption).
In the 2014 draft, EPA chose to assume the "worst case" scenario and derived an egg/ovary ECio of
15.91 mg/kg using the results from that scenario and the deformity alone endpoint. In the 2015 draft
EPA acknowledged the uncertainty with how to best address the loss of fish, and decided to calculate an
ECio for survival only during the first portion of the test, prior to the accident. This resulted in calculation
of an ECio of 18.09 mg/kg egg dw for larval survival.

We still believe that all appropriate data from this study should be used and that the "optimistic"
scenario should be considered, as it reflects what was observed in the remaining population (i.e., the
fish not lost to overflow). EPA did not provide any reasons for why use of all data with the "optimistic"
assumption is not a valid approach. As stated previously, throughout GEI's 25 years of conducting
toxicity tests in our laboratory, we have observed that dead fish actually collect at the bottom of the
aquarium not at the top, where fish would be more likely to be lost in an overflow event. Therefore,
based on what has been observed in standard toxicity tests EPA should use their "optimistic" (i.e., what
we consider to be realistic) assumption, and continue to use the entire data set that provides both
survival and deformity information, rather than just using survival data from one portion of the test.

We also believe there is no valid reason to use the deformities endpoint alone when the combined
survival and deformities endpoint is available. In fact, this is more in line with the EPA's previous
approach in the 1999 ammonia criteria document where they used the combined survival and growth
endpoint, termed "biomass". Additionally, there were no difficulties in using TRAP to fit the combined
data, and these data and resulting graphs are provided in Appendix C, Figure 1 of the EPA 2015 draft Se
criteria document. When these combined data are used, the ECio for the "optimistic" assumption is
21.16 mg/kg for egg/ovary, which we recommend as an appropriate and protective Brown Trout chronic
value.

1.3.3 White Sturgeon

In the 2015 draft criteria document EPA included data from the Linville (2006) White Sturgeon study.
These data were not included in the 2014 draft, but White Sturgeon is now the most sensitive species in
the database. This study was a dietary exposure in which adult female sturgeon were fed a Se spiked
diet, and effects on larvae were measured. Larval effects were observed for edema and deformities.

EPA calculated an ECio for total deformities (edema + skeletal) of 16.3 mg/kg using TRAP. This value was
calculated based on a partial dose response. When data from this study are analyzed, TRAP warns that
data should only be used for "exploratory purposes." Due to the partial response in the data, the
calculation is highly dependent on the initial guess used for the slope in TRAP. In fact, in Appendix C (EPA
2015), EPA shows how choice of initial slope can affect these calculations, with ECio values ranging from
16.3 mg/kg to 19.1 mg/kg when using different slopes. EPA selected the most conservative value for use
in criteria development, even while acknowledging there was no scientific reason one value was more
valid than another (and, in fact, all calculated values were statistically identical).

176

-------
Minnow data from the chronic dataset because of an "insufficient response". EPAs reasoning for using
the White Sturgeon data that also had only a partial response is because there are data that suggest that
the federally-listed Green Sturgeon is also sensitive to selenium. The Green Sturgeon study consisted of
effects on survival and percent body weight in juvenile sturgeon and is not comparable to the
reproductive endpoints in all other studies used for criteria development. However, we realize that EPA
is concerned with protecting any threatened or endangered species, and the criteria document states
that the White Sturgeon serves as a surrogate for other sturgeon as well.

Although the White Sturgeon data are somewhat questionable, they are important data due to the
threatened or endangered listings of other species of sturgeon. However, we do recommend revisions
to the ECio value used for criteria calculations. As stated previously, the ECio is based on only a partial
response, therefore when the threshold sigmoid nonlinear regression model in TRAP is used, several
curves may be fit by varying the slope used in the calculation. In Appendix C of the 2015 EPA criteria
document, four ECio values have been calculated, all with the same goodness of fit. While EPA chose the
most conservative value for use in criteria calculations, we recommend use of the geometric mean of
the four ECi0s, as they are all equally valid. This results in an egg/ovary ECio of 17.8 mg/kg dw for White
Sturgeon (Table 2).

4. Conclusions

Overall, the 2015 draft Se criteria document, including the tissue-based criteria approach, is a significant
improvement over the 2004 draft criteria document. We strongly support EPA's decision to develop
tissue-based Se criteria that are toxicologically and ecologically relevant. While we support the overall
approach and core of the 2015 draft Se criteria document, we have several recommended revisions and
considerations that, if considered by EPA, could significantly improve the document and resulting Se
criteria.

5.1 Reevaluation of Acceptable Studies

We reviewed all of the reproductive toxicity study data deemed acceptable by EPA in the 2015 draft Se
criteria document. Data were presented for ten fish genera. Overall, we concur with most of the data
usage decisions made by EPA, but have comments and suggestions on some of the data that were used
to develop the egg/ovary chronic criterion (and subsequently, the whole-body and muscle criteria).

Specifically, we had comments on usage of specific White Sturgeon, Bluegill and Brown Trout data in the
criteria calculations.

• White Sturgeon - We recommend using the geometric mean of all the ECi0s calculated by EPA,
as the goodness-of-fit is equal for all curves, therefore all ECi0s are equally appropriate. The
recommended egg/ovary ECio value, which is equivalent to the SMCV, is 17.8 mg/kg.

• Bluegill - We have concerns about the data used from the Hermanutz et al. (1992, 1996)
studies. We recommend rejecting data from all Hermanutz studies. Using only data from the
other two studies, the updated Bluegill egg/ovary GMCV is 22.57 mg/kg, with an accompanying
whole-body value of 10.58 mg/kg using regression-based CF.

• Brown Trout - We recommend using the Brown Trout ECio calculated under the
"optimistic"/i"ealistic assumption for dealing with the lab accident that resulted in loss of study
organisms in the Formation Environmental (2011) study. The recommended ECio value, which is

177

-------
equivalent to the SMCV, is 21.16 mg/kg, with an accompanying whole-body value of 14.59
mg/kg based on updated CFs.

• We developed regression-based CFs for translating between egg/ovary and whole-body. When
the regression relationship is strong, we recommend using regression-based CFs instead of
median ratio-based CFs. Incorporation of these suggested changes results in updated
egg/ovary, whole-body, and muscle chronic criteria that are even more scientifically defensible
and consistent with EPA's other data-usage decisions (Table 13).

EPA-HQ-0W-2004-0019-0400-A1; The National Mining Association (NMA); Posted 10/15/2015

We do have some concerns with several of the data decisions made by EPA in calculating the chronic
tissue criterion concentrations. We have provided recommendations on revisions to the White
Sturgeon, Bluegill and Brown Trout genus mean chronic values (GMCVs) or species mean chronic values
(SMCVs). Should our recommendations be accepted, the result would be small, but significant, changes,
with an egg/ovary criterion to 17.9 mg/kg, the whole-body criterion to 9.5 mg/kg, and the muscle
criterion to 12.0 mg/kg. These revisions are based on our reanalysis of the studies, recalculation of
conversion factors, or both.

1.6 Reliance on Maternal Transfer Rather Than Juvenile Survival Data

We would like to point out one additional point which we strongly agree with. We support EPA's
decision to not rely on juvenile survival data, including overwinter survival, in developing the Se criteria.
Although winter-stress may be a valid hypothesis, there are no data supporting its occurrence in the
field (Janz 2008). Additionally, if other commenters have concerns with the exclusion of winter-stress
data, we would recommend directing them to the Bluegill studies by Hermanutz et al. (1992, 1996),
which included Bluegill exposed to year-round seasonal conditions in an outdoor test system, and thus
include "winter stress" under natural conditions.

1.6.1 Bluegill

In the 2015 draft Se criteria document, EPA utilized three Bluegill studies in the derivation of the tissue-
based criteria: Doroshov et al. (1992), Coyle et al. (1993), and Hermanutz et al. (1992, 1996). While we
agree with the use of the Doroshov et al. (1992) and Coyle et al. (1993) studies, we have several
concerns regarding the use of the Hermanutz et al. (1992, 1996) data.

EPA reported an egg/ovary ECio of 12.68 mg/kg for the Hermanutz et al. (1992, 1996) studies. This value
was derived through a reanalysis of the data using TRAP and combining results from Studies I and II
(Hermanutz et al. 1992, 1996) - in other words, they actually combined data from studies conducted
over two different years, a data usage practice rarely used in criteria development. EPA elected to
exclude Study III and several streams from Study II because they were considered to be "recovering"
streams and were no longer receiving aqueous Se dosing. However, fish in these streams were obviously
exposed to dietary Se throughout the study, which is the same exposure route that was used in other
studies considered acceptable by EPA, such as the Linville (2006) White Sturgeon study, the Doroshev et
al. (1992) Bluegill study, and the CP&L (1997) Largemouth Bass study. It is unclear why these dietary
studies were included in the criteria development and the "recovering" streams from Studies II and III,
which showed no effects, were not.

178

-------
In addition, we previously expressed several concerns about the results of Studies I and II and the
resulting analysis:

• The studies were conducted 1 year apart, which resulted in significant variation in all of the
measured water quality parameters between the two studies.

• There were differences in how EPA analyzed and used the data from the two studies in
Appendix C (see pages C-107 and C-108 of the 2014 draft Se criteria document).

• There were differences in egg and larvae survival between Studies I and II (see pages C-107 and
C-108 of the 2014 draft Se criteria document).

1.6.1.1 Use of Only Egg Data from Bluegill Studies with Egg and Ovary Data Available

Both Doroshov et al. (1992) and Coyle et al. (1993) present tissue data for both ovaries and eggs. It is
unclear why EPA only used egg data from these studies and did not include both egg and ovary tissues
as an average value. On page 78 of the 2014 draft Se criteria document, EPA suggests that it is
appropriate to use the average of egg and ovary concentrations when both tissues are reported when
deriving CFs. We recommend EPA include data for both ovaries and eggs, or at the very least provide
details about its preference for egg versus ovary tissues, as this information will be important to make
sure future toxicity testing and criteria implementation include measurement of Se in appropriate
tissues.

1.6.2 Brown Trout

In the 2015 draft Se criteria document, EPA utilized Brown Trout data from Formation Environmental
(2011). During this study, a tank overflow accident occurred which resulted in the loss of several study
fish. In the 2014 draft, EPA presented two approaches for dealing with this loss of these study
organisms: 1) assume that all fry lost were dead or deformed ("worst case" assumption) and 2) assume
that fry lost had the same rates of mortality and deformities as those not lost ("optimistic" assumption).
In the 2014 draft, EPA chose to assume the "worst case" scenario and derived an egg/ovary ECio of
15.91 mg/kg using the results from that scenario and the deformity alone endpoint. In the 2015 draft
EPA acknowledged the uncertainty with how to best address the loss of fish, and decided to calculate an
ECio for survival only during the first portion of the test, prior to the accident. This resulted in calculation
of an ECio of 18.09 mg/kg egg dw for larval survival.

179

-------
We still believe that all appropriate data from this study should be used and that the "optimistic"
scenario should be considered, as it reflects what was observed in the remaining population (i.e., the
fish not lost to overflow). EPA did not provide any reasons for why use of all data with the "optimistic"
assumption is not a valid approach. As stated previously, throughout GEI's 25 years of conducting
toxicity tests in our laboratory, we have observed that dead fish actually collect at the bottom of the
aquarium not at the top, where fish would be more likely to be lost in an overflow event. Therefore,
based on what has been observed in standard toxicity tests EPA should use their "optimistic" (i.e., what
we consider to be realistic) assumption, and continue to use the entire data set that provides both
survival and deformity information, rather than just using survival data from one portion of the test.

We also believe there is no valid reason to use the deformities endpoint alone when the combined
survival and deformities endpoint is available. In fact, this is more in line with the EPA's previous
approach in the 1999 ammonia criteria document where they used the combined survival and growth
endpoint, termed "biomass." Additionally, there were no difficulties in using TRAP to fit the combined
data, and these data and resulting graphs are provided in Appendix C, Figure 1 of the EPA 2015 draft Se
criteria document. When these combined data are used, the ECio for the "optimistic" assumption is
21.16 mg/kg for egg/ovary, which we recommend as an appropriate and protective Brown Trout chronic
value.

1.6.3 White Sturgeon

Generally, use of an ECio based on a partial response is not appropriate for national criteria
development, especially when it results in a value becoming the "most sensitive" and thereby driving
the final criterion. In fact, on page 33 of the 2015 draft criteria document EPA states that "an ECio based
on only one partial response would not ordinarily be included in the data set." A similar partial response
was observed in the Fathead Minnow data from the GEI 2008 study; EPA excluded these Fathead
Minnow data from the chronic dataset because of an "insufficient response." EPA's reasoning for using
the White Sturgeon data that had only a partial response is that there are data that suggest that the
federally-listed Green Sturgeon is also sensitive to selenium. The Green Sturgeon study consisted of
effects on survival and percent body weight in juvenile sturgeon and is not comparable to the
reproductive endpoints in all other studies used for criteria development. However, we realize that EPA
is concerned with protecting any threatened or endangered species, and the criteria document states
that the White Sturgeon serves as a surrogate for other sturgeon as well.

180

-------
Although the White Sturgeon data are somewhat questionable, they are important data due to the
threatened or endangered listings of other species of sturgeon. However, we do recommend revisions
to the ECio value used for criteria calculations. As stated previously, the ECio is based on only a partial
response, therefore when the threshold sigmoid nonlinear regression model in TRAP is used, several
curves may be fit by varying the slope used in the calculation. In Appendix C of the 2015 EPA criteria
document, four ECio values have been calculated, all with the same goodness of fit. While EPA chose the
most conservative value for use in criteria calculations, we recommend use of the geometric mean of
the four ECi0s, as they are all equally valid. This results in an egg/ovary ECio of 17.8 mg/kg dw for White
Sturgeon (Table 2).

3.3 Derivation of Tissue Criterion Elements
3.3.1 Updates to Egg/Ovary Criterion

Implementing the data usage modifications discussed above (Section 3.1.1) results in changes to the
criteria calculations. The following is a summary of our recommended modifications:

• Update the White Sturgeon SMCV to 17.8 mg/kg

• Result of calculating the geometric mean of the four EClOs calculated by EPA

• Update the Bluegill GMCV to 22.57 mg/kg

• Result of excluding the Hermanutz et al. (1992, 1996) studies

• Update the Brown Trout SMCV to 21.16 mg/kg

• Result of using all data (survival and deformity) with the "optimistic"/i"ealistic assumption for
addressing the issue of the fry lost during the study

• Include three additional fish species to increase N based on observed insensitivity to Se
The order and chronic values for the top four most sensitive species change as a result of the
modifications to the White Sturgeon, Bluegill and Brown Trout GMCVs (Table 1). In addition, the sample
size increases from N=15 to N=18. Using the data presented in Table 3, an updated egg/ovary criterion
of 17.9 mg/kg can be derived using EPA criteria calculation methodology (Stephan et al. 1985; Table 2).
We recommend EPA considers this recalculated criterion, as it is based on sound data for relevant
species.

Original letter contains Table 1 - Modified version of Table 3.2 of the 2015 draft Se criteria document
following adjustments to chronic values for White Sturgeon, Bluegill, and Brown Trout based on review
and analysis of study data. See original letter.

Original letter contains Table 2 - Updated calculation of egg/ovary fish tissue-based Se criterion based
on modifications in Table 2 (N= 14 genera, R = sensitivity rank in database). See original letter.

4. Conclusions

5.1 Reevaluation of Acceptable Studies

181

-------
We reviewed all of the reproductive toxicity study data deemed acceptable by EPA in the 2015 draft Se
criteria document. Data were presented for ten fish genera. Overall, we concur with most of the data
usage decisions made by EPA, but have comments and suggestions on some of the data that were used
to develop the egg/ovary chronic criterion (and subsequently, the whole-body and muscle criteria).

Specifically, we had comments on usage of specific White Sturgeon, Bluegill and Brown Trout data in the
criteria calculations.

Incorporation of these suggested changes results in updated egg/ovary, whole-body, and muscle chronic
criteria that are even more scientifically defensible and consistent with EPA's other data-usage decisions
(Table 13).

182

-------
1 ii 11ii 11' ii iim I. ii ' ii 11ii 11 ii 11 > ii i II ¦. iriiviiii i II ;-Ovairy Criterion

Ell

3.3 Summary

This section contains comments about EPA's method of deriving the egg-ovary criterion element.
Specifically, the derivation of Species Mean Egg-Ovary Chronic Values (SMCVs), Genus Mean Egg-Ovary
Chronic Values (GMCVs) and the Final Egg-Ovary Chronic Value (FCV). Included are comments
concerning the identification of the four most sensitive Egg-Ovary GMCVs and the use of 5th percentile
projection to select the FCV. Some commenters requested clarification on how the FCV was calculated
using the GMCVs.

3.3 Response to comments

3.3 Responses concerning the derivation of the egg ovary criterion element value
Egg-ovary Equivalence

Regarding the comment on egg-ovary equivalence, EPA notes that the text in the criterion document
actually states: "In this document, concentrations of selenium in ovaries are considered equivalent to
concentrations of selenium in eggs because most studies measured selenium in the ovaries prior to
spawning." The last part of the sentence is the relevant part as to why EPA considers theses tissues
equivalent in the document.

From Section 6.1.8. Researchers often report concentrations of selenium in fish eggs or ovaries (e.g.,
Formation Environmental 2011; Formation Environmental 2012; Holm et al. 2005; Osmundson et al.
2007). Osmundson et al. (2007) found reduced levels of selenium in ovaries after spawning, presumably
due to the loss of selenium through spawning and release of eggs with relatively high concentrations of
selenium. Of the 14 chronic values determined from the maternal transfer reproductive studies, 11
values represent selenium measured in eggs. Three values represent selenium measured in the ovaries:
Schultz and Hermanutz (1990), Hermanutz et al. (1992, 1996) and Carolina Power & Light (1997).
However, information in two of these studies indicates that the concentrations of selenium in the
ovaries were similar to concentrations in eggs. Schultz and Hermanutz (1990) measured selenium in
fathead minnow ovaries at the end of the study from fish that presumably had spawned. The authors
found the concentrations of selenium in the ovaries and embryos of the fathead minnows exposed to
the same treatments to be similar. Hermanutz et al. (1992, 1996) sampled adult female bluegill just prior
to spawning and at the end of the test (post spawning) and found no decreases in the concentration of
selenium in the post-spawned fish. In the Carolina Power & Light (1997) study, the third study, selenium
in ovaries of largemouth bass (Carolina Power & Light 1997) was measured from fish sampled just after
spawning. No comparison to prespawning fish or selenium in eggs can be made for the largemouth bass
study, however, the EC10 of 26.3 mg Se/kg ovary dw was mid-range in the SSD indicating this test was
not overly conservative due to lower selenium measurements in post spawning ovaries. Based on the
indications that the selenium concentrations in the ovaries were similar to that in eggs in the Schultz and
Hermanutz (1990) and Hermanutz et al (1992, 1996) study observations stated above, egg selenium and
ovary selenium were considered equal for the toxicity data set. Any potential error resulting from this

183

-------
assumption would be conservative since the effect of spawning only lowers the selenium concentration
in the ovary. EPA recognizes selenium ovary concentrations may vary in field collected samples due to
fish reproductive cycles and will address such concerns in the implementation information. Although
there is a strong correlation between predicted and observed egg-ovary concentration values, Figure 16
does show more data points above the y = x (observed egg-ovary concentration vs predicted egg- ovary
concentration) line at low selenium concentrations. This result suggests the model underestimates
bioaccumulation at low selenium concentrations. However, within the range of concentrations near the
egg-ovary criterion element value, the relationship between predicted and observed selenium
concentrations are evenly dispersed around the y = x line. Thus the model is unlikely to result in biased
estimates near the egg-ovary criterion concentration, the focus of the criteria development.

GMCVs

Regarding the comment on the logic for setting the number of GMCVs to 14, the goal of aquatic life
criteria is to ensure protection of populations of species representing the entire aquatic community, and
not just fish, as described in the methodology for criteria development (Stephan et al. 1985). Estimated
egg-ovary concentrations were originally calculated in 2014 for invertebrates as a point of reference to
show that they were typically less sensitive then the fish represented in the SSD, and would therefore be
protected by a tissue criterion based on fish. In the 2015 draft and 2016 final document, fish and
invertebrate SMCVs and GMCVs are listed in separate tables, and whole body and estimated egg- ovary
values for invertebrates have been removed from the two tables showing the distribution of GMCVs
used in the Sensitivity Distribution for egg-ovary and whole body criteria, respectively.

A separate Table (5b) was added showing the measured whole body SMCVs and GMCVs for the three
invertebrate taxa with relevant test data, as well as what the predicted whole body fish tissue
concentrations would be after multiplying the measured invertebrate SMCV and GMCV by the median
tropic level 2 to trophic level 3 trophic transfer function of 1.27. The results described in sections 3.1
demonstrate that invertebrates are less sensitive to selenium in fish, and while they are used implicitly
in the SSD to fulfill taxonomic minimum data requirements, they are not included with fish values. These
studies are taken into account as part of the total "N" in the criterion calculations, in accordance with
the standard methodology for criteria development (EPA 2008).

Criteria are developed to protect the entire aquatic community, such that the available data should
reflect the sensitivity range for various components (e.g., fish, invertebrates) of the aquatic system.
Towards this end, invertebrate sensitivity to selenium was evaluated in terms of both measured whole
body concentrations as well as in terms of what the whole body tissue concentration of a representative
fish would be were it to consume each invertebrate with a whole body concentration at the SMCV and
GMCV, by multiplying each invertebrate GMCV by 1.27, the median TTF for all fish species. When
evaluated with or without the trophic level biomagnification, the available data demonstrate that
compared to fish, invertebrates are not as sensitive to selenium, and do not comprise any of the four
lowest GMCVs. They are used to fulfill the taxonomic minimum data requirements, and are counted in
the total number of genera ("N") in the calculations. We note the numeric impact of increasing "N"
though inclusion of the invertebrate data in the "N" for the criterion calculation is minimal, reflecting an
approximately 3% change in the egg-ovary criterion element value.

184

-------
Regarding the comment on the number of GMCVs, EPA agrees that the invertebrate and the Gambusia
data should not be represented in the egg-ovary sensitivity distribution, and they have been removed
from that distribution.

3.3 Specific comments

EPA-HQ-OW-2004-0019-0390-A1; West Virginia Coal Association (WVCA); Posted 10/14/15
Calculation of Fish Body Burden Elements

The 1985 Guidelines provide two options for preparation of a numeric water quality criterion. Both
processes begin with the compilation of a database of all available, qualifying aquatic life toxicity studies
meeting the requirements of the 1985 Guidelines. In its 2014 peer review draft selenium criteria, EPA
published an updated selenium database and received extensive comments regarding its data decisions.
While EPA made some adjustments based on the comments received, EPA's database utilized for
calculation of the body burden elements remains crippled by improper data decisions.

The draft body burden numbers are based on Genus Mean Chronic Values (GMCVs) calculated from the
database, and the inclusion or exclusion of a specific study can materially affect the calculated selenium
criteria. The egg-ovary element is calculated directly from available toxicity studies, whereas the whole
body number is calculated from conversion of the egg-ovary concentrations for each genus to whole-
body concentrations. The chronic criterion is calculated through a multi-step process following a
statistical methodology based on qualifying toxicity tests for all available aquatic life species. However,
the four lowest egg-ovary GMCVs largely determine the resultant numeric chronic criterion elements for
fish tissue.

EPA-HQ-OW-2004-0019-0391-A1; North American Metals Council (NAMC); Posted
10/14/2015

2.2.6 Egg/Ovary Selenium Criterion Should Be at Least 20 mg/kg dw

As outlined above, the three studies that drive the egg/ovary selenium criterion below 20 mg/kg are
technically flawed as is the historic winter stress syndrome that also limits the whole body criterion to
8.0 mg/kg whole body or 11.3 mg/kg muscle.

Eight of the ten fish species for which the EPA developed SMCVs/GMCVs are the same fish species for
which DeForest et al. determined predicted no effect concentrations (PNEC).47 The values are all >20
mg/kg dw other than the EPA's brown trout and bluegill data (Table 1). As previously noted, those data
should also be >20 mg/kg dw.

Original letter contains Table 1 - Comparison of EPA SMCVs/GMCVs with DeForest et al. (2012) PNECs.
See original letter. Table 1 contains footnote 48 in reference to "cutthroat trout."

The egg/ovary PNEC developed by DeForest et al. of 20 mg/kg dw is above the highest documented
reference concentration of 19 mg/kg dw.49 It was based on a species sensitivity distribution (SSD), was

185

-------
published in an international peer-reviewed journal, and as the authors note, "should be considered a
conservative, broadly applicable guideline."

The 2015 EPA Selenium Draft considered two fish species not considered by DeForest et al., the desert
pupfish, at 27 mg/kg dw; and, the white sturgeon at 16.27 mg/kg dw. As detailed in Section 2.2.3, the
white sturgeon study is flawed and should not have been included in the criterion derivation.
Accordingly, the available, technically defensible data clearly indicate that the egg/ovary criterion should
be at least 20 mg/kg dw and the whole body/muscle criterion should be proportional.

Field evidence also supports an egg/ovary criterion >20 mg/kg dw. For example, pages E31 to E36 in the
2015 EPA Selenium Draft document present a study of a contaminated creek (Denver, CO) that clearly
demonstrates that fathead minnows have a relatively high tolerance to selenium based on the
deformity data, which provide a fingerprint of selenium effects. That study's findings relative to
selenium, in which acute effects occurred related to other stressors, mirror data from Belews Lake,
where fathead minnow were relatively insensitive to selenium as they were one of the few fish species
left after other fish species had been lost from the lake.50 Thus, the Schultz and Hermanutz study used
by EPA to establish the proposed egg/ovary criterion was likely overly conservative because that study
did not provide a clear endpoint as it was unbounded for fathead minnows.51

EPA-HQ-OW-2004-0019-0385-A2; Kentucky Division of Water (KDOW); Posted 10/13/2015

Acceptable study data representing twelve fish species were available to calculate the SMCV (Species
Mean Chronic Value) and nine fish genera to calculate the Genus Mean Chronic Value (GMCV). The EPA
considered fourteen genera to calculate the GMCV, but because the data indicate invertebrates are
tolerant of high selenium concentrations, the invertebrate values were not included in deriving the FCV
which is determined from the four most sensitive GMCVs.

The EPA also incorporated the 1985 Guidance (Stephan, et al. 1985) recommendations which include
consideration of a commercially or recreationally important warm water species when determining
appropriate data to calculate the GMCVs. This important consideration was also essential in developing
Kentucky's selenium criteria. Like Kentucky, the EPA concluded that the Doroshov, et al. (1992) catfish
(Ictaluridae) study contained unusable data because the mode of exposure was injection of the test
fishes, rather than through diet (cf. the inclusion of Linville, 2006)

186

-------
1 ii 11ii 11^ ii ii 1 i1 ii ! 1 ii nil in ii ii * II ¦> riving V, II i ¦!< B I' hi ¦ > l ll n• lie

Criterion Element Values
3.4 Summary

This section focuses on comments about EPA's method of deriving the whole body criterion and muscle
element. It includes comments concerning using egg-ovary and the measured egg-ovary/whole body or
measured egg-ovary/ muscle ratios for calculating the whole body and muscle chronic values, the
derivation of a generic muscle/whole body ratio, and the use of a taxonomic classification method to
estimate conversion factors. Also presented are comments concerning derivation of a final whole body
and muscle chronic values, including the identification of the four most sensitive whole body and muscle
GMCVs and the use of the 5th percentile projection to select the final whole body and muscle criterion
values. Some commenters recommended that regression-based conversion factors should be used
instead of the median ratio-based approach. Also some commenters did not agree with the concept of
generic muscle/whole body information, citing studies that illustrate variability associated with the
spawning cycle, seasonal dietary changes, etc.

3.4 Response to comments

The majority of the data for the egg-ovary to whole body selenium relationship analysis came from
Osmundson et al (2007) who did have egg data and whole body selenium concentration data from the
same fish. The whole body selenium concentration was calculated by adding back the egg selenium that
was removed for analysis. Osmundson et al (2007) had 9 of the 10 species in EPA's data set for this
analysis. Coyle et al (1993) also added back egg selenium for the whole body same fish comparison.
Formation (2011) and Doroshov et al. did not specify how the whole body selenium concentration was
determined. Hermanutz (1996) and Hardy (2005) apparently measured whole body and egg selenium
concentrations in different fish with the same exposure. EPA has added clarifying discussion to the
section discussing fish tissue relationships.

Regarding the definition of a whole body tissue sample, the entire fish (carcass and visceral tissue) is
homogenized, and then a sample of the homogenized tissue is collected and analyzed for selenium

3.4 Description of refined approach to deriving Conversion Factors (CFs), for conversions to
whole body or muscle selenium concentrations

In the 2016 final document, EPA retained the methodology incorporated into the 2015 draft using
species-specific data or most-taxonomically-proximate muscle to whole body, and egg-ovary to muscle,
conversion factors to calculate CF values when matched pairs of selenium measurements in eggs and/or
ovaries and whole body tissue or muscle were not available.

Species-specific data were used preferentially over other available data. When species-specific data
were not available, EPA estimated CFs using sequentially higher taxonomic classifications (i.e., species,
genus, family, order, then class) until one or more taxa for which a calculated CF value was available
matched the taxon being considered. If the lowest matching taxon was common to more than one
species with CF value available, EPA used the median CF from the matching species. EPA has decided to

187

-------
retain the use of the median rather than a higher percentile as the general range of CFs is relatively
narrow with the exception of the mountain whitefish, so use of a higher percentile threshold would
allow the CF to be unduly influenced by this species. (See response below, and 2016 final document,
Section 3)

3.4 EPAs Response to Conversion Factor Comments

EPA noted that both GEI and NAMC commented that conversion factors (CFs) derived using a median
were inappropriate; GEI comments did not provide specific reasons for why medians were
inappropriate, but rather, that regressions would be more appropriate, whereas NAMC noted that "in
many cases, assumptions of linear regression were violated, and thus the results were unreliable." No
specific details were provided by the commenters to describe how this conclusion was reached.

Given the commenters concern over the statistical basis of the derivation of the CFs, EPA examined the
issue thoroughly.

First, EPA incorporated comments regarding the use of directly calculated muscle or whole body tissue
selenium concentrations rather than an estimated value using conversion factors. For whole body, this
resulted in the use of empirically-derived chronic toxicity values for bluegill, and brown trout. For muscle
this resulted in the use of empirically derived chronic values for bluegill, cutthroat trout, white sturgeon
and northern pike. These empirically derived whole body or muscle toxicity values were used
preferentially over CF-derived chronic values.

Second, EPA evaluated an approach using ordinary least squares (OLS) regression. For each species 2
regressions were performed: standard OLS regression (xy OLS), and then a second regression with the
axes switched (yx OLS). EPA noted that OLS regression only considers vertical fitting error resulting in
uncertainties in predictions depend on which tissue is assigned to which axis. Given the uncertainty
associated with OLS, EPA then used total least squares regression (TLS) in the derivation of CFs. Unlike
OLS, which only considers vertical fitting error, TLS considers both vertical and horizontal fitting error.
CFs derived using TLS typically fell between the values derived using xy OLS or yxOLS.

In general, the median and the TLS predicted CFs will be similar under either of the following conditions:
(a) log-log slope near 1.0, or (b) criterion near the middle of the observed data range of tissue
concentrations for the species, irrespective of slope or R-square. They will be distant from each other
when both of the following conditions simultaneously occur: (c) log-log slope distant from 1.0, and (d)
criterion distant from the center of the data range, again irrespective of R-squared.

Regarding the median, EPA applied statistical criteria to the available data plots to determine whether
the relationship is strong enough to obtain a median. The commenter's statement that a ratio is
analogous to a constrained regression is incorrect. This is because a median is a non-parametric
measure of central tendency, and is not subject to assumptions of linearity. It is in not analogous to a
constrained regression, except for the fact that it is a single parameter, and does not have a y-intercept
term.

Based on EPA's in-depth analyses as a result of public comments, it was determined that all 4
approaches yielded CFs that were similar. EPA decided to retain the use of the median to derive CFs in

188

-------
the criteria document for several reasons: because it is a non-parametric measure of central tendency
that does not required an assumption of linearity; it is direct and simple analysis that can easily be
verified and replicated with simple spreadsheet tools, and the more complex analyses such as TLS did
not yield results that were different or superior. A chart comparing the CFs derived by each method is
shown in Appendix N.

Regarding variability in egg-ovary to whole body CFs, in the EPA dataset, variability in egg-ovary to
whole body CFs was relatively low for the majority of species, ranging from 1.38-2.44 for 16 of 17
species. The inter-species range of CFs for egg-ovary to muscle was comparable.

Regarding inter-species variability in conversion factors, while not all fish taxa have been studied, major
freshwater fish families that are phylogenetically distinct and diverse are represented, and yet there is
only roughly a two-fold variability between them. Given the inherent variability in sensitivity observed
even among species with repeated toxicity tests, 2-fold variability in CFs is a small level of uncertainty
compared to uncertainty associated with other pollutants. EPA agrees that more data for the tissue
conversion factors are always desirable. EPA is moving forward with the current database to ensure
protection of aquatic life based on the current state-of-the- science.

For those species of fish with neither sufficient data to directly calculate an egg-ovary to whole body CF,
nor data to calculate a conversion factor for egg-ovary to muscle or whole body to muscle, EPA
estimated CF following the approach described for the estimation of TTF values in the 2016 Selenium
Criterion document. In this approach, EPA sequentially considered higher taxonomic classifications until
one or more taxa for which a calculated CF value was available matched the taxon being considered, and
if the lowest matching taxon was common to more than one species with a CF value available, EPA used
the median CF from the matching species. For fish species without sufficient data to directly calculate
an egg-ovary to whole body CF but which had sufficient data to calculate a conversion factor for either
egg-ovary to muscle or whole body to muscle, EPA followed a two-stage approach based on taxonomic
similarity. If a fish species had a species- specific egg-ovary to muscle conversion factor, but no whole-
body data with which to calculate an egg to whole body CF, then available data for other species were
used to estimate a muscle- to-whole-body conversion factor for that species based on taxonomic
relatedness. The estimated muscle-to-whole-body factor would be multiplied by the directly measured
egg-ovary- to-muscle factor to estimate an egg-ovary-to-whole-body CF for that species.

Typographical errors in CF values were corrected based on comments.

3.4 Specific comments

EPA-HQ-OW-2004-0019-0387-A1; Colorado Department of Public Health & Environment;
Posted 10/14/2015

EPA used conversion factors to derive whole-body and muscle concentrations from egg/ ovary values.
Were there any studies that provided egg/ ovary data in addition to whole-body or muscle data? If so, it
seems appropriate to use actual measured values instead of translated values, which have an added
element of uncertainty.

189

-------
EPA-HQ-0W-2004-0019-0401-A1; Colorado Wastewater Utility Council; Posted 10/15/2015

3.3.2 Updates to Whole-body Criterion

In Section 3.1.5.2 of the 2015 draft Se criteria document, EPA presents its approach for deriving a whole-
body-based criterion. To develop whole-body chronic values, EPA translated the egg/ovary chronic
values used to develop the egg/ovary criterion using egg/ovary to whole-body CFs (Table 3.5 of the 2015
draft Se criteria document). These converted values were then used to calculate a whole-body criterion
of 8.0 mg/kg. EPA again used a sample size of 15, which included the three invertebrate-based values,
the two assumed crustaceans, and Gambusia.

As we previously commented, in the past, EPA has used regression-based CFs (e.g., Bluegill CF from
EPA's 2004 draft Se criteria document). However, in the 2015 draft Se criteria document, EPA developed
CFs based on the median of available matched egg/ovary and whole-body Se data. We believe a more
appropriate method would be to use regression-based egg/ovary to whole-body translators when
appropriate (i.e., when the regression relationship had an R2 value >0.70). We translated the egg/ovary
database to whole-body and derived an updated whole-body criterion (Table 3 and Table 4). For
Oncorhynchus, individual CFs were used to translate each species individually (O. mykiss and O. clarkii),
but an overall regression using data for both species was used to convert the Oncorhynchus egg/ovary
GMCV to a whole-body GMCV of 14.99 mg/kg (Table 6). In addition, as discussed in Section 4.3.1, we
corrected and updated the median-based CFs based on our review of EPA's data and addition of our
data; these updates were incorporated here. Many of these recommendations, which are based on data
EPA had in 2014 or provided in our earlier review, were not used, and with no reasons provided.

The whole-body criterion should also be adjusted as a result of our updates to the White Sturgeon,
Bluegill, and Brown Trout egg/ovary chronic values.

Incorporating our modifications to the egg/ovary to whole-body translators, updated toxicity values for
White Sturgeon, Bluegill, and Brown Trout, and increasing the sample size from 15 to 18 results in an
updated whole-body criterion of 9.55 mg/kg which can be derived using EPA criteria calculation
methodology (Stephan et al. 1985; Table 7). We recommend EPA considers this recalculated criterion
with regression-based CFs in place of median-based conversion factors when possible, as it is based on
sound data for relevant species.

Original letter contains Table 3 - Modified version of Table 3.5 of the 2015 draft Se criteria document
following adjustments to conversion factors and chronic values for White Sturgeon, Bluegill, and Brown
Trout. See original letter.

Original letter contains Table 4 - Updated calculation of whole-body fish tissue-based Se criterion based
on modifications, including use of EO/WB regressions (N = 18 genera, R = sensitivity rank in database).
See original letter.

3.3.3 Updates to Muscle Criterion

In Section 3.1.5.3 of the 2015 draft Se criteria document, after presenting its approach for developing
the egg/ovary criterion, EPA presents its approach for deriving a muscle-based criterion. To develop
muscle chronic values, EPA translated the egg/ovary chronic values used to develop the egg/ovary
criterion using egg/ovary to muscle CFs (Table 3.7 of the 2015 draft Se criteria document). These

190

-------
converted values were then used to calculate a muscle-based criterion of 11.3 mg/kg. EPA again used a
sample size of 15, which included the three invertebrate-based values, the two crustaceans and
Gambusia.

As a result of our recommended updates to the White Sturgeon, Bluegill, and Brown Trout egg/ovary
chronic values, the muscle criterion will also need to be adjusted.

Incorporating our suggested modifications results in several changes in the top four most sensitive
species of the muscle-based toxicity database (Table 5). In addition, the sample size increases from N=15
to N=18 as a result of the additional fish species discussed previously. Using the data presented in Table
6 an updated muscle criterion of 11.96 mg/kg can be derived using EPA criteria calculation methodology
(Stephan et al. 1985). We recommend EPA considers this recalculated criterion, as it is based on sound
data for relevant species.

Original letter contains Table 5 - Modified version of Table 3.7 of the 2015 draft Se criteria document
following adjustments to chronic values for White Sturgeon, Fathead Minnow, Bluegill, and Brown Trout
based on review and analysis of study data. See original letter.

Original letter contains Table 6 - Updated calculation of muscle fish tissue-based Se criterion based on
modifications to Table 8 (N = 18 genera, R = sensitivity rank in database). See original letter.

4.3 Additional Discussion of Tissue Criterion Elements

4.3.1 Conversion Factors

As part of our previous review, we conducted an evaluation of the CFs developed by EPA in the 2014
draft Se criteria document, we reviewed all of the data used and corrected values where mistakes were
found. A detailed evaluation of this issue is also presented by NAMC-SWG (2014).

In addition to reviewing EPA's data and calculations, we also compiled matched tissue data from studies
conducted by GEI to supplement the CF database (GEI Appendix A, 2014). We also used the matched
egg/ovary, whole-body, and muscle Se data provided in the 2014 draft Se criteria document, which was
further updated by GEI as described above, and developed regression-based CFs (Table 9; Appendix A).
When the regression has a relatively high goodness of fit (i.e., when R2 is at least 0.70), we recommend
using the regression equation in place of the median (or geometric mean) ratios, as the regression
better predicts tissue concentrations, particularly at the high and low ends of the spectrum. Where the
strength of the regression is not as high (e.g., creek chub, Fathead Minnow, mountain whitefish), it may
be more appropriate to use the median or geometric mean CF to represent the central tendency of the
relationship. As shown in Section 3.2.2, we used the regression-based CFs where appropriate to
translate the updated egg/ovary criterion database for these species to whole-body for the purposes of
deriving the updated whole-body criterion. For the remaining species, we used the updated and new
median ratio-based CFs.

EPA-HQ-0W-2004-0019-0400-A1; The National Mining Association (NMA); Posted 10/15/2015

3.3.2 Updates to Whole-body Criterion

191

-------
In Section 3.1.5.2 of the 2015 draft Se criteria document, EPA presents its approach for deriving a whole-
body-based criterion. To develop whole-body chronic values, EPA translated the egg/ovary chronic
values used to develop the egg/ovary criterion using egg/ovary to whole-body CFs (Table 3.5 of the 2015
draft Se criteria document). These converted values were then used to calculate a whole-body criterion
of 8.0 mg/kg. EPA again used a sample size of 15, which included the three invertebrate-based values,
the two assumed crustaceans, and Gambusia.

As we previously commented, in the past EPA has used regression-based CFs (e.g., Bluegill CF from EPA's
2004 draft Se criteria document). However, in the 2015 draft Se criteria document, EPA developed CFs
based on the median of available matched egg/ovary and whole-body Se data. We believe a more
appropriate method would be to use regression-based egg/ovary to whole-body translators when
appropriate (i.e., when the regression relationship had an R2 value >0.70). We translated the egg/ovary
database to whole-body and derived an updated whole-body criterion (Table 3 and Table 4). For
Oncorhynchus, individual CFs were used to translate each species individually (O. mykiss and O. clarkii),
but an overall regression using data for both species was used to convert the Oncorhynchus egg/ovary
GMCV to a whole-body GMCV of 14.99 mg/kg (Table 6). In addition, as discussed in Section 4.3.1, we
corrected and updated the median-based CFs based on our review of EPA's data and addition of our
data; these updates were incorporated here. Many of these recommendations, which are based on data
EPA had in 2014 or were provided in our earlier review in 2014, were not used, with no reasons
provided.

The whole-body criterion should also be adjusted as a result of our updates to the White Sturgeon,
Bluegill, and Brown Trout egg/ovary chronic values.

Original letter contains Table 4 - Updated calculation of whole-body fish tissue-based Se criterion based
on modifications recommended by GEI Consultants, Inc., including use of EO/WB regressions (N= 14
genera, R = sensitivity rank in database). See original letter.

3.3.3 Updates to Muscle Criterion

192

-------
sample size of 15, which included the three invertebrate-based values, the two crustaceans and
Gambusia.

As a result of our recommended updates to the White Sturgeon, Bluegill, and Brown Trout egg/ovary
chronic values, the muscle criterion will also need to be adjusted.

Original letter contains Table 6 - Updated calculation of muscle fish tissue-based Se criterion based on
modifications to Table 8 (N= 14 genera, R = sensitivity rank in database). See original letter.

4.3 Additional Discussion of Tissue Criterion Elements

4.3.1 Conversion Factors

As part of our previous review, we conducted an evaluation of the conversion factors (CFs) developed by
EPA in the 2014 draft Se criteria document wherein we reviewed all of the data used and corrected
values where mistakes were found. A detailed evaluation of this issue is also presented by NAMC-SWG
(2014).

In addition to reviewing EPA's data and calculations, we also compiled matched tissue data from studies
conducted by GEI to supplement the CF database (GEI Appendix A, 2014). We also used the matched
egg/ovary, whole-body, and muscle Se data provided in the 2014 draft Se criteria document, which was
further updated by GEI as described above, and developed regression-based CFs (Table 9; GEI Appendix
A). When the regression has a relatively high goodness of fit (i.e., when R2 is at least 0.70), we
recommend using the regression equation in place of the median (or geometric mean) ratios, as the
regression better predicts tissue concentrations, particularly at the high and low ends of the spectrum.
Where the strength of the regression is not as high (e.g., creek chub, Fathead Minnow, mountain
whitefish), it may be more appropriate to use the median or geometric mean CF to represent the central
tendency of the relationship. As shown in Section 3.3.2, we used the regression-based CFs where
appropriate to translate the updated egg/ovary criterion database for these species to whole-body for
the purposes of deriving the updated whole-body criterion. For the remaining species, we used the
updated and new median ratio-based CFs.

EPA-HQ-OW-2004-0019-0411-A2; American Petroleum Institute (API); Posted 10/28/2015

2.2.2 Updates to the Whole-Body Criterion

193

-------
In Section 3.1.5.2 of the 2015 draft Se criteria document, EPA (2015) presents its approach for deriving a
WB-based criterion. To develop WB chronic values, EPA translated the egg/ovary chronic values used to
develop the egg/ovary criterion using egg/ovary to WB CFs (i.e., Table 3.5 of EPA (2015)). These
converted values were then used to calculate a WB criterion of 8.0 mg/kg. EPA again used a sample size
of 15, which included the three invertebrate-based values, the two 'assumed' crustaceans, and the
genus, Gambusia.

In the past, EPA has used regression-based CFs (e.g., Bluegill CF from EPA's 2004 draft Se criteria
document). However, in the 2015 draft Se criteria document, EPA developed CFs based on the median
of available matched egg/ovary and WB Se data. As noted in comments on the 2014 draft (GEI 2014a, b),
we believe that a more appropriate method would be to use regression-based egg/ovary to WB
translators when the regression relationship has an R2 value >0.70. We translated the egg/ovary
database to WB and derived an updated WB criterion. For the genus Oncorhynchus, individual CFs were
used to translate each species individually (i.e., O. mykiss and O. clarkii), but an overall regression using
data for both species was used to convert the Oncorhynchus egg/ovary GMCV to a WB GMCV of 14.99
mg/kg

In addition, we updated the median-based CFs based on our review of EPA's data and the addition of
our data; these updates were incorporated here. Many of these recommendations are based on data
that EPA had included in the 2014 draft (EPA 2014) or were provided in an earlier review of the criteria
document (including tissue-to-tissue data for fathead minnows in the GEI (2008) study cited by EPA),
and were not used; no reasons for this were provided.

The WB criterion should also be adjusted as a result of our updates to the egg/ovary chronic values
(Table 2). We recommend that EPA consider a recalculated criterion with regression-based CFs in place
of median-based CFs when possible, as it is based on scientifically-defensible data for relevant species.

Original letter contains Table 2 - Modified version of Table 3.5 of (EPA 2015) following adjustments to
conversion factors and chronic values recommended herein. See original letter.

2.2.3 Updates to Muscle Criterion

In Section 3.1.5.3 of EPA (2015), after presenting its approach for developing the egg/ovary criterion,
EPA presents its approach for deriving a muscle-based criterion. To develop muscle-based chronic
values, EPA translated the egg/ovary chronic values used to develop the egg/ovary criterion using
egg/ovary to muscle CFs (Table 3.7 of EPA (2015)). These converted values were then used to calculate a
muscle-based criterion of 11.3 mg/kg. EPA again used a sample size of 15, which included the three
invertebrate-based values, the two 'assumed' crustaceans and the genus, Gambusia.

As a result of our recommended updates to the egg/ovary chronic values, the muscle-based criterion
will also need to be adjusted (Table 3). We recommend that EPA consider this recalculated criterion, as
it is based on scientifically-defensible data for relevant species.

Original letter contains Table 3 - Modified version of Table 3.7 of EPA (2015) following adjustments to
chronic values based on review and analysis of study data recommended herein. See original letter.

194

-------
Comment Category 3.6 - Comments Concerning Other Toxicity Data
Full fill ing Minimum Data Needs (N=15)

¦11 mm airy

This section contains comments about chronic effects studies of non-fish organisms, including
invertebrates, crustaceans and amphibians, and their use in the minimum data requirements to fulfill
EPA's guideline recommendations for calculating the criterion. Some commenters questioned the
decision and method used to translate invertebrate chronic concentrations to a predicted fish tissue
concentration. Some commenters noted that invertebrate data would be useful in establishing site
specific criteria for streams that do not support fish populations.

,)onse to comments

3.6 Responses regarding use of other toxicity data

EPA agreed that the estimated egg-ovary concentrations for invertebrates should be deleted and did so
in the 2015 draft and also in the 2016 final selenium Criterion document. The data available do indicate
that the invertebrates are somewhat less sensitive to fish on a whole-body basis and thus invertebrates
are included in the "N" for the criterion derivation. Invertebrates are portrayed explicitly in the whole
body Sensitivity Distribution (SD) with the converted whole-body fish tissue concentrations for
comparative purposes.

The selenium fish tissue criterion is expected to be protective of invertebrates, based on limited data
available. EPA further agrees that additional data on invertebrate sensitivity to selenium would be useful
to further support this conclusion. EPA is aware of the available invertebrate data that reported
selenium impacts at low concentrations, however our review of these data uncovered data quality flaws
that limit the quantitative use of these studies for criteria derivation.

In the criterion document, fish and invertebrate SMCVs and GMCVs are no longer presented on the
same tables and invertebrate SMCVs and GMCVs are presented as whole body concentrations.

However, criteria are developed to protect the entire aquatic community, such that the available data
should reflect the sensitivity range for various components (e.g., fish, invertebrates) of the aquatic
system. Towards this end, invertebrate sensitivity to selenium was evaluated in terms of both measured
whole body concentrations as well as in terms of what the whole body tissue concentration of a
representative fish would be were it to consume each invertebrate with a whole body concentration at
the SMCV and GMCV, by multiplying each invertebrate GMCV by 1.27, the median TTF for all fish
species. When evaluated with or without the trophic level biomagnification, the available data
demonstrate that compared to fish, invertebrates are not as sensitive to selenium, and do not comprise
any of the four lowest GMCVs. They are used to fulfill the taxonomic minimum data requirements, and
are counted in the total number of genera ("N") in the calculations. We note the numeric impact of
increasing "N" though inclusion of the invertebrate data in the "N" for the criterion calculation is
minimal, reflecting an approximately 3% change in the egg-ovary criterion element value.

Regarding the comment on the number of GMCVs, EPA agreed that the invertebrate and the Gambusia
data should not be represented in the egg-ovary sensitivity distribution, and they have been removed

195

-------
from that distribution. EPA 304(a) criteria are developed to protect the aquatic community, and so the
"N" should consider the other less sensitive taxa data implicitly; even though they are not represented
in the SD, they are still protected by the criterion. We also evaluated new studies, and so the "N"
reflects the number of studies that are used quantitatively, and represented explicitly in the SD; as well
as those less sensitive taxa that cannot be represented in the SD, but are part of the total number of
taxa where data are available that show that the criterion is protective.

EPA identified limited data on amphibians and characterized effects in comparison to fish in the 2015
draft. Data indicate that the selenium fish tissue criterion is expected to be protective of the aquatic
community including invertebrates and amphibians.

¦pecific comments

EPA-HQ-OW-2004-0019-0387-A1; Colorado Department of Public Health & Environment;
Posted 10/14/2015

• In Table 3.9 (page 53), it would be helpful to list the species or genera used for each category,
rather than just counts.

• Should Xenopus be included in the total number of GMCVs? While not a native species, it is a
surrogate for amphibians in general. In addition, including Xenopus would be consistent with
EPA's approach of including "waived" genera in the "N" used derive the criterion.

EPA-HQ-OW-2004-0019-0412-A2; Utility Water Act Group (UWAG); Posted 10/28/2015
3.1.4 Summary of Relevant Invertebrate Tests

UWAG agrees with EPA's conclusion that there is implicit protection for invertebrates if the fish-based
chronic tissue criteria are attained. EPA reviewed all of the available literature concerning chronic
toxicity effects to invertebrates and found that:

The relative insensitivity of invertebrates when compared with the fish whole-body concentrations
demonstrates that invertebrates are generally protected by selenium criterion values derived from fish.
Therefore, the invertebrates are considered implicitly in the species sensitivity distribution, and are
counted toward the number of values available to calculate the fish tissue criterion elements...

Draft Criterion at 45.

Considering acute water column exposure only, invertebrates (notably amphipods and planktonic
crustaceans) are more sensitive than fish when exposed to either selenite or selenate (US EPA 1995).
However, when invertebrate species are exposed to selenium via the diet during chronic tests, the
resulting chronic values are higher than the four most sensitive fish species GMCVs based on
reproductive effects (Draft Criterion Tables 3.3 and 3.4 at 46-47). Additional studies support that
invertebrates can accumulate selenium to very high levels in field settings, with seemingly no
population-level effects. For example, in a fly ash-impacted receiving stream in West Virginia (Little
Scary Creek), Hydropsychid caddis flies and the hellgrammite Corydalus had mean whole-body selenium
concentrations of 31 mg/kg and 25 mg/kg, respectively, without any apparent population-level adverse

196

-------
effects (Reash et al. 1999). These studies lend additional support to EPA's decision to assume the
criterion is protective of invertebrates at the proposed levels.

3.1.6 Selenium Fish Tissue Toxicity Data Fulfilling Minimum Data Needs

EPA's "Guidelines for Deriving Numerical National Water Quality Criteria for the Protection of Aquatic
Organisms and Their Uses" ("1985 Guidelines")(US EPA, 1985) specify certain data that should be
available if a numerical criterion is to be derived. If all the required data are not available, then "usually
a criterion should not be derived." 1985 Guidelines at 3. In developing the selenium criterion, EPA found
there were no available data on the chronic toxicity to two invertebrate groups, planktonic crustaceans
and benthic crustaceans. EPA's Guidelines specifically require toxicity data representing these two
faunal groups. EPA made a policy decision to waive this requirement. While UWAG normally would not
support derivation of aquatic life criteria in the absence of the minimum data specified in the 1985
Guidelines, in this case UWAG agrees with EPA's policy decision.

EPA used available toxicity data for organisms from the phylum Arthropoda (which includes insects) as a
surrogate for the two missing faunal groups. Those data indicate that invertebrates are less sensitive
than the aquatic species having the four lowest Genus Mean Acute Values. Because crustaceans are
arthropods, the agency concluded that there are no data that might indicate that planktonic and benthic
crustaceans are more sensitive to selenium than insects. The agency also relied on field studies and
consensus of the experts that it is the egg-laying (oviparous) vertebrates that are most at risk. UWAG
agrees with this conclusion. Since crustaceans are less likely to transfer selenium via the egg, EPA
determined the data requirements to be adequately met. EPA's conclusion is consistent with results
from field studies that suggest amphipods are tolerant to chronic selenium exposure. For example, in
Connor Run, a fly-ash pond receiving stream with average water concentrations of 73 ng/L, composite
samples of Hyallela azteca had a mean concentration of 25.4 mg/kg (American Electric Power
unpublished data).

EPA-HQ-0W-2004-0019-0400-A1; The National Mining Association (NMA); Posted 10/15/2015
3.6.1 Recommendation for Invertebrate Tissue Criterion

In cases where fish populations are not present due to flow limitations the aquatic life to be protected
are the macrobenthos. In EPA's analysis of invertebrate data, the most sensitive species had a GMCV of
24.2 mg Se/kg dw wb. All other effect concentrations for invertebrates for which chronic data were
available were substantially higher than 24.2 mg/kg. Therefore, based on these data, an invertebrate Se
tissue concentration of 24.2 mg/kg dw would be protective of these invertebrates and an appropriate
tissue-based criterion for fishless waters.

197

-------
1 viiilliii> iiii 1 hi. hi ' > iiiii" . * II l...i 11 -I nil II •. riv;liii'ii

of Fish Tissue Criterion Elements
3.7 Summary

This section contains comments on reproductive and non-reproductive studies that were not used in
numeric criterion derivation. Included are comments concerning juvenile salmonids studies.

3.7 Response to comments

3.7 Responses concerning studies not used by EPA

EPA did not use Gambusia data directly to derive the selenium egg-ovary criterion, given they are a
viviparous fish. EPA did use these data to calculate the whole body criterion, as well as counted in the
"N" for derivation of the egg-ovary criterion. The available data indicates that Gambusia is insensitive
compared to other species in the database.

EPA examined Buchwalter's (Conley et al. 2013) recent study. There were 2 studies with the same
exposures but different food rations (lx and 2x). There were major differences in the effects levels
observed, which seemed to be a diet effect, rather than a selenium toxicity effect. These data are
discussed in the selenium criterion document. EPA has also examined the Conley data and have included
the data that has met the data quality requirements for quantitative consideration in the criteria
derivation process.

Regarding consideration of new data, EPA has acquired new data through the peer review and public
process and has included these data, as appropriate, in the derivation of the criterion.

EPA has not previously accepted injection studies as a valid exposure method for aquatic life criteria
development, and this exposure route is considered particularly critically regarding selenium for several
reasons.

1. A 100% selenomethionine exposure does not reflect natural dietary sources.

2. Microinjection does not include the natural metabolic detoxification and storage processes
occurring in the female over time in the diet, as opposed to a bolus dose of a single form of
selenium which likely overwhelms the body's metabolic processes.

We note that most of the peer-reviewers of the 2014 criterion agreed with the exclusion of the injection
route of exposure, and several provided additional lines of reasoning for excluding injection studies from
consideration. EPA has reexamined the Doroshov et al. (1992) study data and determined that it will not
accept this injection study as a valid exposure method for selenium aquatic life criterion development.
The majority of the peer reviewers' comments on the 2014 criterion support this decision.

Regarding the exclusion of the 90-day endpoint in Hamilton et.al. 1990, EPA had significant data quality
concerns with this data, most significantly that the 90-day control survival (67%) was below toxicity test
acceptability thresholds, such that we could not use those effects data. We did consider and use the 60-
day time point for this study, and concluded that the egg-ovary transformed whole body criteria of 8
mg/kg dw would protect against growth effects in juvenile salmonids. Compromising data analysis to

198

-------
include very high control mortality data despite the availability of other more acceptable data within the
same test does not meet EPA's data quality expectations, and would not constitute modernizing
methods.

Regarding the relative sensitivity of reproductive and non- reproductive (e.g., larval growth) effects, EPA
completed a new analysis and found that the reproductive endpoint is protective of juvenile survival and
growth (see pages 134 and 135 in the revised 2015 document, and Section 6 of the final 2016 Selenium
Criterion document). EPA however has noted in the criterion document that the whole body tissue is
the sample of choice to assess selenium exposure in freshwater environments in juvenile anadromous
salmonids (i.e., Pacific Northwest species) from the larval swim up to the smolt stage prior to
outmigration to the marine environment.

We have added a figure to Section 6 where the reproductive effects and non-reproductive effects are
compared, demonstrating that the whole body criterion element translated from egg-ovary criterion
threshold is expected to be protective against non-reproductive effects.

Regarding the comment on the inclusion of additional genera, the Catostomus and Xyrauchen studies,
presented in Appendix D of the 2014 draft, EPA determined these studies are not of sufficient quality for
quantitative use, therefore they have not been included in setting N. During development of the
document, EPA considered increasing the N used in the criterion calculation to reflect inclusion of
Catostomus, based on the suggestive evidence from the de Rosemond et al. study, but decided that it
would be more prudent not to, given that Crutchfield (2000) states that "By the early 1980s, the fish
community had collapsed in the Hyco Reservoir and species such as [names of several taxa]... and
suckers (Catostomidae) became much reduced throughout the reservoir." Regarding the Hamilton et al.
study, EPA does not believe it can come to any reliable conclusion about the relative sensitivity of
razorback suckers.

Regarding the comment on adding new studies, the authors of the mountain whitefish study were
contacted by EPA; however, the authors declined to provide these data to EPA at this time. The
Formation Environmental Yellowstone cutthroat trout study has been analyzed and added to the
database.

Regarding the zebrafish studies, EPA has evaluated the available zebrafish data and identified several
important issues with the quantitative use of those data. This information is included and fully discussed
in the revised 2015 draft criterion document and the 2016 Final Selenium Criterion document. This issue
was also highlighted in the FR Notice in July 2015 with a request for additional data on zebrafish and
other cyprinids. A major problem is that the concentration-response curve was so unusually shallow
that zebrafish could be interpreted to be among the most or least sensitive species depending on the
level of effect considered. Further, high control mortality (47%) at the end of the study raised concerns
about the health of the fish at the time of testing. In addition, since the zebrafish is a non-native
cyprinid species, EPA assessed the information available on zebrafish sensitivity to selenium compared
to the sensitivity of native cyprinid (minnow) species across the United States (Appendix D in the criteria
document), including several studies where native cyprinids were investigated in selenium-impacted
waters. Data from these studies suggest that native cyprinids are likely less sensitive to selenium than
the currently available non-native zebrafish data suggest. The results of these analyses, particularly a
comparison of the concentration response relationships of zebrafish versus all of the other fish species

199

-------
for which we have similar data, raises a concern. Given these concerns, EPA has not used the zebrafish
data quantitatively in the derivation of the 2016 criterion.

3.7 Specific comments

EPA-HQ-OW-2004-0019-0385-A2; Kentucky Division of Water (KDOW); Posted 10/13/2015

The exclusion of bioassay studies using only water-column exposure is appropriate. Chapman, et al.
(2010) found the measurement of selenium in fish tissue appropriate given its close ties to chronic
toxicity, namely embryo mortality and teratogenic effects.

EPA-HQ-OW-2004-0019-0387-A1; Colorado Department of Public Health & Environment;
Posted 10/14/2015

In Section 3 of the criteria document, the Gambusia data are only mentioned in footnotes to Tables 3.2
and 3.5 and shown in figures. It would be helpful to have more information in the text about its status
and use in the criteria. It appears to have been included in the "N" used to derive the different elements
of the criterion, as it is included in the genera list on page 54, but its use should be explained in the main
text.

EPA-HQ-OW-2004-0019-0413-A2; Federal Water Quality Coalition (FWQC); Posted
10/28/2015

In its Notice, EPA also asks for comment on its decision not to use studies of zebrafish (a non-native
minnow species) in developing the draft criteria. While these studies were submitted to EPA for
consideration in developing the criteria, the Agency identified a number of concerns, with the studies,
including issues regarding the concentration response curve and a high control mortality rate. EPA also
concluded that native minnow species are probably less sensitive to selenium than the zebrafish data
suggest. We agree that these reasons are well-founded. Therefore, we support EPA's decision not to use
the zebrafish data in deriving the new criteria.

EPA-HQ-OW-2004-0019-0411-A2; American Petroleum Institute (API); Posted 10/28/2015
1.1.4 Exclusion of Juvenile Survival

We support EPA's decision to focus on studies using dietary uptake and maternal transfer to young,
rather than juvenile survival data.

EPA-HQ-OW-2004-0019-0391-A1; North American Metals Council (NAMC); Posted
10/14/2015

200

-------
We support strongly EPA's decision to not use juvenile survival, in particular juvenile overwinter survival,
as an endpoint in developing the proposed criterion. We note, however, that the reproductive criterion
is inappropriately compared to endpoints from juvenile overwintering survival in Section 2.2.4 of the
2015 EPA Selenium Draft.

NAMC supports EPA's decision not to include data on zebrafish for the reasons outlined in the Federal
Register. The data in question are from an academic study on a non-native fish with a very strange dose-
response curve that is not typical of the selenium dose-response curve.

EPA-HQ-OW-2004-0019-0396-A1; National Marine Fisheries Service Office of Protected
Resources Interagency Cooperation Division; Posted 10/15/15

Special Considerations for Pacific Salmonid Juveniles: While NMFS appreciates EPA's consideration of
the effects of selenium on Pacific salmonid juvenile growth rates in both the 2014 and 2015 drafts, the
handling of the 90 day exposure data from Hamilton et al. 1990 should reflect a more up to date
perspective. EPA states that these data were excluded based on control survival requirements for
ambient aquatic toxicity data established in the 1985 Aquatic Life Guidelines (USEPA, 1985) and the
Manual of Instructions for Preparing Aquatic Life Water Quality Criteria Documents (Stephan, 1987). The
guidance documents EPA referenced were developed at a time when approaches to tease out
interacting factors or incorporate complex aspects of exposure were not readily performed due to
limitations in computational technology. EPA has already circumnavigated these limitations in its
development of the copper guidelines based on biotic ligand binding and the dietary exposure based
selenium guideline currently under review. EPA also recently hosted an "Invited Expert Meeting on
Revising U.S. EPA's Guidelines for Deriving Aquatic Life Criteria" to collect information on alternative
approaches to selecting and analyzing data, with a view to expanding the data used in guideline
development. This selenium guideline should reflect EPA's intent to modernize its assessment
approaches in arriving at water quality guidelines. Eliminating data based on these valuable, but
outdated guidances, is not consistent with EPA's current vision for guideline development.

EPA-HQ-OW-2004-0019-0412-A2; Utility Water Act Group (UWAG); Posted 10/28/2015
6.1.5 Reproductive studies not used in the numeric criterion derivation

EPA considered two studies conducted using zebrafish (a non-native minnow species) but found a
number of anomalies with each. The study by Penglase et al. (2014) was not used for criterion
calculation purposes due to only one treatment level causing reproductive effects. The study by Thomas
and Janz (2014) was considered but not used by EPA due to the extremely low reproductive-based EC10
value (7.0 mg/kg). The agency then evaluated the relative sensitivity of native minnow species based on
laboratory and field studies (analysis found in Appendix E). UWAG agrees with EPA's decision not to use
the zebrafish data and offers the following comments on native minnow selenium sensitivity.

In a field study, Reash et al. (2006) reported an average selenium whole-body concentration for
bullhead minnows (Pimephales vigilax) in a fly-ash pond receiving stream of 44.55 mg/kg. For
comparison, bluegill collected in the same stream had an average whole body concentration of 17.26

201

-------
mg/kg. The bullhead minnow population in the stream was considerable, and the species fed largely on
filamentous algae. Creek chub (Semotilus atromaculatus) and blacknose dace (Rhinichthys atratulus)
collected from a different fly ash pond receiving stream also accumulated high levels of selenium
(Reash, 2012). The findings in these studies suggest that some minnow species are relatively tolerant to
long-term selenium exposure.

EPRI (2001) reported results of a selenium uptake and depuration study using fathead minnows. This
study is attached as Appendix 1 and UWAG requests that this study be added to the Draft Criterion
section "Studies of Non-Reproductive Effects" in Appendix D. Juvenile fathead minnows were exposed
to selenium via water (sodium selenite), diet (fish chow spiked with selenomethionine), or both water
and diet. Fish were exposed to selenium in flow-through tanks for 144 days; one group of fish were then
analyzed for selenium content in muscle, carcass, liver, and gonad tissue, while another group were
provided a depuration period of approximately 60 days after which tissue samples were analyzed. In a
parallel study, minnows were exposed to selenium at four discrete temperatures: 10, 15, 20, and 25° C.
The principal results of the study were:

• Original letter contains additional information in the form of an attached appendix, EPRI (2001)
report: Selenium Cycling and Impact in Aquatic Ecosystems: Defining Trophic Transfer and
Water-Borne Exposure Pathways. See original letter 0412-A3.

• Fish exposed via the diet accumulated significantly higher levels of selenium compared to fish
exposed via water. This finding is consistent with several other studies referenced by EPA. Tissue
selenium concentrations were highest in fish exposed to selenium via both water and diet.
Selenium concentrations in the various tissues were liver > muscle > gonads.

• Depuration of tissue selenium was observed starting four days after the exposures were
discontinued. Within 30 days post-exposure, the reduction in tissue selenium was 73% in ovary
samples, 49% in liver samples, and 19% in muscle samples.

• The effect of temperature on fish growth and selenium accumulation was not consistent. As
expected, growth was lowest at the coldest temperature (10° C). The selenium content in
muscle and liver samples was highest at 20°C, however the lowest level of selenium in these
tissues occurred at different temperatures.

• Fish survival and weight was not different between control fish and fish exposed to 30 mg/kg Se
in their diet. And the concentration of selenium in the various tissues was not elevated (relative
to control fish) at a dietary concentration of 13 mg/kg Se; markedly higher tissue levels were
evident in fish exposed to a diet concentration of 30 mg/kg. The mean concentration of
selenium in ovary and muscle tissues, at this dietary exposure concentration, was 36.9 and 28.5
mg/kg, respectively. Though the sample sizes of fish in the various tests were not substantial,
the results of this study suggest a non-reproductive effect level for this species as > 30 mg/kg
(expressed as selenium in the diet) for fish growth and survival and an ovary concentration
effect level of 25.7 mg/kg (the mean ovary concentration in fish exposed to 30 mg/kg in their
diet).

• Whatever the exact effect levels are based on this study, there is strong evidence that fathead
minnows are more tolerant to selenium compared to largemouth bass (the species having the
4th lowest reproductive effect Species Mean Chronic Level).

• UWAG agrees with EPA's assessment that the zebrafish reproductive effect EC10 value of 7.0
mg/kg Se should not be used to derive a reproductive effect-based chronic value. The

202

-------
information provided above, and in Appendix E of the Draft Criterion, clearly indicates a
reasonable weight-of-evidence conclusion that native minnow species are considerably more
tolerant to chronic selenium exposure relative to zebrafish.

203

-------
Comment Category 4.1 - Comments on Translation Equation (Eq. 18)
4.1 Summary

This section contains comments about EPA's equation that translates tissue concentration to an
equivalent water concentration. It includes comments on the USGS bioaccumulation model (Equation 1),
EPA's decision to simplify Equation 1 by not including growth and selenium uptake, and the method for
deriving of single-species TTF and jjFcomP°slte. Comments on structure and applicability of Equation 18
are also included here. Many commenters agree that the translation equation was a valid approach.

4.1 Response to comments

4.1 Responses to comments on translation equation

Regarding the translation of the tissue criterion to water, EPA re-evaluated conversion factors used and
refined the approach to reflect the most taxonomically proximate data in its criterion calculations in
2015, strengthening its previous draft analyses as noted above. EPA also re-evaluated the lentic and lotic
water column values, and developed values that are based on the best available, scientifically defensible
science to yield water column values that are protective of aquatic life based on available data.

The water-column element is a translation of the egg-ovary element and thus is intended to provide the
same level of protection as the egg-ovary element. However, some level of uncertainty in translating the
egg-ovary element to the water-column element is unavoidable. To address this uncertainty, EPA chose
the 20th percentile of translated water-column values using the most bioaccumulative food web present
at each site to select a protective national water column criterion element. Limitations of available data
allowed translation of the egg-ovary criterion element at 69 unique aquatic sites. The 20th percentile
was selected because it results in a low probability of false negatives even for the most bioaccumulative
species in ecosystems, (i.e., low probability of failure to indicate exceedance of the egg-ovary criterion
element via use of the water column element).

It would be impossible to complete a national-level kinetic analysis because sufficient data do not exist
to allow for a rate-based dynamic model using speciated selenium data. Thus, EPA uses total dissolved
selenium to reflect the selenium concentration in water that is available for uptake and assimilation at
TL1. The 2014 document's narrative on the uptake and transformation of selenium has been rewritten
to reflect the actual processes in nature.

EPA has followed upon the recommendation from a peer reviewer of the 2014 draft that:

"all that is needed is a model that covers dissolved to first trophic level interactions, and from
there the existing biodynamic part of the Presser and Luoma (2006; 2010) could be employed.

In this case, using Equations 4-6, and 7, in the Meseck and Cutter (2006) paper (and related
equations in the Appendices) could suffice."

EPA coupled the above with consideration of the reviewer's recommendation to:

"Use ... simple first order rate equations (and values) described in the literature."

204

-------
Based on input from the peer review of the 2014 criterion, with first-order kinetics and steady-state
assumptions, particulate organic selenium, PSe(-ll), in the primary producers is proportional to dissolved
organic selenium, DSe(-ll), which in turn is proportional to dissolved Se(IV), which in turn is proportional
to dissolved Se(VI). Consequently, under these model assumptions, selenium in the primary producers is
proportional to total dissolved selenium in the water column. The recommended approach from a peer
reviewer of the 2014 criterion thus appeared to be a more detailed articulation of the approach EPA
applied, provided that the first-order kinetic assumption is taken to be one of the acceptable
approaches, as the reviewer seemed to suggest, and if the commonly used steady-state modeling
assumption is applied, as was done for all other trophic levels.

In site-specific situations where ample measurements are available for all forms of selenium, EPA
recognizes the value of a detailed articulation of the organic-inorganic selenium processes. However,
because such data are rarely available, the detailed approach is not feasible to capture the wide
diversity of sites represented via the derivation of the national criterion. EPA believes that confining its
assessment to a very small number of well-studied sites where such measurements are available would
reduce rather than increase confidence in the appropriateness of the national criterion. Likewise, an
analysis based on a few site- or lab-specific measurements would also decrease rather than increase
confidence, when compared to an approach that homogenizes the rate coefficients but uses the end-
resulting ratios of concentrations observed in water and primary producers at a wide diversity of sites.

The model prediction itself is correct with respect to the translation calculations. In applying the
translation model, the egg-ovary criterion element concentration is used to translate that value to a
water concentration by dividing that egg-ovary criterion element by the product of a species-specific
composite TTF, a species specific egg-ovary to whole body to CF, and a site specific EF). The model
values in Table 12 of the 2015 draft, as well as in the final 2016 Criterion, are correct calculations that
have been verified through a data quality analysis.

EPA disagrees that the previously peer-reviewed and published methodology (Presser and Luoma, 2010)
applied by EPA to derive the criteria needs to be reformulated. EPA worked with the scientists who
published the methodology in applying their model in the context of a national criterion development to
ensure appropriate application of their model. Further, none of the other six peer reviewers
recommended or indicated that EPA needs to reformulate the methodology used for setting water
column criteria.

Notably, an entirely independent analysis (DeForest et al 2014, as provided as an appendix to the NAMC
2015 public comments) based on a regression approach directly comparing empirical water column data
to empirical egg/ovary data, resulted in a water column criterion magnitudes for lotic and lentic systems
that were less than 1 ng/L different than EPA's bioaccumulation modeling approach, when the same
toxicity data used by EPA were applied (15.1 mg/kg). This further substantiates the rigor and outcome of
EPA's approach.

Regarding the inclusion of sulfate interactions in the criterion derivation, the Deforest et al 2014 report
indicated that a sulfate-dependent selenium criteria would apply only to selenate dominated, well-
oxygenated streams, which is a small subclass of waters in the US. The publication also noted that "due
to methodological challenges and high costs, it is difficult to comprehensively evaluate the influence of

205

-------
sulphate on bioconcentration and transfer up the food chain." For reasons including these uncertainties,
EPA has decided not to include a sulfate correction factor in the 2016 selenium criterion.

EPA has noted previously in its response to Peer Review Comments document

(https://www.epa.gov/sites/production/files/2015-10/documents/epa-response-external-peer-review"
comments-selenium.pdf) and in its 2015 draft criterion document that a "trophic transfer model that is
dynamic and rate/kinetically based" is not feasible due to lack of availability of kinetic data. The kinetic
analyses recommended in the comment would be impossible to complete nationally because sufficient
data do not exist to allow for of a rate-based dynamic model using speciated selenium data. Thus, EPA
uses total dissolved selenium to reflect the selenium concentration in water that is available for uptake
and assimilation at trophic level 1.

As noted in the criterion document, states can develop their own site-specific criterion. EPA has
provided substantial assistance for state interested in developing site-specific criteria, including
descriptions of two different approaches to developing SSCs, with example calculations, and several
appendices contain data with geographical specification that could be applied to develop SSC's.

4.1 Specific comments

EPA-HQ-OW-2004-0019-0392; Anonymous Commenter; Posted 10/14/15

First, although the mathematical structure presented in the document appears to be sound, there is no
way to know whether the computations were set up correctly in the spreadsheet or whatever
computational tool EPA used. Consequently, EPA needs to release that computational tool (such as a
spreadsheet) to the public for inspection, in order to assure that the equations, as described in the
document, were incorporated correctly into the computational tool used for the criterion derivation.
Because of the complexity of the water criterion derivation from its database, this is not a trivial matter.
In general, QA/QC checking of computational tools is not a trivial matter.

EPA-HQ-OW-2004-0019-0390-A1; West Virginia Coal Association (WVCA); Posted 10/14/15
Development of Water Column Concentrations

WVCA is most concerned with the methodology utilized by EPA to convert the egg-ovary criteria into
water column criteria for lentic and lotic waters. As described more fully below, we believe the USGS
methodology described in Section 3.2 of the Draft Selenium Criterion and most recently published by
Presser (2013) is fatally flawed and cannot be utilized for this purpose. We will focus our comments on
EPA's use of this troubling publication.

The water column criteria are based largely upon the model presented in the recent USGS publication
Selenium in Ecosystems within the Mountaintop Coal Mining and Valley-Fill Region of Southern West
Virginia-Assessment and Ecosystem-Scale Modeling (Presser 2013) and previous publications supporting
the same concepts regarding selenium uptake in the food web. WVCA has grave concern with EPA's use
of this publication and the validity of the base assumptions in the USGS ecosystem-scale model.

206

-------
The USGS model claims to integrate a predator's dietary exposure pathway to develop a link to selenium
toxicity and, thus, to species vulnerability. West Virginia has conducted extensive water column
monitoring, fish tissue analysis, and deformity analysis for State waters. The field evidence and
laboratory data entirely contradict the USGS model. The route of selenium uptake into aquatic species in
West Virginia waters cannot be accounted for by the USGS model.

EPA entirely departs from the 1985 Guidelines in preparing the water column elements in the Draft
Selenium Criterion. Instead, EPA develops tropic transfer function (TTF) values for a number of
freshwater fish and attempts to use these numbers to back-calculate conversion factors (CF) to prepare
water column criteria. This procedure is complex and is fraught with inconsistent internal assumptions
and cross-species extrapolation to yield the calculated water column elements of the chronic criterion.

The derived water column numbers are entirely at odds with the criteria that would be calculated based
upon the methodology set forth in the 1985 Guidelines and empirical data. EPA's novel approach has
not received public input separate and apart from its use for selenium, requiring the public to challenge
both the TTF methodology as well as the data decisions by EPA that are specific to selenium.

WVCA urges EPA to reconsider the draft water column elements of the selenium criterion. This effort is
too broad, which leads to numbers that are likely inaccurate all of the time. While WVCA understands
and appreciates EPA's effort to create a shortcut for States who do not want to develop water column
numbers, the development of generic numbers based on inaccurate assumptions that contradicts
empirical data is not an acceptable approach.

EPA-HQ-0W-2004-0019-0404-A2; Kentucky Coal Association (KCA); Posted 10/15/2015

KCA is also concerned that EPA's methodology for calculating the water column criterion from fish tissue
is unprecedented, not properly supported by sound science, and subject to significant uncertainties.

Although KCA strongly supports the fish tissue approach and urges EPA to reconsider the concept of
alternative water column concentrations proposed in the criterion, if a water column number is included
in the criterion in any form, it must be based on scientifically defensible methods. Here, KCA does not
believe the record adequately demonstrates that EPA has used reliable science to derive water column
values. The criterion states that EPA derived the water column values through use of a complex formula
that estimates bioaccumulation and trophic transfer by relying on many variables that will differ widely
in the field based on site-specific conditions. There is no clear indication that EPA's methods have been
tested or proven to be reliable. And aside from the apparent unreliability of its approach, by layering
conservative assumption upon conservative assumption in the models, EPA's final derived water column
criteria is necessarily overly conservative and fails to accurately reflect actual conditions in real-life
receiving streams. Given these concerns, the water column values should, at most, be used as a "trigger"
as in the case of Kentucky's 5 ug/l standard, particularly given that the Kentucky experience
demonstrates that such a water column trigger has been sufficiently protective of aquatic life.

207

-------
EPA-HQ-OW-2004-0019-0412-A2; Utility Water Act Group (UWAG); Posted 10/28/2015

1. EPA Should Recalculate the Draft Water Column Criterion Using Appropriate Data and Good
Science.

UWAG has serious concerns with the science underlying the draft chronic water column values. While
the bioaccumulation model that EPA selected to back-calculate site-specific water quality criteria from
egg/ovary thresholds and site-specific factors is a scientifically valid approach when used on a site-
specific basis, application of the model to derive nationally recommended criteria for lotic and lentic
settings has significant uncertainty in terms of all input variables being available for all sites. EPA's
reliance on a limited data set and selection of a conservative 20th percentile protection value resulted in
water column criterion that is overly conservative for most sites and will compel many permittees to
conduct needless fish tissue studies. UWAG recommends that EPA withdraw and recalculate the
proposed water column criterion considering the additional data made available through comments
submitted by GEI Consultants in comments on the 2014 Draft. This approach was strongly
recommended by several of the external expert peer reviewers and should not be ignored by EPA. See,
e.g., 2014 Peer Review at 21-22. Failure to recalculate the water column values will result in
unnecessarily over-conservative criterion while achieving no environmental benefit.

Water Column Translation:

• "I am very concerned that EPA is placing too much value on extrapolated and modeled values.
The translation approach [calculating site water criteria based on a bioaccumulation model]
involves building food models for 69 sites that in many cases have significant data gaps (e.g.,
dietary composition, extrapolated TTFs, extrapolated CFs, etc.)." (2014 Peer Review at 7).

• "Related to this was the "translation" of the tissues (all)-based criteria to the water column (14%
of comments), and most of these comments were directed to the inappropriate use of the
Presser and Luoma model. Considering my review above and the community response, it would
seem that the EPA needs to reformulate their methodology for setting water column criteria."
(2014 Peer Review at 21-22).

EPA-HQ-OW-2004-0019-0406-A2; Kentucky Chamber of Commerce; Posted 10/15/2015

With respect to the derivation of the proposed water column concentration criteria for lotic and lentic
waters, the Chamber also believes U.S. EPA has not adequately demonstrated the reliability of the
science that was used to derive those values. As stated above, EPA derived the water column values
using a complex model that attempts to estimate bioaccumulation and trophic transfer through aquatic
system food-webs. There are many variables, which largely depend on site-specific conditions. (Criterion
Document at Section 3.2.1). The model appears to be untested and is not proven to be reliable. By
definition, water column values cannot be as reliable an indicator of the threshold for impact as the fish
tissue concentrations from which they were derived. The derived water column values are more
stringent than the 5 ng/l total selenium value previously adopted by many states, including Kentucky,
based upon U.S. EPA's prior recommended criterion. Based upon available (albeit limited) fish flesh test

208

-------
data, it appears that the 5 mg/l value has been protective of aquatic life in Kentucky. Given the
documented superiority offish tissue as a more reliable indicator of protection of the aquatic life use,
there is no justification or record support to require states to adopt the unproven proposed lotic and
lentic water concentration criteria particularly where existing state water column selenium standards
have been shown to be protective of aquatic life. At most, any such water column concentrations should
only apply as a "reasonable potential" trigger for conducting fish tissue testing. The record does not
support the establishment of those concentration values as reliable independent criteria.

EPA-HQ-OW-2004-0019-0385-A2; Kentucky Division of Water (KDOW); Posted 10/13/2015
Water Column Criterion

KDOW believes that EPA's approach to deriving water column elements, considering lotic and lentic
waters separately, appropriately considers how aquatic habitat affects selenium speciation, water
residence time, and selenium accumulation in the food chain. However, EPA has not adequately
explained its approach to deriving the water column criteria for lentic and lotic waters from the fish-
tissue values. As much as any aspect of this proposed criterion the method for deriving the water-
column values should be transparent and fairly scrutinized. Using the (extremely low) water-column
values as proposed by EPA, many waters currently meeting their intended uses and that do not have
selenium concentrations in fish tissue at a level of concern would otherwise be deemed as not meeting
designated uses, if based solely on water chemistry. That empirical disconnect may be the result, in part,
of the inappropriate inclusion of the Linville (2006) study. However, as the methodology for deriving
water column values has not been adequately explained any conclusion is speculative.

EPA-HQ-OW-2004-0019-0388-A2; J. R. Simplot Company; Posted 10/14/2015

C. Miscellaneous Comments

C.l. Simplifying the Bioaccumulation Model

[Page 55, Section 3 .2.1.1]

In ERG's (2014) External Peer Review of the 2014 Draft Criterion, Dr. Gregory Cutter noted the following
with respect to the first step of trophic transfer:

"The overall approach of considering selenium's pathway from the water column , dissolved
state, through trophic levels, and into tissues such as reproductive organs is well justified ,
particularly the trophic transfer model that is dynamic and rate/kinetically based (uptake rate *
assimilation efficiency/elimination rate); the trophic transfer approach largely developed by Nick
Fisher and collaborators. However, the water to first trophic level approach is completely
unacceptable in that it is not dynamic or rate-based (actually assumes equilibrium) and
completely ignores the effects of speciation. The latter is curious in that they seem to be relying
on the Chapman et al (2009 and 2010) recommendations from the SETAC Pellston workshop
which specifically states, "Understanding Se speciation is critical to understanding its mobility,

209

-------
transformation, partitioning in the environment, and potential risk to aquatic ecosystems and
'The single largest step in the bioaccumulation of Se occurs at the base of food webs,
characterized by an "enrichment function"; thermodynamic or equilibrium-based principles are
not appropriate for predicting Se bioaccumulation at the base of food webs." The choice of the
Presser and Luoma model used in this EPA document is completely contrary to these
recommendations since the water/particle ratio called the Enrichment Factor (EF) is only a
renamed equilibrium distribution coefficient (Kd) that was used long ago for metal cations."

He goes on to state,

"The use of the Presser and Luoma (2006, 201 0) model for any aquatic ecosystem to predict
dissolved or particulate concentrations is questionable for the simple reason that while it
acknowledges the importance of chemical speciation , and the rates of processes (kinetics as
opposed to equilibrium thermodynamics), it largely ignores them in application."

Dr. Cutter recommends, at least for the first step in the trophic transfer modeling approach for
selenium, that "biological uptake of dissolved nutrients such as nitrogen, and metals, is best (most
accurately) modeled using Michaelis-Menten kinetics, or at least pseudo-first order rate expressions."
He suggests that simple Michealis-Menten equations and values in the literature (i.e., Riedel et al.,
1991), and simple first order rate equations (and values) described in the literature (i.e., Cutter, 1982;
Cutter and Bruland, 1984; Reinfelder et al., 1993) would suffice for the EF stage of the Presser Luoma
Model. He concludes by stating,

"After the dissolved to first trophic level particulate selenium part of the model that I am
criticizing above, the rest of the Presser and Luoma (2006) model (including the derivation of
TTFs) is excellent and accurately predicts bioaccumulation through the various parts of the food
web (and earlier documented in the Luoma and Rainbow (2005) peer-reviewed paper). The
reason here is that once into the first trophic level, the primary speciation of particulate
selenium is organic selenide, and the concepts of assimilation efficiency, trophic transfer factors,
ingestion and depuration (egestion) work well for selenium (and any other metal or nutrient)."

This concept is empirically demonstrated by using much of the field data generated for selenium to
derive an aqueous concentration. Very low selenium concentrations in water result in high EF values,
that drive the ultimate aqueous value derivation to a value as low or lower than the original field
measured value. Conversely, high aqueous selenium values in water often times result in low uptake in
algae and thus low EF values. Resulting aqueous values using EFs from these types of scenarios will
result in higher concentrations due to the low EFs. It is clear that the EFs drive the Presser Luoma model
and more than one investigator has suggested that the uncertainty in this single (large) step results in an
uncertain aqueous value (Deforest and Cutter, in ERG 2014).

210

-------
EPA-HQ-OW-2004-0019-0395-A1; CONSOL Energy; Posted 10/15/15
Water Column Concentration Recommendations

CONSOL recommends that EPA reconsider the draft water column elements of the selenium criterion.
The methodology utilized by EPA to convert the egg-ovary criteria into water column criteria for lentic
and lotic waters is too broad, and could lead to inaccuracies. These nationwide lentic and lotic water
column-based criteria were derived to supplement the tissue-based criteria. However, it is not
appropriate to derive a single nationwide standard for water column based criteria for only two water
body types (lentic or lotic); and, such an effort is not supported by EPA's own analysis. The water column
criterion from the egg/ovary criterion is a valid approach, but should be used on a site-specific basis. In
addition, we recommend that EPA evaluate direct use of tissue-based criteria in NPDES permitting,
which would eliminate the need for the back-calculation to a water value.

EPA-HQ-0W-2004-0019-0401-A1; Colorado Wastewater Utility Council; Posted 10/15/2015
3.4.3 Alternative Approach to Chronic Water Column Standard

An alternative to the EPA approach for the water column criterion, would be to use an alternative
chronic water column criterion, developed using the approach described in DeForest et al. 2014. The
authors derived an equation for developing protective water concentrations that, like EPA's water
column criteria, are based on attaining a tissue-based criterion, but are also dependent on sulfate (S04)
concentrations. In addition to sulfate limiting the acute effects of Se exposure, it has been shown that
increasing sulfate concentrations reduce the bioavailability of Se to organisms at the base of the food
web, which in turn reduces dietary concentrations for higher trophic level organisms (DeForest et al.
2014). This protective effect of sulfate occurs in selenate-dominated lotic systems. The equation used to
calculate the water column value is:

[Original letter contains Equation - not numbered. See original letter.]

Where 'Water Se concentration' is in ng/L, 'fish Se cone' is the egg/ovary criterion in mg/kg, and S04 is
in mg/L.

Using this equation with the egg/ovary value of 17.9 mg/kg and sulfate concentrations ranging from 50
mg/L to 250 mg/L would result in protective water column criteria ranging from 5.7 ng/L to 40.0 ng/L,
respectively (Table 7). These values fall within the range of protective lotic water column concentrations
of 1.2 ng/L to 40.6 ng/L calculated by EPA, and well within the protective concentrations calculated by
GEI of 1.2 to 81.0 Hg/L.

[Original letter contains Table 7 - Protective water column Se concentrations at various sulfate
concentrations using the equation from DeForest et al. 2014. See original letter. ]

211

-------
EPA-HQ-OW-2004-0019-0411-A2; American Petroleum Institute (API); Posted 10/28/2015
2.3.6 Overall Comment on Water Column Criteria Elements

The previous sections highlight our concerns with the methodology for back-calculating the water
column criterion element, discuss the resulting conservatism in the draft criteria for lentic and lotic
waters, and identify alternative methodologies for deriving water column criteria that we believe to be
more robust. Several concerns with the methodology were also raised by the external peer reviewers of
the EPA's 2014 draft AWQCfor Se (Eastern Research Group Inc. 2014), but negligible revisions were
considered in the revised draft. The most substantive change from 2014 to 2015 in this regard was that
the number of data points included in the evaluation was reduced, which had the subsequent effect of
lowering the draft lentic and lotic Se criteria further. At a minimum, we request that the EPA clarify that
states may derive water Se criteria based on alternative methods and using site-specific data for state
water bodies.

Water Column Criterion Element

• We advocate for flexibility in translating between tissue and water: guidance, not prescription.

• We believe that the model used to derive lentic and lotic water column Se criteria is technically
flawed. The unusual use of site-specific EFs, along with various amounts of site-specific food
web information and TTFs compiled from a variety of non-site-specific sources, results in
translated water Se concentrations that have no ecological relevance in actual systems. The fact
that the model predicts that the water column concentration at a reference lake would need to
be lowered from 0.32 ng/L to 0.23 ng/L to meet the draft fish egg Se criterion of 15.8 mg/kg-dw,
despite fish tissue Se concentrations being nowhere near levels of concern in the lake, highlights
the fundamental flaw in the model and its lack of ecological relevance.

• Five of seven external peer reviewers of the 2014 draft AWQC document (EPA 2014) identified
concerns in the EPA's approach for deriving lentic and lotic water column Se criteria; yet, no
effort was made to reconsider the methodology. We do not suggest that the EPA abandon the
food web modeling approach, but highly recommend that the EPA reconsider how the EFs and
TTFs are compiled and applied. Use of regression relationships, rather than constant EF and TTF
assumptions when translating between different Se concentration magnitudes, would remove a
lot of the concerns with the current approach.

• EPA should adopt an alternative model to derive lentic and lotic water column selenium criteria.
The model EPA used to derive the proposed criteria is technically flawed.

212

-------
> ' 'lull II li < ir

Function (TTF) Values
4.2 Summary

This section contains comments on deriving trophic transfer function (TTF) values using physiological
coefficients found in literature and the empirical relationship of matched pairs, including EPA's hybrid
approach using selenium tissue concentration to food concentration ratios, and ordinary least squares
for statistical verification.

4.2 Response to comments

4.2 Responses concerning derivation of TTF values

In effect, EPA is positing that although partitioning from water to particulate matter is highly variable
and can best be evaluated using data gathered from the specific site under consideration (if available),
selenium partitioning among tissues (CF), and from particulate matter to fish through the food chain
(TTF) is relatively similar for a given species regardless of site. Although this introduces some
uncertainty to the translated values, CFs and TTFs are less variable across sites than EFs based on
available data. EPA also notes that site-specific data necessary to evaluate the partitioning of selenium
from water to particulate matter is often not available.

EPA used TTFs derived from both lab and field data, but field data were considered superior to lab data
for several reasons, such as: having representative diets, reflecting assumed steady-state conditions and
an absence of artifacts that can be observed in laboratory exposures (e.g., poor nutrition, selenium
speciation issues). TTFs derived from field data are expected to be sufficiently protective of aquatic life
in the field since they are estimates based on real- world conditions.

The dynamic nature of selenium transfer from one trophic level to another is better represented in field
data than in laboratory or mesocosm data. EPA did however include TTFs from the multiple Conley et al.
publications based on the life-cycle exposure of selenium to mayflies via a complex periphyton diet.
There were some concerns regarding the Conley mayfly TTFs when insufficient food resulted in different
TTF (and toxicity) measurements, which illustrate the potential bias with laboratory studies. The
decision to include the Conley studies was based on the weight of evidence of similar TTFs for most of
the exposures and the need to fill a data gap for a TTF of an important fish prey item.

Regarding regression versus median approaches, ultimately, a hybrid approach (median ratios with
regression statistics used to screen data quality) was selected by EPA because the use of ratios was less
sensitive to outliers or from regressions where the y-intercept was notably different from zero, and the
slope of a constrained regression (following the approach used by Presser and Luoma) had a poor fit.
Every approach (median ratios, constrained regression, and conventional ordinary least squares (OLS)
regression) has inherent strengths and weaknesses, and ultimately, the hybrid approach was
determined to be the most robust.

213

-------
EPA has carefully considered its assumption that EFs and TTFs do not vary with concentration. One
advantage of total least squares (TLS) is that the regression relationship is unaffected by which variable
is assigned to the x and y axes. Another advantage is that introducing sampling noise into an underlying
relationship does not consistently flatten the TLS slope (in contrast to its flattening of the OLS slope).
Because concentrations in both media have essentially equal uncertainty, total least squares might be
the preferred approach.

EPA used a median of concentrations of lower trophic organisms since most predators are opportunistic,
preying on the organism that they encounter at random. Site-specific studies would reveal predator
prey preference or prey abundance relationships that may warrant consideration of a different
approach.

EPA used a time limit threshold of 1 year to characterize samples as "matched" to maximize the
available data that are likely to be temporally similar. This approach was deemed necessary in order to
obtain sufficient data from sufficient sites in order to develop water column criteria. It is more
appropriate to collect site-specific data for a known impacted site at more proximate time intervals to
insure that both temporal and spatial considerations at the site are accounted for. This will be addressed
in implementation materials.

Again, EPA supports the development of site-specific values to reflect bioaccumulation of selenium in
specific ecosystems when data are available, and EPA has provided supporting information for SSC
development in appendices to the 2016 criterion document.

4.2 Specific comments

EPA-HQ-OW-2004-0019-0411-A2; American Petroleum Institute (API); Posted 10/28/2015
2.3.3 Constant TTFs Used in the EPA's Food Web Models

The EPA's food web model assumes that TTFs are constant over a range of dietary Se exposure
concentrations. However, as for EFs, there is typically an inverse relationship between TTF and exposure
concentration, which the EPA acknowledges in the draft document (EPA 2015). The EPA suggests that
use of median TTF data helps to account for this issue: "EPA evaluated the effect of very high and very
low selenium concentrations on the calculation of TTF values using the hybrid approach described above
by excluding selenium measurements above various minimum and/or below various maximum selenium
concentrations. EPA found that using the median ratio effectively attenuates any effects of selenium
concentration on the calculation of TTF values using the hybrid approach described above without the
need to introduce additional arbitrary exclusion criteria."

The EPA did explore the use of a regression-based approach, but noted disadvantages that included an
assumption that the underlying data are normally distributed and that one or a very few high values can
have a disproportionate influence on the slope of fitted line. Ultimately, the EPA chose to use median
TTFs, by species, if the ordinary least squares (OLS) linear regression of those data resulted in a
significant (i.e., p < 0.05) fit and positive regression coefficient. Ultimately, in developing the water-
column criteria elements, the EPA derived TTF composite values that ranged from 1.73 to 3.51, with a

214

-------
median of 2.75 (Table 3.13; EPA 2015). These magnitudes do not seem unreasonable, but it would be
interesting to see a sensitivity analysis of whether the median TTFs tend to be biased high or low relative
to the exposure concentrations to which they are being applied.

For example, the first row in Table 3.13 (EPA 2015) is for Iowa Darter in East Allen Reservoir, Wyoming
The TTFcomposite is 3.08, which is based on a median TTF of 1.62 for the darter and an invertebrate TTF of
1.90 [calculated from: (2.14insects X 0.70) + (1.22amphipods X 0.16) + (1.46crayfish X 0.08) +
(1.41planktonic_crustaceans X 0.06) = 1.90]. If we accept the WB-to-egg CF of 1.45, the draft egg Se
criterion of 15.8 mg/kg-dw translates to a WB Se concentration of 10.9 mg/kg-dw. The median TTF of
1.62 for the darter was based on data for fish in the family Percidae. Based on the relationship between
WB Se in fish within this family and co-located invertebrates, the invertebrate Se concentration
predicted to result in a WB Se concentration of 10.9 mg/kg-dw is 9.3 mg/kg-dw (Figure 4). This results in
a WB TTF of 1.18 and a revised TTFcomposite of 2.24 (for the purposes of this example we are assuming
the same invertebrate TTF used by the EPA, but that could be adjusted as well). Use of this revised TTF
would result in a translated water Se concentration of 2.11 mg/L (i.e., 15.8 mg/kg-dw / [2.31 L/kg x 1.45
x 2.24]). This translated water Se concentration of 2.11 ng/L is 38% greater than the concentration of
1.53 ng/L reported in Table 3.13 (EPA 2015). Accordingly, selection of the median TTF, rather than a
regression-based TTF, can potentially have important implications.

This is just one example, but it raises the question of whether it was premature to abandon a regression-
based approach, or at least evaluate the median TTFs relative to the exposure concentrations in the
food web models to ensure that they appear to be of an appropriate magnitude.

EPA-HQ-0W-2004-0019-0401-A1; Colorado Wastewater Utility Council; Posted 10/15/2015
4.4 Derivation of Water Column-based Criterion Elements

In Section 3.2 of the 2015 draft Se criteria document, EPA derives an equation to be used to translate
the egg/ovary tissue criterion into a water column criterion. The result is Equation 18 on page 63:

[Original letter contains Equation - not numbered. See original letter. ]

There are several components of this equation that require further analysis and consideration as
discussed below.

4.4.1 Discussion of Trophic Transfer Functions (TTF)

4.4.1.1 Importance of Site-specific TTFs

Ranges of TTFs can vary widely due to site-specific factors. To demonstrate this in our previous review
(GEI 2014a), we reviewed data in Appendix B of the 2014 draft Se criteria document and compiled
information on the ranges of TTFs for each species with data derived from field studies. We also verified
the median TTF values and found several errors in Tables 9 and 10 on pages 76 and 77 of the 2014 draft,
these errors still remain in the 2015 draft and should be fixed. Additionally, we recalculated TTF's
including additional data that were submitted for addition to the TTF database in our previous review,

215

-------
but were not incorporated in the 2015 draft. These values are included in Table 8 (below) and the
corrections are shown in bold with the incorrect values shown as strikeouts.

[Original letter contains Table 8 - TTF median ratios from Table 3.10 and 3.11 in the 2015 draft Se
criteria document and calculated ranges and geomean ratios. See original letter.]

EPA-HQ-0W-2004-0019-0400-A1; The National Mining Association (NMA); Posted 10/15/2015
4.4 Derivation of Water Column-based Criterion Elements

[Original letter contains Equation - not numbered. See original letter. ]

There are several components of this equation that require further analysis and consideration.
4.4.1 Discussion of Trophic Transfer Functions (TTF)

4.4.1.1 Importance of Site-specific TTFs

[Original letter contains Table 7 - TTF median ratios from Table 3.10 and 3.11 in the 2015 draft Se
criteria document and calculated ranges and geomean ratios. See original letter.]

EPA-HQ-OW-2004-0019-0390-A1; West Virginia Coal Association (WVCA); Posted 10/14/15

The use of the composite trophic transfer function (TTF) and the calculated enrichment factor (EF) is
highly questionable. WVCA believes that site-specific or State-specific studies are necessary to
implement the fish tissue criteria as water column concentrations.

EPA-HQ-OW-2004-0019-0391-A1; North American Metals Council (NAMC); Posted
10/14/2015

Further, the method used to apply EFs and TTFs (trophic transfer functions) to back-compute
waterborne concentrations used as much of the available data as possible, but ignored important
processes such as, for example, time variable accumulation processes (uptake and depuration kinetics)
and the effects of site chemistry (e.g., selenium speciation, S04). Site-specific considerations should

216

-------
allow for a mechanistically detailed site-specific analysis, which can be compared to the simplified
approach used in the subject document.

217

-------
< ' 'lull II li < 2 measurements and <1 sediment) and the EF calculations for 69 sites.

4.4 Response to comments

4.4 Responses concerning derivation of site-specific enrichment factor values

It is technically incorrect to call EFs an equilibrium concept. EFs, like TTFs and BAFs, are better
characterized as steady-state concepts. Steady state does not require equilibrium or process
reversibility. For example, irreversible uptake coupled with growth dilution would still yield a steady-
state EF.

EPA carefully re-examined the distribution of EFs and corrected the data for overweighting by EFs
measured in systems with unusually low or high waterborne selenium. Appendix (H) of the 2016
criterion document lists all of the EF values, by site, and the values used to make the calculations.

To reduce uncertainty in estimating site-specific EF values, EPA limited its analysis to those aquatic sites
with at least two particulate selenium measurements with corresponding water column measurements,
and only used sediment measurements if there was at least one other measurement from either algae
or detritus. That is, EPA would calculate an EF from algae (or detritus) alone; however, in order to
qualify, at least 2 algal (or detritus) samples from a site were required. In contrast, sediment data alone
were insufficient to calculate an EF. In order for sediment to be included, additional algal or detrital
data were also required.

EPA added the data from Bowie et al 1996, Casey 2005, Fan et al 2002, McDonald and Strosher 1998 and
Zhang and Moore 1996 to the EF data set. EPA already had Minnow Environmental 2007 in the EF data
set which includes the data contained in Orr et al 2006.

Regarding the nonlinear EF comments, some commenters opined that because EFs are not constant
over a range of water-column selenium concentrations, EPA developed water column values that are
too conservative (in a protective sense) because the analysis resulted in high EFs in low selenium
concentration water driving the water column value too low. Others commented, on the other hand,
that because there is higher selenium bioaccumulation in low selenium waters, EPA developed water
column values that were too high and did not account for this potentially higher relative impact in low
selenium waters. EPA used empirical data and a peer-reviewed, published model to generate protective
water column values based on the best available data.

218

-------
Stakeholders desiring to develop site-specific criteria based on local, site relevant data are encouraged
to do so. EPA provides information in Appendix K of the 2016 final criterion document, and other
appendices also contain data that would support such approaches.

One commenter (NAMC) submitted an analysis by DeForest el al (2014) regarding factors such as the
influence of exposure concentration on EF estimation. Following extensive analysis of this and other
factors, DeForest et al (2014) derived water column recommendations of 5.4 ng/L for lotic and 2.1 ng/L
for lentic systems at an egg-ovary value of 20 mg/kg, based on a regression analyses. The Deforest
water column values would equal approximated recommendation of 4 ng/L and 1.5 ng/L for lotic and
lentic systems respectively, at the EPA 15 mg/kg egg-ovary recommendation, based on an proportional
reduction in egg/ovary and water column values to EPA's values. These DeForest-approach-based
calculated water column values are strikingly similar to EPA's 3.1 lotic and 1.5 ng/L lentic
recommendations, with the lotic water value being less than 1 ng/L different than EPA's and the lentic
value being identical to EPA's value. This result further substantiates and supports EPA's water column
values, despite the use of different analytic approaches and slightly different protection levels for water
column values (75th percentile versus EPA's 80th percentile).

Regarding the comment on timing of sample collection within a year and that it somehow reflects a
"weak tie between dietary exposure and related tissue accumulation," there is a well know strong tie
between dietary exposure, fish tissue concentrations and adverse effects of selenium, and that
selenium, once in an ecosystem, persists in the food web for an extended period of time. Numerous
publications noting this relationship are cited in the criterion document.

Regarding the treatment of growth (g) in the bioaccumulation modeling, Luoma and Rainbow (2005)
suggest that for selenium, growth rate is a relatively inconsequential parameter under most
circumstances. Food consumption is typically high during periods of high growth rate. Because food
consumption is the primary route of selenium uptake in aquatic organisms (Ohlendorf et al. 1986a, b;
Saiki and Lowe 1987; Presser and Ohlendorf 1987; Lemly 1985a; Luoma et al. 1992; Presser et al. 1994,
Chapman et al. 2010), high consumption rates of selenium-contaminated food may counteract the
selenium dilution that occurs with the addition of body tissue during periods of fast growth. EPA
evaluated the effect of removing the parameter g in the by performing a sensitivity analysis. EPA
concluded from that analysis that omitting the growth rate parameter g is an appropriate simplification
of Equation 1. A more detailed description of this sensitivity analysis is provided in Appendix J of the
2016 criterion document.

Regarding requests for more discussion on uncertainty, EPA has conducted additional analyses and has
included a more extensive discussion of uncertainty in Section 6.5 of the 2016 final criterion document
in Section 6.

4.4 Responses to other issues brought up in these comments

Specifically with regards to EPA-HQ-OW-2004-0019-0400-A1 and additional data for EFs and, EPA
contacted data stewards identified by commenters (e.g., DeForest, DeBruyn), and requested data sets
from them. EPA then independently evaluated those data sets for quality, and then used those data

219

-------
that met Agency data quality standards. In some cases, EF data in particular could not be used since
there was no paired fish data collected with the water and particulate matter at a site. This is crucial for
calculating water values used in the distributional analyses in the derivation of the water column
elements. Useable quality data was entered into the models' data sheets for use in deriving EFs. Data
used to calculate EFs in DeForest et al. 2014 were received by the author. Studies were examined, and
data for EFs that met EPA data requirements, such as the requirement of at least two particulate
measurements, one of which must be either algae or detritus, for example, were added to the EPA
dataset. A total of 22 EFs were added, increasing the total number of EFs in the EPA dataset to 91. This
included the addition of 6 EFs from data separately obtained from Adrian DeBruyn of Golder Associates.
Of the additional added EFs, examination of the source material associated with the EF data resulted in
the addition of 11 sites for which water could be translated, increasing the total from 53 to 64. Data
were added from Orr et al. (2006), Muscatello et al. (2008), Muscatello and Janz (2009), Orr et al. (2012),
and Golder (2011); Teck Coal (2013). EPA thanks the data stewards for providing requested data.

GEI Consultants were contacted and paired fish and invertebrate data were obtained for fourteen
species, which were added to the EPA dataset and used to update TTF values. Additional TTF data were
added from Muscatello and Janz (2009), Muscatello et al. (2008), and Orr et al. (2012) as a result of the
inclusion of additional of EF data. In addition to changes in TTFs resulting from additional data, TTFs for
some species have changed slightly as a result to changes in conversion factors between tissue types.
Finally, discrepancies in TTF values reported by EPA in the 2014 draft criterion document and those
calculated by GEI in their review of that draft were examined and updated if needed. In some instances,
EPA disagreed with the recommended changes. For example, EPA reported a TTF of 2.88 for damselflies
and 1.97 for dragonflies, in contrast to GEIs recalculated TTFs of 0.89 and 1.30. In both cases, TTFs
differed by a factor of 2.21. This is because EPA incorporated an additional trophic level accounting for
the TTF between the invertebrate prey of damsel and dragonflies and their prey's food, which had a
median TTF of 2.21. As a second example, the difference between the striped bass TTF of 1.48
calculated by EPA and 0.62 calculated by GEI was the result in how data across studies were grouped.
EPA calculated the median AE, IR, and Kd data across all studies, while it appears GEI calculated TTF
separately for each study, and then calculated the median of the two TTFs for the two studies.

Regarding the issue of the "constant EF approach", the relationship between EF and Se in water, that is
that EF decreases as Se in water increases, is recognized by EPA and is presented in Section 3.2.4. The
effect of this decreasing trend in bioaccumulation (including EF) is discussed on page 145 of the 2016
final document as it pertains to the ability of the model to predict observed concentrations. The model
accurately predicts observed values around the criterion concentration (15.1). See Figure 6.4 of the
2016 criterion document. This demonstrates that at the criterion, the model works. At low
concentrations, the model underestimates because of the nonlinearity of EF, TTF and CF - due to
saturation. EPA discusses the application of the model (which includes EF) on page 145 in that
statement. The linear assumption is environmentally protective.

EPA does not believe it has erred in the wet weight/dry weight conversions for the Lemly data.

Regarding the role of sediment and its variability, EPA only used sediment when either algae or detritus
were also available, to form a geometric mean. Regarding the use of reference sites like Badin Lake, EPA

220

-------
asserts that the use of this lake is defensible; however, commenters on the previous 2014 draft pointed
out the methodology used by EPA to assemble a distribution of lotic and lentic values was flawed.
Commenters stated that some waters (including Badin Lake) were over-represented in the distribution
due to unequal weighting of water values available based on the number of fish species sampled at a
site. For Badin Lake, this was 6 species at 3 sites, so 18 of 44 water measurements were derived from
this waterbody, skewing the distribution. EPA corrected this methodological flaw in subsequent
analyses by using only the species most sensitive to selenium bioaccumulation when more than 1 fish
was sampled, resulting in EPA using only 1 fish per site in the criterion document. Therefore in both the
2015 and 2016 analyses, Badin Lake data consists of only 3 measurements in the database.

Because of its high BAF, EPA examined the Lemley study carefully and found the data acceptable.

As discussed in Appendix I, EPA also considered the influence of the linearity assumption: that is, the
assumption, implicit in the derivation, that BAFs remain constant if a site's observed tissue
concentrations increased or decreased to the criterion level. The assumption has a protective influence
as low concentrations, but has the opposite influence at high concentrations. Overall, because the
criterion is set based on the 20th percentile calculated BAF, it has a protective influence.

EPA examined the sensitivity of the criterion to the Badin Lake data point. Even if a reason could be
found to reject it, the lentic criterion element would change from 1.53 to only 1.57 ng/L (recognizing an
additional significant digit for comparison purposes), indicating the lack of substantial impact of this data
on the magnitude of the lentic criterion element, furthering the support for its sustained inclusion in the
criterion derivation.

For examining the influence of the potential outliers, it is likely better to consider the effect of
simultaneously trimming both ends of the distribution. If one data point was trimmed from each end of
the lentic and lotic distributions (while calculating criteria to three significant digits), the lentic criterion
would increase by 0.02 ng/L and the lotic criterion by 0.22 ng/L. If two data points were trimmed from
each end, the lentic criterion would increase by 0.18 ng/L and the lotic criterion by 0.37 ng/L. This
information is intended for understanding the sensitivity of the result to the values at the ends of the
site distributions. EPA does not believe it has a sound reason to actually exclude those data points from
the calculation, but is simply providing this information as a point of reference.

4.4 Specific comments

EPA-HQ-0W-2004-0019-0409-A2; OC Public Works; Posted 10/20/2015
5. Calculation ofKd and Treatment of Se Speciation

In the bioaccumulation model EFs are not based on real Kds, which by definition are the ratio between
selenium concentrations in particulate form (w/w) and dissolved form (w/v) for a single chemical
species. Instead, fine sediment or organic-rich debris, or often bedded sediment were used to substitute
for particulates in the calculation of EFs. Furthermore, Kds are species-specific (i.e. different selenium
species, if measured separately, will have drastically different Kd values) and aggregating together all
species in calculating EFs is inappropriate for developing a national Criterion. While this approach is due

221

-------
to a severe lack of selenium speciation data, the result appears to be a large spread of EF values.

Without a good estimate of EFs, which introduce the largest errors, it is impossible to make the
bioaccumulation model sufficiently robust.

EPA-HQ-OW-2004-0019-0388-A2; J. R. Simplot Company; Posted 10/14/2015
Simplifying the Bioaccumulation Model

EPA derived a chronic water column selenium criterion element by translating the egg-ovary
concentration to an equivalent water concentration through the use of a bioaccumulation model
(Presser and Luoma 2010). However, a close examination of the first step (i.e., enrichment factor) of this
trophic transfer model shows that there are high uncertainties associated with the calculated water
column values. Other approaches may result in a more robust calculated water column value.

EPA-HQ-0W-2004-0019-0419-A2; Lima Refining Company; Posted 11/03/2015

COMMENT #1B. Kinetic Artifacts -The use of a steady state modeling framework for translation of the
egg — ovary criterion to a water column criterion introduces poorly quantified kinetic artifacts that
further add to the unexplained variability in the results, including field measurements of trophic transfer
factors. For example, some sites exhibit well defined patterns that reflect seasonal variations in annual
rainfall — runoff processes that tend to repeat from one year to the next (e.g., San Louis Drain; Elk River
Valley). During months of the year when instream concentrations are increasing, biota levels should
tend to increase, but will lag the increasing water column concentrations (steady state is an elusive
condition, a moving target that may never be achieved). During such times the measured tissue Se levels
will tend to understate the steady state concentration associated with the current water concentration
and/or the Se concentrations in their diet. During other months, the reverse occurs as water column
concentrations are decreasing while tissue concentrations continue to lag behind, perhaps even
continuing to increase. This is because their body burden reflects the previously higher instream
concentrations. At this point, measured tissue concentrations may overstate the steady state residues
associated with the current, descending water concentrations. The reverse occurs when water column
concentrations are approaching an annual minimum. The EPA simulations of Figure 2 (Figure J5 from
Appendix J, inserted below) illustrate trends of this sort quite nicely. For example, note how the
maximum fish tissue concentration (brown line) occurs at about day 310, approximately 25 days after
the occurrence of the maximum 30-d average water concentration (blue dashed line), and not far in
advance of the time when the minimum 30-day average water concentration occurs. Such patterns are
relatively easily understood when a kinetic model is applied in time variable mode, but they become
subsumed in the data when a steady state analysis is used to interpret the results.

[Original letter contains Figure 2 - Time Variable Simulation of Normalized Selenium in Fish Tissue
Relative to Daily Water Column Selenium (Figure J5 of Appendix J in USEPA, 2015). See original letter.]

In regard to a somewhat related matter, EPA presents an analysis used to establish an appropriate time
period with which to define matched pairs of Se measurements to use to evaluate trophic transfer

222

-------
factors (TTFs; Section 3.2.2.1). They conclude from the consistency in correlation coefficients over a
wide range of lag-times that timing is "insensitive to relative collection time within a one year time
period". Whether this somewhat surprising result is a reflection of the insensitivity of the TTFs to timing
of sample collection, or some other possibility (e.g., a mis-match between paired concentrations in the
predator and presumed food item(s)) is never addressed. Regardless, it would seem to be the case that
this sort of weak tie between dietary exposure and related tissue accumulation introduces considerable
uncertainty in the interpretation of results.

As a check on the veracity of the predictions of translated water column Se concentrations EPA uses the
same database that was used to derive the translated WQC values in the first place. Although the
predicted egg - ovary concentrations are correlated to the observed values, they tend to underestimate
observed tissue Se levels at low concentrations. EPA recognizes that this deviation likely reflects the
simplifying assumption of "constant bioaccumulation rates regardless of selenium concentration" in
spite of the commonly observed inverse relationship between accumulation (e.g., enrichment factor)
and water column selenium (Figure 3.7B of USEPA, 2015). Otherwise, the predicted and observed
concentrations tend to increase in a qualitatively consistent manner, though the unexplained variation
displayed by the overall trend in the data, and about the line of agreement, is considerable. EPA
attributed the uncertainty to several sources, including small sample sizes ("two particulate
measurements are only marginally sufficient"), temporal and spatial variability in selenium exposure,
and local variability in food webs.

Finally, it is important to recognize that g, in combination with Ke, also controls the response time of
biota tissue Se to changes in Se concentration in water or food. Thus, it is necessary to consider the
magnitude of g relative to Ke before one can neglect g (i.e., Ke » g for the effect of g on response time to
be neglected). EPA should also consider the possible effect of g on response times, and hence on the
averaging period analysis, as presented in Appendix J. Since a tissue residue-based criterion is being
proposed, the response time for tissue Se accumulation is expected to be important (see Paquin &
Bubnyte, 2014, a SETAC presentation on averaging period for Se).

EPA-HQ-OW-2004-0019-0411-A2; American Petroleum Institute (API); Posted 10/28/2015
2.3 Derivation ofLentic and Lotic Water Column Criteria Elements

In EPA (2015), lentic and lotic water-column criteria elements were back-calculated from the draft
egg/ovary Se criterion of 15.8 mg/kg-dw using the following equation:

[Original letter contains Equation - not numbered. See original letter.]

The EPA's first step was to compile enrichment factors (EFs) from sites for which information was
available on the fish species and the food web. For each EF, a composite trophic transfer factor
f|—ppcompositej value was identified for each of the fish species present. This jjFcomP°slte considered both
the fish species present and the "nature" of the food web. Because the TTF for fish was based on Se
concentrations in WB fish, a CF was needed in order to convert the WB Se concentration to an

223

-------
egg/ovary Se concentration. Below we provide narrative comments on specific components of this
approach, followed by comments on the overall approach.

We have concerns with this approach, the most important of which is that the measured EF for each site
used in the evaluation is assumed to be constant. However, Se EFs are clearly not constant over a range
of water-column Se concentrations (DeForest et al. 2014; in revision). Rather, they tend to be inversely
related with higher EFs observed at low water Se concentrations and lower EFs observed at high water
Se concentrations. The EPA made this same observation in Fig. 3.7 of EPA (2015), when comparing the
magnitudes of EFs for lentic and lotic sites; yet, the implications of this inverse relationship on the
translation of the draft fish egg Se criterion of 15.8 mg/kg-dw to water Se concentrations was not
accounted for. The end result is that sites with low water Se concentrations, such as reference water
bodies, are the drivers for both the draft lentic and lotic criterion elements, as the EFs for these sites are
relatively high. The problem is that these relatively high EFs should decrease as the water Se
concentration is "translated up."

The following narrative provides two examples of how the "constant EF approach" used by the EPA
provides counterintuitive results, followed by an alternative interpretation of the same dataset used in
EPA (2015). For the latter, the relationship between particulate Se and water Se concentrations is
considered, instead of assuming a constant EF at each site.

2.3.1 Examples of Counterintuitive Results Using "Constant EF Approach" for Translating Water
Column Se Concentrations Within a Site

2.3.1.1 Lentic Example

The lowest translated water Se concentration in Table 3.13 (EPA 2015) is 0.23 ng/L in Badin Lake, NC. In
other words, a water Se concentration of 0.23 ng/L is predicted to result in the draft fish egg Se criterion
of 15.8 mg/kg-dw based on a site-specific EF of 12.48 L/g, a WB-to-egg conversion factor (CF) of 2.00,
and a jjFcomP°slte of 2.77. The EF of 12.48 L/g is very high because the water Se concentration reported
for Badin Lake is very low (i.e., 0.32 ng/L). We can evaluate the accuracy of the EF of 12.48 L/g in
translating from a fish egg Se criterion of 15.8 mg/kg-dw by reviewing the co-located measured fish Se
concentrations in Badin Lake. Six fish species were identified for this site, and the EPA selected the
Fathead Minnow as the most conservative species. Lemly (1985) reported Se concentrations of 0.57 and
1.38 mg/kg-ww, respectively, in Fathead Minnow muscle and viscera (i.e., 2.28 and 5.52 mg/kg-dw;
assuming 75% moisture). For the sake of comparison, if it is assumed that the WB Se concentration is
near the mean of these two tissue components (i.e., 3.9 mg/kg-dw) and using the WB-to-egg Se
concentration factor of 2.00 (Table 3.13), the predicted egg Se concentration would be 7.9 mg/kg-dw —
less than one-half the draft egg Se criterion of 15.8 mg/kg-dw.

In other words, the empirical data for Badin Lake indicate that a water Se concentration of 0.32 ng/L is
associated with a fish Se concentration that is well below the draft tissue criterion; however, the model
applied by the EPA (2015) predicts that a lower water Se concentration of 0.23 ng/L in this lake will
result in a fish egg Se concentration of 15.8 mg/kg-dw despite the fact that the measured water Se
concentration is higher (0.32 ng/L) and the associated Se concentration in Fathead Minnows is well
below the draft fish tissue-based criterion. This erroneous conclusion is strongly driven by the

224

-------
assumption that the EF of 12.48 L/g is a constant that equally applies to higher Se concentrations in
Badin Lake, which is clearly not the case. This highlights the fact that the EPA's overall approach for
translating from the egg/ovary-based Se criterion to water column concentrations is flawed and has no
bearing on actual ecosystems. This is especially problematic because examples like this one (and similar
for the High Rock Lake reference lake) are the drivers for the lentic water column criterion element.

2.3.1.2 Lotic Example

Another counterintuitive example, this time for lotic sites, are the translated water Se concentrations of
1.19 and 8.15 ng/L derived for Deerlick Creek and Luscar Creek, respectively (both in the McLeod River
Basin in West-Central Alberta, Canada) (Casey 2005). Deerlick Creek is a reference creek with a water Se
concentration of 0.2 ng/L and an EF of 2.24 L/g and Luscar Creek is a Se-exposed creek with a water Se
concentration of 10.7 ng/L and an EF of 0.33 L/g (as reported in Appendix H; EPA (2015)). Consistent
with the inverse relationship between EF and water Se concentration that is typically observed, the
Deerlick Creek EF is relatively high (i.e., about 7-fold greater than the Luscar Creek EF). Use of the
relatively high EF for the reference Deerlick Creek results in a relatively low translated water Se
concentration of 1.19 ng/L, while use of the relatively low EF for the Se-exposed creek (Luscar Creek)
results in a relatively high translated water Se concentration of 8.15 ng/L. In other words, the EPA model
predicts that a water Se concentration of 1.19 ng/L in Deerlick Creek would result in the draft fish egg Se
criterion of 15.8 mg/kg-dw, while a water Se concentration of 8.15 ng/L would be required in Luscar
Creek.

Given the overall similarity of these creeks in terms of location and size, and the fact that both creeks
support Rainbow Trout, it is counterintuitive that the reference creek could only tolerate a water Se
concentration up to 1.19 ng/L before exceeding the draft fish tissue-based criterion, while the Se-
exposed creek could tolerate a water Se concentration up to 8.15 ng/L and not exceed the draft fish
tissue-based criterion. Again, as in the above lentic example, the only reason for this difference in
translated water column concentrations is the inverse relationship between EF and water Se
concentration that is not being accounted for in the EPA model. As also noted above, this becomes
especially problematic because this incorrect application of the EF results in a reference site being the
lowest translated water column concentration.

2.3.1.3 Overall Relationship Between Empirical and Translated Water Se Concentrations

Sections 2.3.1.1 and 2.3.1.2 provide two specific examples in which the simplified application of
uncertain EFs from reference sites results in very low water Se concentrations predicted to result in the
draft fish egg Se criterion of 15.8 mg/kg-dw. We also looked at the relationship between the paired
empirical and translated water Se concentrations for all 53 sites (i.e., 20 lentic and 33 lotic) used by the
EPA (translated water Se concentrations are those back-calculated from the egg Se criterion and
reported in Table 3.13). As shown in Figure 2, there is a positive relationship between the empirical and
translated water Se concentrations (i.e., translated water Se concentrations increase as empirical water
Se concentrations increase). Again, this is reflecting the influence of water Se magnitude on the EF and
highlights a deficiency in the "constant EF" approach used in EPA's model.

225

-------
[Original letter contains Figure 2 - Relationship between translated water Se concentrations at various
field sites that are predicted to result in a fish egg Se concentration of 15.8 mg/kg-dw (Table 3.13) vs.
measured water Se concentrations at those sites (Appendix H; EPA 2015). See original letter. ]

2.3.2 Alternative "Non-constant EF" Evaluation ofEPA's Dataset

Given our concerns for the "constant EF" approach used in the EPA model, we provide an alternative
evaluation of the EPA's dataset. First, using the CF and jjFcomP°slte assumptions in Table 3.13 of EPA
(2015), particulate Se concentrations predicted to result in the draft fish egg Se criterion of 15.8 mg/kg-
dw could be back-calculated. In the first row of Table 3.13, for example, the particulate Se concentration
was calculated as follows:

[Original letter contains Equation - not numbered. See original letter.]

This calculation was applied for all 53 sites in Table 3.13 and the particulate Se concentrations ranged
from 2.3 to 4.6 mg/kg-dw. This means that particulate Se concentrations of 2.3 to 4.6 mg/kg-dw, based
on the food web models compiled by the EPA, are predicted to result in the draft fish egg Se criterion of
15.8 mg/kg-dw. Then, using the co-located particulate and water column Se concentrations compiled in
Appendix H of EPA (2015), we evaluated the water Se concentrations in lentic and lotic water bodies
predicted to result in the back-calculated particulate Se concentration range of 2.3 to 4.6 mg/kg-dw.

This approach allows for consideration of the relationship between particulate and water Se
concentrations without the use of a model that relies on a constant EF assumption. This approach also
allows for a direct comparison to the draft water column criterion elements of 1.2 and 3.1 ng/L for lentic
and lotic sites (EPA 2015), respectively, based strictly on the data already compiled and used by the EPA.

Using the paired particulate and water Se data compiled by the EPA (2015) in Appendix H, the inflection
points where particulate Se concentrations begin to exceed the particulate Se concentration range of
2.3-4.6 mg/kg-dw is at about 3 ng/L for lentic waters and 6 ng/L for lotic waters (Figure 3). For lentic
waters, note that there are three data points in the particulate Se range of 2.3 to 4.6 mg/kg-dw that are
to the left of the vertical dashed green line (Figure 4a). Two of these are the reference lakes in Lemly
(1985), which were discussed above in Section 1.1.1. The other is a pond within the Mancos River Basin,
CO (Site MNP3; Butler et al. 1997). The reported water Se concentration in this pond was 1 ng/L and the
associated particulate Se concentration was 5.15 mg/kg-dw. In a nearby pond (MNP2), the water Se
concentration was 3 ng/L and the associated particulate Se concentration was 6.01 mg/kg-dw. The
higher water Se concentration in the nearby pond (MNP2), which had very similar particulate Se
concentrations, suggests that the single water Se concentration of 1 ng/L measured in MNP3 may not be
reflective of the true Se exposure over time in this pond. This raises the question of whether the
particulate Se concentration is linked to the appropriate waterborne Se concentration (or the particulate
Se concentration may simply be of a magnitude that is similar to the two reference lakes evaluated in
Lemly [1985]).

Interestingly, these lentic and lotic Se concentrations of about 3 and 6 ng/L that begin to exceed the
critical particulate Se range, are similar to recommended lentic and lotic guidelines of 3.5 and 6.0 ng/L
that have been previously developed (DeForest et al. 2014; in revision). In the latter evaluation, Se EF

226

-------
and TTF data were compiled from a large number of both field and laboratory studies, and quantile
regression methodology was used to relate a fish tissue Se guideline back to water Se guidelines.

The methodology of using site-specific EF data and then often generic TTFs and WB-to-egg Se CFs in
order to estimate what the site-specific water Se concentration would need to be to result in a criterion
of 15.8 mg/kg-dw is unusual and, as shown above, inappropriate, given the assumptions of constant EFs
and TTFs that are required over broad ranges of Se exposure concentrations. Furthermore, several of
the TTFs WB-to-egg CFs used in the evaluation are not even species-specific, so it appears that the final
number of sites evaluated (53) is artificially limited. If the model is sometimes based on generic fish and
food web information anyway, a much larger number of sites could be incorporated into the evaluation.
Finally, it seems that it would be better to compile the "universe" of available Se partitioning data
between water and particulates and among different trophic levels, and then consider the complete
relationships over a range of exposure concentrations, rather using constant EF and TTF ratios. From
these partitioning relationships, reasonable and appropriately-conservative estimates of water Se
concentrations that may result in the fish egg Se criterion can be derived (e.g., DeForest et al. 2014).
This, in our view, would be an improvement over the EPA's current approach in which EFs and TTFs are
separately compiled and defined without consideration of how they may or may not relate to one
another in actual field scenarios. As indicated in our comments, the EPA model often produces results
that are counterintuitive with the site-specific data from which they were partially derived.

[Original letter contains Figure 3 - Relationships between particulate and water Se concentrations for (a)
lentic and (b) lotic water bodies. See original letter.]

EPA-HQ-OW-2004-0019-0392; Anonymous Commenter; Posted 10/14/15

Third, most importantly, EPA has relied on a small number of sites (69) where EF values are available.
This approach exacerbates the problem that EF values are highly uncertain and highly variable across
sites, as discussed on page 74. In EPA's first draft, released in 2014, this small dataset having EF values
was supported by a much larger database where fish tissue and water concentrations were measured
but algae-detritus-sediment concentrations were not measured. In the 2014 draft this is discussed
starting on page 90: 2,588 lotic and 596 lentic sites. Oddly, EPA removed this large database from the
2015 document. The removal of so large a body of highly relevant scientific evidence has the
appearance of scientific misconduct. EPA must consider the large database of 2,588 lotic and 596 lentic
sites and make that database available to the public.

EPA-HQ-OW-2004-0019-0415-A1; Idaho Department of Environmental Quality (DEQ); Posted
10/28/2015

EPA's derivation of protective water column concentrations from the root egg-ovary criterion is logical.
However, the dataset used is skewed toward waters that would likely be judged as impaired. While EPA
acknowledges that "selenium accrual in particulate matter is lower at aquatic sites with a higher water
concentration of selenium compared to aquatic sites with a lower water concentration of selenium"

227

-------
they don't appear to have taken this into account. We urge EPA to seek out and incorporate data sets
from waters with lower selenium levels to assure the relations relied upon in their criterion derivation
hold at lower environmental concentrations as well. As an example we provide the figure below plotting
empirical selenium bioaccumulation versus ambient selenium concentrations in water in Idaho's major
rivers.

[Original letter contains an unnumbered figure - Se BAFs i/s SE in H20. See original letter. ]

While representing a variety of species, these data hint that selenium bioaccumulates more strongly in
environments with less selenium in water, as would be expected with bio-regulation of a metabolically
important element. These data represents a probabilistic slice of waters across Idaho, thus a more
typical range of water quality than EPA examined in their criteria development. The implication is that if
a median BAF, or EF, were used it would lead to a translation to a water column criterion that would be
under-protective at low environmental selenium concentrations and over-protective at high
environmental selenium concentrations.

EPA-HQ-0W-2004-0019-0401-A1; Colorado Wastewater Utility Council; Posted 10/15/2015
4.4.2 Enrichment Factors (EF)

Pages 74 of the 2015 draft Se criteria document state that "The single most influential step in selenium
bioaccumulation occurs at the base of the aquatic food webs (Chapman et al. 2010)" and "The
availability of selenium measurements from particulate material was limited." Given these statements
and how critical it is to generate valid EFs, EPA should consider inclusion of an uncertainty discussion
related to the particulate material Se data and how representative the calculated EF values may (or may
not) be to all sites. We specifically note that reliance on only suspended particulate Se in the water
column (as described in the document) is a poor measure of Se at the base of the food web in flowing
waters. EPA should specifically note that inclusion of periphyton, biofilm, or sediment Se can and should
also be considered when developing EFs for lotic systems.

Additional data to add to the EF database from GEI studies were included in Appendix A of our previous
review. These additional data were not incorporated into the 2015 draft; therefore, calculated EF values
for GEI studies are shown in Appendix A.

EPA-HQ-0W-2004-0019-0400-A1; The National Mining Association (NMA); Posted 10/15/2015
4.4.2 Enrichment Factors (EF)

228

-------
were included in Appendix A of our previous review. These additional data were not incorporated into
the 2015 draft; therefore, calculated EF values for GEI studies are shown in Appendix A.

229

-------
> ' 'lull III <
-------
¦pecific comments

EPA-HQ-0W-2004-0019-0401-A1; Colorado Wastewater Utility Council; Posted 10/15/2015
4.4.1.2 Use of Composite TTFs

The composite TTF is the product of the TTFs that represent dietary pathways of Se exposure for a given
species within an aquatic system. As we commented in our previous review, the jjFcomP°slte approach
would require prey item-specific consumption rate/frequency data (ideally site-specific) to best
characterize exposure. Species-and site-specific dietary preference data are likely not available for many
combinations of organisms. In those scenarios, assumptions will need to be made for each prey item,
introducing additional and potentially unrealistic uncertainty to derive a TTF composite value.

For more accurate analysis of a specific aquatic system, the best approach is to collect and analyze tissue
and water samples to measure actual Se concentrations rather than making assumptions and using
highly uncertain composite TTFs.

EPA-HQ-0W-2004-0019-0400-A1; The National Mining Association (NMA); Posted 10/15/2015
4.4.1.2 Use of Composite TTFs

231

-------
Comment Category 4.6 - Comments on Classifying Categories of Aquatic
Systems

¦ummary

This section contains comments on the characteristics EPA used to define lotic and lentic systems. Also
included are comments on influence of residence time on selenium bioaccumulation and EPA's
classification of the 69 EF sites. Comments also request that EPA specify a tissue-only criterion and
guidance on fishless waters.

,)onse to comments

4.6 Responses on classifying categories of aquatic systems

Based on the comments of peer reviewers as well as stakeholders in the public process, the EPA went
through each waterbody used in its analysis to develop lotic and lentic water column values, and
determined whether it was lentic or lotic, being careful to note specific issues like run of the river
reservoirs which could cause misclassification.

EPA revised the lotic and lentic section to better identify the difference between lotic and lentic in
Section 3.2.4. Unfortunately, a parameter like residence time was not available for most of the waters in
the dataset, thus EPA cannot provide this information in response to comments.

Regarding the somewhat conflicting comments that "it is known that bioaccumulation of Se is
significantly different in lentic and lotic systems (Adams et al. 2000)." ... and "it would be more
appropriate to develop site-specific criteria rather than create artificial groupings of waterbody types
that mask the site-specific differences so important to ensuring attainment of the tissue criterion"
(emphasis added), EPA's goal was to provide national guidance on water column values that reflect the
agreed-upon differences in selenium bioaccumulation between lotic and lentic systems. We agree the
classification of lotic and lentic waterbodies could be nuanced in some instances, but in many cases such
water body classification will be clear and self-evident (i.e., flowing and non-flowing waters,
respectively). Residence time information for the waterbodies included in the 2016 criterion document
analysis was not available to EPA, so EPA cannot provide specific residence time values associated with
the lotic and lentic classifications used in the criterion, or that could be applied more generally, as some
commenters are seeking.

We assert that decision to generate 2 sets of values to reflect the "known" fact "that bioaccumulation of
Se is significantly different in lotic and lentic systems" to be superior and more appropriate than
developing one national water column value which could be overprotective of lotic waters and
underprotective of lentic water bodies, in some cases. Interested parties can develop site-specific
criteria to reflect their particular water body conditions, as noted in the selenium criterion document.
EPA has also provided helpful information regarding the development of site-specific criteria in
Appendix K of the 2016 criterion document. Water column criterion element values were developed and
are important to support adoption and implementation of the selenium criterion.

232

-------
¦pecific comments
EPA-HQ-0W-2004-0019-0409-A2; OC Public Works; Posted 10/20/2015
6. Unclear Definition ofLentic and Lotic Waters

The water column portion of the Draft Se Criterion has very different concentrations for lotic and lentic
waters. This appears reasonable because selenium speciation, thus its transport, fate, and
bioaccumulation, are all affected by residence time, reaeration rate, and other biogeochemical
characteristics in which lentic and lotic waters are distinct from each other. However, it is not clear how
these two water types are defined, how transitional waters (i.e. those with characteristics between
lentic and lotic) and seasonally-variable waters (i.e. those switching from lentic to lotic between
seasons) and specially-variable waters (i.e. those with lentic and lotic conditions at different sections of
the water body) are treated. Without this distinction, the development process as depicted in figures
3.6-3.9 appears arbitrary and significant issues will be raised when the Draft Se Criterion is implemented
in these three types of waters.

For comments 2 through 7, USEPA should clarify the definition of lotic and lentic waters and use water
column concentrations as a screening tool. For sites where selenium levels in tissue is near or above the
Draft Se Criterion, site-specific water column guidelines should be derived by the bioaccumulation
model used in this Draft Se Criterion. In this approach, only site-specific data should be used to reduce
the uncertainty and improve scientific vigor of the model. With the tissue-based criterion to assess final
attainment, the water column guidelines should be used to gauge the progress toward selenium load
reduction and attainment of tissue-based criterion. This approach will also solve the 'new or increased
inputs' issue (comment 7).

EPA-HQ-OW-2004-0019-0415-A1; Idaho Department of Environmental Quality (DEQ); Posted
10/28/2015

Bifurcating water column criteria into lotic and lentic has merit based on the information presented on
differences in selenium speciation, food webs, and residence in these different waterbody types (section
3.2.4). But this also presents implementation challenges EPA should better address. Although EPA
speaks of residence time as a defining characteristic, no definitive residence time is offered. EPA does
not even provide descriptive information on the residence time of the waters used in their criteria
development. A clear definition or method of classifying waterbodies into lentic and lotic is crucial to
appropriate application of these separate criteria. Examples of waters that may be hard to classify
include so called "run of the river" impoundments and the flooded outlet streams of dammed natural
lakes. Furthermore there will always be transition areas, the inflows to lakes and reservoirs that may
behave more stream-like. And outlet streams whose water quality, if not ecology, will for some distance
be dictated by the water leaving the lake or reservoir. EPA should address how these transition zones
are to be handled, especially in the light of their national push for clearer downstream waters
protection.

233

-------
EPA-HQ-OW-2004-0019-0391-A1; North American Metals Council (NAMC); Posted
10/14/2015

2.3 Other Comments

2.3.1 Definition ofLentic and Lotic

NAMC again requests that EPA provide a clear and unambiguous definition of lotic and lentic areas and
clearly state that, since the draft criteria are based on freshwater data, they do not apply to estuaries or
other transitional waters (e.g., coastal lagoons).

Other issues that will also result in a very high incidence of false alarms include, but are not restricted
to: a clear definition of lentic and lotic is notably absent; protection goals are not clearly defined; there
are flaws in the translations from tissue to water; a key exposure and toxicity modifying factor is
ignored: the inappropriate requirement to "never exceed a tissue criterion"; and the current unrealistic
approach to fishless waters and increasing inputs.

EPA-HQ-0W-2004-0019-0401-A1; Colorado Wastewater Utility Council; Posted 10/15/2015
4.4.3 Classification of Aquatic Systems - Lotic vs. Lentic

It is known that bioaccumulation of Se is significantly different in lentic and lotic systems (Adams et al.
2000). EPA attempts to differentiate between system types using EFs and residence time. As discussed
in our previous review, this makes sense in theory, however, the resulting data have substantial overlap,
indicating the differences are not that clear. It would be more appropriate to develop site-specific
criteria rather than create artificial groupings of waterbody types that mask the site-specific differences
so important to ensuring attainment of the tissue criterion.

5.1 Discussion of Final Criterion

As stated above, we strongly support EPA's decision to develop tissue-based Se criteria that are
toxicologically and ecologically relevant. The tissue-based criteria, including our recommended
modifications and updates, reflect the best science and are protective of fish.

EPA also derived nationwide lentic and lotic water column-based criteria to supplement the tissue-based
criteria. However, as discussed in Section 3.3 it is not possible or appropriate to derive a single
nationwide standard for water column-based criteria for only two water body types (lentic or lotic), and
such an effort is not supported by EPA's own analysis. While we agree that use of Equation 18 (or a
bioaccumulation factor) to translate a water column criterion from the egg/ovary criterion is a valid
approach, it should only be used on a site-specific basis. In addition, as noted earlier, we strongly urge
EPA to evaluate direct use of tissue-based criteria in NPDES permitting, which would obviate the need
for the back-calculation to a water value, thus eliminating the potential errors and uncertainty that arise
in such calculations.

234

-------
EPA-HQ-0W-2004-0019-0400-A1; The National Mining Association (NMA); Posted 10/15/2015
4.4.3 Classification of Aquatic Systems - Lotic vs. Lentic

It is known that bioaccumulation of Se is significantly different in lentic and lotic systems (Adams et al.
2000). EPA differentiates between system types using EFs and residence time. As discussed in our
previous review, this makes sense in theory, however, the resulting data have substantial overlap,
indicating the differences are not that clear. It would be more appropriate to develop site-specific
criteria rather than create artificial groupings of waterbody types that mask the site-specific differences
so important to ensuring attainment of the tissue criterion.

5.1 Discussion of Final Criterion

EPA also derived nationwide lentic and lotic water column-based criteria to supplement the tissue-based
criteria. However, as discussed in Section 3.3 it is not possible or appropriate to derive a single
nationwide standard for water column-based criteria for only two water body types (lentic or lotic), and
such an effort is not supported by EPA's own analysis. While we agree that use of Equation 18 or other
approaches could be used to translate a water column criterion from the egg/ovary criterion, this type
of modeling should only be done with the appropriate site-specific data and not rely on default
generalized model parameters. In addition, as noted earlier, we strongly urge EPA to evaluate direct use
of tissue-based criteria in NPDES permitting, which would obviate the need for the back-calculation to a
water value, thus eliminating the potential errors and uncertainty that arise in such calculations.

EPA-HQ-OW-2004-0019-0413-A2; Federal Water Quality Coalition (FWQC); Posted
10/28/2015

1. Classification of Lentic and Lotic Waters

The draft criteria contain different water column levels for the two categories of lentic and lotic waters.
However, it is not entirely clear how waters will be classified as either lentic or lotic. The Agency
indicates that it has confidence in how it classified various waters that it included in its evaluation for
development of the criteria document. However, that does not tell regulated parties and other
stakeholders how those waters were classified, or how other waters would be placed into one or the
other category. EPA should provide clear definitions of the "lentic" and "lotic" terms that can be readily
applied, considering factors such as flow rate, hydraulic residence time, selenium speciation, and (if
appropriate) differing biological conditions.

III. CLASSIFYING WATERS AS LOTIC OR LENTIC

In the water column element of the criterion, there are separate values for lotic and for lentic waters.
Because those levels differ substantially, it is critical for the criterion structure to clearly differentiate
between those two types of waters.1 Unfortunately, EPA has not provided a clear differentiation, and

235

-------
the peer reviewers raised this as a concern. The Agency's responses to those comments shed no clearer
light on the question. If the water column levels are retained as elements of the criterion, then EPA
needs to provide a clear, quantitative way to determine whether a waterbody is classified as lotic or
lentic.

Here are some of the statements by peer reviewers on the lotic/lentic issue:

"Presumably the definitions for lotic and lentic systems would be based on residence time of
water or some related criteria, but the Draft Document does not contain an explicit definition of
either type of system. Back calculating from egg/ovary to muscle/whole body and then down
through trophic levels to derive allowable water column criterion for each of these types of
aquatic systems is not scientifically valid, because of the use of generic conversion factors and
broadly based trophic transfer factors. These generic terms do not incorporate site specific
information, including concentration dependent uptake kinetics and consideration for important
influencing factors (e.g., sulfate). The water based criterion is therefore, conservative and
variable. As evidence for this, the monthly average exposure value for lentic systems is 1.3 ug/L.
This value is at the upper end of background values for freshwater and may be exceeded even in
the absence of industrial inputs in areas receiving runoff from seleniferous soils. The value is
also lower than recently recommended lentic values based on similar analysis (2 and 2.1 ug/L
respectively (Deforest et al., 2104, BC MoE 2014). (pp. 51-52)

"Derivation of Enrichment Factors (EF) based on paired concentrations of selenium determined
in water and particulate would have been influenced by the practice of allowing data to be
paired if they were collected up to a year apart. In terms of application of EF to categories for
lentic and lotic systems it is difficult to judge because of the lack of specific criteria to distinguish
between the two types of systems in the Draft Document. While the US-EPA acknowledges the
importance of residence time for defining aquatic systems as either lentic or lotic, the criterion
for their initial assignment to each category is not apparent (Page 82). Despite statistical
comparisons that support their aggregation, it is very likely that lakes, reservoirs, ponds and
marshes will have vastly different selenium kinetics, and yet they are all designated as lentic
systems. Likewise, selenium uptake into aquatic food-webs of creeks, drains, washes, rivers and
streams may differ markedly. The wide range of variability in the aggregated categories (Figure
10, page 84) is compelling evidence in support of this point. Additional specific guidance is
required to distinguish between the two types of aquatic systems and the applicability of EFs for
each." (p. 118)

"I agree with several commenters that EPA must develop rigorous definitions of lentic and lotic
as guidance for regulators." (p. 125)

"Again related to implementation, is the question of whether the lotic and lentic water Se
criteria can be replaced by a different metric, such as residence time. In my opinion, the latter
would be worthy of further consideration by the EPA, although I wonder whether more reliable
categories could be developed based on existing datasets." (p. 128)

236

-------
"Lentic lotic clarification: The public comments express thoughtful concerns and practical
questions that can serve as prompts to draft additional guidance and supporting information, (p.
135)

EPA responded to each of these concerns in the same basic way, expressing this thought:

"Regarding the lack of clarity in the definitions of lotic and lentic water bodies.

Based on comments received, EPA examined the potential for residence time to classify the water
bodies in the database, but available data were extremely limited. EPA also examined each waterbody
to ensure that the available characteristics and sample location indicated that it was either a lentic or
lotic site. EPA's follow-up analyses ensured that sites such as run-of-the-river-reservoirs, an example
discussed in the reviewers comment, were not misclassified as lotic." (pp. 30-31)

In this response, EPA seems to be answering only part of the question. The Agency is explaining how it
tried to ensure that among the waterbodies covered in its studies, it did not mischaracterize a
waterbody as lentic when it should have been lotic, or vice versa. But the main concern being expressed
by the commenters is that guidance is needed to determine whether other waters - ones that are not
covered in the studies - are lentic or lotic. It does no good, in addressing that issue, for EPA to say that it
did work to ensure that studied waters were properly classified. More is needed.

In addition, the FWQC disagrees with EPA's statement that "sites such as run-of-the-river- reservoirs"
would be "misclassified as lotic." To the contrary, we believe that river systems and impoundments on
river systems should be clearly classified as "lotic." For many such large river systems, there is a large
flow between the related reservoirs each year, and the reservoirs are an integral part of the river
systems. Based on the amount of water moving through these river systems and related reservoirs, they
should be considered as lotic rather than lentic.

IV. CONCERNS ABOUT LENTIC VALUE

In addition to voicing concern about the Agency's lack of clarity in distinguishing between lotic and lentic
waters in the water column element of the criterion, the peer reviewers also identified specific
problems with the development of the lentic value, which is 1.3 ug/l. EPA has recognized these issues,
but the way in which they have addressed the problems does not adequately respond to the comments.

Here is what the peer reviewers said about the lentic value:

"The draft water column selenium criteria are 4.8 and 1.3 ng/L for lotic and lentic waters,
respectively. In general, I do not agree with the approach used by the EPA in deriving these
water column criteria. Although I do not agree with the approach, I do believe that the draft
criterion of 4.8 ng/L for lotic waters is reasonable and consistent with our understanding of the
range of Se bioaccumulation potential into fish across a wide range of lotic sites. However, for
the draft lentic Se criterion of 1.3 ng/L, the approach used by the EPA results in this criterion
being almost exclusively driven by data for two reference locations. This in turn is mostly due to
what I perceive as a flaw in the approach, where site-specific Se data in invertebrates and fish
are ignored and instead non-site-specific TTFs and CFs are applied that are inconsistent with the
site-specific data. This resulted in cases where erroneously high modeled Se concentrations in

237

-------
fish tissue are linked with low water Se concentrations (i.e., reference site concentrations), and
then these become the "drivers" for the draft lentic criterion of 1.3 ug/L." (p. 42).

"For the water column based criterion, two separate elements are prescribed in the Draft
Document: a monthly average and a separate element for intermittent (discontinuous)
exposures. Each of these is further delineated to apply to either lentic or lotic systems.
Presumably the definitions for lotic and lentic systems would be based on residence time of
water or some related criteria, but the Draft Document does not contain an explicit definition of
either type of system. Back calculating from egg/ovary to muscle/whole body and then down
through trophic levels to derive allowable water column criterion for each of these types of
aquatic systems is not scientifically valid, because of the use of generic conversion factors and
broadly based trophic transfer factors. These generic terms do not incorporate site specific
information, including concentration dependent uptake kinetics and consideration for important
influencing factors (e.g., sulfate). The water based criterion is therefore, conservative and
variable. As evidence for this, the monthly average exposure value for lentic systems is 1.3 ug/L.
This value is at the upper end of background values for freshwater and may be exceeded even in
the absence of industrial inputs in areas receiving runoff from seleniferous soils. The value is
also lower than recently recommended lentic values based on similar analysis (2 and 2.1 ug/L
respectively (Deforest et al., 2104, BC MoE 2014)." (pp. 51-52)

"A limitation of this modeling approach is that it ignored site-specific information on Se
bioaccumulation in fish and their diets. The EFs used were site-specific, but Se modeling up the
rest of the food chain and into fish was based on assumed model parameters. This becomes
particularly important when considering the data "drivers" for the draft lentic Se criterion of 1.3
Ug/L. This value is driven almost exclusively by data for two reference lakes (Badin Lake and High
Rock Lake, NC, USA). Badin Lake was reported to have a water Se concentration of 0.32 ug/L and
High Rock Lake a water Se concentration of 0.67 ug/L (Lemly 1985). For comparison, the mean
water Se concentrations translated from a fish egg/ovary Se criterion of 15.2 mg/kg dw were
0.54 ug/L for Badin Lake and 1.2 ug/L for High Rock Lake. The former falls between the water Se
concentrations reported for these two reference lakes and the latter almost equals the draft
lentic criterion of 1.3 ug/L. Since six fish species were assumed to represent each of these two
sites, these two reference sites are the drivers for the draft lentic Se criterion of 1.3 ug/L.

In addition to two reference sites being the drivers for the draft lentic Se criterion of 1.3 ug/L, the model
for translating a fish egg/ovary Se criterion of 15.2 ug/L to a water Se concentration does not appear to
be correct for these two sites. Although fish egg/ovary Se concentrations were not reported for Badin
Lake and High Rock Lake, muscle Se concentrations were. Those muscle Se concentrations were
reported on a wet weight basis and converted to a dry weight basis by assuming a moisture content of
75%. The muscle-to-egg CFs reported in Table 12 of the draft AWQC document were then used to
estimate fish egg Se concentrations. These estimated fish egg Se concentrations for the two reference
sites were, on average, less than one-half of the draft fish egg/ovary Se criterion of 15.2 mg/kg dw.
Further, the muscle Se concentrations at the references sites ranged from 2.3 to 5.8 mg/kg dw, which
are well below the draft muscle Se criterion of 11.8 mg/kg dw. The above demonstrates that the food

238

-------
web model for these two reference sites does not accurately reflect Se bioaccumulation potential at
these two sites and in fact greatly overestimates Se bioaccumulation potential."(pp. 93-94)

In response to these serious concerns with the lentic value, EPA says as follows:

"Regarding the influence of Badin and High Rock Lakes, EPA agrees that they have a strong
influence on the derived lentic value. Of the 44 site-species used for the lentic derivation via the
20th percentile value, 12 of the lowest 13 values are for Badin and High Rock. These lakes each
have one EF, but each of its EFs is used six times, once for each of six fish species. The
particulate concentrations measured in both of these lakes are near the median observed in
EPA's lentic database, but their water concentrations are among the lowest. As a result of the
peer reviewer's comments, EPA completed a reanalysis of the data to remove any
overweighting of a few key high and low end sites in the calculations. To account for
overweighting, EPA used one fish species per site - the species most sensitive to selenium
bioaccumulation, to yield an appropriately protective water column criterion element." (pp. 93-
94)

While it is helpful that EPA took action to address the overweighting problem, that is not a complete
response to the concerns being raised. The peer reviewers indicated several problems with the way in
which the lentic value was calculated, but the Agency addressed only one small part of the overall
problem. Until it has considered and taken action to remedy the issues that have been identified, a
lentic value should not be adopted as an element of the selenium criterion.

1. We recognize that this clear differentiation is difficult for some waterbodies. The uncertainty
that results is yet another reason why it is important for EPA to clearly state the primacy of the
fish tissue elements of the criterion over the water column elements.

239

-------
< ¦ 'lull iii <
-------
95% of aquatic genera, employing fish as the most sensitive taxa. EPA expects the water column values
developed through bioaccumulation modeling to be protective against selenium impacts in the most
bioaccumulative food web in 80% of lentic and lotic water bodies, respectively.

Regarding the number of studies upon which this approach is ultimately reliant, EPA evaluated 80
studies on selenium toxicity to aquatic organisms, identified in appendices C, D, and E; and in section
3.1.3. The 13 GMCVs (not including the 2 "placeholder" GMCVs used to fulfill missing taxonomic
minimum data requirements [MDRs] included in the sensitivity distribution [SD]) were calculated from
15 SMCVs, which were calculated from 19 chronic values obtained from 24 studies. An additional 21
non-reproductive toxicity values were obtained from 20 studies for 10 species, including 5 species that
were not used in the SD. Fish reproductive and non-reproductive summaries are included in Appendix C
and D, respectively, and were used to demonstrate that the egg-ovary based criterion protects against
both reproductive and non-reproductive effects in aquatic organisms. An additional 21 toxicity values
from 22 studies encompassing 18 species, seven of which were not included among the reproductive or
non-reproductive studies listed above, were evaluated and are included in Appendix E (other data).
Three field studies with multiple species were also evaluated qualitatively to assess the relative
sensitivity of Cyprinidae to selenium, and are included in Appendix E.

Finally, 11 studies encompassing 11 species were qualitatively evaluated to assess selenium nutritional
requirements and are included in Appendix E. While EPA agrees that additional studies are always
desirable, the selenium criterion is based upon a substantial dataset.

Regarding the comment that notes that there are a few reproductive studies for which ECi0s can be
directly calculated as muscle or whole-body concentrations measured during the study, EPA opted to
use the egg-ovary ECio from such studies and then convert the SMCV or GMCV from egg-ovary to whole
body or muscle using all available whole body/egg-ovary or muscle/egg-ovary ratios. EPA used the
current approach because it is also scientifically defensible.

There were over 3000 whole- body fish tissue measurements that could be coupled with water
concentrations measured. The 3000 measurements include same- site values of (a) different species
measured at the same time, (b) the same species measured at different times, and (c) different species
measured at different times.

4.7 Responses to other issues brought up in these comments

Regarding the comment on the biokinetic approach, EPA's kinetic analysis was presented in Appendix G
of the 2014 draft. It was similar to the work of Brix and DeForest (2008). It has since been revised to
make it even more similar in structure to Brix and DeForest (2008). It continues to have a similar
response time as Brix and DeForest because the limiting kinetic rate used in both Brix and DeForest and
the 2014 Appendix G is based on the same fathead minnow study by Bertram and Brooks (1986).

Regarding the biokinetic model and intermittent criteria, the kinetic analysis of Appendix G of the 2014
draft specifically addressed intermittent exposure. When applied to intermittent exposure, the analysis
demonstrated the protectiveness of the 30- day averaging period that EPA is recommending.

241

-------
The intermittent criterion is a rearrangement of the water chronic criterion and its kinetic model-based
30-day averaging period. The 2014 intermittent criterion thus hinged on the kinetic model of the 2014
Appendix G. In the final 2016 criterion document Appendix G is now Appendix J. Appendix J uses a
single set of kinetic parameters. Its different figures address different assumptions about the time
series of water concentrations.

Regarding the comment that "EPA's approach suffers a weakness in relying on model-predicted
fish/water concentration ratios/' rather than on the measured fish /water ratios, as noted above, such
an empirical fish/water ratio approach was used by DeForest et al in 2014. Recalculating the DeForest
et al water concentrations based on EPA's 2016 15.1 mg/kg dw bw egg-ovary criterion element value
(instead of the authors' proposed 20 mg/kg egg-ovary value), the outcome of the DeForest analysis
matched EPA's 2016 lentic water column element criterion value exactly, and was within 1 ng/L of the
EPA's 2016 water column element criterion value. This information further corroborates and
substantiates the validity of EPA's bioaccumulation modeling approach.

4.7 Specific comments

EPA-HQ-OW-2004-0019-0392; Anonymous Commenter; Posted 10/14/15

Second, irrespective of the soundness of the mathematical structure, EPA's approach suffers a weakness
in relying on model-predicted fish/water concentration ratios for the 69 sites in its database, rather than
on the actual measured ratios. This seems peculiar. Reliance on model predictions can be appropriate
when observations are lacking. But there appears to be no good reason for opting for model-predicted
fish/water concentration ratios for sites where the actual fish and water concentrations have been
measured.

EPA-HQ-0W-2004-0019-0409-A2; OC Public Works; Posted 10/20/2015
4. Deviation of Established Aquatic Life Criterion Development Process

In Figure 3.9 on page 84, the probability distributions with lentic and lotic conditions no longer
correspond to the 20th percentile protection as claimed. Due to the large uncertainties in the
calculation of water column numbers, as explained above, the lower end data points simply represent
water bodies with the lowest observed water column selenium concentrations, not necessarily fish with
the lowest egg/ovary levels. Therefore these 20th percentile cutoff points do NOT protect 80% of the
fish species. More importantly, USEPA selected the lowest water column level for each site with more
than one fish species, while higher water column numbers at these sites corresponding to other species
were not represented at all in Figure 3.9. This is a clear violation of the established protocol and has
resulted in artificially lowered water column criteria. This approach is not justified and Figure 3.9 needs
to be replotted.

242

-------
EPA-HQ-OW-2004-0019-0388-A2; J. R. Simplot Company; Posted 10/14/2015
A.2. Conservatism in Calculating Lentic and Lotic Values

[Page 80, Section 3.2.5]

This section presents EPA's approach to deriving protective water column concentrations for lentic and
lotic categories. Translated aqueous concentrations are derived for each site evaluated. The resulting
water quality lotic and lentic criterion values are derived from the 20th percentile of translated water
column values using the egg/ovary criterion as the target criterion. Table 3.13 (pages 82-83) of the
document compiles the dataset for lotic and lentic translated chronic criterion values from a number of
different field studies. The primary issue that arises in review of this section is that both the lentic and
lotic values are very low and may lead to conclusions of impairment when in fact none exists.

As noted in Simplot's previous comments on the 2014 Draft, EPA has continued to use the 20th
percentile of a distribution of chronic criterion values for the lentic and lotic systems, which are each
individually based on an EC10 response. This is an example of conservatism that is compounding
uncertainty by ultimately deriving a value that is too low.

EPA-HQ-0W-2004-0019-0401-A1; Colorado Wastewater Utility Council; Posted 10/15/2015

3.4 Derivation of Protective Water Column Concentrations

3.4.1 Use of Probability Distribution of Water Column Concentrations

As stated previously, we do not agree with the general approach used to derive the two default water
column concentrations meant to be protective offish-tissues. Table 3.13 on page 82 of the 2015 draft Se
criteria document presents site-specific data for 20 lentic and 33 lotic species-site combinations and
includes site-specific enrichment factors (EF), species-specific whole-body to egg/ovary CF, and
composite trophic transfer function (TTFcomposite) values based on expected trophic levels at the site.
As will be discussed in Section 4.4 of our review, use of site-specific EFs and TTFs is critical when deriving
a translated water quality criterion.

The final water quality criteria selected for lotic and lentic systems were based on Figure 3.96, page 84
of the 2015 draft Se criteria document. This figure is a probability distribution of the water column
concentrations for lentic and lotic sites after being translated from the final egg/ovary tissue criterion
(data from EPA Table 3.13). As we stated in our previous review, it appears this figure and the choice of
a 20th percentile were used as if these represented sensitivity distribution curves, in which
protectiveness can be predicted based on selecting a certain percentile value from the curve. However,
these are not sensitivity distribution curves - in fact, as we pointed out last year, each value on these
curves is a translated water concentration value that was specifically calculated to be protective of the
egg/ovary tissue criterion. Therefore, each and every point on this graph (and those values in the far-
right column of Table 12) is protective of the egg/ovary criterion based on the site-specific parameters
at that site (given site-specific EF, CF, TTF).

243

-------
Thus, the analysis by EPA actually demonstrates that water concentrations that are protective of the
tissue criterion can range from 0.23 ng/L to 50.4 ng/L for lentic sites, and 1.2 ng/L to 40.6 ng/L for lotic
sites, depending on the site-specific factors used in Equation 18. By selecting a 20th percentile value to
use as the water column criteria, as EPA did, 80% of the sites in Table 3.13 would be overprotected, and
20% of the sites would be underprotected, resulting in a water column criterion that is wrong virtually
100% of the time - with the exception being the particular sites that fell right on the 20th percentile
lines.

We agree that use of Equation 18 (or a bioaccumulation factor, as discussed later) to translate a water
column criterion from the egg/ovary criterion is one of many valid approaches, but only on a site-
specific basis. In addition, the dataset used to derive these water column values is very limited,
especially when considering this is supposed to be the basis for a national criterion. Using additional
data from several GEI projects (and other GEI updates previously discussed in this document), paired
data were available for 47 additional sites, which more than doubles the database used by EPA. Using
this expanded dataset, we recalculated the lotic water column value using EPA's method and derived a
value of 4.2 ng/L, roughly 35% higher than EPA's default. Protective values for these lotic sites range
from 1.18 ng/Lto 81.03 Hg/L.

While we still believe this method for calculating a nationwide water column value is over-conservative,
this recalculation demonstrates how additional data can substantially affect the final value. If EPA
intends to use this method to calculate a nationwide water column value, there should be more effort
put into collecting site-specific data to include in the database, and not just relying on a few published
literature values, most of which were data collected more than 20 years ago.

EPA-HQ-0W-2004-0019-0400-A1; The National Mining Association (NMA); Posted 10/15/2015

3.4 Derivation of Protective Water Column Concentrations

3.4.1 Use of Probability Distribution of Water Column Concentrations

As stated previously, we do not agree with the general approach used to derive the two default water
column concentrations meant to be protective offish-tissues. Table 3.13 on page 82 of the 2015 draft Se
criteria document presents site-specific data for 20 lentic and 33 lotic species-site combinations and
includes site-specific enrichment factors (EF), species-specific whole-body to egg/ovary conversion
factors (CF), and composite trophic transfer function (jjFcomP°slte) values based on expected trophic
levels at the site.

244

-------
curves is a translated water concentration value that was specifically calculated to be protective of the
egg/ovary tissue criterion. Therefore, each and every point on this graph (and those values in the far-
right column of Table 12) is protective of the egg/ovary criterion based on the site-specific parameters
at that site (given site-specific EF, CF, TTF).

Thus, the analysis by EPA actually demonstrates that water concentrations that are protective of the
tissue criterion can range from 0.23 ng/L to 50.4 ng/L for lentic sites, and 1.2 ng/L to 40.6 ng/L for lotic
sites, depending on the site-specific factors used in Equation 18. By selecting a 20th percentile value to
use as the water column criteria, as EPA did, 80% of the sites in Table 3.13 would be overprotected, and
20% of the sites would be underprotected, resulting in a water column criterion that is wrong virtually
100% of the time - with the exception being the particular sites that fell right on the 20th percentile
lines.

We believe there are a number of valid approaches to translate an egg/ovary criterion to a water
column value, such as Equation 18 or a bioaccumulation factor (as discussed later), but only if the proper
data are used in the equations. The key is to properly characterize the base of the food chain, which for
lotic systems includes incorporating sediment and/or periphyton data, rather than relying on suspended
particulate Se data. Other modeling approaches that also capture the various aspects of the system
could also be used to translate a water column criterion.

In addition, the dataset used to derive EPA's water column values is very limited, especially when
considering this is supposed to be the basis for a national criterion. Using additional data from several
GEI projects (and other GEI updates previously discussed in this document), we recalculated the lotic
water column value using EPA's method and derived a value of 4.2 ng/L. Paired data from GEI projects
were available for 47 additional sites, which more than doubles the database used by EPA. Protective
values for these lotic sites range from 1.18 ng/L to 81.03 ng/L. While we still believe this method for
calculating a nationwide water column value is over-conservative, this recalculation demonstrates how
additional data can substantially affect the final value. If EPA intends to use this method to calculate a
nationwide water column value, there should be more effort put into collecting site-specific data to
include in the database, and not just relying on a few published literature values, most of which were
data collected more than 20 years ago.

In addition, we also concur with comments provided by (API 2015) regarding the counterintuitive results
calculated when using constant EFs and TTFs. We would agree that the alternative approaches they
provide for these calculations would result in more appropriate protective water column criteria.

245

-------
1 ii 11 ii 11 ^ iiii 'i. ii ! 1 ii nil in hi* in II ¦> iriiviii I - ii i nil II - ii ik-1 hi I'll

IIiiIi< i mili< hi I
-------
Brix and DeForest (2008). It continues to have a similar response time as Brix and DeForest because the
limiting kinetic rate used in both Brix and DeForest and the 2014 Appendix G is based on the same
fathead minnow study as referred to in the comment: Bertram and Brooks (1986).

The intermittent criterion is a rearrangement of the water chronic criterion and its kinetic model-based
30-day averaging period. The 2014 intermittent criterion thus hinged on the kinetic model of the 2014
Appendix G. In the revised document Appendix G is now Appendix J.

4.8. Specific comments

EPA-HQ-OW-2004-0019-0392; Anonymous Commenter; Posted 10/14/15

Fourth, EPA has indicated that the water criteria were derived from site median water concentrations.
To accord with how EPA derived them, each of the two water criteria must be expressed as a site
median water concentration, not as a site 30-day once-in-three-year concentration. If EPA wants a
criterion that applies to once-in-three-year concentrations, then EPA's derivation must use water
concentrations corresponding to the upper tail of observed concentrations at each site having an EF
measurement.

EPA-HQ-OW-2004-0019-0387-A1; Colorado Department of Public Health & Environment;
Posted 10/14/2015

While the WQCD appreciates a simplified approach such as that used to develop the intermittent
criterion, there are other methods available (e.g., biokinetic-based model [Brix and DeForest 2008]) that
may be more relevant and scientifically valid. Did EPA consider use of any other type of model? The
draft criteria document would benefit from a detailed discussion on this topic, including an evaluation of
the benefits of using one approach over the other. It may be possible to recommend multiple options
for acceptable approaches that would answer the same question, as both regulators and the regulated
community appreciate knowing criteria are as precise and relevant as possible.

The WQCD looks forward to seeing implementation guidance and additional information from EPA on
this element of the criterion.

247

-------
EPA-HQ-0W-2004-0019-0401-A1; Colorado Wastewater Utility Council; Posted 10/15/2015
3.4.4 Intermittent-exposure Element

The intermittent exposure component of the water column-based criterion attempts to address pulses
of elevated Se concentrations that could contribute to chronic effects. The equation to calculate the
intermittent exposure criterion seems to be an oversimplification as it is essentially just a
rearrangement of the equation to calculate a 30-day average concentration. A more appropriate way to
determine limits for short-term elevated pulsed Se exposures would be to use a scientifically-based
biokinetic model as discussed in Appendix G of the 2014 draft Se criteria document, and described in
more depth in DeForest et al. (in press). We have also provided further recommendations on calculation
of this element in our previous review (GEI 2014a).

EPA-HQ-0W-2004-0019-0400-A1; The National Mining Association (NMA); Posted 10/15/2015
3.4.2 Intermittent-exposure Element

248

-------
Comment Category 4.9 - Comments on EPA's Validation of the

Train slat ion Equation (Equation 18)

¦ummary

This section contains comments on the method EPA used to validate the translation equation.

)onse to comments

4.9 Response concerning the validation of the translation equation

Regarding the translation of the tissue criterion to water, EPA re-evaluated conversion factors used and
refined the approach to reflect the most phylogenetically proximate data in its criterion calculations (see
section 6.2 of the 2016 final criterion document), strengthening its previous draft analyses. As noted
above EPA also re-evaluated the lentic and lotic water column values, and developed values that are
based on the best available, scientifically defensible science to yield water column values that are
protective of aquatic life based on available data.

The water-column element is a translation of the egg-ovary element and thus is intended to provide the
same level of protection as the egg-ovary element. However, some level of uncertainty in translating the
egg-ovary element to the water-column element is unavoidable. To address this uncertainty, EPA chose
the 20th percentile of translated water-column values using the most bioaccumulative food web present
at each site to select a protective national water column criterion element. Limitations of available data
only allowed translation of the egg-ovary criterion element at 69 unique aquatic sites. The 20th centile
was selected because it results in a low probability of false negatives, (i.e., low probability of failure to
indicate exceedance of the egg-ovary criterion element via the water column element).

4.9 Specific comments

EPA-HQ-OW-2004-0019-0418; Anonymous public comment; Posted 11/02/2015

The Selenium aquatic life criterion is proposed to be reduced from the current 5 g/L water to new
criteria of 1.2 f g/L in lentic water and 3.1 g/L water in lotic water. These significant reductions of 76%
and 38% respectively are based on the correlation in Figure 6.4 on page 125 of the document. That
figure shows a chart that has a stated coefficient of correlation (r) of 0.81. This means it has a coefficient
of determination (r2) of 0.65. Thus only 65% of the variance of one variable can be explained by the
other variable.

This chart with the low 65% factor of predictability is being used to make significant reductions in a
criterion that will eventually be applied to direct and indirect dischargers of waste water containing
selenium. There is no practical large-scale waste water treatment method to achieve the proposed 1.2
and 3.1 g/L levels if they were ever imposed by permits on direct and indirect waste water discharges.

249

-------
Additional testing should be conducted in order to increase the model's coefficient of determination (r2)
before making such a significant reduction in a criterion that will have real-world impacts on direct and
indirect waste water discharges. Thank you for letting me comment.

250

-------
> "lull iii <
-------
EPA-HQ-OW-2004-0019-0386-A2; Asarco LLC; Posted 10/13/2015

Like NAMC, Asarco supports the option for development of site-specific criteria based on appropriate
studies, to reflect instances of naturally elevated selenium and/or the presence of less sensitive
populations.

EPA-HQ-OW-2004-0019-0413-A2; Federal Water Quality Coalition (FWQC); Posted
10/28/2015

One concern is the use of species sensitive distributions (SSDs) to generate predictions and quantify
stressor-response relationships. While we support the use of SSDs when used to generate relative
species sensitivities, it is important to recognize that SSDs are predictive toxicological responses of a
range of test organisms as well as some surrogate species. Since concern about selenium is not based on
its acute or chronic toxicity of its inorganic form in the water column per se, but rather, the
accumulation of its organic form via dietary exposure that is subsequently passed to eggs resulting in
juvenile deformities and/or embryonic mortality, the automatic incorporation of SSDs as default values,
without consideration of other information, is inappropriate. We suggest that one should have the
opportunity to collect and apply site-specific fish tissue and toxicity information, which could be used
instead of relying entirely on simple SSD toxicity relationships. This approach will address key variables,
including, but not limited to: geologic substrate, temporal differences, hydrological and chemical
conditions, watershed land-use, geomorphology, and natural vegetation.

EPA-HQ-0W-2004-0019-0401-A1; Colorado Wastewater Utility Council; Posted 10/15/2015
1.1.1 Incorporation of Background Conditions into Site-specific Criteria

Appendix K of the 2015 draft Se criteria document, which describes methods for deriving site-specific
criteria, does not discuss inclusion/consideration of data from reference or background sites. This is
especially critical at sites with background levels of Se that would be considered elevated, comparatively
speaking. In Appendix A, Examples 1-6 include calculation of site-specific criteria that includes
background Se concentrations of 5.0 ng/L. Working through the calculation results in a site-specific
water column criteria that are always lower than the original background concentration. It is unclear
how these calculations could be applied to sites with naturally elevated background Se and result in a Se
criterion that would be appropriate for that site.

As discussed in Section 4.1 of this review, ambient site-specific criteria may be appropriate in many parts
of the country containing underlying geology with elevated Se levels. This naturally elevated Se may lead
not only to elevated water concentrations, but also to naturally elevated fish tissue concentrations,
resulting in the need for not only site-specific water column criteria but also a need for an approach that
acknowledges there could be site-specific tissue criteria. In our previous review a more in depth
discussion of this issue was provided along with specific examples (GEI 2014a).

252

-------
4.5 Site-specific Criteria

4.5.1 Deriving Site-specific Water Concentration Values from the Egg/Ovary Criterion

Appendix K of the 2015 draft Se criteria document describes a methodology to derive site-specific
criteria using a mechanistic modeling approach. However, Appendix K seems to simply be a more
detailed presentation of information presented in the main text. It does not describe the process or data
requirements that a state or tribe would need to follow/generate in order to have their site-specific
criteria considered by EPA. In our previous review we discussed several questions and concerns that
should be addressed to enable states and tribes to appropriately develop site-specific criteria (GEI
2014a).

EPA-HQ-0W-2004-0019-0400-A1; The National Mining Association (NMA); Posted 10/15/2015
1.1.1 Incorporation of Background Conditions into Site-specific Criteria

Appendix K of the 2015 draft Se criteria document, which describes methods for deriving site-specific
criteria, does not discuss inclusion/consideration of data from reference or background sites. This is
especially critical at sites with background levels of Se that would be considered elevated, comparatively
speaking. In Appendix A, Examples 1-6 include calculation of site-specific criteria that includes
background Se concentrations of 5.0 ng/L. Working through the calculation results in a site-specific
water column criteria that are always lower than the original background concentration. It is unclear
how these calculations could be applied to sites with naturally elevated background Se and result in a Se
criterion that would be appropriate for that site.

4.5 Site-specific Criteria

4.5.1 Deriving Site-specific Water Concentration Values from the Egg/Ovary Criterion

253

-------
EPA-HQ-OW-2004-0019-0402-A2; The Stakeholders Implementing TMDLs in the Calleguas
Creek Watershed; Posted 10/15/2015

For the Development ofSSOs, the Methodology to Derive Water Concentrations from Whole-Body Fish
or Muscle Tissue Criterion, Rather than Egg/Ovary Criterion, Should Be Clearly Presented.

The Stakeholders strongly support the flexibility provided by the 2015 Draft Selenium Criterion on Page
95, Section 5 to adopt SSOs which use a peer-reviewed model to derive water concentrations from fish
whole-body or muscle tissue criterion rather than egg/ovary criterion:

"Using either the EPA national recommended egg-ovary, whole-body, or muscle criterion
concentration element or a site-specific egg-ovary, whole-body, or muscle criterion element,
translation of the fish tissue criterion to a water concentration can be performed in a manner
that accounts for site-specific conditions."

While the Stakeholders acknowledge that selenium toxicity occurs primarily through transfer to the eggs
and subsequent reproductive effects, using a peer-reviewed model to derive water column
concentrations from fish whole-body or muscle tissue criterion is necessary during development of SSOs
because the site-specific validation of the model may prove to be infeasible if fish eggs and/or ovaries
are needed for model validation. In many areas in Southern California, obtaining fish eggs/ovaries is
infeasible, whereas collection of whole-body fish tissue can be consistently expected and is therefore
significantly more practical.

For these reasons, efforts in Southern California have focused on whole-body fish tissue for regulatory
purposes, including development of site-specific objectives. In Southern California (and most likely
throughout the arid western United States), all model validation efforts will be based on whole-body fish
tissue. As a result, it may be necessary and/or preferable for the model to be validated using fish whole-
body or muscle tissue data. Therefore, SSOs should not be limited to relying on calculating water column
concentrations from egg/ovary and the 2015 Draft Selenium Criterion should explicitly provide the
process for such a translation in Appendix K. Currently, Appendix K only identifies the steps necessary to
translate egg/ovary concentrations into water column concentrations (using both the mechanistic
model and a field derived bioaccumulation factor (BAF)). While the steps may be very similar, identifying
the process specifically for whole-body fish tissue would aid in making the appropriateness of such a
translation more explicit, as well as supporting local efforts to implement such a translation.

Requested Actions:

Revise Section 1.1 and 1.2 of Appendix K to include the process for deriving a site-specific water column
concentration value from whole-body and/or muscle tissue criterion.

EPA-HQ-OW-2004-0019-0415-A1; Idaho Department of Environmental Quality (DEQ); Posted
10/28/2015

The criterion proposed is fish-centric. While there is good reason for this, i.e. greater sensitivity of fish
than other aquatic taxa, this can be problematic for states like Idaho whose fish species diversity is

254

-------
naturally low. The national criterion is based on a number of fish species that are not native to Idaho,
but excluding those non-native species in a criterion recalculation may not leave enough data to do a
calculation. A broader species sensitivity distribution, including less sensitive macro-invertebrates might
address this but would yield a quite different criterion that could leave the more important and more
sensitive fish suffering. How does EPA view this conundrum? What do you recommend for waters with
limited fish species diversity? Would EPA accept a criterion be based on a single native 'most-sensitive'
fish species?

Tailoring the criterion to site-specific aquatic communities should be built into the criterion. EPA
acknowledges different sites will have different communities with different species selenium
sensitivities (e.g. lentic and lotic, warm versus coldwater), yet of necessity the national criterion throws
all species into one sensitivity distribution. While recalculation of a site-specific criterion is offered as a
route to an alternative, selectively dropping species not present at a site, this will require rulemaking
and EPA action over and over. We suggest EPA build into the criterion, much like the translation of the
root egg-ovary criterion to other bases, a preset recalculation procedure that would not require
rulemaking to tailor the criterion to species present at a site. For example, like the lentic-lotic bifurcation
of the water criterion, EPA could develop a cold-warm water fisheries bifurcation as well.

We recommend EPA make an explicit statement that implementing the chapter 6 "site-specific criteria"
is a translation of the national criteria using site-specific conditions, and as such can be implemented
without additional rulemaking. Site-specific translations have been around for decades with the pH and
temperature dependent ammonia criteria, hardness-dependent metals criteria, and more recently the
BLM-based copper criteria. Such "on-the-spot" calculations of national criteria that are dependent on
site-specific data are quite different administratively than are "site-specific criteria," which must be
adopted through rulemaking. If the chapter 6 procedures are intended to be "site-specific translation"
— an approach we strongly recommend — then consistent use of the term "site-specific translation"
and removal of all instances of "site-specific criteria" throughout the document would be helpful.

EPA-HQ-0W-2004-0019-0417-A1; Utility Water Act Group (UWAG); Posted 11/02/2015

a. EPA should clarify what is meant by "ecological significance".

UWAG requests EPA clarify what it means by "ecological significance" and identify the ecological
parameters to consider in relation to identifying the appropriate target fish species in aquatic systems
with fish of unknown selenium sensitivity and bioaccumulation potential. See Draft Criterion at K-10.
Ecological significance could refer to a variety of traits, such as trophic level or habitat preferences.
Clarity on this is important to ensure appropriate traits are considered. Absent clarity, the default may
be to use trophic level as the deciding trait for determining ecological significance, which may be
inappropriate because, unlike mercury, selenium levels normally do not increase with fish trophic level.

EPA should include flexibility to use data from species other than the targeted species.

UWAG recommends that EPA include flexibility to collect and use data on more than the targeted fish
species to derive a water-column concentration. Such flexibility is necessary because there is likely to be

255

-------
situations where collecting representative data on the most sensitive species is not possible. For
example, a warm water stream may contain three or four species of sunfish (family Centrarchidae),
including bluegill. UWAG believes that documenting the selenium tissue level of the related sunfish
species should be allowed. Such data would provide useful information on the potential range of
selenium concentrations in fish belonging to the genus Lepomis, which in turn would improve the
validity of the derived water-column concentration.

If EPA revises the Draft Criterion to include the requested flexibility, we recommend that EPA also
provided guidance on how the data from multiple species should be evaluated, particularly if used to
determine compliance with the criterion. We recommend the following approach, consistent with EPA's
recommended approach for the methylmercury human health fish tissue criterion:

• Calculate a sample size-weighted geometric mean selenium concentration for each genus.

• If fish from only one genus are targeted, use the geometric mean concentration to derive a
water-column concentration or to compare to the appropriate tissue criterion.

• If fish from two or more genera are targeted, calculate the sample size-weighted geometric
mean concentration for all genera combined and use it to derive a water-column concentration
or to compare to the appropriate tissue criterion.

The methods for deriving water-column concentrations from fish tissue concentrations must take into
consideration the variability of the data.

In Appendix K of the Draft Criterion, EPA discusses two options for translating the egg-ovary final chronic
value (FCV) into site-specific water-column concentrations - (1) a mechanistic model of
bioaccumulation, and (2) a site-specific, field-derived bioaccumulation factor (BAF). Both are serviceable
methods, but they could benefit from additional consideration of the variability in the data as well as
clarification and flexibility to ensure derivation of an appropriate water-column concentration.

UWAG's supplemental comments focus on EPA's guidance for developing a site-specific criterion by
translating fish tissue concentrations into a water-column concentration as presented in Appendix K of
the Draft Criterion.2 UWAG also identifies topics that are important for EPA to address in the
forthcoming implementation guidance. Finally, UWAG is clarifying some information presented in its
October 13, 2015 comments.

EPA-HQ-OW-2004-0019-0387-A1; Colorado Department of Public Health & Environment;
Posted 10/14/2015

On Pages K-10 and K-12, the concept of "ecological significance" for choosing target species is
presented. Does EPA plan to provide guidance for how "ecological significance" might be determined?
There are multiple ways this concept could be considered (e.g., endangered species, keystone species,
species with economic value) and it would be useful for EPA to present some acceptable options for
considering this approach.

Can EPA provide information regarding the use of median, geomean, and average in the analyses? That
is, what was the rationale for using one statistic versus another for different components of the criteria

256

-------
derivation (e.g., ITF, EF, water translations, etc.)? Similarly, what statistics does EPA recommend for site-
specific calculations? For instance, on page K-19, EPA suggests "calculating the mean or median of all the
ratios." It would be helpful to have guidance on what statistic EPA finds most appropriate for each
calculation that would be needed to develop site-specific criteria.

EPA-HQ-OW-2004-0019-0411-A2; American Petroleum Institute (API); Posted 10/28/2015
3.2 Site-specific Criteria

3.2.1 Deriving Site-specific Water Concentration Values from the Egg/Ovary Criterion

Appendix K of the 2015 draft Se criteria document describes a methodology to derive site-specific
criteria using a mechanistic modeling approach. However, Appendix K seems to simply be a more
detailed presentation of information presented in the main text. It does not describe the process or data
requirements that a state or tribe would need to follow/generate in order to have their site-specific
criteria considered by EPA. This type of information is very important to those that need to develop site-
specific criteria and should be provided in a timely fashion in implementation guidance.

EPA-HQ-OW-2004-0019-0376-A2 [Comment 0381-A3 is a duplicate of 0376-A2]; Texas
Commission on Environmental Quality (TCEQ); Posted 09/24/2015

D. EPA is requested to provide guidance on the development of site-specific criteria to account for
regional differences in naturally-occurring selenium.

EPA states in the draft criterion that underlying geology may contain deposits of selenium-rich materials,
such as organic-enriched, sedimentary shales and petroleum source rocks. In locations where elevated
selenium is naturally occurring, native fish species may have lower sensitivities to these naturally
occurring concentrations. The draft criterion does not account for variability in selenium levels based on
region or aquatic species' sensitivity. Guidance for the development of site-specific bioaccumulation
factors for selenium is needed to account for regional differences and elevated background
concentrations in lotic and lentic environments. Site-specific studies evaluating bioaccumulation
certainly seem possible; however, it is unclear how criterion will affect permit reviews and effluent limits
until implementation guidance is provided by EPA.

257

-------
Comment Category 6.1 - Comments Concerning Correct lions and Typos
6.1 Summary

Several typos and corrections were noted by commenters.

6.1 Response to comments

6.1 Response to concerns with typographical corrections

EPA has noted suggested edits and corrected the typographical errors. EPA has clarified phrases and
terms to enhance readability.

Regarding the reviewer's recommendation to periodically redefine acronyms in the document, EPA has
adopted that practice for those terms that are that are heavily used in the document.

Regarding the missing "dw," it has been added.

6.1 Specific comments

EPA-HQ-OW-2004-0019-0387-A1; Colorado Department of Public Health & Environment;
Posted 10/14/2015

Typos

• Page 4, second to last line: Consider replacing "... is the mining of metals, minerals and
refinement and use ..." with "...is the mining of metals and minerals, as well as refinement and
use... ".

• Page 5, second line: Correct the reference error.

• Page 9, third line in Section 2.2.1: Remove comma before "to transformed inorganic and organic
species ... "

• Page 34, third line: The sentence starting with "This species, which is listed as endangered ..." is
incomplete.

• Page 41, notes: Add space between "Concentration" and "reported".

• Page 46, third line in Section 3.1.5: Should this be 9 fish genera and 10 fish species?

• Page 50, third line: instead of "egg-ovary concentrations to whole-body concentrations", this
should read "egg-ovary concentrations to muscle concentrations".

• Page 53, sixth line: Delete the space between "arthropods" and the comma.

• Page 54, first line of Last paragraph: The number should be 15 instead of 154.

• Page 64, third line in Section 3.2.2: There is a period missing after "literature as described
above".

• Page 80, third line: Change "EPA uses" to "EPA used".

• Page 98, last line: "GEI Associates (2008)" is not in the references.

• Page 101, first line in Section 6.1.2.3.2: The Kennedy et al. (2000) study has not been discussed
yet. Consider adding something like (see 6.1.5.1.2 below) to help guide the reader.

• Page 110, second to last line: Delete the apostrophe from "ECio's".

• Page 131, third line from end of the second paragraph: Delete the comma after the period.

258

-------
• Page 159: There are three "GEI Consultants. 2008" references. Consider using letters (e.g.,
2008a, 2008b, and 2008c) to clarify which reference is intended in the text.

• Page D-2: Italicize the scientific names.

• Page K-23, third line from the bottom: The word "alsites" appears to be a typo.

• Pages K-30 and K-31: The allowable water column concentrations (4.2 and 0. 9 ng/L) used in the
examples are different from the proposed water column criteria (3.1 and 1.2 ng/L). Is this
intentional or a typo? If intentional, how were the values in the examples derived?

EPA-HQ-OW-2004-0019-0412-A2; Utility Water Act Group (UWAG); Posted 10/28/2015
3.1.5.3 Fish Muscle Criterion Element Concentration

On page 50, a correction should be made in the 2nd sentence:

Table 3.7 shows the conversion of reproductive-effect egg-ovary concentrations to whole-body
concentrations.

The word "whole-body" should be changed to "muscle".

EPA-HQ-0W-2004-0019-0417-A1; Utility Water Act Group (UWAG); Posted 11/02/2015
a. EPA should make the following corrections.

On the middle of page K-22 of the Draft Criterion (§ 1.2.5.2), EPA states:

"Alternatively, a state or tribe could derive a CF value by verifying the statistical relationship
between paired particulate and water concentrations through linear regression, calculating the
ratios of paired Cegg-ovary to Cwhoie-body, and then taking the median ratio of the paired values as the
CF (see Appendix B)."

UWAG believes that highlighted language is incorrect and should be revised as follows:

"Alternatively, a state or tribe could derive a CF value by verifying the statistical relationship
between paired egg or ovary and whole-body concentrations through linear regression,
calculating the ratios of paired Cegg-ovary to Cwhoie-body, and then taking the median ratio of the
paired values as the CF (see Appendix B)."

At the bottom of page K-23 of the Draft Criterion (§ 1.3), there is a typographical error as highlighted
which EPA needs to correct:

Though not required, the effectiveness of effluent limits and waste load alsites of selenium that are
based on water concentration values derived from the egg-ovary FCV should be confirmed whenever
practical using appropriate fish tissue assessment methods.

259

-------
1 ii11ii11^ iiii 'i. ii 1 1 ii11ii11 iiii II ' «ii' mil • II ii iiin:-1iii-Ii

¦ummary

Several commenters requested better definitions of specific terms.

,)onse to comments
6.2 Response to requests for clarification
EPA has clarified phrases and terms to enhance readability.

¦pecific comments

EPA-HQ-OW-2004-0019-0387-A1; Colorado Department of Public Health & Environment;
Posted 10/14/2015

In Appendix C, it would helpful to denote which studies were used to derive the criterion.

EPA-HQ-0W-2004-0019-0417-A1; Utility Water Act Group (UWAG); Posted 11/02/2015

In addition to addressing the variability, UWAG requests that EPA incorporate the following
clarifications, flexibility and correction into the final criterion.

a. EPA should clarify what is meant by steady state".

Several of the factors in the mechanistic model (TTF, EF and CF) are defined as "steady state." UWAG
requests that EPA clarify what is meant by "steady state" in relation to the sampling necessary for
deriving the factors. In other words, what amount of variability (spatial and temporal) is acceptable in
the underlying data for the derived factor still to be considered steady state. This clarification is
particularly important for the enrichment factor (EF), which, as EPA acknowledges, is subject to
significant variability and has the "greatest potential to introduce uncertainty" in the translation. Draft
Criterion at K-18. EPA provides three procedures3 to determine an appropriate EF but does not address
what is meant by steady state.

EPA-HQ-OW-2004-0019-0402-A2; The Stakeholders Implementing TMDLs in the Calleguas
Creek Watershed; Posted 10/15/2015

Key Definitions are Absent from Table 4.1 and Need to be Added or Modified in order to Interpret the
Proposed Criterion.

The 2015 Draft Selenium Criterion lacks clarity in that many terms are used but not defined until much
later in the document. As a result, it is very difficult to interpret, evaluate, or implement (once

260

-------
promulgated) the criterion without searching through the 740 pages of the 2015 Draft Selenium
Criterion to find the definitions and intentions of key terms.

For example, footnote 4 of Table 4.1 of the 2015 Draft Selenium Criterion defines the Cbkgmd variable
as the "average background selenium concentration." However, the definition for Cbkgmd is absent
from the table. The definition is located on Page 87 of the 2015 Draft Selenium Criterion and identified
as "the average daily background concentration occurring during the remaining time" (when elevated
selenium concentrations are not occurring). Further, the use of this element is to capture situations
whereby the average monthly concentrations attain the criterion, but intermittent excursions of
concern may occur. Therefore, it would be helpful to add to the definition or explanation of this element
that it only applies when the average monthly concentrations are not attained. As the definition of
Cbkgmd is necessary to interpret Table 4.1, a modified and more precise definition needs to be added.

Additionally, footnote 5 of Table 4.1 of the 2015 Draft Selenium Criterion provides justification for the
duration of the fish tissue elements of the 2015 Draft Selenium Criterion being expressed as an
"instantaneous measurement". However, it is unclear whether this "instantaneous measurement"
should consist of either an individual fish sample or a composite sample. Fish tissue samples typically
consist of composite samples to provide spatial representation of the conditions at a site. Clarification
regarding the type of sample to be collected needs to be added to Table 4.1.

Requested Actions:

• Define each term upon initial use and revise the language in Table 4.1 of the 2015 Draft
Selenium Criterion to read as follows:

• Modify footnote 4: "...Cbkgrnd is the average background selenium concentration occurring
during the remaining time when elevated selenium concentrations are not occurring..."

• Modify footnote 5: "Instantaneous measurement. Fish tissue data provide point measurements
that reflect integrative accumulation of selenium over time and space in the fish at a given site.
Selenium concentrations in fish tissue are expected to change only gradually over time in
response to environmental fluctuations. Fish tissue data are to be collected as composite
samples."

The Term "Dissolved Total Selenium" is Ambiguous and Should Be Avoided.

Footnote 3 of Table 4.1 of the 2015 Draft Selenium Criterion states that water column values are based
on "dissolved total selenium (includes all oxidation states, i.e., selenite, selenite, organic selenium and
any other forms) in water". However, "dissolved total selenium" is an ambiguous term as typically, total
recoverable selenium is the terminology employed. The Stakeholders request the term "dissolved total
selenium be eliminated from the 2015 Draft Selenium Criterion and that the species and fraction (i.e.,
dissolved or total) be clearly identified and defined, and consistently applied, throughout the document.

Requested Actions:

Revise the language in Table 4.1 of the 2015 Draft Selenium Criterion to read as follows:

Modify footnote 3: "Water column values are based on dissolved total recoverable selenium (includes
all oxidation states, i.e., selenite, selenate, organic selenium, and any other forms) in water."

261

-------
Comment Category 6.3 - Comments Concerning the Additional Data,
Informational Sources, and Alter in at live Approaches

ummary

Several commenters recommended EPA incorporate additional studies in their analyses.

;esponse to comments

6.3 Response to concerns with additional data

Regarding consideration of new data, EPA has acquired new data through the peer review and public
process and has included these data, as appropriate, in the derivation of the criterion.

262

-------
Comment Category 6.4- Comments Concerning Failure to Address 2014

II nihil" .iilli'I I'll I i. IIIImI h ' II I ' • II' 11 < III'-

6.4 Summary

Several commenters requested that EPA provide responses to specific public and/or external peer
review comments provided on the 2014 draft selenium criterion.

6.4 Response to comments

EPA carefully considered the 2014 public and external peer-review comments in revising the 2015 draft
document and subsequently the final 2016 document. Additionally EPA had provided the 2014 external
peer reviewers with the 2014 public comments and solicited their feedback on the 2014 public
comments. The external peer reviewers' feedback on the 2014 public comments was also considered
directly in developing the 2015 draft selenium criterion document. The EPA response to external peer
review comments was released previously. EPA will provide the Response to 2014 Public Comments on
the EPA website.

6.4 Specific comments

EPA-HQ-OW-2004-0019-0384-A2; Cameco Resources; Posted 10/13/2015

The Federal Register (Vol 80, No 143, July 27, 2015) states that US EPA responses to comments on the
2014 draft would be available in the docket with the 2015 draft. However, they are not available and we
have not received responses to comments from the US EPA in this regard. Thus, it would appear that our
comments on the 2014 draft have not been considered. This raises concern about not only the
transparency of the development of the selenium criterion, but also the public review process itself.
From our review, it appears that revisions in the 2015 criterion may have been selective in nature rather
than based on full consideration of the science-based perspectives submitted.

Furthermore, it is challenging to comment on the 2015 draft because many of the same comments that
we made on the 2014 draft would still apply given the lack of change in the draft selenium criterion.

In the result, we want to encourage the US EPA to consider and fully respond to the comments Cameco
submitted during the public review period for the 2014 draft document. In addition, a further
opportunity to comment on any revisions made to the criterion document prior to finalization would
enhance this process.

EPA-HQ-OW-2004-0019-0402-A2; The Stakeholders Implementing TMDLs in the Calleguas
Creek Watershed; Posted 10/15/2015

The Stakeholders previously submitted a comment letter on the External Peer Review Draft Aquatic Life
Ambient Water Quality Criterion for Selenium - Freshwater 2014 (2014 External Peer Review Draft

263

-------
Criterion); however, many of the Stakeholder's comments included on the 2014 External Peer Review
Draft Criterion have not been addressed and are still applicable. In addition to the Stakeholder's
comments on the 2014 External Peer Review Draft Criterion which have not been addressed, the 2015
Draft Selenium Criterion contains modification that are also of concern for the Stakeholders.

Comments Previously Submitted on the 2014 External Peer Review Draft Criterion and Not Addressed in
2015 Draft Selenium Criterion

• Clarification is needed regarding the applicability of the four elements of the criterion to ensure
that fish-tissue elements supersede the water column elements.

• A universal, nationally applicable water column number is inappropriate due to the site-specific,
bioaccumulative nature of selenium. As such, the Draft Selenium Criterion should only be based
on fish tissue elements, with water column concentrations used as a tool for implementation of
the criterion.

• States should clearly be allowed to adopt Site-Specific Objectives (SSOs) that not only modify
each of the four elements of the criterion, but that also allows States to opt to eliminate aspects
of the criterion (e.g., water column concentrations).

• For the development of SSOs, the methodology to derive water concentrations from the whole-
body fish or muscle tissue criterion, rather than egg/ovary criterion, should be clearly presented.

• Justification and explanation for the selection of exceedance frequencies is absent and must be
provided in order to comment on the appropriateness of such frequencies. However, even
without any provided explanation or justification, the exceedance frequency for the fish tissue
elements of the criterion of "never to be exceeded" is inappropriate, impractical and contrary to
USEPA guidance.

• Key definitions are absent from Table 4.1 and need to be added or modified in order to interpret
the proposed criterion.

• The term "dissolved total selenium" is ambiguous and should be avoided.

EPA-HQ-OW-2004-0019-0411-A2; American Petroleum Institute (API); Posted 10/28/2015

EPA released a previous draft of the selenium criteria on May 14, 2014 (79 Fed. Reg. 27601, May 14,
2014, hereafter "2014 Notice"), in response to which API submitted detailed comments (API letter, R.
Claffto K. Gallagher, July 28, 2014). API is pleased EPA accepted public comment in response to the 2014
Notice, and conducted a contractor-led expert external peer review, leading to the current Notice. API
fully supports EPA's efforts to obtain input from the public and from technical experts. But EPA did not
respond in full to the comments submitted by API and others. In addition to API's comments in and
attached to today's letter and the comments submitted by the NAMC-SWG in response to the current
Notice, EPA should review, respond to, and address API's and NAMC-SWG's comments submitted in
response to the 2014 Notice.

API's detailed comments and concerns about the present Notice are attached. A summary of API's
comments follows.

264

-------
• EPA should consider and address peer-review comments on the criteria in the 2014 Notice. Not
all of the technical comments provided by expert peer reviewers in response to the 2014 Notice
were addressed in the current Notice, including comments pertaining to: site-specificity of
conversion calculations between tissue and water column selenium concentrations;
transparency of uncertainties and the need for less emphasis on modelled data; sensitivity
analysis of the EC10 method; conflicting comments on the use of injection studies; and the need
for detailed implementation guidance.

• EPA should consider and address public comments on the criteria in the 2014 Notice. Not all of
the technical comments provided by the public in response the 2014 Notice were addressed in
the current Notice, including concerns over food web models used to convert tissue selenium
concentrations to water column concentrations; the designations of water bodies as lotic vs.
lentic and the application of freshwater criteria to estuarine environments; the primacy criteria
for selenium thresholds, including the determinations of tissue and water thresholds; concern
about the flexibility to preferentially choose criteria for site-specific objectives (application of
tiered criteria); and concern about the application of various statistical methods such as the
percentile model to derive criteria.

1. Introduction

The United States Environmental Protection Agency (EPA) approach in the 2015 draft selenium (Se)
criteria document (EPA 2015) is more consistent with standard water quality criteria development
methodology (Stephan et al. 1985) than previous attempts at revising and updating the Se criteria (i.e.
EPA 2004, 2014), and includes a critical evaluation of 16 studies using various fish species and results in
Se tissue thresholds for twelve fish species within ten genera. Fish tissue-based criteria calculations
follow recommendations by Stephan et al. (1985), including the use of the 5th percentile (HC5) of genus
mean chronic values (GMCVs). This approach is more scientifically-defensible than the earlier 2004 draft
tissue criterion (EPA 2004), which was based on a single study.

We acknowledge the extensive effort that EPA and others put into development of these updated
chronic Se criteria and realize that all attempts were made to develop and establish a scientifically-
sound criteria document. However, even though a substantial number of interested parties provided
written comments pertaining to the first draft of this criteria document (EPA 2014), the majority of
these comments appear to have largely been ignored; as of the preparation of this technical
memorandum, there was no response provided as to the reason why particular comments were not
considered in the 2015 revision. Having ecologically-relevant water quality criteria based on current
science is of great importance to the scientific and regulated community and we appreciate this
opportunity to provide our comments and recommendations pertaining to the 2015 draft Se criteria
document, with hopes that they will be reviewed and considered in the next draft of the document.

2. Key Concerns and Recommendations

As indicated above, while the 2015 draft Se criteria are a significant improvement over previous criteria
(e.g., EPA 2004), there are some key areas of the criteria that we feel still need closer examination and
revision. We should also note that many recommendations provided in comments on the 2014 draft (for

265

-------
instance, GEI 2014a, b), were not incorporated in the latest draft, with no explanation provided as to
why.

4. EPA Response to Comments on the 2014 Draft

Subsequent to the publication of the 2014 Draft (EPA 2014), comments and suggestions for
improvement were solicited from both peer reviewers (Eastern Research Group Inc. 2014) and multiple
public stakeholders (see below) over a period between May 14, 2014 and June 13, 2014 (originally). Due
to numerous requests by many stakeholders - based on the volume and complexity of the document -
the original 30-day period was extended by a total of an additional 45 days, with final comments being
due to EPA by July 28, 2014.

Part of our review of the 2015 draft included the review of the above-mentioned peer review and public
comments and comparing it to the modifications made in the 2015 draft; however, this specific
information was not provided by EPA (e.g., in a comment/response tracking table).

During the review, we corresponded with EPA staff to inquire about this, and they communicated to us
that the current 2015 draft (EPA 2015) reflected a number of modifications, including: improvements in
calculations; analyses of additional data based on references obtained from the peer and public
reviewers; the inclusion of updated (and higher quality) data, as appropriate; enhanced discussions of
uncertainty throughout the document; and, other improvements based on suggestions raised during the
peer review and via comments from the public. Finally, it was indicated that the structure of the tiered
criterion and the structure of the document had not changed between the two (EPA 2014 and EPA 2015)
drafts.

4.1 Peer Review Comments

A total of seven (7) expert reviewers, selected by EPA, were asked to complete a questionnaire provided
by the Eastern Research Group Inc. (2014). The result of the peer review process comprised a series of
comments and perspectives provided by the reviewers in response to a series of specific questions — in
4 parts — relating to a number of issues about the clarity and scientific defensibility of the [first] draft
EPA Se criteria document (EPA 2014). We consolidated the peer review comments, by reviewing all of
the responses for questions/concerns raised by each reviewer, with the identified elements of the [first]
draft EPA Se criteria document (EPA 2014). These were then categorized by subject, with the intention
of determining the level of agreement (consensus)/ disagreement among reviewers.

Not all of the technical comments provided by expert peer reviewers were addressed in the 2015 draft
(EPA 2015). The suggested modifications — some of which were addressed and some of which were not
- related to the following major themes:

• The conversions (i.e., 'back-calculation') between tissue and water-column Se concentrations do
not necessarily take into account differing kinetics of Se uptake and elimination from an aquatic
system. Site-specific information and species-specific regressions are needed in order to reduce
uncertainty and variability. Some reviewers suggested that biokinetic models would be a better
option than the current model.

266

-------
• Given site-specific differences in Se behavior, it was suggested that uncertainties be made more
transparent and that less emphasis should be placed on the modelled data without further
exploring the variability and uncertainty inherent in each of the parameters.

• There were mixed feelings among reviewers regarding the use of the EC10 method. It was
highlighted by some as arbitrary and by others, as scientifically sound. A sensitivity analysis
would be need to be conducted to address this and such an analysis has not been done by EPA,
to our knowledge.

• There were conflicting comments from peer reviewers on the use of injection studies. In
particular, some reviewers felt that results from these studies should be removed from the
guideline derivation, as injection does not represent a realistic environmental route of exposure.

• Reviewers felt that better guidance was needed on the application of the proposed criteria for
various purposes: small streams, permitting, implementation, frequency limits, and acute
releases. Intermittent discharge was mentioned as being arbitrary.

We are unclear as to why the EPA did not address the concerns of the majority of peer reviewers. As an
example, we reviewed the responses relative to the back-calculation of the water Se criteria for lentic
and lotic. The following were the general comments from the reviewers (numbers identify reviewer):

1. Concerned with the approach;

2. Questioned the approach;

3. Concerned with the approach;

4. Didn't provide a clear statement of support;

5. Thought the approach was modern and valid, but a safety factor should be applied and call it a
trigger;

6. Supports the approach;

7. Concerned with the approach and the conservatism inherent in the values.

Based on this, it would appear that 5 of the 7 peer reviewers were generally not satisfied with the
approach, for various reasons and to different degrees. Nevertheless, the EPA moved forward with the
approach, with the major change being to actually remove data from their analyses, which in turn
resulted in even lower lentic and lotic criteria.

4.2 Public Comments

A substantial number (i.e., > 100) of stakeholders including representatives of: governments (e.g.,
federal, state, regional, municipal); industry associations; private companies; academics; environmental
non-governmental organizations; and private citizens, provided written comments — in some cases,
detailed technical comments - pertaining to the first draft of this criteria document (EPA 2014).

Some of the major themes of the comments provided — and it is not clear which of the comments were
addressed/incorporated in EPA (2015) - included the following:

• Concerns were raised by a number of stakeholders over the food web models used to back-
calculate from tissue Se concentrations to water concentrations. Comments focused on the

267

-------
nature and magnitude of trophic transfer/conversion factors (e.g., TTFs), selected organisms,
and differences in bioaccumulation factors.

• A number of technical comments were made regarding the designation of water bodies as lotic
vs. lentic. While several paragraphs were added to the text of the 2015 draft regarding this issue
(making it clearer how EPA classified systems as either lotic or lentic), many comments were not
addressed. As examples, will estuarine environments be classified as lentic or lotic? Also, how
will fish tissue criteria for selenium derived from freshwater species be applied to fish species
that live in an estuarine environment?

• Concerns were also expressed regarding the primacy criteria for Se thresholds, including the
determinations of tissue and water thresholds. There were many comments relating to
variability based on the way in which thresholds were calculated.

• There was concern across different regions in the US regarding how the criteria are meant to be
applied, how much flexibility there will be to preferentially choose criteria in order to create
site-specific objectives (e.g., ensuring there is a hierarchy/tiering in the application of thresholds
for the various media (e.g., water, WB/muscle, egg/ovary); e.g., the state of Kentucky employs a
tiered approach to the application of criteria)).

• Concerns were also raised regarding the use and application of various statistical methods in
deriving the criteria, including some suggestions that the percentile model used by EPA is out-
dated; in addition, there were requests for disclosure of the variability inherent in all
parameters used to calculate/derive the various thresholds.

Based on our review of the comments in comparison to the changes between the 2014 and 2015 drafts,
it appeared that the majority of these comments were not addressed (i.e., resulting in changes to the
criteria document, and in particular to the derivation and establishment of thresholds/criteria for
different media (e.g., water, WB/muscle, egg/ovary)).

We corresponded with EPA staff to inquire about this, and they communicated that the case with the
review of the Se drafts is an extraordinary situation (i.e., the public having the opportunity to evaluate
how the peer review and public comments were addressed prior to final document being released). In
other words, there is typically no response to peer review or public comment released until the criteria
document is final. Based on this response, we assume that a clearer understanding of the way in which
EPA has addressed the various stakeholder comments will not be available until the publication of the
final criteria document.

5. Conclusions and Recommendations

We have highlighted several technical concerns with the EPA's draft 2015 AWQCfor Se, with key
conclusions and recommendations provided below:

Previous Comments

Even though a substantial number of interested parties provided written comments pertaining to the
first draft of the selenium criteria document, the majority of these comments appear to have been
largely ignored. EPA should consider and respond to all of the substantive comments received, both on
the first draft and on the current one.

268

-------
EPA-HQ-OW-2004-0019-0391-A1; North American Metals Council (NAMC); Posted
10/14/2015

NAMC notes that the Federal Register stated: "EPA will make the external peer review and public
comments, as well as Agency responses to these comments on the previously published External Peer
Review Draft Aquatic Life Ambient Water Quality Criterion for Selenium - Freshwater 2014 (EPA 822-P-
14-001) (External Peer Review Draft), available in the docket with the revised draft selenium criteria
document at www.regulations.gov." This, however, has not occurred. It is difficult to comment on the
2015 EPA Selenium Draft without knowing how our previous comments on the 2014 EPA draft were
addressed, particularly given that there appear to be no substantive changes to the 2015 EPA Selenium
Draft based on our previous, extensive technical comments (i.e., the differences between the 2014 and
2015 drafts are relatively minor), and there is clear evidence that some, if not all, of our technical review
comments were ignored. For example, one of our major recommendations, to consider new information
that fills existing data gaps on the translation of tissue-based criteria to water-based selenium screening
levels, appears to have been ignored, as has information related to fish feeding over winter. Thus, as
Attachment 1 to the present comments, NAMC resubmits its comments on the previous draft,
requesting review and response from EPA on both our 2014 and 2015 comments.

1.0 INTRODUCTION

This document, submitted by the North American Metals Council (NAMC), comprises integrated
comments provided by individual Members and Associates of the North American Metals Council-
Selenium Work Group (NAMC-SWG). It is provided in response to the U.S. Environmental Protection
Agency's (EPA) July 27, 2015, request for public comment on its "External Peer Review Draft Aquatic Life
Ambient Water Quality Criterion for Selenium - Freshwater 2015" (2015 EPA Selenium Draft).1

We note that the Federal Register stated: "EPA will make the external peer review and public
comments, as well as Agency responses to these comments on the previously published External Peer
Review Draft Aquatic Life Ambient Water Quality Criterion for Selenium - Freshwater 2014 (EPA 822-P-
14-001) (External Peer Review Draft), available in the docket with the revised draft selenium criteria
document at www.regulations.gov."2 This, however, has not occurred. It is difficult to comment on the
2015 EPA Selenium Draft without knowing how our previous comments on the 2014 EPA draft were
addressed, particularly given that there appear to be no substantive changes to the 2015 draft based on
our previous, extensive technical comments (i.e., the differences between the 2014 and 2015 drafts are
relatively minor). Accordingly, NAMC resubmits its comments on the previous draft as Attachment 1 to
the present comments, requesting review and response from EPA on both our 2014 and 2015
comments. We note that, to date, EPA has not denied our technical arguments nor provided evidence
that our arguments are not technically defensible. As detailed below, it appears that some, if not all, of
our previous 2014 comments (Attachment 1) and supporting evidence were ignored.

We thank EPA for the opportunity to provide these comments that primarily focus on the fish tissue
criteria derivations. We trust EPA will seriously consider them as well as our previous comments and
look forward to the eventual adoption of a technically defensible selenium aquatic life criterion that is
appropriately but not, as is now the case, unnecessarily protective.

269

-------
3.0 CONCLUSIONS

The primary conclusions from our review of the 2015 EPA Selenium Draft are that:

• Our previous, detailed comments on the 2014 EPA draft do not appear to have been given
serious consideration. There is clear evidence that at least some, if not all, of our technically
valid review comments were ignored. Accordingly, the previous comments are resubmitted as
Attachment 1 for consideration. EPA has neither denied our technical arguments and evidence
nor provided evidence that our arguments and evidence are not technically defensible.

EPA-HQ-0W-2004-0019-0401-A1; Colorado Wastewater Utility Council; Posted 10/15/2015

We would like to acknowledge the extensive effort that EPA and others put into development of these
updated chronic Se criteria and realize that all attempts were made to create a scientifically sound
criteria document. However, it seems that even though a substantial number of interested parties
provided written comments to the first draft of this criteria document in 2014, the majority of these
comments appear to be largely ignored, with no responses provided as to the reason why particular
comments were not considered in the revision beyond responding to those provided by the peer
reviewers. Having ecologically relevant water quality criteria based on current science is of great
importance to the scientific and regulated community and we appreciate this opportunity to again
provide our comments and recommendations on the 2015 draft Se criteria document, and hope they
will be reviewed and considered in the next draft of the document.

1. Key Issues and Recommendations

While the 2015 draft Se criteria are an improvement over previous criteria, there are some key areas of
the criteria that we feel still need closer examination and revision. We presented many of these
recommendations and revisions to these key issues in our review of the 2014 draft, but most appear not
to have been addressed. We provide additional comments on these areas below, and a further
discussion of new concerns on the 2015 draft.

EPA-HQ-0W-2004-0019-0400-A1; The National Mining Association (NMA); Posted 10/15/2015

We would like to acknowledge the extensive effort that EPA and others put into development of these
updated chronic Se criteria and realize that considerable attempts were made to create a scientifically
sound criteria document. However, it seems that even though a substantial number of interested parties
provided written comments to the first draft of this criteria document in 2014, the majority of these
comments do not appear to be addressed in the criteria, with no responses provided as to the reason
why particular comments were not considered in the revision beyond responding to those provided by
the peer reviewers. Having ecologically relevant water quality criteria based on current science is of
great importance to the scientific and regulated community and we appreciate this opportunity to again
provide our comments and recommendations on the 2015 draft Se criteria document.

1. Key Issues and Recommendations

270

-------
While the 2015 draft Se criteria are an improvement over previous criteria, there are some key areas of
the criteria that we feel still need closer examination and revision. We presented many of these
recommendations and revisions to these key issues in our review of the 2014 draft, but most appear not
to have been addressed. We provide additional comments on these areas below, and a further
discussion of new concerns on the 2015 draft.

EPA-HQ-OW-2004-0019-0413-A2; Federal Water Quality Coalition (FWQC); Posted
10/28/2015

I. PROCESS PROBLEMS

Many of our concerns regarding the Draft Criteria Document are the same as, or related to, concerns
that we raised in 2014 as to the previous version of this document. The fact that these issues do not
appear to have been addressed, or have been addressed only partially, touches on a fundamental
procedural problem with EPA's Notice and its Draft Criteria Document. The Agency has provided no
indication of how it has addressed - if at all- the significant issues that were raised as to the 2014 draft,
by the FWQC or other commenters. In addition, while EPA had an independent peer review panel
conduct a review of its 2014 draft document, the Agency did not- until just a week ago- give the public
any information about how it was addressing (or not addressing) the comments of that panel. The public
simply has not been given the information needed in order to perform an adequate review of the new
draft and to develop useful, complete comments.

The FWQC, for example, raised a series of specific concerns about particular studies and the manner in
which those studies were used in deriving the criterion. It appears that some of those comments were
not accepted, since the discussion of these studies in the Draft Criteria Document is the same, or nearly
the same, as it was in the 20 14 draft. But nowhere in any of the EPA documents is there a clear, direct
explanation of why those comments by the FWQC were not accepted. In some instances, the discussion
of studies in the Draft Criteria Document has changed, indicating that comments were considered - but
again, there is no explanation of the Agency's thought process in making these changes. As a result, the
commenters are left without any ability to comment effectively on the new draft, since the Agency's
rationale for the new decisions is entirely opaque. Before moving forward with the new criteria, EPA
should make available, for public review, a detailed response to comments on the 2014 draft document,
so stakeholders can more fully understand the basis for the 2015 proposal.

This lack of explanation, made available in a timely manner, is particularly troubling as to the comments
submitted by the peer review panel. This group of seven non-EPA scientists was convened to review the
2014 draft of the Criteria Document, and they submitted a final report in September 2014- over a year
ago. The peer review panel's individual comments, which are included in the final report, raise a number
of serious questions as to the scientific basis for the proposed criteria. Yet, until October 6 - one week
ago - EPA had not made available any document that reviews the panel's comments and explains how
they were addressed, or whether they were not appropriate to address. As a result, commenters could
not determine what EPA thought of the peer review panel's comments, and had no ability to assess how
those issues have been dealt with by the Agency in the Draft Criteria Document.

271

-------
Then, on October 6, EPA finally made available its responses to the peer review comments, in the form
of a 145-page document. Surely the Agency did not expect that the interested stakeholders would be
able, in those seven days, to carefully review all 145 pages, determine how the new document affects
the comments that they had already been drafting, and make all necessary modifications and additions
to address issues raised in the new 145 pages. That is simply not possible. Therefore, upon learning of
the new document, the FWQC submitted a request for an extension of the comment period by 60 days,
to give the public enough time to fully absorb the new information and supplement or revise their
comments accordingly. As of October 13, we have not received a response from the Agency to that
request, despite the impending end of the comment period. Therefore, we are forced to submit these
comments without being able to perform a careful review of the new EPA document. The FWQC, and all
other commenters, deserve that opportunity. Therefore, before finalizing new criteria, EPA needs to
provide a 60-day comment period on the new document, so we can submit a supplemental set of
comments that are considered before any final criteria decisions are made.

272

-------
" ' 'lull > II li < < i« 1 ' 'II II " ' 'lull II ill » alb - Ill I I.I. - > II III k-l II > Ih /

an , s

7.1 Summary

Commenters expressed a concern that a 30-day comment period was not adequate.

7.1 Response to comments

EPA extended public comment periods for both the 2014 and 2015 draft criterion documents, as
commenters requested.

For the 2014 external peer review draft criterion, EPA held a 30-day public comment period and
extended the public comment period for another 30 days based on public comments. The 2014 public
comment period totaled over 60 days due to Federal Register Notice issuance lag time. Thus, the 2014
external peer review criterion document the docket was open from May 14 to July 28, 2014,
approximately 2.5 months

For the 2015 criterion document, EPA held a 60-day public comment period, with two 15 -day
extensions of the public comment period. Thus, for the 2015 draft selenium criterion document, the
docket was open from July 28, 2015 to October 30, 2015, approximately 3 months

In total, EPA provided a total of approximately 5.5 months of public comment on the recommended
selenium national criterion draft documents.

EPA carefully considered the 2014 public comments in revising the 2014 document for the 2015 draft
criterion document, and additionally provided the 2014 external peer reviewers with the 2014 public
comments and solicited their feedback on the public comments. The external peer reviewers' feedback
on the 2014 public comments was also considered directly in developing the 2015 draft selenium
criterion document. EPA will provide the Response to 2014 Public Comments on the EPA website.

7.1 Specific comments

EPA-HQ-OW-2004-0019-0390-A1; West Virginia Coal Association (WVCA); Posted 10/14/15
Conclusion

WVCA agrees with commenters that have already requested an extension of the comment deadline. The
new methodology for calculation of the water column elements deserves a separate comment period
apart from consideration of the Draft Selenium Criterion. EPA's attempt to circumvent the approach
required by the 1985 Guidelines is not appropriate. While WVCA has not had sufficient time to develop
detailed comments regarding all aspects of the Draft Selenium Criterion, we also have general concern
regarding the following additional components of the document:

273

-------
EPA-HQ-0W-2004-0019-0414-A1; Colorado Mining Association (CMA); Posted 10/28/2015

- I had understood that the comment period for ID: EPA-HQ-OW-2004-0019-0367 would be open
through October 13; I attempted to submit comment through Regulations.gov but the website was
down much of the day. When it came back on-line, I saw two notices of extension of time in the docket,
but one listed "comments closed" as of October 10. The second listing under the "Extension of time -
Request for Scientific Views: Draft Recommended Aquatic Life Ambient Water Quality Chronic Criterion
for Selenium - Freshwater 2015" describes a completely different regulation on the Safe Drinking Water
Act was listed.

I am therefore submitting the Selenium comments made on behalf of the Colorado Mining Association
by e-mail in the event the comments are still being received on the selenium criterion.

EPA-HQ-OW-2004-0019-0412-A2; Utility Water Act Group (UWAG); Posted 10/28/2015

On behalf of the Utility Water Act Group (UWAG), we write to request that the public comment period
for the U.S. Environmental Protection Agency's (EPA) Draft Aquatic Life Ambient Water Quality Criterion
for Selenium - Freshwater 2015 be extended 30 days, from September 25, 2015 to October 26, 2015.
The Draft Criterion and External Peer Review Comments documents were published on July 27 providing
stakeholders just 60 days to review and develop comments. Our technical experts believe that a
minimum of 30 additional days is needed to thoroughly review EPA's revised interpretation of the
science and methodologies used to derive the proposed criterion. Sufficient additional time is also
needed to review comments of 7 expert external peer reviewers and analyze how EPA considered these
comments in developing the revised draft criterion.

Introduction

These are the supplemental comments of the Utility Water Act Group (UWAG)l on EPA's Request for
Scientific Views: Draft Recommended Aquatic Life Ambient Water Quality Criterion for Selenium-
Freshwater 2015 ("Draft Criterion"). 80 Fed. Reg. 44,350 (Jul. 27, 2015). Previously, UWAG filed
comments on October 13, 2015. EPA subsequently extended the comment period until October 30,
2015. 80 Fed. Reg. 63,552 (Oct. 20, 2015). UWAG appreciates the additional time to file comments on
the Draft Criterion and to review EPA's responses to the external peer review comments, which were
made available on September 25, 2015.

The importance of science-based water quality criterion development cannot be overstated. For over
two decades, the best science and approach for assessing the risks of selenium exposure has been
widely debated. The final criterion will have significant regulatory impact as EPA is encouraging states
and Tribes to adopt EPA's recommended 304(a) criteria directly into their water quality standards. It is
therefore vital that such recommendations be based on sound science and proper data. To help the
agency achieve that aim, UW AG respectfully requests an additional 30 days, until October 26, 2015, to
review the revised draft criterion so that we and other interested parties may have sufficient time to
provide meaningful substantive technical comments.

274

-------
Attached are the Comments of the Utility Water Act Group (UWAG) in response to the Request for
Scientific Views: Draft Recommended Aquatic Life Ambient Water Quality Chronic Criterion for Selenium
— Freshwater 2015. 80 Fed. Reg. 44350 (Jul. 27, 2015) (providing notice and request for comments by
September 25, 2015). On September 24, 2015, EPA announced that the deadline for comments was
extended for an additional 15 days, from September 25, 2015 to October 10, 2015. 80 Fed. Reg. 57065.
Subsequently, the agency confirmed, on October 1, 2015, that the agency would accept comments
through Tuesday, October 13, 2015, to avoid closing the comment period on Saturday, October 10 or
Monday, October 12, a federal holiday. See, email from Betsy Behl, Director, Health and Ecological
Division, to UWAG counsel, Karen Bennett, Hunton and Williams, LLP, October 1, 2015 (attached).

1. EPA Should Reopen the Public Comment Period to Afford Time to Review and Comment on the
Agencies Responses to the 2014 Peer Review.

During 2014, the agency formed an external peer review group of seven experienced scientists to review
EPA's External Peer Review Draft Aquatic Life Ambient Water Quality Criterion for Selenium-
Freshwater 2014, 79 Fed. Reg. 27,601 (May 14, 2014) ("2014 Draft").2 The final external peer review
report, External Peer Review of the Draft Aquatic Life Ambient Water Quality Criterion for Selenium -
Freshwater 2014 ("2014 Peer Review"), was issued on September 25, 2014 and made available to the
public in July 2015. The 2014 Peer Review is relevant to the agency's development of sound,
scientifically-defensible water quality criteria. Equally important is the public's ability to understand how
EPA considered the 2014 Peer Review in developing the 2015 Draft Criterion. It is troubling that EPA
waited until just days before the close of the comment period to make the agency's response to the
peer reviewers' comments available to the public. See EPA's Response to Comment Document for the
external peer review of the "External Peer Review Draft for the Chronic Aquatic Life Criterion for
Selenium: Freshwater 2014" (September 25, 2015).

In the recent update to the federal water quality standards rule, EPA emphasized how important
transparency is in relation to adoption of criteria. See, e.g., 80 Fed. Reg. 51,020, 51,022 (Aug. 21, 2015).
Since that new rule requires states to either adopt EPA's Clean Water Act Section 304 (a) criteria or
provide an explanation for not doing so, transparency in how EPA adopted the criteria is equally
important, if not more so. Yet, for the draft selenium aquatic life criteria, EPA has not been transparent
with the scientific community or the public by providing this information so late in the comment period.
There simply was not enough time provided for stakeholders to understand and comment on how EPA
dealt with important aspects of the science underlying the criteria development.

EPA should re-open the comment period, providing additional time needed to thoroughly review and
adjust comments in light of EPA's delay in making the agency's responses to the peer reviewers'
comments available to the public. The following are a few examples of relevant external peer review
comments that warrant additional time for consideration of EPA's response:

• The agency also solicited public comments on the 2014 Draft with the intent of making those
comments available to the expert peer reviewers. UWAG filed comments on the 2014 Draft and
those comments are attached to this document and incorporated by reference.

275

-------
EPA-HQ-OW-2004-0019-0383; Federal Water Quality Coalition (FWQC); Posted 10/13/2015

The Federal Water Quality Coalition (the "FWQC" or the "Coalition") hereby submits the following
response to EPA's Reopening of Request for Scientific Views on the Draft Aquatic Life Ambient Water
Quality Criterion for Selenium - Freshwater 2015 (80 Fed. Reg. 63552, October 20, 2015), which is
referred to in these comments as the "Selenium Notice" or the "Notice").

The Notice reopens the comment period concerning EPA's draft recommended freshwater aquatic life
criterion for selenium. The previous deadline for submitting comments on the criterion was October 13,
2015, and the FWQC submitted comments on that date. On that same date, EPA granted the FWQC's
earlier-submitted request for an extension of the comment period, setting a new deadline of October
30, 2015.

The primary basis for the FWQC extension request was that EPA had just made available a significant
new document that bears directly on the scientific basis for EPA's draft criterion: a detailed, 145-page
response to comments submitted by the Agency's peer review panel concerning the draft selenium
criterion (referred to here as the "Response to Peer Review Comments"). The FWQC appreciates EPA's
attention to, and granting of, its request. We believe that allowing this opportunity for public input
regarding new scientific information can only serve to strengthen the technical basis for the decision
that is eventually made by the Agency.

On behalf of the Federal Water Quality Coalition (FWQC), I am requesting an extension of time for the
comment period on EPA's draft selenium criteria document. We understand that the initial comment
deadline has recently been extended by about two weeks, so that comments are due on October 13.
However, the Agency has still not made the information available that is needed in order for us to
comment effectively. This is so in several respects. First, it is important to recognize that the criteria
document is a revised version of the draft that was sent out for comment in 2014. The FWQC and other
parties submitted detailed comments on the 2014 document. We think, based on our review of the
2015 document, that EPA has taken action on some of the comments, and has not taken action on
others. But there is no explanation, anywhere in the 2015 document or elsewhere in the record, of what
changes EPA decided to make - or not make - and why it made or didn't make those changes. We simply
cannot tell what the Agency thinks of the issues raised in the 2014 comments. Even a line-by-line
comparison of the 2014 and 2015 documents will not tell us what EPA's rationale is for any changes that
were made, and it certainly cannot tell us why some of our suggested changes were not made.

In addition to those problems in assessing the Agency's action or inaction on the 2014 public comments,
there is another lack of information that hampers our ability to formulate new comments. EPA formed
an independent peer review panel to assess the 2014 document. That panel delivered a final report in
September 2014, which included numerous specific comments, concerns and recommendations. Yet
until yesterday, the Agency had provided no indication at all of how it was addressing - or not addressing
- the issues raised by the panel, in developing the 2015 document. Then yesterday (October 6), just
seven days before the end of the comment period on the 2015 document, EPA issued a 145-page
response to the peer review comments. Apparently EPA expects that commenters will now review the
entire 145-page document, assess how the Agency responded to each of the peer review comments,
and incorporate our thoughts on those issues into the comments that we have to file in six days. That

276

-------
expectation by EPA contravenes every principle of adequate public participation, and is simply
inconsistent with due process and with the Agency's own public participation regulations.

EPA needs to do two things. First, the Agency needs to extend the comment period by at least 60 days.
Second, within 15 days, EPA needs to produce a detailed response to the public comments on the 2014
draft criteria document, so the commenters can understand how their concerns were (or were not)
addressed in the 2015 document. Then, the stakeholders can review that response-to-comments
document, and the newly available response to peer review comments, and incorporate their remaining
concerns into their comments on the 2015 document. This process is necessary in order to ensure that
stakeholders will be able to effectively assess the Agency's decision-making process and then submit
comments that reflect a full understanding of the science regarding EPA's actions.

277

-------
1 ii11ii11^ iiii 1 i. ii 1 iiIii11 iiii * II ' «ii' » mil II •. II nil llss ¦ -I

Criterion

7.2 Summary

Commenters requested that issuance of the final criterion recommendations should be delayed until
comment is received on the informational material being provided to aid state and tribal adoption of the
rule as discussed in the Federal Register notice.

7.2 Response to comments

The 2016 final selenium criterion scientific recommendations are being issued, not delayed, to provide
states and the public with the benefit of EPA's recommendations, based on the best available science,
on the development of water quality standards to protect aquatic life designated uses. The 2016
criterion document contains substantial information in the appendices, including recommended
approaches to calculating site-specific criteria and data that could be applied in developing site-specific
criteria. EPA intends to issue implementation materials and implementation-related technical support
documents for comment following the issuance of the 2016 final.

7.2 Specific comments

EPA-HQ-0W-2004-0019-0389-A1; Thunder Basin Coal Company, LLC (TBCC); Posted 10/14/15
Conclusion

Promulgation of this rule should be delayed until comment is received on the informational material
being provided to aid state and tribal adoption of the rule as discussed in the Federal Register notice.

Final standards must be derived and applied in an appropriate, site specific and attainable manner
tailored to the aquatic life being protected. TBCC supports the selenium comments from the National
Mining Association and their consultant GEI as well as comments from Wyoming Mining Association.
Thank you for your consideration of these comments.

EPA-HQ-0W-2004-0019-0416-A1; Wyoming Mining Association (WMA); Posted 11/02/2015

WMA also believes that promulgation of the rule should be delayed until comment is received on the
informational material being provided to aid state and tribal adoption of the rule as discussed in the
Federal Register notice.

278

-------
Comment Category 8.1 - References
8.1 Response to comments

EPA reviewed submitted references and data and included them in the 2016 final selenium criterion
document as appropriate.

8.1 Specific comments

EPA-HQ-OW-2004-0019-0388-A2; J. R. Simplot Company; Posted 10/14/2015
References

AECOM. 2012. Reproductive success study with brown trout (Salmo trutta). Data quality assurance
Report. Final. December 2012.

Chapman P.M., W.J. Adams, M.L. Brooks, C.G. Delos, S.N. Luoma, W.A Maher, H.M. Ohlendorf, T .S.
Presser and D.P. Shaw. 2009. Ecological Assessment of Selenium in the Aquatic Environment: Summary
of a SETAC Pellston Workshop. Pensacola FL (USA): Society of Environmental Toxicology and Chemistry
(SETAC).

Chapman P.M., W.J. Adams, M.L. Brooks, C.G. Delos, S.N. Luoma, W.A Maher, H.M. Ohlendorf, T.S .
Presser and D.P. Shaw (eds). 2010. Ecological Assessment of Selenium in the Aquatic Environment.
SETAC Press, Pensacola, FL, USA.

Cutter, G.A. 1982. Selenium in reducing waters. Science 217:829-83 1.

Cutter, G. A. and Bruland, K. W. 1984. The marine biogeochemistry of selenium: a re-evaluation Limnol.
Oceanogr. 29: 1179-1192.

DeForest OK, Brix KV, Elphick JR, Rickwood CJ, DeBruyn AMH, Tear LM, Gilron G, Hughes SA, Adams WJ.
2014. Selenium partitioning between water and fish tissue in freshwater systems: Development of
water-based selenium screening guidelines. Prepared for the North American Metals Council - Selenium
Working Group, Washington ,D.C. http://www.namc.org/docs/Selenium%201ntegrated%20Report%20-
%20Final%20(2014-05-20).pdf

ERG (Eastern Research Group, Inc.) 2012. External Peer Review of the Interpretation of Results of a
Study on the Effect of Selenium on the Health of Brown Trout Offspring. EPA Office of Science and
Technology. Contract No. EP-C-12-021.

ERG (Eastern Research Group, Inc.). 2014. External Peer Review of the Draft Aquatic Life Ambient Water
Quality Criterion for Selenium- Freshwater 2014. EPA Office of Science and Technology. Contract No. EP-
C-12-021.

Formation Environmental. 2011. Brown Trout Laboratory Reproduction Studies Conducted in Support of
Development of a Site-Specific Selenium Criterion. Prepared for J.R. Simplot Company by Formation
Environmental. Revised October 2011.

279

-------
Formation Environmental. 2012. Technical Support Document: Proposed Site-Specific Selenium
Criterion, Sage and Crow C reeks, Idaho.

Great Lakes Environmental Center (GLEC). 2014. External Peer Reviewer Comments on Review of Draft
EPA Report, 'Analysis of the Brown Trout Selenium Toxicity Study Presented by Formation
Environmental and Reviewed by U.S. Fish and Wildlife Service'. Prepared for the USEPA Office of Water,
Washington, D.C.

Linville, R.G. 2006. Effects of Excess Selenium on the Health and Reproduction of White Sturgeon
(Acipenser transmontanus): Implications for San Franscisco Bay Delta. Dissertation. University of
California at Davis.

Luoma, S.N. and T.S. Presser. 2009. Emerging opportunities in management of selenium contamination:
Environmental Science and Technology. 43:8483-8487.

McDonald, B.G. and P .M. Chapman. 2009. The need for adequate quality assurance/quality control
measures for selenium larval deformity assessments: Implications for tissue residue guidelines.
Integrated Environmental Assessment and Management, Vol. 5, Number 3. Pp. 470-475.

Presser, T.S. and S.N. Luoma. 2006. Forecasting selenium discharges to the San Francisco Bay-Delta
estuary: ecological effects of a proposed San Luis drain extension. Professional Paper 1646, U.S.
Department of Interior, U.S. Geological Survey, Reston, Virginia.

Presser, T.S. and S.N. Luoma. 2010. A Methodology for Ecosystem-Scale Modeling of Selenium.
Integrated Environmental Assessment and Management. Volume 6, Number 4, pp. 685-710.

Reinfelder, J.R., N.S. Fisher, S .W. Fowler and J.L. Teyssie. 1993. Release rates of trace elements and
protein from decomposing planktonic debris 2. Copepod carcasses and sediment trap particulate
matter. J. Mar. Res. 51 (2): 423-442.

Riedel, G. F., D.P. Ferrier and J.G. Sanders. 1991. Uptake of selenium by freshwater phytoplankton.
Water Air Soil Pollut. 57-58: 23-30.

Simplot. 2013. J .R. Simplot Company (Simplot) Responses to US Fish and Wildlife Service (USFWS)
Comments on the Draft Interpretive Report, August 2010. March 2013.

USFWS. 2012. Technical Review: Smoky Canyon Mine Site-Specific Selenium Criterion Report. US
Department of Interior, Fish and Wildlife Service, Division of Environmental Quality, Environmental
Contaminants Program.

USEPA. 1997. Ecological Risk Assessment Guidance for Superfund: Process for Designing and Conducting
Ecological Risk Assessments. Interim Final. EPA 540- R-97-006. Environmental Response Team, Edison,
NJ.

USEPA. 1998. Guidelines for Ecological Risk Assessment. Final. Risk Assessment Forum. EPA/630/R-
95/002F.

USEPA. 2014. External Peer Review Draft- Aquatic Life Ambient Water Quality Criterion for Selenium-
Freshwater, 2014. Office of Water. USEPA-822- P-14- 001.

280

-------
USEPA. 2015a. Draft Aquatic Life Ambient Water Quality Criterion for Selenium Freshwater, 2015. Office
of Water. USEPA-822-P-15-001.

USEPA. 2015b. External Peer Review Draft Aquatic Life Ambient Water Quality Criterion for Selenium -
Freshwater 2015. USEPA-822-P-15-001.

USEPA. 2015c. Toxicity Relationship Analysis Program (TRAP Version 1.20). NHERL. Duluth, MN.

USEPA. 2012. Toxicity Relationship Analysis Program (TRAP Version 1.3A). Mid-Continent Ecology
Division. http://www.epa.gov/med/Prods Pubs/trap.html

EPA-HQ-OW-2004-0019-0385-A2; Kentucky Division of Water (KDOW); Posted 10/13/2015
Literature Cited

Chapman, P.M., W.J. Adams, M.L. Brooks, C.G. Delos, S.N. Luoma, W.A. Maher, H.M. Ohlendorf, T.S.
Presser and D.P. Shaw (eds). 2010. Ecological assessment of selenium in the aquatic environment. SETAC
Press, Pensacola, FL.

Doroshov, S., J. Van Eenennaam, C. Alexander, E. Hallen, H. Bailey, K. Kroll and C. Restrepo. 1992.
Development of water quality criteria for resident aquatic species of the San Joaquin River, Part I,
bioaccumulation of selenium in broodstock of channel catfish (Ictalurus punctatus) and its effects on
reproduction. Final report to State Water Resources Control Board, State of California. Contract number
7- 197-250-0.

Linville, R.G. 2006. Effects of excess selenium on the health and reproduction of white sturgeon
(Acipenser transmontanus): implication for San Francisco Bay-Delta. PhD dissertation, University of
California at Davis.

Stephan, C.E., D.I. Mount, D.J. Hansen, J.H. Gentile, G.A. Chapman and W.A. Brungs. 1985. Guidelines for
deriving numerical national water quality criteria for the protection of aquatic organisms and their uses.
PB85-227049. National Technical Infonnation Service, Springfield, VA.

U.S. EPA 19978. Report on the peer consultation workshop on selenium aquatic toxicity and
bioaccumulation. Office of Water. USEPA, Washington, DC. EPA-822-R-98-007.

EPA-HQ-OW-2004-0019-0387-A1; Colorado Department of Public Health & Environment;
Posted 10/14/2015

References

Brix, K.V., and D.K. DeForest. 2008. Selenium. Pages 123-172 in Gensemer RW, Meyerhoff RD, Ramage
KJ, Curely EF, eds. Relevance of ambient water quality criteria in ephemeral and effluent-dependent
watercourses of the arid western US. SETAC Press, Pensacola, FL, USA.

Chapman, P.M., W.J. Adams, M.L. Brooks, C.G. Delos, S.N. Luoma, W.A. Maher, H.M. Ohlendorf, T.S.
Presser, and D.P. Shaw. 2009. Ecological assessment of selenium in the aquatic environment: Summary

281

-------
of a SETAC Pellston Workshop. Pensacola FL (USA): Society of Environmental Toxicology and Chemistry
(SETAC).

EPA-HQ-OW-2004-0019-0370-A2; USGS Leetown Science Center; Posted 09/08/2015

Hitt, N.P. and D.R. Smith. 2014. Threshold-dependent sample sizes for selenium assessment with stream
fish tissue. Integrated Environmental Assessment and Management 11:143-149.

Presser, T.S. 2013. Selenium in ecosystems within the mountaintop coal mining and valley-fill region of
southern West Virginia - assessment and ecosystem-scale modeling. US Geological Survey Professional
Paper 1803. 86 pg. http://pubs.usgs.gov/pp/1803/

EPA-HQ-0W-2004-0019-0419-A2; Lima Refining Company; Posted 11/03/2015
References

CH2M Hill, 2010. Ottawa River Biological and Water Quality Study and Ecological Risk Assessment.
Prepared for Lima Refining Company. March 2010.

Hill, W.R., and I.L. Larsen, 2005. "Growth Dilution of Metals in Microalgal Biofilms," Environmental
Science and Technology, 39(6): 1513-1518.

HydroQual, 2009. Ottawa River Sampling Program for Water Quality Analyses to Determine the
Selenium Load Upstream of the Husky Refinery in Lima, Ohio: Survey results and Waste Load Allocation
Analysis. Prepared for the Lima Refining Company.

Luoma, S.N., and P.S. Rainbow, 2005. "Why is metal bioaccumulation so variable? Biodynamics as a
unifying concept, a critical review." Environmental Science & Technology, 39: 1921-1931.

Paquin R., and R. Bubnyte, 2014. "Considerations for Evaluating an Averaging Period for Selenium
Bioaccumulation," presented at the 2014 SETAC meeting, Vancouver, BC, CA.

Thomann, R.V., 1981. "Equilibrium Model of the Fate of Microcontaminants in Diverse Aquatic Food
Chains," Canadian Journal of Fisheries and Aquatic Sciences, 38: 280-296.

USEPA, 2015. Draft Aquatic Life Ambient Water Quality Criterion for Selenium — Freshwater 2015.

Office of Water, Office of Science and Technology, Washington, D.C. EPA 822-P-15-001. July 2015.

EPA-HQ-OW-2004-0019-0410-A2; West Virginia Department of Environmental Protection (DEP); Posted
10/28/2015

USEPA. 2014. External peer review draft aquatic life ambient water quality criterion for selenium-
freshwater 2014. Federal Register, Vol. 79, No. 93. pp. 27601-27604. http://www.gpo.gov/fdsys/pkg/FR-
2014-05-14/htm 1/2014-11080. htm.

282

-------
USEPA. 2015. Draft Aquatic Life Ambient Water Quality Criterion for Selenium - Freshwater 2015.

http://water.epa.gov/scitech/swguidance/standards/criteria/aqlife/selenium/loader.cfm?csModule=sec

urity/getfile&PagelD=718002.

EPA-HQ-OW-2004-0019-0412-A2; Utility Water Act Group (UWAG); Posted 10/28/2015
Literature Cited

Coyle, J.J., D.R. Buckler, C.C. Ingersoll, J.F. Fairchild, and T.W. May. 1993. Effect of dietary selenium on
the reproductive success of bluegills Lepomis macrochirus. Environmental Toxicology and Chemistry 12:
551-565.

Crutchfield, J.U., Jr. 2000. Recovery of a power plant cooling reservoir ecosystem from selenium
bioaccumulation. Environmental Science and Policy 3: S145 - S163.

Cumbie, P.M., and S.L. Van Horn. 1980. Selenium accumulation associated with fish mortality and
reproductive failure. Proceedings of the Annual Conference of Southeast Association of Fishery and
Wildlife Agencies 32: 612 - 624.

Doroshov, S., J. Van Eenennaam, C. Alexander, E. Hallen, H. Bailey, K. Kroll, and C. Restrep. 1992.
Development of Water Quality Criteria for Resident Aquatic Species of the San Joaquin River, Part II:
Bioaccumulation of Dietary Selenium and its Effects on Growth and Reproduction in Bluegill (Lepomis
machrochirus). Final Report to State Water Resources Control Board, State of California.

EPRI. 2001. Selenium cycling and impact in aquatic ecosystems: defining trophic transfer and water-
borne exposure pathways. Technical Report 1005217. Electric Power Research Institute, Palo Alto, CA.

Hermanutz, R.O., K.N. Allen, T.H. Roush, and S.F. Hedtke. 1992. Effects of elevated selenium
concentrations on bluegills Lepomis macrochirus in outdoor experimental streams. Environmental
Toxicology and Chemistry 11: 217 - 224.

Hermanutz, R.O., K.N. Allen, N.E. Detenbeck, and C.F. Stephan. 1996. Exposure to bluegill (Lepomis
machrochirus) to selenium in outdoor experimental streams. U.S. EPA Report. Mid- Continent Ecology
Division, Duluth, MN.

Hoang, T.C., and S.J. Klaine. 2008. Characterizing the toxicity of pulsed selenium exposure to Daphnia
magna. Chemosphere 71: 429 - 438.

Penglase, S., K. Hamre, and S. Ellingson. 2014. Selenium and mercury have a synergistic negative effect
on fish reproduction. Aquatic Toxicology 149: 16 - 24.

Petrov, P.K., J.W. Charters, and D. Wallschlager. 2012. Identification and determination of selenosulfate
and selenocyanate in flue gas desulfurization waters. Environmental Science & Technology 46: 1716 -
1723.

Reash, R.J., T.W. Lohner, K.V. Wood, and V.E. Willet. 1999. Exotoxicological assessment of bluegill
sunfish inhabiting a selenium-enriched fly ash stream. Pp. 423 - 445 in Environmental Toxicology and
Risk Assessment: Standardization of Biomarkers for Endocrine Disruption and Environmental

283

-------
Assessment: Eighth Volume, ASTM STP 1364. American Society for Testing and Materials, West
Conshohocken, PA.

Reash, R.J., T.W. Lohner, and K.V. Wood. 2006. Selenium and other trace metals in fish inhabiting a fly
ash stream: implications for regulatory tissue thresholds. Environmental Pollution 142: 397 - 408.

Reash, R.J. 2012. Selenium, arsenic, and mercury in fish inhabiting a fly ash exposure gradient:
bioaccumulation patterns and elemental associations. Environmental Toxicology and Chemistry 31: 739
-747.

Riscassi, A.L., A.D. Converse, K.J. Hokanson, andT.M. Scanlon. 2010. Evaluation of automated
streamwater sampling during storm events for total mercury analysis. Journal of Environmental
Monitoring 12: 1833 - 1839.

Steven, D., and O. McAdam. 2015. Retrospective weight-of-evidence analysis identifies substrate change
as the apparent cause of recruitment failure in the upper Columbia River white sturgeon (Acipenser
transmontanus). Canadian Journal of Fisheries and Aquatic Sciences 72: 1208 - 1220.

Tashjian, D.H., S.J. The, A. Sogomonyan, and S.S.O. Hung. 2006. Bioaccumulation and chronic toxicity of
dietary L-selenomethionine in juvenile white sturgeon (Acipenser transmontanus). Aquatic Toxicology
79: 401-409.

Thomas, J.K., and D.M. Janz. 2014. In ovo exposure to selenomethionine via maternal transfer increases
developmental toxicities and impairs swim performance in F1 generation zebrafish (Danio rerio). Aquatic
Toxicology 152: 20 - 29.

US EPA. 1985. Guidelines for Deriving Numeric National Water Quality Criteria for the Protection of
Aquatic Organisms and their Uses. PB85-227049. US EPA Office of Research and Development, Duluth,
MN.

US EPA. 1995. Great Lakes Water Quality Initiative Criteria Documents for the Protection of Aquatic Life
in Ambient Water. EPA-820-B-95-004. US EPA Office of Water, Washington, DC.

US EPA. 1991. Technical Support Document for Water Quality-based Toxics Control. EPA/505/2- 90-001.
US EPA Office of Water, Washington, DC.

US EPA. 2010a. Guidance for Implementing the January 2001 Methylmercury Water Quality Criterion.
EPA 823-R-10-001. US EPA Office of Science and Technology, Washington, DC.

US EPA. 2010b. U.S. Environmental Protection Agency NPDES Permit Writer's Manual. EPA-833- K-10-
001. US EPA Office of Wastewater Management, Washington, DC.

284

-------
EPA-HQ-OW-2004-0019-0411-A2; American Petroleum Institute (API); Posted 10/28/2015
6. References

Brix KV, DeForest DK. 2008. Selenium. Pages 123-172 in Gensemer RW, Meyerhoff RD, Ramage KJ,
Curely EF, editors. Relevance of ambient water quality criteria in ephemeral and effluent-dependent
watercourses of the arid western US. SETAC Press, Pensacola, FL, USA.

Burau, R.G. 1985. Environmental Chemistry of Selenium. California Agriculture pp. 16-18.

Butler DL, Osmundson BC, Krueger RP. 1997. Field screening of water, soil, bottom sediment, and biota
associated with irrigation drainage in the Dolores Project and the Mancos River Basin, southwestern
Colorado, 1994. US Geological Survey, Denver, CO, USA.

Canton S. 2010. Appendix B: Commentary: Persistence of some fish populations in high-Se
environments. Pages 293-323 in Chapman PM, Adams WJ, Brooks ML, Delos CG, Luoma SN, Maher WA,
Ohlendorf HM, Presser TS, Shaw DP, editors. Ecological assessment of selenium in the aquatic
environment. SETAC Press, Pensacola, FL, USA.

Casey R. 2005. Results of aquatic studies in the McLeod and Upper Smoky River systems. Alberta
Environment, Edmonton, AB, Canada.

Chapman PM. 2015. Including or excluding toxicity test data for development of a geometric mean.
Environ Toxicol Chem 34: 1691-1692.

Chapman PM, Adams WJ, Brooks ML, Delos CG, Luoma SN, Maher WA, Ohlendorf HM, Presser TS, Shaw
DP. 2010. Ecological assessment of selenium in the aquatic environment. SETAC Press, Pensacola, FL,
USA.

Coyle JJ, Buckler DR, Ingersoll CG, Fairchild JF, May TW. 1993. Effect of dietary selenium on the
reproductive success of bluegills (Lepomis macrochirus). Environ Toxicol Chem 12:551-565.

CP&L (Carolina Power & Light). 1997. Largemouth bass selenium bioassay. Environmental Services
Section, NC, USA.

DeForest DK, Adams WJ. 2011. Selenium accumulation and toxicity in freshwater fishes. Pages 193-229
in Beyer WN, Meador JP, editors. Environmental contaminants in biota: Interpreting tissue
concentrations (Second Edition). CRC Press, Boca Raton, FL, USA.

DeForest DK, Gilron G, Armstrong SA, Robertson EL. 2012. Species sensitivity distribution (SSD)
evaluation for selenium in fish eggs: Considerations for development of a Canadian tissue-based
guideline. Integr Environ Assess Manag 8:6-12.

DeForest DK, Brix KV, Gilron G, Hughes SA, Tear LM, Elphick JR, Rickwood CJ, DeBruyn AMH, Adams WJ.
2014. Selenium partitioning between water and fish tissue in freshwater systems: Development of
water-based selenium screening guidelines. Poster presentation at the Society of Environmental
Toxicology and Chemistry 35th Annual Meeting, Vancouver, BC, Canada. November 9-13, 2014.

285

-------
DeForest DK, Brix KV, Gilron G, Hughes SA, Tear LM, Elphick JR, Rickwood CJ, DeBruyn AMH, Adams WJ.
In Revision. Selenium partitioning between water and fish tissue in freshwater systems: Development of
water-based selenium screening guidelines. Submitted to Environ Sci Technol.

De Riu, D., L. Jang-Won, Huang, S., Monielloa, G., and Hung, S. 2014. Effect of dietary selenomethionine
on growth performance, tissue burden, and histopathology in green and white sturgeon. Aquat. Toxicol.
148:65-73.

Doroshov S, Van Eenennaam J, Alexander C, Hallen E, Bailey H, Kroll K, Restrepo C. 1992. Development
of water quality criteria for resident aquatic species of the San Joaquin River. University of California-
Davis, CA, USA.

Eastern Research Group. 2014. External Peer Review of the Draft Aquatic Life Ambient Water Quality
Criterion for Selenium — Freshwater 2014. Contract No. EP-C-12-021, Work Assignment 1-43.
September 25, 2014.

EPA. 2004. Draft aquatic life water quality criteria for selenium - 2004. Office of Water, Washington, D.C.
EPA 822-D-04-001.

EPA. 2013. Aquatic life ambient water quality criteria for ammonia — freshwater 2013. Office of Water,
Washington, D.C. EPA 822-R-13-001.

EPA. 2014. External peer review draft aquatic life ambient water quality criterion for selenium —
freshwater 2014. Office of Water, Washington, D.C. EPA 822-P-14-001.

EPA. 2015. Draft aquatic life ambient water quality criterion for Selenium — freshwater 2015. Office of
Water, Washington, D.C. EPA 822-P-15-001.

Finley K, Garrett R. 2007. Recovery at Belews and Hyco Lakes: Implications for fish tissue selenium
thresholds. Integr Environ Assess Manag 3:297-299.

Formation Environmental. 2011. Brown trout laboratory reproduction studies conducted in support of
development of a site-specific selenium criterion. Prepared for J.R. Simplot Company by Formation
Environmental. Revised October 2011.

GEI Consultants, Inc. (GEI). 2007. Aquatic Biological Monitoring and Selenium Investigation of the
Arkansas River, Fountain Creek, Wildhorse Creek, and the St. Charles River. Prepared for City of Pueblo
WRP.

GEI Consultants. 2008. Maternal transfer of selenium in fathead minnows, with modeling of ovary tissue
to whole body concentrations. Project 062790. Chadwick Ecological Division, Littleton, CO.

GEI Consultants, Inc. (GEI). 2013. Use Attainability Analysis for the St. Charles River, Middle Arkansas
River Segment 6, Colorado. Prepared for Public Service Company of Colorado.

GEI Consultants, Inc. (GEI). 2014a. Review of EPA 2014 draft Se criteria document EPA 822-P14- 001.
Submitted to Docket ID EPA-HQ-OW-2004-0019.

GEI Consultants, Inc. (GEI). 2014b. Supplemental comments on EPA 2014 draft Se criteria document EPA
822-P-14-001. Submitted to Docket ID EPA-HQ-OW-2004-0019.

286

-------
Hermanutz RO, Allen KN, Roush TH, Hedtke SF. 1992. Effects of elevated selenium concentrations on
bluegills (Lepomis macrochirus) in outdoor experimental streams. Environ Toxicol Chem 11:217-224.

Hermanutz RO, Allen KN, Detenbeck NE, Stephan CE. 1996. Exposure of bluegills (Lepomis macrochirus)
to selenium in outdoor experimental streams. U.S. Environmental Protection Agency, Duluth, MN, USA.

Herring, J.R., and K. Walton-Day. 2007. Selenium and Other Elements in Water and Adjacent Rock and
Sediment of Toll Gate Creek, Aurora, Arapahoe County, Colorado, December 2003 through March 2004,
58 p.

Janz DM, DeForest DK, Brooks ML, Chapman PM, Gilron G, Hoff D, Hopkins WA, Mclntyre DO, Mebane
CA, Palace VP, Skorupa JP, Wayland M. 2010. Selenium toxicity to aquatic organisms. Pages 141-231 in
Chapman PM, Adams WJ, Brooks ML, Delos CG, Luoma SN, Maher WA, Ohlendorf HM, Presser TS, Shaw
DP. 2010. Ecological assessment of selenium in the aquatic environment. SETAC Press, Pensacola, FL,
USA.

Lemly AD. 1985. Toxicology of selenium in a freshwater reservoir: Implications for environmental hazard
evaluation and safety. Ecotoxicol Environ Saf 10:314-338.

Linville RG. 2006. Effects of excess selenium on the health and reproduction of white sturgeon
(Acipenser transmontanus): Implications for San Francisco Bay-Delta. Doctoral Dissertation. University of
California-Davis, CA, USA.

Lo BP, Elphick JR, Bailey HC, Baker JA, Kennedy G. 2015. The effect of sulphate on selenate
bioaccumulation in two freshwater primary producers: a duckweed, Lemna minor, and a green alga,
Pseudokirchneriella subcapitata. Environ Toxicol Chem DOI: 10.1002/etc.3127.

Skorupa JP. 1998. Selenium. United States Department of the Interior. National Irrigation Water Quality
Program Information Report No. 3: Guidelines for interpretation of the biological effects of selected
constituents in biota, water, and sediment.

Stephan CE, Mount Dl, Hansen DJ, Gentile JH, Chapman GA, Brungs WA. 1985. Guidelines for deriving
numerical national water quality criteria for the protection of aquatic organisms and their uses. U.S.
Environmental Protection Agency, Washington, D.C. NTIS No. PB85- 227049. 98 pp.

Williams MJ, Ogle RS, Knight AW, Burau RG. 1994. Effects of sulfate on selenate uptake and toxicity in
the green alga Selenastrum capricornutum. Arch Environ Contam Toxicol 17:449-453.

EPA-HQ-0W-2004-0019-0401-A1; Colorado Wastewater Utility Council; Posted 10/15/2015
5. References

Adams, W.J., J.E. Toll, K.V. Brix, L.M. Tear and D.K. DeForest. 2000. Site-specific approach for setting
water quality criteria for selenium: differences between lotic and lentic systems. Proceedings Mine
Reclamation Symposium: Selenium Session; Sponsored by Ministry of Energy and Mines, Williams Lake,
British Columbia, Canada, June 21-22, 2000.

287

-------
Brix, K and D. De Forest. 2008. Selenium. In: Relevance of ambient water quality criteria in ephemeral
and effluent-dependent watercourses of the arid western US (Gensemer RW, Meyerhoff RD, Ramage KJ,
Curely EF, eds.). Pensacola, FLSETAC Press, pp. 123-172.

Carolina Power & Light (CP&L). 1997. Largemouth Bass Selenium Bioassay- Report. Carolina Power &
Light Company, Environmental Services Section, NC. December 1997.

Chapman P.M., W.J. Adams, M.L. Brooks, C.G. Delos, S.N. Luoma, W.A Maher, H.M. Ohlendorf, T.S.
Presser, and D.P. Shaw. 2009. Ecological Assessment of Selenium in the Aquatic Environment: Summary
of a SETAC Pellston Workshop. Pensacola, FL (USA): Society of Environmental Toxicology and Chemistry
(SETAC).

Chapman P.M., W.J. Adams, M.L. Brooks, C.G. Delos, S.N. Luoma, W.A Maher, H.M. Ohlendorf, T.S.
Presser, and D.P. Shaw (eds). 2010. Ecological Assessment of Selenium in the Aquatic Environment.
SETAC Press, Pensacola, FL, USA.

Coyle, J.J., D.R. Buckler, C.G. Ingersoll, J.F. Fairchild, and T.W. May. 1993. Effect of dietary selenium on
the reproductive success of bluegills Lepomis macrochirus. Environmental Toxicology and Chemistry
12(3):551-565.

DeForest D.K., and W.J. Adams. 2011. Selenium accumulation and toxicity in freshwater fishes, pp. 193-
229. In: Beyer, W. N., and J. P. Meador, editors. Environmental contaminants in biota: interpreting tissue
concentrations. 2nd edition. Boca Raton (FL, USA): Taylor and Francis.

DeForest, D.K., K.V. Brix, J.R. Elphick, C.J. Rickwood, A.M.H. deBruyn, L.M. Tear, G. Gilron, S.A. Hughes,
and W.J. Adams. 2014. Final Report, Selenium Partitioning between Water and Fish Tissue in Freshwater
Systems: Development of Water-based Selenium Screening Guidelines. Prepared for: North American
Metals Council -Selenium Working Group, Washington, D.C. May 2014.

DeForest, D.K., S. Pargee, C. Claytor, S. Canton and K. Brix. In press. Biokinetic food chain modeling of
waterborne selenium pulses into aquatic food chains: implications for water quality criteria. Integrated
Environmental Assessment and Management.

Doroshov, S., J.V. Eenennaam, C. Alexander, E. Hallen, H. Bailey, K. Kroll, and C. Restreppo. 1992.
Development of water quality criteria for resident aquatic species of the San Joaquin River. Draft final
report to the California State Water Resources Control Board for Contract No. 7-197-250-0. Department
of Animal Science, University of California, Davis, CA.

Formation Environmental. 2011. Yellowstone cutthroat trout laboratory reproduction studies conducted
in support of development of a site-specific selenium criterion. Draft. Prepared for J.R. Simplot
Company. Pocatello (ID): Smoky Canyon Mine.

GEI Consultants, Inc. (GEI). 2008. Maternal transfer of selenium in fathead minnows, with modeling of
ovary tissue to whole body concentrations. Project 062790. Chadwick Ecological Division, Littleton, CO.

288

-------
GEI Consultants, Inc. (GEI). 2013a. Use Attainability Analysis for the St. Charles River, Middle Arkansas
River Segment 6, Colorado. Prepared for Public Service Company of Colorado. Corporation.

GEI Consultants, Inc. (GEI). 2013b. Bioassessment of the Streams in the Vicinity of the Catenary and
Hobet Mine Complexes, West Virginia. Prepared for Patriot Coal

GEI Consultants, Inc. (GEI). 2014a. Review of EPA 2014 draft Se criteria document EPA 822-P-14-001.
Submitted to Docket ID EPA-HQ-OW-2004-0019.

GEI Consultants, Inc. (GEI). 2014b. Supplemental comments on EPA 2014 draft Se criteria document EPA
822-P-14-001. Submitted to Docket ID EPA-HQ-OW-2004-0019.

GEI Consultants, Inc. (GEI). 2014c. Use-Attainability Analysis for Selected Tributaries to the Yampa River,
Routt County, Colorado. Prepared for Seneca Coal Company, Peabody Sage Creek Mine, LLC and
Twentymile Coal, LLC.

Hermanutz, R.O., K.N. Allen, T.H. Roush, and S.F. Hedtke. 1992. Effects of elevated Se concentrations on
bluegills (Lepomis macrochirus) in outdoor experimental streams. Environmental Toxicology and
Chemistry 11:217-224.

Hermanutz, R.O., K.N. Allen, N.E. Detenbeck, and C.E. Stephan. 1996. Exposure of bluegill (Lepomis
macrochirus) to selenium in outdoor experimental streams. U.S. EPA Report. Mid-Continent Ecology
Division. Duluth, MN.

Janz, D.M. 2008. A critical evaluation of winter stress syndrome. In Selenium tissue thresholds: Tissue
selection criteria, threshold development endpoints, and potential to predict population or community
effects in the field. Washington (DC, USA): North America Metals Council - Selenium Working Group.

Linville, R.G. 2006. Effects of Excess Selenium on the Health and Reproduction of White Sturgeon
(Acipenser transmontanus): Implications for San Francisco Bay-Delta. Dissertaiton. University of
California at Davis.

North American Metals Council-Selenium Working Group (NAMC-SWG). 2014. Comments of the North
American Metals Council-Selenium Working Group in Response to EPA's Draft National Recommended
Aquatic Life Criterion for the Pollutant Selenium. Submitted to Docket ID EPA-HQ-OW-2004-0019.

Stephan, C.E., D.I. Mount, D.J. Hansen, J.H. Gentile, G.A. Chapman, and W.A. Brungs. 1985. Guidelines
for Deriving Numerical National Water Quality Criteria for the Protection of Aquatic Organisms and Their
Uses. PB85-227049. National Technical Information Service, Springfield, VA.

U.S. Environmental Protection Agency (EPA). 2004. Draft Aquatic Life Water Quality Criteria for Selenium
- 2004. EPA-822-D-04-001. Office of Water, Washington, DC.

U.S. Environmental Protection Agency (EPA). 2014. External Peer Review Draft Aquatic Life Ambient
Water Quality Criterion for Selenium - Freshwater 2014. EPA 822-P-14-001. Office of Water 4304T,
Washington, D.C.

289

-------
EPA-HQ-0W-2004-0019-0400-A1; The National Mining Association (NMA); Posted 10/15/2015
5. References

Brix, K and D. De Forest. 2008. Selenium. In: Relevance of ambient water quality criteria in ephemeral
and effluent-dependent watercourses of the arid western US (Gensemer RW, Meyerhoff RD, Ramage KJ,
Curely EF, eds.). Pensacola, FLSETAC Press, pp. 123-172. Carolina Power & Light (CP&L). 1997.
Largemouth Bass Selenium Bioassay- Report.

Carolina Power & Light Company, Environmental Services Section, NC. December 1997.

Chapman P.M., W.J. Adams, M.L. Brooks, C.G. Delos, S.N. Luoma, W.A Maher, H.M. Ohlendorf, T.S.
Presser, and D.P. Shaw. 2009. Ecological Assessment of Selenium in the Aquatic Environment: Summary
of a SETAC Pellston Workshop. Pensacola FL (USA): Society of Environmental Toxicology and Chemistry
(SETAC).

290

-------
GEI Consultants, Inc. (GEI). 2008. Maternal transfer of selenium in fathead minnows, with modeling of
ovary tissue to whole body concentrations. Project 062790. Chadwick Ecological Division, Littleton, CO.

GEI Consultants, Inc. (GEI). 2013a. Use Attainability Analysis for the St. Charles River, Middle Arkansas
River Segment 6, Colorado. Prepared for Public Service Company of Colorado.

GEI Consultants, Inc. (GEI). 2013b. Bioassessment of the Streams in the Vicinity of the Catenary and
Hobet Mine Complexes, West Virginia. Prepared for Patriot Coal Corporation.

GEI Consultants, Inc. (GEI). 2014a. Review of EPA 2014 draft Se criteria document EPA 822-P-14-001.
Submitted to Docket ID EPA-HQ-OW-2004-0019.

GEI Consultants, Inc. (GEI). 2014b. Supplemental comments on EPA 2014 draft Se criteria document EPA
822-P-14-001. Submitted to Docket ID EPA-HQ-OW-2004-0019.

U.S. Environmental Protection Agency (EPA). 2004. Draft Aquatic Life Water Quality Criteria for Selenium
- 2004. EPA-822-D-04-001. Office of Water, Washington, DC.

291

-------
II Data

8.2 Summary

Several commenters provided EPA additional data for consideration in updating selenium criterion
documentation.

8.2 Response to comments

EPA has acquired new data through the peer review and public process and has included these data, as
appropriate, in the derivation of the criterion.

8.2 Specific comments

EPA-HQ-OW-2004-0019-0413-A2; Federal Water Quality Coalition (FWQC); Posted
10/28/2015

Original letter contains additional information in the form of an attached comment letter submitted by
Barnes & Thornburg LLP, on behalf of Federal Water Quality Coalition (FWQC) for the 2014 selenium
comment response. See original letter.

EPA-HQ-OW-2004-0019-0412-A3; Utility Water Act Group (UWAG); Posted 10/28/2015

Original letter contains additional information in the form of an appendix, EPRI (2001) report: Selenium
Cycling and Impact in Aquatic Ecosystems: Defining Trophic Transfer and Water-Borne Exposure
Pathways. See original letter.

EPA-HQ-OW-2004-0019-0391-A1; North American Metals Council (NAMC); Posted
10/14/2015

Original letter contains additional information in the form of an attached comment letter submitted by
North American Metals Council (NAMC) for the 2014 selenium comment response. See original letter.

EPA-HQ-0W-2004-0019-0401-A1; Colorado Wastewater Utility Council; Posted 10/15/2015

Original letter contains additional information in the form of supplemental data tables. See original
letter.

EPA-HQ-0W-2004-0019-0400-A1; The National Mining Association (NMA); Posted 10/15/2015

Original letter contains additional information in the form of supplemental data tables. See original
letter.

292

-------