&EPA
United States
Environmental Protection
Agency
Office of Water
4304T
EPA-822-R-22-002
March 2022
Response to
External Peer Review Comments on
EPA's Metals CRADA Phase I Report:
Development of an Overarching Bioavailability
Modeling Approach
to Support US EPA's
Aquatic Life Water Quality Criteria for Metals
March 2022
i
-------
1.0 Introduction
In December 2017, the U.S. Environmental Protection Agency (EPA) signed a Cooperative Research and
Development Agreement (CRADA) with eight metals associations (Aluminum Association, Aluminum REACH
Consortium, Cobalt Institute, International Copper Association, Copper Development Association, International
Lead Association, International Zinc Association, NiPERA Inc.) in order to leverage the knowledge and
resources of scientists inside and outside of the agency to better protect aquatic life. EPA's Office of Science
and Technology within the Office of Water (OW) is the Agency's technical lead on this CRADA which supports
EPA's FY 2018-2022 Strategic Plan Goal: Provide for Clean and Safe Water: Protect and Restore Water Quality.
EPA is using a two-phased approach to address the CRADA.
In the first phase, EPA has worked with external technical experts from the metals associations to develop a
proposed modeling approach to predict the bioavailability and toxicity of metals under the range of water
chemistry conditions found in aquatic environments common in freshwaters of the United States. The results
of work conducted in this phase are captures in the CRADA Phase I Report: Development of an Overarching
Bioavailability Modeling Approach to Support US EPA's Aquatic Life Water Quality Criteria for Metals.
Subsequently, in the second phase, EPA will work with the metals associations to develop bioavailability
models for individual metals using the overarching modeling approach. Using the resulting peer-reviewed
models, EPA plans to develop updated, externally-peer reviewed Aquatic Life Ambient Water Quality Criteria
for metals to better support states, territories and tribes with criteria that reflect the latest science and are
easier to implement than more complex, previous approaches using metals bioavailability modeling for criteria
development.
This document provides EPA responses to the results of an independent, external peer review of the U.S.
Environmental Protection Agency's (EPA's) Metals CRADA Phase I Report: Development of an Overarching
Bioavailability Modeling Approach to Support US EPA's Aquatic Life Water Quality Criteria for Metals
(hereafter, CRADA Phase I Report). The peer reviewers were external expert scientists with expertise in 1)
Modeling as applied to the characterization of metals bioavailability in aquatic systems, 2) Aquatic inorganic
chemistry, hydrogeology, and biogeochemistry of metals in aquatic systems, 3) Aquatic toxicology of metals,
aquatic ecology, and physiology of aquatic organisms, 4) Statistical analyses and data interpretation for the
determination of data acceptability and/or 5) Knowledge of the Clean Water Act, especially water quality
standards (WQS). Eastern Research Group, Inc. (ERG), a contractor to EPA, organized this external peer review
for EPA's Office of Water (OW) and developed the external peer review report.
Section 2.0 of this document provides individual reviewer comments on the CRADA Phase I Report and EPA's
responses to the peer reviewer comments. Section 3.0 provides additional reviewer comments and EPA
responses.
EPA's contractor identified, screened, and selected the following five experts who met technical selection
criteria provided by EPA and had no conflict of interest in performing this review:
• David Buchwalter, Ph.D.: Professor, Department of Biological Sciences, North Carolina State University
• Claude Fortin, Ph.D.: Professor, Institut National de la Recherche Scientifique (INRS), Canada
• Erin M. Leonard, Ph.D.: NSERC Post-Doctoral Fellow, Integrative Biology, University of Guelph
• Christopher A. Mebane: Water Quality Specialist, U.S. Geological Survey
• Wilhelmus Peijnenburg, Ph.D.: National Institute of Public Health and the Environment (RIVM), Centre
for Safety of Substances and Products, The Netherlands
2
-------
2.1 Please provide your scientific feedback of the strengths and weakne the MLR (Multiple Linear Regression) and BLM (Biotic
Ligand Model) approaches for estimating the effects of water chemistry/toxicity modifying factors on the bioavailability and
2.1 General comments on the strengths and weaknesses of the MLR and BLM approaches.
Reviewer
Comments
EPA Response
Reviewer
1
In general terms, the MLR and BLM approaches that are presented in the
documents are clearly the state of the art. It is to be noted that a major part of
the models have developed in close cooperation between scientists and
industry, as assisted by regulatory institutions. This cooperation has been
successful and resulted in a number of sophisticated models that are suited for
the derivation of water quality criteria. A pragmatic question that arises is
associated to the fact that the development of the key models has been
performed by a relatively small cross section of the researchers active in the
field of metal bioavailability. It is therefore essential to warrant sufficient
academic support regarding the scientific foundations of the models and the
justification for use in regulation.
Strengths:
The approaches represent the state-of-the-art with regard to the scientific
aspects of metal bioavailability quantification.
A proper combination of mechanism-based knowledge (as exemplified for
instance by model development based on first principles) and pragmatic
approaches (as exemplified by MLR approaches) is used and integrated in the
broad spectrum of models available. The basic approaches supplement each
other, and the BLM approach can for instance be used to inform the
correctness of the MLR approach.
The overall concept is applicable to a multitude of metals and to an array of
biological species of different trophic level: it is clear that the same basic
principles apply across the universe of water chemistries and across the
Thank you for your comment. EPA agrees that the
approaches are currently the state-of-the-science.
EPA acknowledges your concern regarding the
expert researchers involved in developing the
models which is why we have conducted this
external expert peer review of the models and
associated information. EPA will also have the
individual selected models for each metal
externally peer reviewed before any use in criteria
derivation. The criteria documents will also
undergo external peer review.
EPA agrees that these two approaches, MLR and
BLM, complement each other, and our objective is
to find a scientifically-defensible approach
applicable to all metals that will be easy for states
and stakeholders to implement.
EPA agrees that metal bioavailability is complex
and would consider non-linear relationships and
interactions as well as additional toxicity
modifying factors if enough data are available for
their inclusion.
EPA also agrees that this effort should result in the
development of user-friendly tools for regulators
1
-------
2.1 General comments on the strengths and weaknesses of the MLR and BLM approaches.
Reviewer
Comments
EPA Response
universe of biological species. This increases the credibility of the basic
hypotheses related to variations in water chemistry modifying metal toxicity.
The validation efforts undertaken to show that the models are capable of
properly predicting toxicity across different water chemistries.
Weaknesses:
A general weakness which is inherent to metal toxicity, is that the general
concept of metal bioavailability is complex. It is complex in the sense that
numerous processes are non-linear and as a consequence the overall impact of
water chemistry on metal toxicity is non-linear. It is therefore important to
make sure that the resulting non-linear relationships as well as the interactions
between the factors modifying toxicity, are properly understood and properly
incorporated in the models.
Although a lot of research has been performed and although various key factors
have been identified, it cannot be ruled out that for specific waters, factors
come into play that have not yet been identified. It is important to keep an
open eye for the possible need of accounting for additional factors in toxicity
assessment. The impact of carbonate that is observed for a limited number of
species is an example of such an additional factor.
The mere fact that numerous models have been developed for various metals
and various biological species make it difficult for non-experts to have an
overview of the models available, their individual strengths and weakness, as
well as their domain of applicability. In practical terms the key weakness is that
overall, the models might be considered as a black box by for regulators with
limited background knowledge on metal bioavailability. This implies that efforts
with regard to communication and development of user-friendly software tools,
need to be optimized.
to be able to predict metal toxicity to aquatic
organisms.
No other reviewers provided general comments on the strength and weaknesses
of MLR and BLM approaches.
2
-------
2.1.a Do you see technical/scientific advantages of using one model over the other for deriving water quality criteria?
2.1.a Technical/scientific advantages of using one model over the other for deriving water quality criteria.
Reviewer
Comments
EPA Response
Reviewer
1
No. The key issue in this respect is my observation that each model has its own
merits and on forehand no model should be ruled out, or be classified as being
better than another model. It is to be realized that each model also has its own
amount of information embedded and this information is used best when
using more than one model in deriving water quality criteria. Actually, a
recommendation with regard to the overall set of models available and with
regard to the overall knowledge available in this overall set of models, is to
investigate whether transfer learning approaches can be applied to improve
model performance.
Thank you for your comment.
Reviewer
2
There is validity of the chemical speciation modeling and modeling of
competition between dissolved ions and complexes for binding to predict the
relationship between water chemistry and metal accumulation and incipient
toxicity. However, there are shortcomings in terms of neglecting that the
kinetics of exposure change over time.
With the BLM, the performance of the model is dependent on the parameters
that are available to predict speciation reactions as well as on those that
define the critical concentration of metal-biotic ligand complex at which
toxicity occurs. In many cases, some of these parameters are not determined
or inaccurate which leads to either inputting estimates or leaving values at the
default settings. Additionally, more input parameters increase the potential
and impact of human error therefore affecting the accuracy of the models.
In addition, in many cases, LA50 values across all species have not been
measured directly, specifically with invertebrates which are the most sensitive
taxa. This should be addressed. Additionally, within the documents, biotic
ligands have been defined as either the gills of fish or the respiratory surface of
invertebrates, however, whole body measurements are used for
determination of LA50 values for these species. For fish, although the gills are
most likely the primary biotic ligand and the one driving toxicity, it should be
Thank you for your comment. EPA also agrees that
it is helpful for bioavailability models to be
informed by a mechanistic understanding of metal
toxicity and of metal speciation but asserts that
the transparency and ease of use of the MLR
outweighs the mechanistic complexity of the BLM.
Thank you for your suggestion to investigate the
gut acting as a biotic ligand in fish as well and the
gills/respiratory surface. As the reviewer
mentioned, the gut can be an important uptake
pathway for marine fish, however, this effort is
currently focused on freshwater models.
Generally, in freshwater fish, gill uptake has been
shown to be more important that the gut (e.g.,
Niyogi et al., 2007) which is one reason the gill has
been the focus of BLM development in these
cases.
EPA agrees that temperature could be an
important toxicity modifying factor for some
3
-------
2.1.a Technical/scientific advantages of using one model over the other for deriving water quality criteria.
Reviewer
Comments
EPA Response
included that the gut, especially in seawater, may add to the complexity by
also acting as a biotic ligand (Alsop et al., 2016 Aquatic Toxicology).
There is strength in an approach that simplifies the BLM model and relies on
extensive toxicity data sets covering wide ranges of water chemistry
parameters and ecotoxicity endpoints. 1 see the benefits of a MLR over a BLM
approach because of its simplicity, the three input parameters (pH, DOC, and
hardness), and therefore less need to collect data (or estimate parameters) on
multiple water chemistry parameters to successfully run the model. However, 1
do see the need to include temperature as a fourth parameter. Metal
accumulation in fish, pond or river water is enhanced by upsurges in
temperature; therefore, it is imperative to study the detrimental effects of
metals in combination with temperature to formulate accurate predictive
models (Kumar et al., 2018 Int. J. Environ. Sci. Technol.).
Overall, although bioavailability models should be informed by mechanistic
understanding of metal toxicity and of metal speciation, 1 think that the
transparency and ease of use of the MLR outweighs the mechanistic
complexity of the BLM.
metals, but there is typically insufficient data on
temperature to incorporate into the models at this
time. Furthermore, while temperature is
considered within the BLM, it is generally not
identified as a major factor. The agreement
between the predictions made by BLM and MLR
models in this report can be used as a verification
that even though the MLR models do not
incorporate temperature, they are still able to
empirically predict toxicity to aquatic organisms.
Reviewer
3
As a scientist 1 philosophically favor the BLM approach to the MLR approach
because it has the most mechanistic validity with reference to acutely sensitive
taxa. At least for the earliest derivations of the BLM, the use of real
experimental data was used to parameterize the model rather than the latter
approaches where they were fitted (fudged) to fit the toxicity outcome data.
However, 1 don't think either approach is particularly defensible for the
derivation of chronic criteria because it neglects the possibility the dietary
metal exposures are toxic.
Thank you for your comment. EPA also agrees that
it is helpful for bioavailability models to be
informed by a mechanistic understanding of metal
toxicity and of metal speciation.
Diet is an additional route of metal exposure that
is generally not considered within bioavailability
models because of a lack of available data and a
mechanistic complexity. Currently, data for many
metals indicate that exposure of the respiratory
organs via water are more sensitive to cationic
metals than exposure through the gut.
Furthermore, these models have been validated
with long-term mesocosm studies in which the
4
-------
2.1.a Technical/scientific advantages of using one model over the other for deriving water quality criteria.
Reviewer
Comments
EPA Response
dietary route of exposure is an operational
pathway (Roussel et al. 2007; Schlekat et al. 2010;
Versteeg et al. 1999). Additionally, in a dietary zinc
toxicity, De Schamphelaere et al. (2004) concluded
that: "the zinc BLM predicts chronic reproductive
zinc bioavailability and toxicity in synthetic and
field surface waters with reasonable accuracy
even without explicitly directly considering the
dietary toxicity pathway". For many metals,
toxicity stemming from waterborne pathway has
shown to occur at similar or lower concentrations
than the dietary route (e.g., Evens et al. 2009 for
nickel, De Schamphelaere et al. 2007 for copper,
Nys et al. 2013 and Alsop et al. 2016 for lead) and
indicate that ALC that are protective of aqueous
metal exposure are also expected to be protective
of dietary exposures.
Reviewer
4
1 see several (dis)advantages to the use of either approach. Among the
arguments presented, the decreased number of input parameters is cited as
an advantage in favour of MLRs. 1 see a hidden disadvantage to that as this
may introduce a bias (see response to Question 3a below).
Another nuance 1 would like to bring forward about the "improved
transparency" of the MLRs is that it may be easy to spot the driving
parameters by simply looking at the equation, but it does not allow the user to
understand why these parameters are important. BLM-based models are more
complicated to use and require training but that results in having more
informed users. MLRs do not incite users to understand the science behind the
equation and in the long run this may represent a loss. It may be a question of
perspective but, from my point of view, MLRs are less transparent than BLMs
because 1 know what the speciation of a metal should be by looking at water
chemistry parameters and can thus expect an output. If this output is far from
Thank you for your comment. EPA understands
your argument that, from a scientific standpoint,
an expert in this field can predict the output of the
BLM from the presented water chemistry
parameters more easily than the MLR equation.
However, both the BLM program and MLR models
can be used without the user understanding the
science behind the equations. Furthermore, since
the 2007 Copper BLM was published, only 6 states
have adopted the Cu BLM statewide and 9 have
adopted it as a means to develop site-specific
criteria. EPA is considering the needs of end-users
of these models, such as regulators and
stakeholders, in our decision and as noted in the
comment, both models provide good results in
5
-------
2.1.a Technical/scientific advantages of using one model over the other for deriving water quality criteria.
Reviewer
Comments
EPA Response
my expectation, 1 would make additional simulations to figure out why and
possibly spot a mistake in data entry for example. On the other hand, using a
long equation does not trigger any expectations in terms of output.
It's not clear to me how easy/hard it is to recompile a new MLR upon the
addition of new data but, intuitively, it seems to me that this requires starting
from scratch. On the other hand, the addition of a binding constant into the
BLM should not require redefining all other constants. Also, the derivation of
an MLR may be different from one user to another and may depend on the
software used. This thus requires a very thorough guideline document to
ensure homogeneity in data treatment and statistical approaches. On this
front, the complexity seems similar.
To circle back to the question, 1 think there is a technical advantage to using
MLRs (ease of use) but a scientific advantage to using BLMs (promotes
knowledge of underlying cause-effects relationships). As a scientist, 1 see the
use of MLRs as a step back, but 1 can understand the motivation of using MLRs
over BLMs. To be fair, they seem to provide just as good results so in terms of
quality of output, they are on the same level. For regulators and stakeholders,
simplicity makes sense.
most instances, the ease of use of the MLR
outweighs the mechanistic complexity of the BLM.
The reviewer is correct that the addition of new
data to the MLR would require performing new
regression analyses. However, models selected
for the derivation of the criteria will be fixed in
time and software to run them will be provided by
EPA to address user differences.
Reviewer
5
Both the BLM and MLR approach are appropriate tools for capturing important
toxicity modifying factors for the metals commonly of concern in
manufacturing, mining, effluents, and runoff. The BLM excels as a research
tool in that it is flexible, not as constrained to the training data as are MLRs,
can be modified to address mixtures, and has good application in ecological
risk assessment and other applied issues. This review provided me the first
view of some of the updates to the Windward BLM software in support of
single metal EU REACH or this CRADA project, and they are impressive.
However, in my view, for regulatory water quality criteria, the BLM approach
has fundamental key disadvantages in terms of transparency and resiliency
overtime. The present BLM software implementations and in some cases, the
speciation models (direct implementation of the WHAM submodel from its
Thank you for your comment. EPA is also
concerned about the transparency and general
useability (see EPA's response to Reviewer 4's
comment on the adoption of the 2007 Cu BLM
above) of the BLM for criteria derivation and
agrees that the MLR equations avoid these issues
without sacrificing performance.
The BLM does generate a detailed output file
showing the concentration of chemical species in
the simulation that can be compared with the BLM
equations that define the model published by US
EPA (2003). The CRADA partners indicate that the
6
-------
2.1.a Technical/scientific advantages of using one model over the other for deriving water quality criteria.
Reviewer
Comments
EPA Response
developers, for example) are the intellectual property of their developers. 1 am
not aware of any open source or public domain version of contemporary
BLMs. The code cannot be directly inspected, and the specific details of
calculations can only be inferred from the narrative descriptions and the
outputs. For EPA to rely on software based BLMs that would require a
sustained commitment to maintaining and updating the software, with
updates to make the software interoperable on different and evolving
computer operating systems, with a software testing and help desk to ensure
it is reliable on different configurations. The push in the corporate IT culture
towards enterprise software, centralized corporate control of whether
individuals can load or modify software, software white lists, and off-site
support can make the use of specialty software such as the BLM a hassle for
many. For instance, 1 had to complete this review at home on personal
computers because of such constraints. While there may be single-shingle
consultants free of such "support" most BLM users are probably in
organizations with IT controls.
Does EPA really want to be in the software business or have to support
software as opposed to putting their finite resources into new criteria or
criteria updates? Or is it fair and reliable to rely on the free services of the
model developers and their employer (or indirectly, their employer's clients)?
The MLRs sidestep all of these issues and perform fine for a wide range of
water chemistries.
BLM software can be modified to include the
details of the equations of each simulation with
the output file to increase transparency.
However, EPA recognizes the reviewer's comment
regarding the proprietary nature of most BLMs
and the issues of maintaining complex model
software.
2.1.b Are the models robust in their ability to accurately predict toxicity as a function of water chemistry? If not, why?
2.1.b. Are the models robust in their ability to accurately predict toxicity as a function of water chemistry?
Reviewer
Comments
EPA Response
Reviewer
1
In general, most models are indeed robust. This can amongst others be
deduced from the statistical parameters provided with each of the models,
and the validation efforts done for each of the models. These validation efforts
Thank you for your comment.
7
-------
2.1.b. Are the models robust in their ability to accurately predict toxicity as a function of water chemistry?
Reviewer
Comments
EPA Response
include internal validation as well external validation, whereas in some cases
additional field samples have been sampled and tested as part of the
validation. It is also to be noted that in most cases the statistical performance
of the models is well above the so-called Setubal-criteria for the acceptance of
predictive models for regulatory application as derived within the OECD.
Reviewer
2
BLM
One of the main concepts of the BLM is that there is a strong overall
correlation between log K values for gill binding and acute toxicity to the
extent that measurement of binding affinity based on gill metal binding is an
acceptable alternative to measurement of toxicity and vice versa. 1 think more
information needs to be obtained to determine whether this concept can be
extended to Ni bioaccumulation in the whole body of invertebrates rather
than bioaccumulation on a theoretical 'biotic ligand' (target site for toxicity)
such as the gills in fish. Although some studies demonstrate relatively good
agreement between the log KNiBL values derived from the ionic component of
the LC50 value (toxicity) with those derived from the ionic component of the
Kd (ionic Ni concentration causing half saturation of Ni bioaccumulation in the
whole organism - invertebrates) suggesting that whole body bioaccumulation
can serve as a surrogate for Ni binding to the theoretical 'biotic ligand' which
causes toxicity, further validation of the modeling approach of the BLM
because estimating the concentration of Ni theoretically bound to the biotic
ligand using the ionic component of the LC50 value (the BLM approach) does
not in all cases correlate with the observed Ni bound to the biotic ligand
(Leonard and Wood, 2013 Comparative Biochemistry and Physiology, Part C).
MLR
We know that invertebrates have greater diversity in ion transport physiology
and differential responses to the TMFs laid out in the documents. Therefore,
gaining more information for multiple invertebrate taxa (e.g., crustaceans,
Thank you for your comment.
For the BLM, whole organism bioaccumulation
typically has not been found to be a good
predictor of toxicity (Amiard et al. 2006)) because
aquatic organisms can sequester metals into
biologically inactive fractions. When calibrating
BLM binding constants, the accumulation is
typically measured during short term exposures on
membranes associated with the "site of action" for
metal toxicity (e.g., the gills of freshwater fish)
rather than whole body accumulation. The BLM
was first developed for fish then applied to
invertebrates and found to be empirically
accurate.
For the MLR, EPA agrees the available data on
invertebrates and algae/aquatic plant taxa within
the scientific literature is limited and, given their
abundance and ecological relevance, additional
toxicity studies on these taxa would benefit the
MLR models.
Regarding DOC, the reviewer is correct that there
are different forms of DOC which may have
different protective capabilities and affect the
bioavailability of metals in an exposure scenario
(Wood et al. 2011). However, validation exercises
8
-------
2.1.b. Are the models robust in their ability to accurately predict toxicity as a function of water chemistry?
Reviewer
Comments
EPA Response
insects, mollusks) is critical. Additionally, much less data is available for algae
and aquatic plants to TMFs and the data is currently limited to a few species
and much like the invertebrates their responses to TMFs is quite variable and
therefore substantial the importance of gaining more insight into these taxa.
General comments:
DOC
Although Brix et al. (2020) briefly alludes to the chemical composition of DOC
affecting the metal binding capabilities and thus its effect on toxicity, there is
no discussion of these difference (e.g., humic acid (HA) vs. fulvic acid (FA)).
Additionally, in the modelling, HA is set to a default of 10%. 1 think this needs
further attention and should be included in the modelling platforms or at
minimum there should be reference to what is currently known regarding the
various forms of DOC and how they differentially affect toxicity. For example,
dark, aromatic-rich compounds of allochthonous origin, with greater humic
acid content, are more effective at protecting organisms against Cu, Ag, and Pb
toxicity (Wood et al., 2011 Aquatic Toxicology). In addition, the specific
absorption coefficient of the DOC in the 300-350 nm range (SAC300-350) is an
effective index of its protective ability. PARAFAC, a multivariate statistical
technique for analysis of excitation-emission fluorescence spectroscopy data,
quantifies humic-like and fulvic-like fluorophores, which tend to be positively
and negatively correlated with protective ability, respectively (Wood et al.,
2011 Aquatic Toxicology).
Temperature
Field temperatures are much more variable than laboratory settings which
may lead to significant under-or overestimation of toxicity. This is an important
component which has been drastically overlooked in the history of metal
toxicity (Kumar et al., 2018 Int. J. Environ. Sci. Technol.).
have been performed using a wide range of
natural waters and do not indicate that there is a
need for bioavailability models to quantify the
forms of DOC in order to accurately (within a
factor of two(Meyer et al. 2018)) predict the
toxicity within an exposure scenario (Besser et al.
2021; Deleebeeck et al. 2008).
Regarding temperature, please see response to
Reviewer 2's comment in section 2.1a.
9
-------
2.1.b. Are the models robust in their ability to accurately predict toxicity as a function of water chemistry?
Reviewer
Comments
EPA Response
Reviewer
3
The models are good for predicting the acute toxicity of metals in the context
of acutely sensitive laboratory models. However, these lab models do not
adequately represent the taxa that typically dominate stream ecosystems -
aquatic insects. If the goal is to predict toxicity in simple lab tests to a limited
set of laboratory models, then models are fine for acute predictions. If the goal
is to protect aquatic life in nature, the models have limited value.
Copper: What is interesting is that there can be substantial differences in HC05
estimates depending on which type of model is employed. 1 looked at ratios of
HC05 estimates generated by the BLM relative to MLR models. Globally
(combining results from synthetic and natural waters, BLMs were more
protective (mean BLM:MLR HC05 = 0.916). These differences were driven by
the results of synthetic water tests (mean BLM:MLR HC05 = 0.569), whereas in
natural waters, the MLR approach appeared more protective (BLM:MLR HC05
= 1.292). Since most data used in the generation of WQC will likely be from
tests in synthetic waters, we can conclude that for Copper, MLRs will be
substantially less protective than BLMs. BLMs were at most 3.04X less
protective (site 51), whereas MLR's were 2 orders of magnitude less protective
at several sites relative to BLMs.
Lead: There appears to be reasonable agreement between BLM and MLR
approaches for HC05 estimates for lead. Globally the mean BLM:MLR HC05 =
1.198, with less protection afforded by the BLM in natural waters (BLM:MLR
HC05 =1.42). In synthetic waters, there is general agreement with the mean
BLM:MLR HC05 = 0.99.
Aluminum: It is interesting that MLR results are slightly more protective than
current EPA guidelines - and that it is shown in this table but not for the other
metals. 1 think this comparison should be made for all of the metals so that it is
transparent how adopting these models would change existing levels of
protection.
Nickel: For Nickel, BLM models were generally less protective than MLR
models. Globally, the mean BLM:MLR HC05 = 1.391, with smaller differences in
Thank you for your review and comment.
EPA agrees the available data on aquatic insects
within the scientific literature is limited and
additional toxicity studies would benefit the
models given aquatic insect abundance and
ecological relevance, however, the available
quality data has been applied to the calibration
and validation datasets of the bioavailability
models.
EPA and the CRADA partners are working towards
a comparison of HC0s values using the BLM and
MLRs to investigate any discrepancies and major
differences between the model predictions.
10
-------
2.1.b. Are the models robust in their ability to accurately predict toxicity as a function of water chemistry?
Reviewer
Comments
EPA Response
synthetic waters (mean BLM:MLR HC05 = 1.27), than in natural waters (mean
BLM:MLR HC05 = 1.51). There were instances where HC05 estimates varied by
3-5 fold between BML and MLR approaches (e.g., sites 25, 26, 27, 29 and 36)
Reviewer
4
As far as 1 can tell from the document summarising the results (Table 3) as well
as from the papers provided in the Appendices, they provide results that are
similar in terms of both precision and accuracy for acute values while there
seems to be an advantage for the MLR for chronic values except for Ni for
which both models gave good results.
1 would expect an MLR to do better than a BLM since there are much less
constraints for the former than the latter.
Based on the documents of Appendix D, the MLR provides better estimates of
Aluminium toxicity than the BLM. Figure 1 of Brix et al. 2020 shows much less
scatter of the data for MLR compared to BLM.
In the case of copper, overall, the BLM seems to be performing slightly better
than MLR for acute tests. However, for chronic data, MLR is best. It seems that
the quantity of data is important. When large data sets are available, both
models perform well, while for smaller data sets, MLR provide much tighter
relationships than BLM (see figures 7 and 8 of Brix et al., 2020; Appendix D).
However, uncertainty increases with less populated data sets.
As for Lead, figures 6 and 16 of DeForest et al., 2020 (Appendix E) indicate that
both models, MLR and BLM, provide similar results and scatter.
Similarly for Nickel, both models seem to perform equally well. Note that in
Table 3 of Croteau et al., 2021 (Appendix F), the reactions are written as
dissociation (ML=M + L) reactions, but the log K value suggest a complexation
(M + L=ML) reaction. Note also that the log K values in the same Table 3 suggest
that the BLM is more empirical than mechanistic. Indeed, it is counter intuitive
that a hydroxo-complex (log K = 4.357) would bind more strongly than the free
metal (log K = 4.00). The same applies to the binding of NiHCCV complexes.
Thank you for your comment.
Nickel CRADA partners have indicated that in
Croteau et al. (2021) the reactions are indeed
complexation reactions (i.e. "ML=M+L" is read as
"ML is formed by M plus L"). Additionally, the
reviewer calls out the log K value for the NiHC03,
but in the supporting documentation for the nickel
BLM, this complex was described as an empirical
adjustment to represent the added toxicity
Ceriodaphnia dubia sees from a combination of
bicarbonate and nickel at pHs above 8. It was not
intended to be mechanistic. However, other parts
of the model are indeed mechanistic in the sense
that they are calibrated to either chemical
speciation data, or metal accumulation data when
those data are available. Furthermore, the
argument that NiOH should bind less strongly than
free Ni because it has a lower charge is not entirely
accurate - the reactions for Ni and NiOH binding
organic matter are given the same constants in
WHAM, thus there is precedence that the binding
constant for NiOH need not necessarily be lower
than that of free Ni.
11
-------
2.1.b. Are the models robust in their ability to accurately predict toxicity as a function of water chemistry?
Reviewer
Comments
EPA Response
The decrease in net charge after complexation (+2 -> +1) should highly
decrease affinity of the complex for the biotic ligand. The formation of these
complexes depend on pH and Ni2+ which are also variables within the BLM.
Adding the binding of these complexes to the biotic ligand seems redundant
(or circular); it's a way to add weight to pH in a manner that pulls away from a
purely mechanistic approach. This being said, the final goal is to have a model
that predicts adequately the effects of metals on aquatic organisms and the
BLM does a great job. Although less empirical than MLRs, the BLM should also
be considered an empirical model.
Reviewer
5
Yes. The performance of all of these model variations has been well described
in the supporting documents, and all function well. I have had some minor
quibbles with Cu BLM versions over the years, such as the handling of
dissolved organic matter (DOM) has never been explained. The BLM describes
implementing WHAM V within the model, which calculates organic
complexation of Cu and other metals with DOM. But the BLM inputs ask for
dissolved organic carbon (DOC), which is not the same as DOM. Since no
adjustment is described, this implies that DOC is treated equal to DOM, which
seems to make the model a little too sensitive to DOC changes (illustrated in
Welsh et al (2008)). The Cu BLM also seems a little too twitchv with pH
changes. By its empirical nature, the Cu MLR does not have these issues. But
these are quibbles. On the whole, all of these models perform well across
diverse taxa and diverse water types.
Thank you for your review and comment.
Copper CRADA partners have indicated Reviewer 5
is correct that DOC concentration does not equal
DOM concentration. Traditionally, DOC
concentration (mg/L) is assumed to equal
0.5*DOM concentration (as mg/L; i.e., mg DOM/L
= 2*mg DOC/L]), and that is how the USEPA BLM
converts from measured DOC concentration to the
DOM input into WHAM. This 2x conversion factor
was explained in USEPA (2003: p. Al), USEPA
(2007: p. C-14), and Farley et al. (2015: p. 744, in
reference to the HDR model, which has the same
BLM framework as the USEPA BLM). However, the
reviewer's point is well taken, and such a
statement could easily be added to the BLM user
guide for more-prominent visibility.
Because mg DOM/L = 2*mg DOC/L in the BLM (and
thus mg DOM/L * mg DOC/L), the model will not,
in the reviewer's words, be "a little too sensitive to
DOC changes".
12
-------
2.1.b. Are the models robust in their ability to accurately predict toxicity as a function of water chemistry?
Reviewer
Comments
EPA Response
If by "twitchy" the reviewer is referring to the
sensitivity of predicted toxicity concentrations as
pH is changed in the Cu BLM, that twitchiness is a
consequence of Cu speciation calculated in the
combination of the CHESS and WHAM speciation
modules in the BLM. Those speciation calculations
are based on fundamental concepts of metal-
ligand interactions and empirical Cu-speciation
results. The predicted toxicity is then directly
related to the speciation-predicted concentrations
of the Cu2+ and CuOH+ bound to the biotic ligand.
In contrast, the acute and chronic Cu MLR
equations contain less "twitchiness" for pH; but as
a consequence, they predict higher toxicity
concentrations and Cu criteria at low pH (Brix et al.
2017).
2.1.c Using the information provided in Appendix G (i.e., models and example water chemistries), please provide feedback on
applying the models for the specific calculations of water quality criteria presented in terms of:
i. Complexity and transparency: are the technical details pertaining to model development and functionality clear to the user?
ii. Representativeness: do the models apply to a sufficient variety of taxa and range of water chemistry conditions?
iii. Rigor: do the modeling approaches reflect the current state-of-the-science regarding robust and unbiased data selection and
analysis?
iv. Usability: are the models sufficiently easy to use?
13
-------
2.1.c Appendix G - applying the models for the specific calculations of water quality criteria presented in terms of:
i. Complexity and transparency
ii. Representativeness
iii. Rigor
iv. Usability
Reviewer
Comments
EPA Response
Reviewer
1
In my opinion, a lot of effort has been put in making the models as transparent
as possible, including their application to specific sets of water chemistry. Any
user with a feeling for the kind of models as developed for the specific
application for setting water quality criteria is likely to be able to work with the
models in a technical sense as the model application in itself is fairly user-
friendly. Hence, the models are sufficiently easy to use. The example water
chemistries span a broad cross section of realistic water chemistries, but it is to
be made sure that in all cases there is a warning when the applicability domain
of the models is exceeded when a specific set of water chemistry is defined
(like: extreme pH-values beyond which the bioavailability models are
operational).1
The models are in general indeed applicable to a sufficient variety of taxa
although the number of taxa for which models are available, is metal-
dependent. Nevertheless, the models cover a broad array of species
representative for most of the aquatic ecosystem. Thereupon, the most
sensitive species are commonly considered.
With regard to the state of the art of the modelling approaches it is to be noted
that the methods chosen (MLR), the models indeed reflect the current state of
the art. Also, essential aspects of model development like model validation
have been properly dealt with. On the other hand, it is to be noted that the
developments within the field of informatics are progressing extremely fast
nowadays and it is recommended to explore whether applications like Artificial
Thank you for your comment. EPA agrees that the
domain of applicability is a key issue for all
bioavailability models and the range should be
published within the model user materials and the
criteria. Whenever possible EPA strives to obtain
data from a wide range of conditions. In previous
models used in criteria, EPA has provided a
warning to alert the user when the predictions fall
outside of the underlying model's water chemistry
data range.
1 In response to a request for clarification from ERG, this reviewer clarified that, by "...it is to be made sure that in all cases there is a warning when the applicability domain of the
models is exceeded when a specific set of water chemistry is defined (like: extreme pH-values beyond which the bioavailability models are operational)," he meant "...it is my
suggestion that the models be equipped with such a warning in order to make sure that the user is aware of the issue of predictions outside of the strict applicability domain of
the model."
14
-------
2.1.c Appendix G - applying the models for the specific calculations of water quality criteria presented in terms of:
i. Complexity and transparency
ii. Representativeness
iii. Rigor
iv. Usability
Reviewer
Comments
EPA Response
Intelligence/Machine Learning or related techniques like Transfer Learning can
be exploited to improve model accuracy and to warrant that the information
present in the impressive datasets, it optimally exploited.
Reviewer
2
i. Complexity and transparency
The information is clear and transparent. Inclusion of the R script significantly
adds to the transparency and functionality of the models. Increasing the
potential for these models to be used for jurisdictions other than the United
States, it may be of interest to include what other endpoints (other than the
FAVs for the U.S.) can be derived from these two models.
ii. Representativeness
The number of taxa included in most of the models (copper and nickel) is
extensive and there is strength with the aluminum model including a wide
range of invertebrates, specifically some of the more sensitive and threatened
species such as Lampsilis. However, it is essential that the life stage assessed is
disclosed because, for example, glochidia (larval stage) are much more sensitive
to metals than juvenile or adult freshwater mussels (Gillis et al., 2010
Environmental Toxicology and Chemistry; Salerno et al., 2020 Environmental
Pollution; Gillis et al., 2008 Aquatic Toxicology; Markich et al., 2017 Science of
the Total Environment). The range of water chemistries nicely brackets
environmentally relevant concentrations and combinations of TMFs.
Although the models estimate the 5th percentile of the SSD (HC5) using a range
of distribution models, one key issue which has not been addressed is Species
at Risk (SARs) or Endangered Species. Have any of these species been included
in the models? Where will they fit on the SSD? The documents should address
limitations/lack of information regarding Endangered Species and their
Thank you for your review and comment.
EPA agrees life stage is an important factor to
consider in bioavailability models and will identify
lifestages for taxa used in model development in
future reports on individual metals models.
Regarding Endangered Species, if quality data are
available for listed species they are included in
EPA's water quality criteria. Frequently no, or only
limited, data are available for listed species. Water
quality criteria for aquatic life that are developed
under the Clean Water Act Section 304(a) are
intended to protect aquatic organisms broadly on
a national scale. Regarding the Endangered
Species Act (ESA), EPA conducts subsequent
evaluations when water quality standards are
submitted by individual states to EPA for approval,
as these approvals are the relevant federal actions
that are taken, and under which consultations
related to ESA requirements are addressed.
See above for EPA's responses to the inclusion of
life stage of taxa in future reports and at section
15
-------
2.1.c Appendix G - applying the models for the specific calculations of water quality criteria presented in terms of:
i. Complexity and transparency
ii. Representativeness
iii. Rigor
iv. Usability
Reviewer
Comments
EPA Response
sensitivities towards metals. This issue needs to be addressed in the Phase 1
document and appendices.
iii. Rigor
Although, the modeling approaches do reflect most of the current state of
science, there are two key areas that need to be addressed: life stage/age of
the species included in the modeling and DOC characteristics which impact
absorption and incipient toxicity. Both issues have been outlined above.
Much of the data implemented into the two frameworks are conducted by a
handful of scientists who also developed the programs. This leads to potential
issues with biased data. Additionally, although this may be the "state-of-the-
science", in terms of an Equity, Diversity, and Inclusion (EDI) standpoint, the
first authors are not representative of the states, territories, and tribes which
these models will be serving.
iv. Usability
There are significant issues downloading the programs and running them on my
computer. Working out the issues took a few hours to manage/mitigate. The
antivirus software (AVG) was triggered with every stage of the download as well
as when the program was running. The program itself once opened and working
is easy to use and well organized. The user guides for all four metals were well
written and helpful, especially with the screen shots. 1 suggest that unzipping
the files before use should be included in every user guide. If this is a common
issue where installing software is onerous, 1 see this as a major hinderance of
using these models to support states, territories, and tribes.
2.1b for EPA's response to DOC characteristics
which impact absorption and incipient toxicity.
EPA acknowledges your concern regarding the
expert researchers involved in developing the
models which is why we have conducted this
external expert peer review of the models and
associated information to assist in identifying any
issues with biased data. EPA plans to also have the
individual selected models for each metal peer
reviewed before use in criteria derivation The
criteria documents will also undergo external
expert peer review.
Thank you for your feedback on the issues you
encountered using the models. EPA will provide a
functional and user-friendly model for criteria
derivation. EPA notes that for the 2018 Aluminum
criteria built upon an MLR approach, Aluminum
criteria calculators (one in Excel, one in R, with
identical underlying code and resultant outputs)
are housed on the EPA website, eliminating the
need for users to download software.
16
-------
2.1.c Appendix G - applying the models for the specific calculations of water quality criteria presented in terms of:
i. Complexity and transparency
ii. Representativeness
iii. Rigor
iv. Usability
Reviewer
Comments
EPA Response
Reviewer
3
i. Complexity and transparency
There is a lack of transparency in these models overall.
ii. Representativeness
This is a significant problem. If one samples a typical flowing water freshwater
ecosystem, one can expect that >90% of the sampled animal life will be insects.
There is a reason that other arms of the Clean Water Act that focus on
ecological integrity rely extensively on aquatic insect communities to make
inferences about ecological conditions. In metals contaminated streams,
alterations of aquatic insect communities are the most common and reliable
source of evidence for metals associated ecological damage. Since these models
likely are not applicable to insects (for reasons that science understands, but
are willfully ignored by both EPA and the industry groups that generated this
approach), the entire exercise is fatally flawed. Work from the Wood lab10
demonstrated that aqueous Cd exposure resulted in the uptake of Cd but not at
the expense of Ca uptake. Therefore, osmoregulatory disturbance was not
associated with aqueous Cd exposure in this tolerant chironomid species. Work
in my lab showed this to be generally true in other aquatic insect species11.
Exposure to metals known to be antagonistic to Ca transport in acutely sensitive
aquatic models (Cd, and Zn) did not affect Ca transport in aquatic insects
described as being highly sensitive to metals exposures in nature
(ephemerellids)12. Similar results were shown for metals associated with Na
transport disturbance (Ag, and Cu)13. Moreover, we showed a limited protective
effect of hardness on metal uptake in aquatic insects14. Science knows that
aquatic insects are generally tolerant to acute aqueous exposures and the
Thank you for your review and comment.
ii. As mentioned previously, EPA agrees the
available data on aquatic insects within the
scientific literature is limited and additional
toxicity studies would benefit the models given
their abundance and ecological relevance.
However, the available data has verified that
current bioavailability models are able to
accurately (within a factor-of-2) predict toxicity to
insects. A recent example is Besser et al. (2021)
where the authors performed toxicity tests
exposing nickel and zinc to the mayfly Neocloeon
triangulifera. In addition, Mebane et al (2020)
illustrated that toxicity tests of Cu, Pb, Ni and Zn
with various insect communities spanned the
entire SSD distribution for all of the metals tested
and were not all found at the sensitive end of the
SSD for any metal tested (Mebane et al. 2020b).
Lastly, mesocosm data using field collected insect
communities (including early life stages) are
available and show HCosS generated using
bioavailability models are protective of insect
communities (Roussel et al. 2007).
iii. The CRADA partners have indicated that there is
no consideration of mode of action in the selection
17
-------
2.1.c Appendix G - applying the models for the specific calculations of water quality criteria presented in terms of:
i. Complexity and transparency
ii. Representativeness
iii. Rigor
iv. Usability
Reviewer
Comments
EPA Response
reasons why6. This entire approach is only suitable for animals sensitive to
acute aqueous exposures.
iii. Rigor
The modelling approach focuses on a very narrow set of possibilities: Taxa that
are acutely sensitive to the surface binding of metals to respiratory surfaces. It
does not consider bioaccumulated metals from ingestion or toxic modes of
action that are not based on ionoregulatory disturbance. There are thousands
of journal articles about the toxicity of metals to animal life. Relatively few of
them focus on osmoregulatory disturbance as a mode of action. Metals are
toxic for a host of reasons - and the biology of cells does not differ enough
between different faunal groups to discount other modes of action and
exposure routes as important.
iv. Usability
This question should be answered by potential end users in state agencies.
of ecotoxicity data used to develop or to validate
the models. The datasets include chronic, full life
cycle, and mesocosm test results. The studies in
these datasets reflect observed acute and chronic
toxicity, regardless of mode of action.
Reviewer
4
i. Complexity and transparency
Aluminium - There were instructions for the use of the BLM but didn't find any
for the MLR. It was not mentioned how hardness was calculated for the MLR
from the raw data set which provided Ca and Mg. The actual equation for the
MLR are not apparent and one has to refer to the Appendices to actually see it.
Transparency could be improved.
1 was able to reproduce the results of the "Answer Key" document. 1 then
plotted the HC5 from both models against one another and it showed a slope of
i. Thank you for the feedback on the individual
models.
Aluminum - Reviewers were initially provided a
file that did not include the MLR equations, but
this oversight was corrected after one reviewer
inquired about the file, and the complete file with
the MLR equations was sent to the peer reviewers
Aluminum CRADA partners indicate that the
reviewer is correct that, in the case of aluminum,
18
-------
2.1.c Appendix G - applying the models for the specific calculations of water quality criteria presented in terms of:
i. Complexity and transparency
ii. Representativeness
iii. Rigor
iv. Usability
Reviewer
Comments
EPA Response
1.33 which means that BLM HC5 values were 33% higher than MLR values. This
suggests that models provide different results.
Copper - Could not find the executable file at first but was able to recover it
from FTP after sending out a request to ERG. 1 was able to reproduce the results
from the answer key without difficulties. A few other observations:
The name of the model suggests that it's chronic only, but the output file
contains headers referring to "acute values". This can be a source of confusion
for users.
Being able to switch from one language to another is a nice option. Thanks!
Program executes smoothly and quickly compared to Al or Ni.
MLR equation easy to spot compared to other metals.
MLR provides higher values, especially in the lower range. Models seem to
agree in the higher range.
Lead - Program (BLM_Ul.exe) won't load. 1 tried two different computers and
using different folder locations. Error message:
BLM and MLR predictions show less agreement
than they do with other metals. This has also been
discussed with within Brix et al. (2021) and
DeForest et al. (2020).
Copper -Thank you for the feedback on the
confusion between the naming of the model and
the output files. EPA will clarify these points in
future iterations of the models so users should not
encounter any potential confusion in the output
from either the acute toxicity option or the chronic
toxicity option.
Lead -Thank you for the feedback so that EPA can
make improvements in future iterations of the
models.
Nickel - Thank you for the feedback so that EPA
can make improvements in future iterations of the
models. The nickel chronic software should say
"FCV" not "FAV" and will be changed in future
model updates.
ii. EPA acknowledges your concern about the
range in diversity of taxa and agrees additional
toxicity studies would benefit the models.
iii. EPA acknowledges concerns about model
applicability to effluent water chemistry conditions
and is evaluating options to expand the range of
19
-------
2.1.c Appendix G - applying the models for the specific calculations of water quality criteria presented in terms of:
i. Complexity and transparency
ii. Representativeness
iii. Rigor
iv. Usability
Reviewer
Comments
EPA Response
_ Impossible d'executer le code, car borlndmm.dll est
Mfcr introuvable. La reinstallation du programme peut corriger ce
probleme.
k
OK
Apparently, 1 am missing a DLL file.
Nickel - the BLM model took about a minute to load, 1 was getting the
impression the computer had crashed or that the program was not responding.
1 didn't have this problem with the Al model.
1 used default settings which specifies "BLM" and "Chronic". The output file was
entitled "Ni test BLM_Chronic.output.xls". The headers of the last two columns
were:
HC5 (Lognormal Dist.) US EPA FAV
There were two confusing elements here. First, this was a simulation for a
chronic exposure so 1 assume that the last header should read "US EPA FCV".
Second, when comparing with the "key" data file, the HC5 columns did not
match those of the output file. But the values given in the output file under the
header "US EPA FAV" had the exact same values as those of the "key" file under
the header "BLM HC5". Either the header of the "key" file is wrong or the one
water chemistries to which the models can be
applied.
iv. Thank you for explaining the issues
encountered when using software in a different
language.
20
-------
2.1.c Appendix G - applying the models for the specific calculations of water quality criteria presented in terms of:
i. Complexity and transparency
ii. Representativeness
iii. Rigor
iv. Usability
Reviewer
Comments
EPA Response
from output file. Or perhaps 1 did something wrong. Same story for the MLR
results.
ii. Representativeness
Taxa: Some models are based on the results of one alga, one invertebrate and
one fish. There is thus lots of room for improvement of diversity.
Chemistry: 1 saw a reasonable range of pH, DOC and Ca values that would
encompass a large range of natural systems. Industrial effluents could be
outside of validation range.
iii. Rigor
Regarding data analysis, the approaches are rigorous, and the authors of the
papers have an outstanding reputation. As for the data selection, 1 can't answer
that. Review of data selection would require weeks (more likely months) of
analysis and backtracking values and literature review. This being said, the
papers were published in reputable journals and there is no reason to think that
there could be a bias in data selection.
iv. Usability
1 had no experience with the end-user BLMs, and 1 found them somewhat easy
to use with the instruction manuals. 1 did run into some problems. When
copying and pasting data from Excel to the AI-BLM software, all values after the
decimal disappeared. 1 only realised after running the program and comparing
results to the Answer Key document. The problem came from the fact that my
Excel program is in French and in French, the decimal mark is a comma instead
of a period for the English format. 1 thus had to modify the default decimal
marker in order to be able to paste values correctly. An error message would
21
-------
2.1.c Appendix G - applying the models for the specific calculations of water quality criteria presented in terms of:
i. Complexity and transparency
ii. Representativeness
iii. Rigor
iv. Usability
Reviewer
Comments
EPA Response
have been useful here. 1 had to investigate to find the source of the
discrepancy. When using the Ni-BLM, this problem got worse. The
comma/period confusion was not limited to the format in Excel. The data 1
copied from Excel was in "period" format but once pasted into the BLM model,
it was changed to a "comma" format. To fix this, this time 1 had to change
Windows settings to English and restart the computer. After that 1 could get the
model to run. Not a huge problem but being forced to switch language of my
operating system was irritating.
Reviewer
5
i. Complexity and transparency
With Al, Cu, and Pb, the MLR models are transparent and reasonably simple to
use. Not so for nickel. 1 could not find a spreadsheet or even the text
description in the articles or SI files describing the complete equation. The
pooled MLR calculates the FCV as a function of hardness and DOC plus an
intercept, but nowhere in the documentation or in the numerous output files
could 1 find a value that the intercept for the HC5 or FCV. For example, the
output file "Ni-inputs.ssdnormalized.xls" in column AC has "MLR intercept"
values but these vary by each test and the intercept for the FCV should not
vary. Obviously the intercept is in the model files somewhere since it works.
This is a minor matter that likely would have quickly been cleared up in an email
with the developers had the review not been explicitly sequestered by the peer
review manager. The explanations of BLM development in the respective
articles is reasonably detailed.
ii. Representativeness
They seem to. The draft report and most of these models may be a bit
overstating the case in that they address "invertebrates" or for the MLRs, that
Thank you for your comments. EPA will display the
Ni MLR in a similar, spreadsheet format as the
other metals for future iterations of the models.
Nickel CRADA partners responded that there was
originally a question of whether the Ni MLR SSD
would be normalized by one pooled model or
multiple models, thus the format (since the whole
SSD would need normalization to each set of
chemistry and then the HC5/FAV/FCV calculated, if
there were multiple models). Because of this, we
did not have a single intercept for the
HC5/FAV/FCV calculated. These are quite easy to
calculate, however, and the final equations using
the "Pooled AN" slopes for each SSD should be:
Acute lognormal HC5
(ug/L)=exp(0.475*ln(Hardness,mg/L)+0.148*ln(DO
C,mgC/L)+2.8220)
22
-------
2.1.c Appendix G - applying the models for the specific calculations of water quality criteria presented in terms of:
i. Complexity and transparency
ii. Representativeness
iii. Rigor
iv. Usability
Reviewer
Comments
EPA Response
they include "invertebrate models" when in fact, the invertebrates tested were
mostly daphnids. The very different phylogeny of crustaceans from aquatic
insects has led to strong criticisms of using crustaceans to represent freshwater
"invertebrates" (Poteat and Buchwalter 2014). All the models are relatively rich
in fish and daphnid data.
To test if the models and associated EPA-style final chronic values (FCV) or 5th
percentile hazardous values (HC5) values calculated from the species sensitivity
distributions (SSDs) compiled as part of the model development were
protective of insects, 1 calculated the FCV/HC5 values from the models and
compared them to Cu and Ni FCV/HC5 values that my colleagues and 1 had
recently updated by added aquatic insect chronic values from community
testing (Mebane et al. 2020b). With Ni, the model FCV/HC5s appeared to scale
appropriately to the test conditions and appeared to be fully protective of the
aquatic insects tested. For the conditions tested (hardness 17.5 mg/L, pH 7.67,
DOC 3 mg/L), the Ni MLR produced a HC5 of 3.3 ng/L Ni and the EPA FCV
equation 1.3 ng/L. The Ni BLM produced similar values (4.7 and 1.4 ng/L) for
the community test water conditions. The lowest NOEC (no observed effect
concentration) with any insect species or insect community metric was 9.5
Hg/L. Algae was affected by nickel at the lowest concentration tested, but the
practice in USA criteria, hazards to algae have not been given the same level of
concern as have effects to aquatic animals
With Cu, the model FCV/HC5s also appeared to scale appropriately, but the
SSDs updated with insect values were lower than the model FCV/HC5s. This
potential underprotectiveness is a function of the different SSDs, not the
models. For the same conditions tested (hardness 17.5 mg/L, pH 7.67, DOC 3
mg/L), the Cu MLR produced a HC5 of 6.2 ng/L Cu and the BLM produced a
lower value (4.7 ng/L). EClOs for reductions in overall taxa richness in the Cu
US EPA
FAV(ug/L)=exp(0.475*lnL'
-------
2.1.c Appendix G - applying the models for the specific calculations of water quality criteria presented in terms of:
i. Complexity and transparency
ii. Representativeness
iii. Rigor
iv. Usability
Reviewer
Comments
EPA Response
tests were 2.6 to 3.4 ng/L (the Cu test was repeated), with some mayfly taxa
EC20 values below the BLM and MLR calculated HC5 values of 4.7 and 6.2 ng/L
(Baetis, Diphetor, Ephemerella). This suggests that the model criteria
adjustments are appropriate but that the Cu criteria SSD should be updated to
account for sensitive insect taxa.
Other non-fish, non-daphnid datasets 1 was familiar with and compared with
include acute mayfly (Baetis) tested in natural waters with a range of hardness
and pH values (Mebane et al (2012), included in the DeForest Appendix E
comparisons) and acute and chronic freshwater mussels with varying hardness,
pH, and DOC (Wang et al. 2009; Wang et al. 2011). The models performed well
with these "nonstandard" taxa. Note also that the Pb and Ni models included
Lymnaea snails in their development.
1 just don't see any major animal taxa for which the model performance gives
great pause, and the BLMs and MLRs have been tested with pretty diverse
artificial and natural waters. While MLRs have been shown to work well with a
wide variety of waters, the power of the BLM approach is that due to its
mechanistic underpinnings, BLMs can often function well beyond their
calibration datasets. This is one more reason BLMs should be kept in the quiver
of potential tools that can be employed in risk assessment or site-specific
criteria development. For instance, BLMs can handle strange Ca:Mg ratios and
other uncommon chemistry reasonably well (Van Genderen et al. 2007). MLRs
fall apart under such scenarios.
iii. Rigor
Yes. 1 think the CRADA crowd should be commended for their work with
primary datasets from the literature and for generating necessary data. In
particular, they avoided the trap that some prominent related efforts have
al. (2020) results and other mesocosm results into
the Cu SSD. A consequence of updating the toxicity
database for Cu or any other metal, with or
without incorporating mesocosm data, is that the
MLR equations would have to be revised to fit the
updated database. That would be more complex
to address than for the BLM, for which only the
critical accumulation value (CAV) might have to be
recalculated to be consistent with the HC5 of the
new SSD.
EPA agrees that the development of empirical
models like MLR can be informed by mechanistic
models like the BLM by helping to identify the key
TMFs and expected mechanistic patterns and by
evaluating MLR models against existing BLMs.
EPA acknowledges the comment that the MLR
should optimally not be used beyond its
calibration dataset and is evaluating options to
expand the range of water chemistries to which
the models can be applied. EPA has in the past, in
limited ranges, applied the MLR model beyond the
calibration dataset when the values yield lower,
more protective criteria than if one used the
24
-------
2.1.c Appendix G - applying the models for the specific calculations of water quality criteria presented in terms of:
i. Complexity and transparency
ii. Representativeness
iii. Rigor
iv. Usability
Reviewer
Comments
EPA Response
fallen into - the incautious reliance on the EPA EcoTox database. Despite the
EcoTox statement that it is "a comprehensive, publicly available Knowledgebase
providing single chemical environmental toxicity data on aquatic life,.." updates
have been ad hoc on a chemical-by-chemical basis and the database does not
appear to have been updated for metals in more than 10 years.
iv. Usability
Yes, mostly. The (not yet public) Windward BLM updates included in this review
were clearly explained and ran without hiccups. The Al, Cu, and Pb MLR models
were straightforward. Rolling the Ni MLR into the BLM software is a nice
comparative touch, but the Ni MLR obviously also needs to be available as a
standalone spreadsheet.
criteria at the limit, in order to address stakeholder
needs (e.g., Al MLR).
EPA notes that the ECOTOX Knowledgebase is
updated periodically, including special targeted
updates that are conducted during criteria
development for specific chemicals, including
metals.
2.2 Please provide your overall review of the approaches used to compare and evaluate the BLM and MLR models for the metals
addressed in the Phase I document and appendices.
2.2 General comments for the approaches used to compare and evaluate the BLM and MLR models.
Reviewer
Comments
EPA Response
Reviewer
4
This is difficult for me to say as 1 am not a specialist in model
performance assessment but as far as 1 know, the approaches
used were convincing and credible. 1 have no alternative
approach to recommend.
Thank you for your comment.
25
-------
2.2.a Are the approaches presented consistent with the state-of-the-science?
2.2.a Consistency with the state-of-the-science.
Reviewer
Comments
EPA Response
Reviewer
1
As far as 1 can judge, the approaches are indeed consistent with
the state-of-the-science with regard to the type of modeling
applied. As already indicated above, nowadays more advances
informatics and bioinformatics tools are becoming increasingly
available and most likely, these tools might be considered more
advanced than for instance MLR models. Nevertheless, in my
opinion the models developed are well suited for the purpose of
quantifying metal bioavailability.
Thank you for your comment.
Reviewer
2
Yes, generally the approaches presented are consistent with the
state-of-the-science, however, 1 feel as though certain aspects
were not addressed adequately. These have been previously
addressed in sections 1 b. DOC and 1. C. ii. and include the
various forms of DOC and how they differentially affect toxicity
and disclosing the life stage/age of the species implemented into
the modeling.
Thank you for your comment. Please see EPA's responses to the
reviewer's concerns above in section 2.1b and 2.1c.
Reviewer
3
The approaches are consistent with the state of the science for
organisms acutely sensitive to aqueous metal exposures only.
The models ignore a large body of science relating to dietary
exposures because this science does suit the goal of relaxing
environmental protection. It is remarkable that the possibility of
dietary exposures is ignored in the main document when these
industry groups have compiled a robust bibliography of
references on the topic (see Appendix 1). Willfully ignoring
science that does not meet set intentions will not make that
science go away. It is incumbent on EPA scientists to appreciate
that these models represent science with a set goal in mind, and
that goal is not purely about protecting aquatic life. The
fundamental underlying premise here is that if a water body can
Thank you for your comment and providing the references for
EPA's review in Appendix 1. Please see responses to Reviewer 3
in sections 2.1a and 2.1c regarding dietary exposures. More
information on the dietary exposure route has been added to
Section III of the report. In addition, Mebane et al. (2020)
suggested there is currently "insufficient evidence to conclude
that bioavailability models would be under-protective if based on
waterborne-only exposures" and recommended that researchers
conduct concurrent exposures to strengthen the literature
surrounding dietary exposure and support the development of a
biodynamic modeling framework that is able to incorporate the
dietary exposure route (Mebane et al. 2020a). Lastly, where it is
well-established that the diet is an important exposure route,
26
-------
2.2.a Consistency with the state-of-the-science.
Reviewer
Comments
EPA Response
absorb more pollution, then more pollution should be
permissible. This is dangerous from the perspective of persistent
contaminants that are very expensive to clean up after the fact.
EPA has considered this information in their criteria
development. For example, the selenium water quality criteria
(US EPA 2016) use fish tissue concentrations as diet is the
primary route of exposure.
Reviewer
5
Yes, the comparisons are consistent with those suggested in the
2017 SETAC experts meeting, and appear to be evenhanded, and
statistically robust.
Thank you for your comment.
2.2.b, Can you identify other approaches that could be used to compare the models?
2.2.b Other approaches that could be used to compare the models.
Reviewer
Comments
EPA Response
Reviewer
1
No doubt, other advanced tools are available from within the
field of (bio)informatics. 1 am, however, not aware of the details
of such alternative tools and approaches. For now, the
comparison made with regard to the performance of the BLM
and MLR models, is sufficient to warrant confidence in the
models and in the selection of the best model.
Thank you for your comment.
Reviewer
2
It would be helpful to provide multiple data sets; some with
common water chemistries and then highlight some more
complex water chemistries for example wastewater effluent
where different combinations of the TMFs are observed.
Thank you for your comment. EPA will consider adding more
datasets (beyond the natural and artificial datasets provided with
this review) to future peer reviews of the individual metals
models.
27
-------
2.2.b Other approaches that could be used to compare the models.
Reviewer
Comments
EPA Response
Reviewer
3
1 don't have any recommendations here but 1 think there could
be more serious treatment about model differences in synthetic
vs natural waters.
Thank you for your comment.
As mentioned previously in response to Reviewer 3's comment in
section 2.1b, EPA and the CRADA partners are working towards a
comparison of HC0s values using the BLM and MLRs to
investigate discrepancies and major differences between the
model predictions.
Furthermore, toxicity tests for metals in both synthetic and
natural waters can be important sources of information for
model development and testing and model application in
estimating protective values under real world conditions. Most
of the toxicity data used in guideline development was
developed in synthetic waters, partially because EPA prefers data
generated in standard toxicity testing regimes (including using
waters with low DOC) to provide consistency across species and
chemicals regarding relative toxicity and to reduce confounding
factor interference,. The consistent exposure conditions that can
be obtained with synthetic waters are also useful for chemical
adjustments of modifying factors to determine how water
chemistry affects metal toxicity (for example, testing over ranges
of hardness, pH, or DOC) and these types of experiments are
useful for model development. Natural waters are useful for
validating the model in real world conditions. The CRADA
partners indicate that the application of the model to synthetic
and natural waters was clearly specified in the papers that were
submitted in support of the nickel BLM and MLR (Appendix F).
Similar breakdowns of synthetic and natural waters have not
been compiled for other metals, but this information is available
in the sources cited for toxicity data.
28
-------
2.2.b Other approaches that could be used to compare the models.
Reviewer
Comments
EPA Response
Reviewer
5
Well yes, there is no end to ways the models could be compared,
but 1 don't know of other approaches that should be used. The
models essentially produce paired groups and there are all sorts
of statistical methods for group comparisons. Likewise, there is
no end of different species and waters and speciated vs.
dissolved metals models, of combined food and water pathways.
1 think the present set of comparisons is at the point of
diminishing returns. Time to move on to other metals.
Thank you for your comment.
2.3 Please comment on the use of a limited set of toxicity modifying factors to estimate toxicity using both the MLR and BLM
approaches (i.e., compared to the full parameter set used to derive ambient water quality criteria for copper in EPA 2007),
a. Please provide feedback on limiting toxicity modifying factors to a set of a priori determined parameters (e.g., pH,
2.3 Use of a limited set of toxicity modifying factors to estimate toxicity using both the MLR and BLM approaches.
Reviewer
Comments
EPA Response
Reviewer
1
There is a wealth of data showing that a limited set of toxicity
modifying factors is capable of capturing most of the impacts of
water chemistry on metal bioavailability. In general terms my
estimate would be that over 90 % of the possible impacts of
water chemistry on metal bioavailability, is properly considered.
This implies that it can never be ruled out for 100 % that in
specific cases not considered so far, additional toxicity modifying
factors might be of importance - even apart from the full
parameter set use in EPA 2007. This is inevitable, and there is no
solution but to accept that models cannot be for 100 % accurate.
Thank you for your comment.
29
-------
2.3 Use of a limited set of toxicity modifying factors to estimate toxicity using both the MLR and BLM approaches.
Reviewer
Comments
EPA Response
Reviewer
2
There is strength in an approach that simplifies the BLM model
from ~10 parameters to 3-4 parameters. In many cases, these
additional parameters are not determined or inaccurate which
leads to either inputting estimates or leaving values at the
default settings. Requiring more variables also increases the
potential and impact of human error for derivation of accurate
water quality criteria for the protection of aquatic life.
However, as mentioned above, there is a need to include
temperature as a fourth parameter. Metal accumulation in fish,
pond or river water is enhanced by upsurges in temperature;
therefore, it is imperative to study the detrimental effects of
metals in combination with temperature to formulate accurate
predictive models (Kumar et al., 2018 Int. J. Environ. Sci.
Technol.). This is an area which has been grossly overlooked in
metal toxicology.
Thank you for your comment.
EPA agrees there is strength in a simplified parameter set for
end-users.
Regarding temperature, please see the response to Reviewer 2's
comment in section 2.1a.
Reviewer
3
There is no doubt that each of these TMFs are important. There
should be balance between TMFs that relax protection with
TMFs that potentially would require additional protections. It
would be great if the influence of temperature was well
understood in metal toxicity, but unfortunately it is not. At this
time of writing the Pacific Northwest is experiencing an
unprecedented heat wave. Does anyone think the effects of
pollutants are not exacerbated under these extreme conditions?
It is progress that temperature is recognized is a potentially
important TMF, but we are nowhere close to being able to
address it at the level of criteria development.
When science emerges that highlight the potential risks of
metals from dietary exposures for example, it is largely ignored
by the metals industry groups that are promoting this modeling
Thank you for your comment.
Regarding temperature, please see the response to Reviewer 2's
comment in section 2.1a.
Regarding dietary exposure, please see the responses to
Reviewer 3's comments in sections 2.1a, 2.1c and 2.2a.
Furthermore, the 1985 Guidelines do not explicitly exclude the
consideration of dietary exposure. EPA believes it is important to
focus on the primary drivers of toxicity, which for the metals
evaluated in this effort, appear to be aquatic exposure (as noted
above).
30
-------
2.3 Use of a limited set of toxicity modifying factors to estimate toxicity using both the MLR and BLM approaches.
Reviewer
Comments
EPA Response
approach. It is remarkable that this work is being sold as state-of-
the-science when there is no recognition of the contributions of
Luoma, Cain, Hare, Fisher, Rainbow and others that do not fit this
aqueous exposure paradigm. This is partially the fault of the
antiquated 1985 guidelines for excluding dietary exposures and
partially a function that considering things that could argue for
strengthening environmental protection is not in the interest of
these metals groups. This effort is all about reducing
overprotection - not protection.
Reviewer
4
When building an empirical model, one must be cautious about
the domain of validity of the model and no extrapolation can be
made. It follows that extensive documentation must be provided
to guide the users for the applicability of the MLR within the
conditions that were used to build the model, even for
parameters that were not considered significant. If a parameter
is not measured and is well outside of the range of values used
for model calibration, the model may be off without the user
being aware of it. For example, if the MLR for Ni does not require
pH as input, it is still an important parameter as some organisms
may not tolerate this pH. The same applies for any parameter
that would be outside of the range of values present in the
calibration data set. In other words, less input data may be
convenient, but it increases the probability of a wrong
conclusion. Range of applicability of water chemistries should
not be limited to the parameters used in the MLR but perhaps
this is already specified, and 1 missed it in my review of the
numerous documents provided.
Temperature - 1 think temperature is only pertinent for Al which
may often exceed solubility. Adequate prediction of the
Thank you for your comment.
EPA acknowledges the need for understanding the range of
conditions that correspond to model development and testing
and that caution and scientific judgement must be applied if
application of the MLR beyond its calibration dataset is
considered. EPA is evaluating options to expand the range of
water chemistries to which the models can be applied.
Furthermore, tables of water chemistry boundaries are included
in the BLM and noted in the user guide, and this information will
be provided with any future models used for criteria derivation.
Lead CRADA partners have indicated that the reviewer is correct
that Pb solubility is affected by the presence of phosphate and
phosphate depletion can cause growth inhibition in plants and
algae. To address this issue, algae and plant tests conducted for
lead toxicity were optimized for phosphate content in the test
media and lead speciation was calculated where needed. The Pb
MLR model for algae was based on 15 Pseudokirchneriella
subcapitata tests reported in De Schamphelaere et al. (2014) and
2 P. subcapitata tests reported in Nys and De Schamphelaere
(2017). In De Schamphelaere et al. (2014), tests were conducted
with organic phosphorous (i.e,. glycerol-2-phosphate) to prevent
31
-------
2.3 Use of a limited set of toxicity modifying factors to estimate toxicity using both the MLR and BLM approaches.
Reviewer
Comments
EPA Response
dissolved concentration is key. 1 don't see any other elements in
the given list of metals for which temperature would be critical.
1 would point out, as an example, that Pb is poorly soluble in the
presence of phosphate. Phosphate has never been mentioned in
the documents (my apologies if 1 missed it) but it is a required
nutrient for plants and usually present at high concentrations in
standard tests for plants and algae. Growth inhibition can be
wrongly interpreted as an effect of Pb while in reality it could be
the lack of available phosphorus that would decrease growth.
Speciation calculations would flag this while an MLR wouldn't.
lead-phosphate mineral precipitation. Those authors conducted
Pb speciation calculations that "strongly indicated" addition of
glycerol-2-phosphate to the alga test waters had no effect on
free Pb2+ activities.
In the Nys and De Schamphelaere (2017) study, Pb toxicity tests
were conducted under both low and high phosphorus conditions
(10 and 100 ng P/L, respectively). Phosphorus was added in
these tests as NaH2P04'2H20. The 2 tests with low phosphorus
concentrations did not meet test validity criteria and were
excluded from the MLR model evaluation. The 2 tests with high
phosphorus concentration did meet test criteria and were
included in the MLR evaluation. The MLR model was "driven" by
the 15 tests from De Schamphelaere et al. (2014), and the
predicted Pb EC20s for the 2 tests from Nys and De
Schamphelaere (2017) were a factor of 1.0 and 1.8 different
from the observed EC20s.
Based on the Pb speciation calculations previously described in
De Schamphelaere et al. (2014), it is not believed that the growth
effects in the Pb toxicity tests with P. subcapitata were
potentially caused by an absence of bioavailable phosphorus due
to formation of insoluble lead-phosphate. Further, the Pb MLR
model that was largely based on tests from De Schamphelaere et
al. (2014) accurately predicted Pb toxicity in 2 tests in which
phosphorus was added as phosphate. This provides an example
where speciation calculations can be used to inform selection of
data sets used for MLR model development.
32
-------
2.3 Use of a limited set of toxicity modifying factors to estimate toxicity using both the MLR and BLM approaches.
Reviewer
Comments
EPA Response
Reviewer
5
Hardness, pH, and DOC have been shown able to capture the
majority of metals toxicity variability in laboratory settings. 1
have never seen a quantitative analysis of why hardness is better
than Ca. No BLM uses hardness. Yes, there is some evidence that
Mg offers some protection to daphnids, but there is lots of
evidence of Ca giving greater protection (Welsh et al. 2000;
Naddv et al. 2002). 1 suspect that the real reason for reiving on
hardness rather than Ca is the policy desire to keep a lineage to
the old hardness-based criteria. 1 also suspect that the empirical
performance of MLRs with Ca or hardness would be similar for
most waters. If this is the case, some quantitative comparison
and a statement of policy heredity might be appropriate.
In regard to temperature, there is evidence that animals may be
more sensitive to metals when tested either well below or well
above their temperature optimums (1 can dig out references
upon request). However, 1 question whether this is a metals
toxicity modifying factor or a multiple stressor, or if this fine
distinction even matters. Adding more factors really complicates
implementation, for temperatures can swing >10°C over the
course of the day, and we already have an underappreciated
problem with daily pH cycles that commonly swing over 0.5 units
in waters and up to at least 2 units. A 0.5 pH swing is a big deal in
any of these models, and diurnal variability in pH has not been
considered in any of these approaches. It should be.
Thank you for your comment.
As a user-friendly modeling version, MLRs use hardness because
most end-users monitor hardness rather than Ca. One line of
evidence that validates the use of hardness instead of Ca and/or
Mg concentrations is the consistency in the result from cross-
validation exercises comparing the BLM and MLR predictions.
Consideration of Ca:Mg ratios could be included in future
guidance for criteria development and implementation, but the
practical utility of models and criteria would be reduced by
introducing the need for additional data that is not typically
collected in state sampling programs. EPA plans to develop
criteria that can be broadly implemented.
Diurnal variability in various parameters including pH (diurnal
fluctuations in temperature was previously discussed by
Reviewer 5 in Section 2.3) can be a consideration within the
EPA's conceptual models being developed as part of the criteria
derivation process.
EPA agrees that temperature could be an important toxicity
modifying factor for some metals, but there is not enough data
on temperature to incorporate into the models at this time.
Please also see response to Reviewer 2's comment in section
2.1a.
2.4 Please provide recommendations on potential software platforms/tools (e.g., Excel, R, or other freestanding programs) that
could/should be used to perform MLR and BLM calculations.
a. Please discuss advantages and disadvantages of any software platforms/tools.
33
-------
2.4 Recommendations on potential software platforms/tools that could/should be used to perform MLR and BLM calculations.
Reviewer
Comments
EPA Response
Reviewer
1
What 1 experience is that the number of data and the number of
models for individual metals (and hence the overall set of data
and models) is increasing. In my experience this means that tools
like Excel cannot be used anymore given their limitations when
dealing with large amounts of (complex) data. Instead, the
number of R-applications as well as the number of advanced
modelling platforms is quickly increasing. Also, modelling
platforms are in development which allow the user to
systematically store data and models, and to use this information
to develop and integrate models and data according to the wish
of the users. It is recommended to explore the new generation of
software platforms and tools which are quickly becoming
increasingly user-friendly.
Thank you for your recommendations. Both an Excel and R
application were provided for use in the 2018 EPA Aluminum
criteria. EPA will also consider other new software platforms.
Reviewer
2
There are many advantages of using R over Excel. R can handle
very large datasets and automate and calculate much faster than
Excel. The reproducibility of R source code is much more
advanced and easier to use than Excel and there are community
libraries of R source code which are available to all. R has more
complex and advanced data visualization. Lastly, which may have
the most significance with broad demographics of people who
will be using these models, R is free and Excel is not.
However, Excel is still a powerful tool for smaller datasets, basic
data entry, simpler functions and formulas, and viewing raw
data. 1 tend to think that more of the general population is
familiar with Excel and will more readily use Excel. R is
overwhelming and may cause more mental barriers in using the
models.
1 cannot comment on programs such as Python, Matlab, SAS, and
SQL which may be arguably better.
Thank you for your recommendations that EPA consider R-
applications for large datasets and potentially Excel for a user-
friendly option for smaller datasets.
34
-------
2.4 Recommendations on potential software platforms/tools that could/should be used to perform MLR and BLM calculations.
Reviewer
Comments
EPA Response
Reviewer
3
1 have no comments or recommendations about which platforms
should be used to make these calculations.
Thank you for your comment.
Reviewer
4
Ideally, online tools should be provided to prevent misuse of
user-owned platforms. This could also prevent issues related to
regional settings (see answer to Question lc above).
Thank you for your recommendation. EPA aims to prevent
misuse and general user issues of the tools.. EPA notes that for
the 2018 Aluminum criteria (MLR approach), criteria calculators
(one in Excel, one in R) are housed on the EPA website,
eliminating the need for users to download software.
Reviewer
5
A major feature of MLRs is that they don't need a specific
software platform. An equation yields the same answer for given
inputs no matter whether it is calculated in an xlsx spreadsheet,
Google Sheets, Open Office, R script, Python, C code, hand
calculator or longhand. It doesn't matter. Imagine if EPA had
provided software to calculate the 1984 Pb criterion. 1 think the
Mac debuted that year, some precursor to MS-DOS was going,
.... Certainly, when it comes time to publish MLR based criteria,
certainly providing some calculation tools such as in xlsx
spreadsheet format and R would be helpful. At present, 1 think
spreadsheet formats have the advantage since they can readily
hold data in most a human-readable format as long as some care
to structure tables in lightly formatted forms that are easily
exported to csv and R. Note that "Excel" and "xlsx" are not the
same thing. "Excel" is a proprietary Microsoft application; "xlsx"
is a non-proprietarv spreadsheet open standad, part of the Open
Office XML standard. At the present, 1 would sav that the "xlsx"
Open XML spreadsheet format would be most widely accessible
and transparent to most users, but that R users are closing the
Thank you for your recommendations and clarification on Excel
vs xlsx format. EPA will consider continuing to provide R-
applications and Excel applications for various users.
35
-------
2.4 Recommendations on potential software platforms/tools that could/should be used to perform MLR and BLM calculations.
Reviewer
Comments
EPA Response
gap. It would not be a big lift for R aficionados to pull information
in from spreadsheets to work with.
2.5 Please provide any additional suggestions that you feel would improve the report.
2.5 Additional suggestions that would improve the report.
Reviewer
Comments
EPA Response
Reviewer
1
My key suggestion is that one overarching approach is chosen for
deriving water quality criteria for metals that take account of the most
important toxicity modifying factors. What is important, if only to gain
sufficient confidence of non-experts, is to not only indicate the merits of
the overarching approach, but to also mention the limitations and the
'domain of applicability' of the models underlying the overarching
approach. These domains may be metal-dependent, and do not include
extreme water chemistries (the more as physiological limitations of
most biota limit the applicability of the models in extreme
environments).
A final suggestion is to take count of interactions between toxicity
modifying factors as such interactions are likely to affect toxicity.
Thank you for your comment. EPA will discuss both the
advantages and limitations of the chosen overarching
approach in both this report and when developing the
individual metals models/criteria.
As exemplified in the case studies presented with the
report (Appendices D, E, and F), EPA will consider
interactions between toxicity modifying factors (as was
also demonstrated in the 2018 Freshwater Aluminum
Criteria) in the development of all metals models.
Reviewer
2
p.3 section a. pH - bioavailability should be changed to bioavailability
(remove extra "i").
In Canada, the government has a dutv to consult (https://www.rcaanc-
cirnac.gc.ca/eng/1331832510888/1609421255810), and where
appropriate, accommodate Indigenous groups when it considers
conduct that might adversely impact potential or established Aboriginal
or treaty rights. The goal is to listen to the views and concerns of
affected Indigenous groups and, where necessary and possible, modify
This edit has been made to the text.
Thank you for your comments.
36
-------
2.5 Additional suggestions that would improve the report.
Reviewer
Comments
EPA Response
the action or decision to avoid unlawful infringement of those rights.
This may be an important consideration when using these models to
support states, territories, and tribes.
Reviewer
3
The report should provide a table showing what the WQC would be
under different water chemistry conditions for the different metals with
columns for the current criteria, what a BLM based criteria would be,
and that the MLR based criteria would be. There should be transparency
about how WQC would be altered from the current values under a wide
range of water chemistry conditions.
1 have never seen any proof or analysis that demonstrate that current
criteria are egregiously over protective. 1 think this is important to show.
This exercise is using taxpayers' dollars to revisit metals criteria yet
again, when the agency is woefully behind in establishing criteria for
thousands of relevant pesticides, industrial pollutants and personal care
products.
On p. 3, section II, there is a statement that toxicity is dependent on
route of exposure, however the entire modeling approach is only based
on direct aqueous exposures. This is a regrettable byproduct of the 1985
Guidelines document's focus on aqueous exposures only. This issue
should be fixed immediately. In Mebane et al, 2020s, there is the
recommendation that "for best practice in the future, that during
chronic tests combined waterborne and dietary matched exposures
should be performed. These should be based on natural live diets that
have undergone full biological equilibration with the waterborne metal
through pre-exposure." These authors comment that very few data of
this type exist. The reason more of these data don't exist is because
there is no market for this information. EPA should require these data
rather than excluding them in the criteria process. My laboratory has
shown a path forward for these type of experiments with a relevant
aquatic insect model4,15"20 as both an end receptor and as a food source,
EPA is considering development of a table like the one
described by the reviewer.
Regarding dietary exposure, please see EPA's responses
to Reviewer 3's comments in sections 2.1a, 2.1c and 2.2a.
Regarding synthetic vs natural waters, please see
response to Reviewer 3's comment in sections 2.1b and
2.2b.
EPA agrees that invertebrates and fish are the most
common species within toxicity databases, however, the
impact of TMFs has been studied in numerous species
including invertebrates, fish, amphibians, insects, plants,
and algae.
37
-------
2.5 Additional suggestions that would improve the report.
Reviewer
Comments
EPA Response
but WQC constructed with the antiquated 1985 guidelines would
exclude these data from consideration for having dietary exposures
associated with them. It is remarkable that a scientific flaw as egregious
as this is allowed to persist in criteria derivation.
There is little attention given to the differences between BLM and MLR
approaches in natural waters vs synthetic waters (e.g., see copper
results above). It is not clear to me what the relative proportions of
toxicity data exist for synthetic vs natural waters, but this should
probably be addressed quantitatively in more detail in a final report.
Finally, there needs to be more attention given to the extrapolation of
TMFs based on 2 taxa to represent thousands of other species. The
distinction between fish and invertebrates is a nice start, but 1 don't
know how people could be comfortable with these extrapolations. 1
have similar discomfort with the application of Acute to Chronic Ratios
(ACRs) in situations where chronic data are limited. Some of Chris
Mebane's work on this area21 needs to be studied by EPA scientists.
Reviewer
4
The document refers to "binding sites on the gill surface or respiratory
surface" on two occasions. This is a too narrow description of the biotic
ligands that only applies to animals. A more generic description would
be "surface binding sites leading to internalization and effect".
On page 2, "...simple linear regression models...", 1 think several of these
are not linear.
On page 3: "The effect of a number of metals on aquatic organisms is
not well predicted by the total metal concentration (or total dissolved
concentration), but rather the bioavailable forms (e.g., the free metal
ion) which is a function of many modifying factors that affect the
speciation, bioavailability, and toxicity of metals." This is an incorrect
wording. Although widely used in the literature, 1 would like to (at least
try to) convince the authors to refrain from using these terms.
Thank you for your comments and suggestions.
Added suggested text on page 1 of the report regarding
the description of biotic ligands.
Text regarding the discussion of bioavailability was
refined on page 3 of the report.
38
-------
2.5 Additional suggestions that would improve the report.
Reviewer
Comments
EPA Response
Bioavailability is a relative concept, not an absolute one. A metal can be
more or less bioavailable depending on ambient conditions, but one
cannot identify a "bioavailable form" or "fraction". In fact, 1 would argue
that all forms are bioavailable because all forms can dissociate. Overall,
there is a mathematical relationship between the free metal ion
concentration and uptake / toxicity, but this does not mean that only
the free species is bioavailable. A metal complex can also react with a
binding site and, by a ligand-exchange reaction, release the original
ligand prior to internalisation. In such a case, the mathematical
relationship between the binding surface and the free ion remains the
same even though the complex was the reacting species. 1 refer the
authors to page 55 of Campbell (1995) for a development of this point:
Putting aside this possible complication for the time being, let us now
consider the implications of assumption (3) (fast transport and adsorption/
desorption kinetics). There are frequent references in the literature to the free-
metal ion as the 'toxic' or 'bioavailable' species.20-25""27 However, if it is assumed
that the cell surface is in equilibrium with the various metal species in the bulk
solution, and that this equilibrium precedes the expression of the biological
response, it follows that the identity of the metal form(s) reacting with the cell
surface is of no biological significance—no single species in solution can be
considered more (or less) available than another. Though this point was made
quite explicitly by Morel,18'28 who referred to the 'profound and widespread
misconception that hydrated metal species is the active one', it has often been
overlooked. In a system at equilibrium, the free-metal ion activity reflects the
chemical reactivity of the metal. It is this reactivity that determines the extent
of the metal's reactions with surface cellular sites, and hence its
'bioavailability'.
Another good paper on this topic is that of Meyer (2002). An easy fix to
this would be to replace "bioavailable form / fraction" by "metal
bioavailability". In other words, one can say that the bioavailability is
39
-------
2.5 Additional suggestions that would improve the report.
Reviewer
Comments
EPA Response
greater / lower in experiment A vs B, but one cannot say that there are
more or less bioavailable forms in A vs B.
Suggested rewording: "The effect of a number of metals on aquatic
organisms is not well predicted by the total metal concentration (or
total dissolved concentration). Metal bioavailability is a function of
many modifying factors that affect the speciation and toxicity of
metals."
Page 3: "In addition, the BLM also accommodates temperature as a
modifying factor for some metals, such as for aluminum (Santore et al.
2018)". It's not clear how temperature influences bioavailability of
Aluminium without reading Santore. This is related to Al solubility which
is sensitive to T in a range pertinent to a natural exposure scenario. Role
of T should be clarified as the reader may think this is a physiological
parameter.
Page 3: "The second way is by competing with metal ions for binding
sites on organisms (e.g., competition from H+, Ca2+, and Mg2+) which
The discussion of temperature as a TMF was moved to
page 4 of the report.
Suggested deletion was made in text on page 3 of the
report.
Sentence was deleted from text on page 4 of the report.
"Usually was added" in the text to provide clarification
on page 4 of the report.
osmoregulation". Somewhat confusing here. Interference with an
essential ion can be a toxicity mechanism but the beginning of the
sentence is about competition between two cations for a binding site;
the sentence is thus deviating from its original purpose. Also, why focus
on H, Ca and Mg if Na and K are the essential ions that are affected?
Deleting this part of the sentence would make the sentence much
clearer.
Page 4: In addition, higher Ca:Mg ratios have a greater protective
effect by modifying toxicity than waters with similar hardness that had
lower Ca:Mg ratios (Welsh et al. 2000)". 1 would delete this sentence.
This repeats the observation about fish being sensitive to Ca and is in
contradiction with the observation about invertebrates.
40
-------
2.5 Additional suggestions that would improve the report.
Reviewer
Comments
EPA Response
Page 4: "An increase in sodium (Na+) cations generally decreases toxicity
by competition at metal binding sites, however Na+ provides less
protection than Ca2+ and Mg2+". For fish and silver, sodium is a better
protecting parameter than calcium. Add "usually" before "provides".
Table 1: First mention of humic acid. This may need an explanatory
sentence perhaps in the DOC section. 1 understand what is meant by the
10% default, but the average reader won't.
Page 5: "The approaches used by these models fall within a continuum
between empirical (e.g., Water Effects Ratio [WER] and hardness
equations) and mainly mechanistic (e.g., biokinetic BLM) (see Textbox 3
in Adams et al. 2020 and Figure 1 in Brix et al. 2020). In the middle of
the continuum are the empirically-based MLR and mechanistically-
based BLM". 1 would argue that MLR are very close to entirely empirical
models and not in the middle of the continuum. It's however reasonable
for the BLM. Although the BLM was initially a purely mechanistic
conceptual model based on the Free-Ion Activity Model, it has evolved
into a more empirical model over time (see also response to Question
lb above).
Table 2:
4 in S04, should be in subscript (also in the main text)
Alkalinity and hardness sometimes have a capital letter, sometimes not
Page 15: "It is important to note that, the Cu BLM is not optimized for
toxicity observations (neither chronic nor acute)". What is it optimised
for? Accumulation?
Per the comment of Reviewer 5 below regarding humic
acid, this portion of the Table 1 footnote was removed.
The continuum presented in Brix et al. (2020) represents
the relative empiricism/mechanistic basis of various
bioavailability models. There may be some variation in
the specific placement of the models along this
continuum, the positions of each model relative to one
another is accurate. In addition, since this diagram has
been published in the peer-reviewed literature, it is
reasonable to keep the text regarding the Figure "as is"
for illustrative purposes.
The suggested changes were applied to Table 2.
The text has been clarified to say, "It is important to note
that, the Cu BLM is optimized for measured Cu
accumulations and not for toxicity observations (neither
chronic nor acute)."
41
-------
2.5 Additional suggestions that would improve the report.
Reviewer
Comments
EPA Response
Page 15: What does "without interactions" mean? 1 found out by
reading the paper in the Appendices, but this should be understandable
for people who read the report only.
Page 16: About bicarbonate toxicity, from reading Santore 2021, this
conclusion lacks nuance. Bicarbonate toxicity is one possible explanation
for the poor reproduction of C. dubia at high pH. It would be preferable
to say that C. dubia does not tolerate pH > 8 and that other factors are
at play and that Santore speculated that this could be due to
bicarbonate toxicity. The reader needs to be guided here.
Clarification was made in text regarding TMF interactions
in both the section referring to MLR models and in the
discussion of the metal case studies.
The text was modified to provide clarification regarding
bicarbonate toxicity in the discussion of the Ni case study.
Reviewer
5
Specific comments on the draft CRADA report
These comments refer to the draft report entitled "Development of an
Overarching Bioavailability Modeling Approach to Support US EPA's
Aquatic Life Water Quality Criteria for Metals" (21 pp) hereafter
"bioavailability report." Appendixes B and C are integral to the report,
and 1 also have some comments on those.
Overall, 1 thought the "bioavailability report" and Appendix B were very
good. They will doubtlessly be influential for years, and so should get
more vetting with attention to referencing and supporting all
statements before final publication. There are some unreferenced
statements that seem like overstatements in Section II.
p. 2, paragraph b, under "Overview of EPA's metals criteria," consider
adding a sentence or so on why some metals have criteria but most do
not. Cobalt is prominent by its absence. Maybe something along these
lines?
'Of the 56 elements commonly classified as metals on the periodic table,
currently EPA has developed recommended AWQC for 9 metals
(aluminum, cadmium, chromium (III and IV), copper, iron, lead, nickel,
silver, and zinc). This list of metals requiring criteria dates to a 1976
Thank you for your comments, your detailed analyses,
and citations on the effect of TMFs across several metals.
42
-------
2.5 Additional suggestions that would improve the report.
Reviewer
Comments
EPA Response
negotiation among parties to a settlement agreement (NRDC et al. vs
Train, 6 ELR 20588, D.D.C. June 9, 1976). In setting priorities for
establishing new or revised criteria EPA may consider the changing
societal uses of metals that could affect potential prevalence in aquatic
environments. For example¦, cobalt has come into wide use in
rechargeable lithium-ion batteries which are ubiquitous in consumer
electronics, electric vehicles, and in other uses that did not exist in 1976.
These demands might increase the prevalence of cobalt mining and
processing, and potential exposure to aquatic life. Likewise, silver uses
have changed. In the 1970s silver was widely used in the photographic
film industry, which has been supplanted by digital imagery. Another
current use of silver, manufactured nanoparticles, did not exist in the
1970s.'
Btw, arsenic (and selenium) are not metals in any periodic table I've
consulted.
Section II. "Metal Toxicity Modifying Factors (TMFs) and their relative
importance", starting on p. 3
p. 3 "These factors include pH, hardness ions (primarily Ca and Mg),
alkalinity, temperature, sodium, chloride, fluoride,..."This statement is
attributed to Adams et al 2020. 1 don't believe that is entirely accurate. 1
did not see the term "hardness ions" in Adams. As noted in my response
to questions, 1 recommend adding some explanation how hardness got
into recent MLRs instead of Ca. 1 have never seen a quantitative analysis
of why hardness is better than Ca. No BLM uses "hardness ions." 1
suspect that the real reason for relying on hardness rather than Ca is the
policy desire to keep a lineage to the old hardness-based criteria. Brix et
al (2017) started this and subsequent MLRs have followed suit. 1 don't
question the approach, but if this is the case, 1 would mention this policy
heredity.
Arsenic has been removed from the text on page 2 of the
report.
The text has been adjusted to "water hardness (primarily
Ca and Mg ions)" rather than hardness ions on page 3 of
the report.
The suggestion that hardness was used in recent criteria
(cadmium 2016, aluminum 2018) was a "policy desire to
keep a lineage to the old hardness-based criteria" is
incorrect. These recent criteria use hardness because
most end-users monitor hardness rather than Ca.
Regarding the use of hardness in MLR models, please see
response to Reviewer 5's comment in section 2.3.
43
-------
2.5 Additional suggestions that would improve the report.
Reviewer
Comments
EPA Response
p.3 "Meyer et al. (2007) described two ways in which these factors can
affect the bioavailability and toxicity of metals" 1 don't follow attributing
this to Meyer et al, as they discuss more than two ways. In particular,
the factors themselves, particularly pH and major ions, affect the vigor
of aquatic organisms. See Meyer et al, (2007), their chapter 6. My
impression of this body of work is that the energy requirements of
osmoregulation is the biggest factor. Fish become leaky in low ionic
strength water requiring much energy to counteract this and maintain
internal mineral balance and metals seem to compound this problem.
The much greater resistance offish to metals in marine waters vs.
freshwaters cannot solely be attributed to competition and
complexation, but that the increased Na marine environment adds
physiological protections. As a practical matter, it matters not to the
organism whether they get killed or not by metals toxicity or whether
they get killed by increased susceptibility to ionic disruption secondary
to metals. People like Chris Wood, Mike Wilkie, Martin Grosell, and
Kevin Brix have published much on this. Most research on this has been
with fish. Mever et al. (2007) have a good discussion of these issues in
their ch. 6. Wood (2012) gives a more recent overview with fish and we
briefly touched on it in our introduction to BLM mechanisms (Mebane
et al. 2020a). Buchwalter touches on this with aquatic insects
(Buchwalter et al. 2008).
p. 3 "Specifically, the effects of the most commonly studied TMFs are
described below (see Meyer et al. 2007for more information)" If this
entire section is attributed to Mever et al (Mever et al. 2007), then the
end of each paragraph should include "(Meyer et al. 2007)." There are
some sweeping statements that presently are either unattributed or
ambiguously attributed to Meyer et al. While the authors may have
considered this an "overview" of metal toxicity modifying factors,
uncluttered by references, rather than a "review" 1 think more precision
on the basis of some of these statements would be helpful
Clarification was made to the text to include "Meyer et al.
(2007) described two ways in which these modifying
factors can affect whether metals result in bioavailable
concentrations that can cause toxicity by affecting the
physiological responses of aquatic organisms" on page 3
of the report.
A more citations were added to the text of Section II of
the report as well as a reference to more detailed
information in Appendix B.
44
-------
2.5 Additional suggestions that would improve the report.
Reviewer
Comments
EPA Response
p. 3 "a. pH" The discussion only addressed speciation changes and not
the role of proton competition. It makes a difference. Al and Cu toxicity
often increase (lower ECx values) at lower pH (but see Cusimano et al
(1986) for an opposite result with Cu) but almost all studies I've seen
show Cd and Zn toxicity increasing at increasing pH, at least within the
range commonly encountered in natural waters, 5.5 to 9 or so (Bradlev
and Sprague 1985; Cusimano et al. 1986; Schubauer-Berigan et al. 1993;
Bervoets and Blust 2000; Hansen et al. 2002; Heiierick et al. 2003; De
Schamphelaere and Janssen 2004a; Tan and Wang 2011). Some studies
showed no consistent effect at all of pH on toxicity, which might be the
two factors (speciation and competition) cancelling each other out
(Nivogi et al. 2008; Clifford and McGeer 2009, 2010). These sorts of
details might better go into Appendix B, but if so the paragraph
attribution should be to Appendix B, and not solely to Meyer et al. 2007.
p. 4 Hardness: "...however Mg2+ is generally as or more protective than
Ca2+ in invertebrates." Generally? That's generally too sweeping. 1 do not
believe there are enough data on this point to say "generally." 1 would
remove this statement, or explicitly support it. From my readings, 1 do
not believe it is supportable. If this refers to Naddv et al. (2002) it
overstates their results. Yes, they found hardness with a 1:1 Ca:Mg ratio
was more protective to Ceriodaphnia and Daphnia compared to the
same hardness with a 4:1 Ca:Mg ratio, but they also tested Gammarus
and found it was better protected at the higher Ca:Mg ratios same as
fish. Gammarus are iust as much invertebrates as daphnids. (Heiierick et
al. 2002; Heiierick et al. 2005) found Ca and Mg were approximately
equal in protectiveness to Daphnia magna from acute Zn toxicity, and
De Schamphelaere and Janssen (2002) found the same for protection
from acute Cu toxicity.
p. 4, Dissolved Organic Carbon - Paragraph is good, but citation needed.
Suggest Wood et al. (2011).
A general reference to find more metal-specific details in
Appendix B was added to the text of Section II of the
report.
This statement was removed from the text in the
hardness section of the report.
45
-------
2.5 Additional suggestions that would improve the report.
Reviewer
Comments
EPA Response
p. 4. d. Other - "... however Na+ provides less protection than Ca2+ and
Mg2+." Citation needed. 1 doubt anyone would challenge that for Ca, but
it's not obvious to me that Na provides less protection than Mg.
Certainly some Na log(K) values in BLMs are lower than Mg, and that
arguments could be invoked if direct evidence is less obvious. 1 looked
through Meyer et al, as that was the implied source. It might be in
there, but 1 did not quickly find it.
Table 1, p4-5. "Table 1 illustrates the relative importance of the most
studied TMFs for several metals."
Table 1 doesn't really do that - capture the relative importance of TMFs.
Most are the same, and since nothing's cited it's hard to evaluate the
evidence behind this interpretation. 1 would change the table as follows:
put it on a three part qualitative scale, instead of the present two parts
(that is, change to +, ++, +++ scale). Shading indicates where 1 removed a
mark that 1 didn't think had strong support in the literature, red marks
are my additions. To show more relative importances, 1 suggest change
the scoring as follows:
The citation Wood et al. 2011 was added to the DOC
section of the report.
Made reviewer's suggested changes in Table 1 to the
qualitative scale and clarified that the scoring illustrates
the relative importance of the TMFs within metals but not
across metals.
46
-------
2.5 Additional suggestions that would improve the report.
Reviewer
Comments
EPA Response
Metal
Type
Most Important Parameters1
Hardness
pH
DOC
Other
Aluminum
Freshwater
+
++
++
temperature
Cadmium
Freshwater
+++
+
+
Cobalt
Freshwater
++
+
+
Copper
Freshwater
+
++
+++
sodium
Copper
Marine
+
+
salinity
Lead
Freshwater
+
+
+++
Nickel
Freshwater
+
+
Silver
Freshwater
+
chromium
reducible
sulfur,
sodium,
chloride
Zinc
Freshwater
+++
++
+
Thank you for the suggestion. This summary information
has been added as Appendix H.
47
-------
2.5 Additional suggestions that would improve the report.
Reviewer
Comments
EPA Response
1 suggest adding a short rationale for the different qualitative rankings
below the table, since many readers won't delve into Appendix B
Aluminum: Hardness has a moderate role in modifying Al toxicity; pH
has a strong role but the direction of effect can change with different
organisms, and DOC consistently reduced Al toxicity (DeForest et al.
2018).
Cadmium: Hardness regressions predict acute and chronic toxicity well
in natural waters (Mebane 2006; USEPA 2016a). pH effect appears weak
and ambiguous (Nivogi et al. 2008; Clifford and McGeer 2010). The
threshold for a DOC effect appears to be >5 mg/L (Nivogi et al. 2008).
Cobalt: Hardness is clearlv important (Diamond et al. 1992; Borgmann
et al. 2005). pH at least affected gill uptake, with uptake increasing with
increasing pH up to 8.7. DOM reduced Co gill binding, but Co-DOM
affinity was much lower than that of Cd, Cu, or Ag (Richards and Plavle
1998).
Copper, freshwater: DOC has a strong binding affinity to Cu and
predictably reduces Cu toxicity, even at low concentrations (Erickson et
al. 1996; Welsh et al. 2008). pH has a strong effect on Cu toxicity, with
toxicity tending to decrease with increasing pH in alkaline conditions,
but toxicity decreasing with decreasing pH in acidic conditions
(Cusimano et al. 1986; Erickson et al. 1996). Hardness is a comparatively
minor factor in natural waters (Markich et al. 2005).
Copper, marine: DOC and salinity tend to reduce Cu toxicity in marine
and estuarine waters (Grosell et al. 2007; Hall et al. 2008).
Pb: Similar to Cu, DOC and pH have strong effects on the bioavailability
and toxicity of Pb (DeForest et al. 2017) Hardness may be an important
factor in natural waters, especially when DOC is low (Mebane et al.
2012).
48
-------
2.5 Additional suggestions that would improve the report.
Reviewer
Comments
EPA Response
Ni: Ni toxicity tends to decrease as hardness increased and decrease
with increasing DOC. pH has inconsistent influence on toxicity (Croteau
et al. 2021; Santore et al. 2021).
Silver: DOC reduces toxicity but pH and hardness influences may be
inconsistent (Naddv et al. 2018).
Zinc: Similar to Cd, hardness has a strong influence on Zn toxicity, with
decreasing toxicity with increasing hardness (Clifford and McGeer 2009;
Mebane et al. 2012; CCME 2018); with fish, toxicity generally increases
with increasing pH but relations may be inconsistent in other taxa (De
Schamphelaere and Janssen 2004a). DOC reduces Zn toxicity but
influence may be nonlinear, with a threshold of >«10 mg/L DOC
reauired to reduce toxicity (Bringolf et al. 2006; Ivev et al. 2019).
Footnote to Table 1. "Additionally, the bioavailability of metals such as
cadmium, copper, nickel, and silver has been shown to be sensitive to
humic acid and scientific advances are beginning to shed light on options
that may be more representative than using the default of 10% generally
recommended for BLM applications (Glover et al. 2005; Nadella et al.
2009; Al-Reasi et al. 2012; Blewett et al. 2016)."
1 recommend deleting this part of the footnote. First, 1 would argue that
if a footnote caution/caveat is warranted, it should first be about pH
which can change by more than a unit depending on the time of day
sampled. A 1-unit change in any of these BLMs or MLR based criteria is
huge - 1 appended an example showing that the BLM Cu chronic criteria
would swing from about 8 to 26 ng/L, just from the time of day that pH
was measured. Regarding DOC, there are lots of practical issues with
DOC in BLMs that might be at least as important as the humic/fulvic -
the DOM/DOC conversion & active fraction, contamination from capsule
filters or tubing. 1 appended an example of likely filter artifacts in USGS
data toward the end of these comments. Further, 1 don't think the
footnote is fully accurate. Three of the 4 references cited studied DOM
This portion of the footnote has been deleted from the
footnote to Table 1.
49
-------
2.5 Additional suggestions that would improve the report.
Reviewer
Comments
EPA Response
with Cu and 1 studied DOM with Ni, so Cd and Ag? True, Nadella found
that NOMs with high humic acid offered less protection to Cu toxicity
than those dominated by fulvic acid, but that is the opposite of the
effect of the humic acid selection in the Windward BLMs. In the
Windward BLMs, higher humic acid fraction adds a slightly greater
protective effect. Plus, it's hard to generalize Nadella's results - testing a
marine species in saltwater with NOMs from different freshwaters.
Table 2: Very nice compilation.
p. 13 "Multiple Linear Regression Models"
Thank you for your comment.
Somewhere in this first paragraph 1 would mention that EPA put out its
first MLR-type criteria in 1984 with ammonia, in which the criteria
varied with a relatively complicated nonlinear equation as a function of
temperature and pH. At least some states (Idaho and Colorado come to
mind) dealt with the calculation complexity by publishing table values of
criteria values for every tenth of a pH unit or degree that could be used
in permitting in lieu of calculating the values directly. While it's a lot
easier now than it was in the 1980s when PCs and spreadsheets were
scarcer, this MLR level of complexity did not seem a big deal with
ammonia.
Thank you for your comment.
p. 17 "... as EPA moves forward with updating the metals AWQC, it is
desirable to have a single software platform." Some people prefer R
scripts, some prefer spreadsheets, overtime something else might
become widely used. At the present, I'd say the "xlsx" format would be
most accessible and it isn't that hard for R users to export carefully
assembled xlsx to a R friendly format. A core, common syntax would be
helpful, but it's easy enough to put out criteria datasets and equations
in say both xlsx and R
50
-------
2.5 Additional suggestions that would improve the report.
Reviewer
Comments
EPA Response
Appendix B comments
Appendix B reflects a big effort and is a very helpful, concise guide to
much relevant information for the subset of metals supported through
the CRADA efforts. While hardness is hard to screw up, 1 do suggest
adding a bit on the importance of data quality in pH and DOC data. pH
probes are notoriously finicky. More importantly, in some waters the
daily cycles of production and respiration can cause pH swings high
enough to skew criteria a lot. Even ~0.2 units can make noticeable
differences in criteria calculations and natural swings of >1 unit aren't
unheard of. Figure 1 gives an example calculation where the criteria
would swing 3-fold from 8 to 25 ng/L over the course of a day. So when
should waters be sampled? Depending on the desired answer? Most
likely, whenever it's most convenient for the person doing the sampling
which might not give the most representative results. In a stream
contaminated with Zn (primarily) and subject to daily Zn and pH swings,
the observed toxicity to trout corresponded best to the daily average
conditions, not the daily maximum (worst case) concentrations
(Balistrieri et al. 2012). 1 recommend saving something about the
uncertainty of daily pH cycles and the need to resolve the most
representative time of day (or daily average) for sampling.
With DOC, there has been lots of research and debate on different
characteristics that affect metal binding and bioavailability, such as that
terrestrial sources with high fulvic/humic acid content reduce Cu
bioavailability more than autochthonous sources such as algae
senescence. However, 1 have seen much less in the BLM and metals
bioavailability literature about the importance of basic QC in collecting
and analyzing DOC. In particular, filtration and tubing can be a real
bugaboo that introduces DOC at biologically and BLM-relevant
concentrations. 1 show a few examples of the issue in figure 2 and figure
3^ In my group, while we think we are reasonably careful and attuned to
the issue, we still sometimes see DOC in filter blanks at 0.2 to 0.3 mg/L,
A brief paragraph about the importance of collecting high
quality data was added to page 5 of the report.
51
-------
2.5 Additional suggestions that would improve the report.
Reviewer
Comments
EPA Response
even though the manufacturer of the organic blank water that we
purchase certifies that the water contains <0.05 mg/LTOC. We're
probably picking up some DOC through the filters and tubing during
filtering. Yoro et al. (1999) is a good citable citation on this point.
9.0 r
8.5
c
3
¦D 8.0
(Z
•o
c
a
C
Ł7-5
7.0
Stalker Creek, Idaho
\ %
's.
o.
v.
°o •%>
15-July-2007
An example of how natural, daily swings in pH can cause wild swings in
criteria that rely on pH as a modifying factor. If the discharger wants a
high criteria value that's easy to comply with, they should sample in late
afternoon (pH 8.7, Cu CCC 26 jig/L). If zealous regulators want a low
criterion value, they should sample late at night or early in the morning
52
-------
2.5 Additional suggestions that would improve the report.
Reviewer
Comments
when pH is low (pH 7.5, BLM based CCC 8 ng/L). So what to do? Take
the average?
EPA Response
Columbia River near Northport, Washington
Neuse River, North Carolina
A couple of examples of differences in DOC concentrations likely
influenced by sampling contamination through filters and bottles, one
from a low DOC river (Columbia River by the US/Canada border) and
one from a high DOC stream (the piedmont Neuse River). In 1993, the
USGS began pushing so-called "clean sampling" methods for trace
metals and this hygiene emphasis seemed to carry over to DOC. We still
see occasional DOC filter blank contamination from modern capsule
filters a biologically and BLM-relevant concentrations.
53
-------
2.5 Additional suggestions that would improve the report.
Reviewer
Comments
EPA Response
Chronic copper criteria: Teton River at St. Anthony, ID
USGS 13055000
90 -i— T 12
80 -
70 -
60 -
50 -
40 -
30 -
20 -
10 -
BLM-based CCC (ng/L, diss.)
-a— doc
-- 10
What's happened in October
1993? Where did the DOC go?
- 6
-- 4
-- 2
O)
E.
u
o
D
0 -| , , , , L 0
Jan-93 Aug-93 Mar-94 Sep-94 Apr-95 Oct-95
Another example of how filtration and cleaning practices can create bad
DOC data, which can be hard to catch on a sample by sample basis. In
this case, DOC contamination was suspected to have been caused
surfactants residual to the capsule filter manufacturing process and
inadequate flushing before the sample was taken.
My point in all this is that either in the main document or in appendix B
it would be prudent to say something about the importance of good
sampling and measurement practices with the inputs to these models,
and in particular pH and DOC. I suggest it could be a lot shorter than my
examples and cite on the pH issue studies like Balistrieri et al (2012) and
maybe Nimick et al (2011), and Yoro et al. (1999) on the DOC issue. As
these models move towards criteria, it would be good to include some
54
-------
2.5 Additional suggestions that would improve the report.
Reviewer
Comments
EPA Response
recommended practices on these mundane but important issues of data
representativeness and quality.
Coooer
Cu and Hardness. "There is a consistent protective effect of water
hardness on Cu toxicity in acute and chronic exposures to fish and
invertebrates ... with equivocal results or no protection in only a few
studies."That seems a little overstated and 1 would reword it to be more
even handed. Something like 'Many studies reviewed have shown some
protective effect of water hardness on Cu toxicity in acute and chronic
exposures to fish and invertebrates (for example, cite; cite; cite;...).
However, inconsistent results or no protection were reported in some
studies, for example (Chapman et al. 1980; Richards and Plavle 1999; De
Schamphelaere and Janssen 2004b; Hvne et al. 2005; Markich et al.
2005; Wang et al. 2009)
Zinc
Zn and Hardness - 1 can't help but chime in with a "us too." In Mebane
et al. (2012), we reported 4 tests with rainbow trout, each with fish
from the same cohort in different natural waters. Hardness explained
between 90% to 99% of the variability in EC50s in these natural waters
where pH was allowed to covary.
Zn and DOC. 1 think the story with DOC protecting against Zn toxicity is
more nuanced and equivocal than this paragraph would lead readers to
believe. In particular the sentence "In freshwaters, dissolved organic
matter (DOM) - quantified as dissolved organic carbon (DOC) -
generally decreases Zn bioavailability (e.g., Hyne et al. 2005; Clifford and
McGeer 2009; Heijerick et al. 2003)." First, that is not what Hyne et al
(2005) reported. Rather, thev reported that the addition of 10 mg/L
DOC only resulted in a very small (1.3-fold) reduction in the toxicity of
zinc to Ceriodaphnia, whereas the same DOC addition resulted in a 45-
Appendix B text was modified based on Copper CRADA
partner's suggestion.
Zinc CRADA partners indicated that they appreciate the
input provided by the reviewer. Indeed, Mebane et al.
(2012) did demonstrate the ameliorative effect of
hardness in natural waters, but as the reviewer indicated:
"pH was allowed to covary". The reviewer is also correct
that the DOC effects on Zn bioavailability are nuanced,
which is why we indicated that DOC "generally decreases
Zn bioavailability". The general trend in tests specifically
investigating the effect of DOC on Zn toxicity is that
increasing DOC concentrations increase Zn effect
concentrations. Although for MLR model development
purposes, it is recognized that the slope of the
relationship between Zn effect concentration and DOC
concentration is shallow (i.e., low slope) for some
datasets (i.e., individual studies). All data suitable for MLR
model development will be used to evaluate the effect of
DOC on Zn bioavailability. The reviewer makes an
55
-------
2.5 Additional suggestions that would improve the report.
Reviewer
Comments
EPA Response
fold reduction in Cu toxicity. 1 have seen no reports of DOC having an
important role in reducing Zn toxicity until DOC concentrations are fairly
high (greater than at least 5 mg/L DOC and probably greater than 10
mg/L DOC). The minimum DOC tested by Heijerick et al was 9 mg/L.
Clifford and McGeer (2009) tested a base condition with 0.6 mg/L DOC,
6-7 mg/L DOC additions, and 10-11 mg/L DOC added. Only the pair of
high DOC additions (10-11 mg/L) reduced toxicity beyond the range of
the base conditions with 0.6 mg/L DOC. In tests of the acute toxicity of
Zn to sturgeon, DOC in the range of 1 to 5 mg/L had no effect (Ivev et al.
2019). In tests with fathead minnow and Zn under different organic
carbon conditions, a threshold concentration of 11 mg/L DOC was
reauired to reduce acute toxicity to (Bringolf et al. 2006). The take home
on Zn-DOC toxicity relations from published research is that DOC
concentrations <10 mg/L are sparse, and from what 1 can find indicates
little protective effect for Zn toxicity.
The significance of this to the MLR approach is that if there is a
threshold effect for DOC reductions at around 10 mg/L, a linear
regression that predicts a linear response may be misleading and
underprotective in the low range between say 0.5 and 10 mg/L. A
regression that fits a straight line from controls with say 0.5 mg/L to 40
mg/L, will show a strong response, and give the same slope in the 0.5 to
10 mg/L DOC range of the regression as in the higher DOC range, even
though no data were in the low range. It's just fitting a straight line. For
instance, in the Heiierick et al. (2003) studv mentioned above, thev have
a very clean plot predicting a linear response between DOC and Daphnia
toxicity (their figure 3). However, the underlying data included test pairs
with huge ranges. One test pair had pH 7.25, hardness 240 and DOC of 2
vs DOC 40 mg/L; one test pair had pH 6, hardness 110 and DOC of 9.7 vs
32 mg/L; and the third test pair was with pH 8, hardness 370 and DOC
9.7 vs 32 mg/L. None of those tests tell us anything about what is going
interesting suggestion regarding a nonlinear or piecewise
nonlinear function to represent the effect of DOC. We
agree that this is something that should be explored
during MLR model development. Ultimately, regardless of
model formulation, if DOC is shown to be an important
TMF, it will be retained in MLR models. The converse will
also be true.
56
-------
2.5 Additional suggestions that would improve the report.
Reviewer
Comments
EPA Response
on at the low 1-5 mg/L DOC values, although one wouldn't immediately
realize that from the pretty model plot in their Figure 5.
The reason for this concern with the potential overextrapolation of
DOC-Zn toxicity relations to the range of « 0 to 10 mg/L, is that that is
the range where the vast majority of flowing waters in the US fall.
USEPA (2016b) included a summary of DOC values collected from 1,392
sites sampled across the 84 ecoregions of the United States using a
probability-based sample design from the EMAP Wadable Stream
Assessment (WSA). The median values for each of the 84 ecoregions
were reported. The 90th percentile of the 84 ecoregions was 8.4 mg/L,
the 75th percentile was 5.2 mg/L, and the national ecoregional median
was 2.7 mg/L DOC (calculated from table 17 of USEPA (2016a)). Thus
>90% of the streams in the United States would be expected to have
DOC values in the range of questionable Zn-DOC relations.
Thus, the usual MLR straight line approach may not be the most
appropriate for Zn and DOC and a nonlinear function or a piecewise
'nonlinear' function may need to be explored.
Appendix C comments
1 just glanced through "Appendix C, Table 2: Supporting Information for
Bioavailability Model Comaprison Table" First, 1 think "comaprison" is a
fine new word that should be added to the spell checker and kept in the
report, applicable to the state of mind in many an office cube. Well,
maybe it should be hyphenated, coma-prison. A couple other items that
caught my eye...
First row, Aluminum BLM: No reference is given, but the version
"3.18.2.42" looks like a Windward numbering version. Santore et al
(2018) describe using CHESS and WHAM V, not WHAM 7. To mv
knowledge, no Windward BLM version has incorporated WHAM 7.
This change has been made to Appendix C.
The Aluminum BLM, similar to other BLMs developed
uses WHAM Model V for speciation calculations. This
change has been applied to Appendix C.
The Cobalt BLM, similar to other BLMs developed uses
WHAM Model V for speciation calculations. This change
has been applied to Appendix C.
57
-------
2.5 Additional suggestions that would improve the report.
Reviewer
Comments
EPA Response
Cobalt BLM says it is "complete" but to my knowledge no Co BLM has
been formally published or publicly released online. The version
"3.15.2.41" also looks like Windward numbering, which makes me
wonder whether it actually used "WHAM 6" for speciation, since as with
WHAM 7, that would have been a big coding project. 1 would check this.
3.0 Additional comments.
Reviewer
Comments
EPA Response
Reviewer
1
General considerations
With much interest 1 have read the documentation that was send
as part of the assignment on the evaluation of EPA's draft report
on the development of an overarching bioavailability modeling
approach to support US EPA's aquatic life water quality criteria
for metals. This brief draft report properly describes the
information available as the basis for the overarching
bioavailability modeling approach.
It is to be noted that the report and the underlying
documentation are a reflections of decades of work by scientists
across the globe on bioavailability modeling. Nevertheless it is
clear from the draft report that proper care needs to be taken
with regard to actual implementation of the various complex
models (independent of them being BLM- or MRL-based) in
derivation of water quality criteria and it is especially clear that it
is essential to make sure that the complexities and the
interactions of the various toxicity modifying factors are properly
incorporated in the software platform that is likely to be the
Thank you for your review. EPA strives to properly incorporate all
the complexities and the interactions of the various toxicity
modifying factors in models and accompanying software
platform.
58
-------
3.0 Additional comments.
Reviewer
Comments
EPA Response
future means of user-friendly implementation of the decades of
metal bioavailability research.
Reviewer
3
Preface:
There is scientific consensus that water chemistry profoundly
affects the bioavailability and toxicity of trace metals in
freshwaters. My own research career started with studies of the
effects of dissolved organic carbon and pH on the speciation and
acute toxicity of Cu to developing amphibian eggs and larvae1. 1
am predisposed to appreciate the development of mechanistic
understanding of how trace metal toxicity occurs from a purely
scientific perspective, and 1 also feel strongly that regulatory
approaches to protecting aquatic life should be based on
defensible science.
1 recognize the scientific achievements and conceptual
advancements embodied by Biotic Ligand Models, and
understand how their complexity contributed to their limited
adoption by regulatory end users. 1 can appreciate the
frustration of the metal industry groups who put substantial
efforts into these scientific developments and not have them
widely adopted. Indeed, the science has progressed considerably
and regulatory approaches for protecting the environment need
to be modernized (see 2). That said, 1 think it is important to
articulate that BLM and MLR models primarily have the shared
goal of accounting for Toxicity Modifying Factors (TMFs) such
that "overprotection" is avoided. As more TMFs are considered,
protection levels will generally be more relaxed. The goal of
these approaches is not protection - it is the avoidance of
overprotection.
Thank you for your review.
In regard to accounting for TMFs to avoid overprotection, the
updated Cu and Al criteria both have values that are lower than
previous criteria for high metal bioavailability conditions and
accounting for TMFs can result in criteria that can more
accurately achieve EPA's targeted level of protection over a wide
range of water chemistry conditions.
Regarding dietary exposure, please see the responses to
Reviewer 3's comments in sections 2.1a, 2.1c and 2.2a.
Furthermore, the 1985 Guidelines do not explicitly exclude the
consideration of dietary exposure. EPA believes it is important to
focus on the primary drivers of toxicity, which for the metals
evaluated in this effort, appear to be aquatic exposure (as noted
above).
59
-------
3.0 Additional comments.
Reviewer
Comments
EPA Response
The models at the heart of this review are driven by the
perspective that metals are problematic or toxic in freshwater
environments as surface-active, aqueous toxicants. While this
perspective is largely accurate and scientifically supported for
acute exposures to many aquatic animals, it is unfortunately not
complete. Dietary exposures are extremely important to aquatic
insects2"7 - the faunal groups that largely drives the ecology of
the ecosystems that EPA is charged with protecting. Aquatic
insects were recognized by Workgroup 2 of the 2017 SETAC
Metal Bioavailability Workshop as a faunal groups that might not
be adequately covered by the models under consideration8 -
likely because dietary exposure pathways predominate from a
toxicity perspective. Thus, the models which are the focus of this
review are likely not applicable to the most ecologically
important faunal group in freshwater ecosystems.
A complete exposure perspective that includes aqueous and
dietary exposure pathways is required for scientifically
defensible Water Quality Criteria. This fact is extremely
problematic in the context of Water Quality Criteria
development because the 1985 Guidance document9 requires
the exclusion of data that deviate from strict aqueous exposures.
Until this changes, even the best aqueous based models will
represent an incomplete understanding of metal toxicity in
aquatic ecosystems.
Reviewer
The End
Thank you for your review.
5
1 realize these comments are longer than 1 intended. 1 hope they
are useful and that they did not come across as giving a negative
perspective on the project. Quite the opposite was intended.
These models in appendices D-F are remarkable and this project
has taken a huge step towards the goal of updating and
60
-------
3.0 Additional comments.
Reviewer
Comments
EPA Response
expanding metals criteria in the US. 1 look forward to seeing good
progress with Co and Zn as well. The summary report and
appendices B and C will be influential and valuable. Well done to
all.
4.0 NEW INFORMATION PROVIDED BY REVIEWERS
This section presents all new information that reviewers provided in addition to or within their specific responses (presented in Section 2, above)
to the charge questions.
4.0 New Information.
Reviewer
Comments
EPA Response
Reviewer
3
As noted in comments above:
(1) Buchwalter, D. B.; Linder, G.; Curtis, L. R. Modulation of Cupric Ion Activity by PH and Fulvic Acid as
Determinants of Toxicity in Xenopus Laevis Embryos and Larvae. Environ. Toxicol. Chem. 1996,
15 (4), 568-573. https://doi.org/10.1002/etc.5620150423.
(2) Buchwalter, D. B.; Clements, W. H.; Luoma, S. N. Modernizing Water Quality Criteria in the United
States: A Need to Expand the Definition of Acceptable Data. Env. Toxicol Chem 2017, 36 (1552-
8618 (Electronic)), 285-291. https://doi.org/10.1002/etc.3654.
(3) Cain, D. J.; Luoma, S. N.; Wallace, W. G. Linking Metal Bioaccumulation of Aquatic Insects to Their
Distribution Patterns in a Mining-Impacted River. Environ. Toxicol. Chem. 2004, 23 (0730-7268
(Print)), 1463-1473.
(4) Xie, L. T.; Lambert, D.; Martin, C.; Cain, D. J.; Luoma, S. N.; Buchwalter, D. Cadmium Biodynamics in
the Oligochaete Lumbriculus Variegatus and Its Implications for Trophic Transfer. Aquat. Toxicol.
2008, 86 (2), 265-271.
Thank you for providing
the following references
for EPA's review.
61
-------
4.0 New Information.
Reviewer
Comments
EPA Response
(5) Xie, L.; Buchwalter, D. B. Cadmium Exposure Route Affects Antioxidant Responses in the Mayfly
Centroptilum Triangulifer. Aquat. Toxicol. 2011, 105 (1879-1514 (Electronic)), 199-205.
(6) Poteat, M. D.; Buchwalter, D. B. Four Reasons Why Traditional Metal Toxicity Testing with Aquatic
Insects Is Irrelevant. Environ. Sci. Technol. 2014, 48 (1520-5851 (Electronic)), 887-888.
https://doi.org/10.1021/es405529n.
(7) Soucek, D. J.; Dickinson, A.; Schlekat, C.; Van Genderen, E.; Hammer, E. J. Acute and Chronic Toxicity
of Nickel and Zinc to a Laboratory Cultured Mayfly ( Neocloeon Triangulifer) in Aqueous but Fed
Exposures. Environ. Toxicol. Chem. 2020, 39 (6), 1196-1206. https://doi.org/10.1002/etc.4683.
(8) Mebane, C. A.; Chowdhury, M. J.; De Schamphelaere, K. A. C.; Lofts, S.; Paquin, P. R.; Santore, R. C.;
Wood, C. M. Metal Bioavailability Models: Current Status, Lessons Learned, Considerations for
Regulatory Use, and the Path Forward. Environ. Toxicol. Chem. 2020, 39 (1), 60-84.
https://doi.org/10.1002/etc.4560.
(9) Stephan, C. E.; Mount, D. 1.; Hansen, D. J. 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. USA, 1985.
(10) Gillis, P. L.; Wood, C. M. Investigating a Potential Mechanism of Cd Resistance in Chironomus
Riparius Larvae Using Kinetic Analysis of Calcium and Cadmium Uptake. Aquat. Toxicol. 2008, 8.
(11) Poteat, M. D.; Buchwalter, D. B. Calcium Uptake in Aquatic Insects: Influences of Phylogeny and
Metals (Cd and Zn). J.Exp.Biol. 2014, 217 (1477-9145 (Electronic)), 1180-1186.
https://doi.org/10.1242/jeb.097261.
(12) Clements, W. H.; Carlisle, D. M.; Lazorchak, J. M.; Johnson, P. C. Heavy Metals Structure Benthic
Communities in Colorado Mountain Streams. Ecol. Appl. 2000, 10 (2), 626-638.
(13) Scheibener, S. A.; Richardi, V. S.; Buchwalter, D. B. Comparative Sodium Transport Patterns
Provide Clues for Understanding Salinity and Metal Responses in Aquatic Insects. Aquat. Toxicol.
2016, 171, 20-29. https://doi.Org/10.1016/j.aquatox.2015.12.006.
62
-------
4.0 New Information.
Reviewer
Comments
EPA Response
(14) Poteat, M. D.; Diaz-Jaramillo, M.; Buchwalter, D. B. Divalent Metal (Ca, Cd, Mn, Zn) Uptake and
Interactions in the Aquatic Insect Hydropsyche Sparna. J.Exp.Biol. 2012, 215 (1477-9145
(Electronic)), 1575-1583. https://doi.org/10.1242/jeb.063412.
(15) Kim, K. S.; Funk, D. H.; Buchwalter, D. B. Dietary (Periphyton) and Aqueous Zn Bioaccumulation
Dynamics in the Mayfly Centroptilum Triangulifer. Ecotoxicology. 2012, 21 (1573-3017
(Electronic)), 2288-2296. https://doi.org/10.1007/sl0646-012-0985-l.
(16)Xie, L.; Funk, D. H.; Buchwalter, D. B. Trophic Transfer of Cd from Natural Periphyton Biofilmsto
the Grazing Mayfly Centroptilum Triangulifer in a Life Cycle Test. Environ. Pollut. 2010, 158,
272-277.
(17) Conley, J. M.; Funk, D. H.; Buchwalter, D. B. Selenium Bioaccumulation and Maternal Transfer in
the Mayfly Centroptilum Triangulifer in a Life-Cycle, Periphyton-Biofilm Trophic Assay.
Environ.Sci.Tech. 2009, 43, 7952-7957.
(18) Conley, J. M.; Funk, D. H.; Cariello, N. J.; Buchwalter, D. B. Food Rationing Affects Dietary Selenium
Bioaccumulation and Life Cycle Performance in the Mayfly Centroptilum Triangulifer.
Ecotoxicology 2011, 20 (1573-3017 (Electronic)), 1840-1851.
(19) Conley, J. M.; Funk, D. H.; Hesterberg, D. H.; Hsu, L. C.; Kan, J.; Liu, Y. T.; Buchwalter, D. B.
Bioconcentration and Biotransformation of Selenite versus Selenate Exposed Periphyton and
Subsequent Toxicity to the Mayfly Centroptilum Triangulifer. Environ. Sci. Technol. 2013, 47
(1520-5851 (Electronic)), 7965-7973. https://doi.org/10.1021/es400643x.
(20) Conley, J. M.; Watson, A. T.; Xie, L.; Buchwalter, D. B. Dynamic Selenium Assimilation, Distribution,
Efflux, and Maternal Transfer in Japanese Medaka Fed a Diet of Se-Enriched Mayflies. Environ.
Sci. Technol. 2014, 48 (1520-5851 (Electronic)), 2971-2978. https://doi.org/10.1021/es404933t.
(21) Mebane, C. A.; Hennessy, D. P.; Dillon, F. S. Developing Acute-to-Chronic Toxicity Ratios for Lead,
Cadmium, and Zinc Using Rainbow Trout, a Mayfly, and a Midge. Water. Air. Soil Pollut. 2008,
188 (1-4), 41-66.
Appendix 1. Diet relevant metals citations compiled by ETAP
63
-------
4.0 New Information.
Reviewer
Comments
EPA Response
References - Survival, Growth, Reproduction, and Feeding Behavior (note: additional silver
references may be added, if confirmed to be of interest to ETAP)
Abdel-Tawwab M, Mousa MAA, Abbass FE. 2007. Growth performance and physiological response of
African catfish, Clarias gariepinus (B.) fed organic selenium prior to the exposure to
environmental copper toxicity. Aquacult 272:335-345.
Ashanullah M, Williams AR. 1991. Sublethal effects and bioaccumulation of cadmium, chromium,
copper and zinc in the marine amphipod Allorchestes compressa. Mar Biol 108:59-65.
Alsop D, Brown S, Van Der Kraak G. 2007. The effects of copper and benzo(a)pyrene on retinoids and
reproduction in zebrafish. Aquat Toxicol 82:281-295.
Ashley LM. 1972. Nutritional pathology. Pages 439-535 in Halver JE, ed. Fish nutrition. Chapter 10.
Academic Press, New York.
Baker RTM, Handy RD, Davies SJ, Snook JC. 1998. Chronic dietary exposure to copper affects growth,
tissue lipid peroxidation, and metal composition of the grey mullet, Chelon labrosus. Mar
Environ Res 45:357-365.
Ball AL, Borgmann U, Dixon DG. 2006. Toxicity of a cadmium-contaminated diet to Hyalella azteca.
Environ Toxicol Chem 25:2526-2532.
Bennett WN, Brooks AS, Boraas ME. 1986. Selenium uptake and transfer in an aquatic food chain and
its effects on fathead minnow larvae. Arch Environ Contam Toxicol 15:513-517.
Berntssen MHG, Hylland K, Bonga SEW, Maage A. 1999a. Toxic levels of dietary copper in Atlantic
salmon (Salmo salar L.) parr. Aquat Toxicol 46:87-99.
Berntssen MHG, Lundebye AK, Maage A. 1999b. Effects of elevated dietary copper concentrations on
growth, feed utilization and nutritional status of Atlantic salmon (Salmo salar L.) fry. Aquacult
174:167-181.
Besser JM, Canfield TJ, La Point TW. 1993. Bioaccumulation of organic and inorganic selenium in a
laboratory food chain. Environ Toxicol Chem 12:57-72.
64
-------
4.0 New Information.
Reviewer
Comments
EPA Response
Besser JM, Brumbaugh WG, Brunson EL, Ingersoll CG. 2005. Acute and chronic toxicity of lead in water
and diet to the amphipod Hyalella azteca. Environ Toxicol Chem 24:1807-1815.
Beyers DW, Sodergren C. 2002. Assessment of exposure of larval razorback to selenium in natural
waters. Arch Environ Contam Toxicol 42:53-59.
Bielmyer GK. 1999. Toxicity of ligand-bound silver to Ceriodaphnia dubia [M.S. Thesis], Pendleton, SC,
USA: Clemson University. 71 p.
Bielmyer GK, Gatlin D, Isely JJ, Tomasso J, Klaine SJ. 2005. Responses of hybrid striped bass to
waterborne and dietary copper in freshwater and saltwater. Comp Biochem Physiol Part C
140:131-137.
Bielmyer GK, Grosell M, Brix KV. 2006. Toxicity of silver, zinc, copper, and nickel to the copepod Acartia
tonsa exposed via a phytoplankton diet. Environ SciTechnol 40:2063-2068.
Bowen L, Werner 1, Johnson ML. 2006. Physiological and behavioral effects of zinc and temperature on
coho salmon (Oncorhynchus kisutch). Hydrobiologia 559:161-168.
Brix KV, Gillette P, Pourmand A, Capo TR, Grosell M. 2012. The effects of dietary silver on larval growth
in the echinoderm Lytechinus variegatus. Arch Environ Contam Toxicol DOI 10.1007/s00244-
012-9757-4.
Bryan GW, Gibbs PE. 1983. Heavy metals in the Fal estuary, Cornwall: a study of long-term
contamination by mining waste and its effects on estuarine organisms. Mar Biol Assoc UK
Occasional Publication 2:1-112.
Bryson WT, Garrett WR, Mallin MA, MacPherson KA, Partin WE, Woock SE. 1984. 1982 Environmental
monitoring studies, Volume II: Hyco Reservoir bioassay studies. New Hill, NC, USA Carolina
Power and Light Company. 65 pp + appendices.
Bryson WT, Garrett WR, MacPherson KA, Mallin MA, Partin WE, Woock SE. 1985a. Hyco Reservoir 1983
bioassay report. New Hill, NC, USA Carolina Power and Light Company. 55 pp.
65
-------
4.0 New Information.
Reviewer
Comments
EPA Response
Bryson WT, MacPherson KA, Mallin MA, Partin WE, Woock SE. 1985b. Hyco Reservoir 1984 bioassay
report. New Hill, NC, USA Carolina Power and Light Company. 51 pp.
Brown BE. 1977. Uptake of copper and lead by a metal tolerant isopod Asellus meridianus Rac. Freshw
Biol 7:235-244.
Canli M. 2005. Dietary and water-borne Zn exposures affect energy reserves and subsequent Zn
tolerance of Daphnia magna. Comp Biochem Physiol Part C 141:110-116.
Chou CL, Uthe JF, Castell JD, Kean JC. 1987. Effect of dietary cadmium on growth, survival, and tissue
concentrations of cadmium, zinc, copper and silver in juvenile American lobster (Homarus
americanus). Can J Fish Aquat Sci 44:1443-1450.
Cleveland L, Little EE, Buckler DR, Wiedmeyer RH. 1993. Toxicity and bioaccumulation of waterborne
and dietary selenium in juvenile bluegill (Lepomis macrochirus). Aquat Toxicol 27:265-280.
Cockell KA, Bettger WJ. 1993. Investigations of the gallbladder pathology associated with dietary
exposure to disodium arsenate heptahydrate in juvenile rainbow trout. Toxicol 77:233-248.
Cockell KA, Hilton JW. 1988. Preliminary investigation on the comparative chronic toxicity of four
dietary arsenicals to juvenile rainbow trout (Salmo gairdneri R.). Aquat Toxicol 12:73-82.
Cockell KA, Hilton JW, Bettger WJ. 1991. Chronic toxicity of dietary disodium arsenate heptahydrate to
juvenile rainbow trout (Oncorhynchus mykiss). Arch Environ Contam Toxicol 21:518-527.
Conley JM, Funk DH, Buchwalter DB. 2009. Selenium bioaccumulation and maternal transfer in the
mayfly Centroptilum triangulifer in a life-cycle, periphyton-biofilm trophic assay. Environ Sci
Technol 43:7952-7957.
Conley JM, Funk DH, Cariello NJ, Buchwalter DB. 2011. Food rationing affects dietary selenium
bioaccumulation and life cycle performance in the mayfly Centroptilum triangulifer. Ecotoxicol
20:1840-1851.
Coughlan DJ, Velte JS. 1989. Dietary toxicity of selenium-contaminated red shiners to striped bass.
Trans Am Fish Soc 118:400-408.
66
-------
4.0 New Information.
Reviewer
Comments
EPA Response
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. 1997. Largemouth bass selenium bioassay. Roxboro, NC, USA. Carolina Power & Light Company.
14 pp.
Croteau MN, Luoma SN. 2008. A biodynamic understanding of dietborne metal uptake by a freshwater
invertebrate. Environ Sci Technol 42:1801-1806.
Dai W, Fu L, Du H, Jin C. 2009. Changes in growth performance, metabolic enzyme activities, and
content of Fe, Cu, and Zn in liver and kidney of tilapia (Oreochromis niloticus) exposed to dietary
Pb. Biol Trace Elem Res 128:176-183.
Dang F, Wang W-X. 2009. Assessment of tissue-specific accumulation and effects of cadmium in a
marine fish fed contaminated commercially produced diet. Aquat Toxicol 95:248-255.
Dang F, Wang WX, Rainbow PS. 2012. Unifying prolonged copper exposure, accumulation, and toxicity
from food and water in a marine fish. Environ Sci Technol 46:3465-3471.
Davis DA, Gatlin III DM. 1996. Dietary mineral requirements offish and marine crustaceans. Rev Fish
Sci 4:75-99.
De Schamphelaere KAC, Janssen CR. 2004. Effects of chronic dietary copper exposure on growth and
reproduction of Daphnia magna. Environ Toxicol Chem 23:2038-2047.
De Schamphelaere KAC, Canli M, Van Lierde V, Forrez 1, Vanhaecke F, Janssen CR. 2004. Reproductive
toxicity of dietary zinc to Daphnia magna. Aquat Toxicol 70:233-244.
De Schamphelaere KAC, Forrez 1, Dierckens K, Sorgeloos P, Janssen CR. 2007. Chronic toxicity of dietary
copper to Daphnia magna. Aquat Toxicol 81:409-418.
Dobbs MG, Cherry DS, Cairns Jr K. 1996. Toxicity and bioaccumulation of selenium to a three-trophic
level food chain. Environ Toxicol Cheml5:340-347.
Doroshov S, Eenennaam JV, Alexander C, Hallen E, Bailey H, Kroll K, Restreppo C. 1992. Development of
water quality criteria for resident aquatic species of the San Joaquin River. Report to the
67
-------
4.0 New Information.
Reviewer
Comments
EPA Response
California State Water Resources Control Board for Contract No. 7-197-250-0. Department of
Animal Science, University of California, Davis, CA, USA.
Ebenso IE, Ologhobo AD. 2009. Effects of lead pollution against juvenile Achatina achatina fed on
contaminated artificial diet. Bull Environ Contam Toxicol 82:583-585.
Eid AE, Ghonim SI. 1994. Dietary zinc requirement of fingerling Oreochromis niloticus. Aquacult
119:259-264.
Erickson RJ, Mount DR, Highland TL, Hockett JR, Leonard EN, Mattson VR, Dawson TD, Lott KG. 2010.
Effects of copper, cadmium, lead, and arsenic in a live diet on juvenile fish growth. Can J Fish
Aquat Sci 67:1816-1826.
Evens RE, De Schamphelaere KAC, Janssen CR. 2009. The effects of dietary nickel exposure on growth
and reproduction of Daphnia magna. Aquat Toxicol 94:138-144.
Evens R, De Schamphelaere KA, Balcaen L, Wang Y, De Roy K, Resano M, Florez Mdel R, Van der
Meeren P, Boon N, Vanhaecke F, Janssen CR. 2011. Liposomes as an alternative delivery system
for investigating dietary metal toxicity to Daphnia magna. Aquat Toxicol 105:661-668.
Evens R, De Schamphelaere KA, Balcaen L, Wang Y, De Roy K, Resano M, Florez M, Boon N, Vanhaecke
F, Janssen CR. 2012. The use of liposomes to differentiate between the effects of nickel
accumulation and altered food quality in Daphnia magna exposed to dietary nickel. Aquat
Toxicol 109:80-89.
Evens RE, De Schamphelaere KAC, De Laender F, Janssen C. 2012. The effects of Zn-contaminated diets
on Daphnia magna reproduction may be related to Zn-induced changes of the dietary P content
rather than to the dietary Zn content itself. Aquat Toxicol 110-111:9-16.
Evens R, De Schamphelaere KA, De Samber B, Silversmit G, SchoonjansT, Vekemans B, Balcaen L,
Vanhaecke F, Szaloki 1, Rickers K, Falkenberg G, Vincze L, Janssen CR. 2012. Waterborne versus
dietary zinc accumulation and toxicity in Daphnia magna: a synchrotron radiation based X-ray
fluorescence imaging approach. Environ Sci Technol 46:1178-1184.
68
-------
4.0 New Information.
Reviewer
Comments
EPA Response
Farag AM, Woodward DF, Brumbaugh W, Goldstein JN, MacConnell E, Hogstrand C, Barrows FT. 1999.
Dietary effects of metals-contaminated invertebrates from the Coeur d'Alene River, Idaho, on
cutthroat trout. Trans Am Fish Soc 128:578-592.
Ferard JF, Jouany JM, Truhaut R, Vasseur P. 1983. Accumulation of cadmium in a freshwater food chain
experimental model. Ecotoxicol Environ Saf 7:43-52.
Finley KA. 1985. Observations of bluegills fed selenium-contaminated Hexagenia nymphs collected
from Belews Lake, North Carolina. Bull Environ Contam Toxicol 35:816-825.
Fort DJ, Stover EL, Rogers RL, Copley HF, Morgan LA, Foster ER. 2000. Chronic boron or copper
deficiency induces limb teratogenesis in Xenopus. Biol Trace Elem Res 77:173-187.
Franklin NM, Glover CN, Nicol JA, Wood CM. 2005. Calcium/cadmium interactions at uptake surfaces in
rainbow trout: waterborne versus dietary routes of exposure. Environ Toxicol Chem 24:2954-
2964.
Galvez F, Wood CM. 1999. The physiological effects of dietary silver sulphide exposure in rainbow
trout. Environ Toxicol Chem 18:84-88.
Gatlin III DM, Wilson RP. 1986. Dietary copper requirement of fingerling channel catfish. Aquacult
54:277-285.
Gatlin III DM, Phillips HF, Torrans EL. 1989. Effects of various levels of dietary copper and zinc on
channel catfish. Aquacult 76:127-134.
Geffard O, Geffard A, Chaumot A, Vollat B, Alvarez C, Tusseau-Vuillemin MH, Garric J. 2008. Effects of
chronic dietary and waterborne cadmium exposures on the contamination level and
reproduction of Daphnia magna. Environ Toxicol Chem 27:1128-1134.
Golding LA, Borgmann U, Dixon DG. 2011. Validation of a chronic dietary cadmium bioaccumulation
and toxicity model for Hyalella azteca exposed to field-contaminated periphyton and lake water.
Environ Toxicol Chem 30:2628-2638.
69
-------
4.0 New Information.
Reviewer
Comments
EPA Response
Gusso-Choueri PK, Choueri RB, Lombardi AT, Melao MGG. 2012. Effects of dietary copper on life-
history traits of a tropical freshwater cladoceran. Arch Environ Contam Toxicol 62 :589-598.
Haesloop U, Schirmer M. 1985. Accumulation of orally administered cadmium by the eel (Anguilla
anguilla). Chemosphere 14:1627-1634.
Hamilton SJ, Holley KM, Buhl KJ, Bullard FA. 2005. Selenium impacts on razorback sucker, Colorado
River, Colorado. III. Larvae. Ecotoxicol Environ Saf 61:168-189.
Hamilton SJ, Buhl KJ, Faerber NL, Wiedmeyer RH, Bullard FA. 1990. Toxicity of organic selenium in the
diet of chinook salmon. Environ Toxicol Chem 9:347-358.
Handy RD. 1993. The effect of acute exposure to dietary Cd and Cu on organ toxicant concentrations in
rainbow trout, Oncorhynchus mykiss. Aquat Toxicol 27:1-14.
Handy RD, Sims DW, Giles A, Campbell HA, Musonda MM. 1999. Metabolic trade-off between
locomotion and detoxification for maintenance of blood chemistry and growth parameters by
rainbow trout (Oncorhynchus mykiss) during chronic dietary exposure to copper. Aquat Toxicol
47:23-41.
Hansen JA, Lipton J, Welsh PG, Cacela D, MacConnell B. 2004. Reduced growth of rainbow trout
(Oncorhynchus mykiss) fed a live invertebrate diet pre-exposed to metal-contaminated
sediments. Environ Toxicol Chem 23:1902-1911.
Hardy RW, Oram L, Moller G. 2010. Effects of dietary selenomethionine on cutthroat trout
(Oncorhynchus clarki bouvieri) growth and reproductive performance over a life cycle. Arch
Environ Contam Toxicol 58:237-245.
Hatakeyama S, Yasuno M. 1981. The effects of cadmium-accumulated Chlorella on the reproduction of
Moina macrocopa (Cladocera). Ecotoxicol Environ Saf 5:341-350.
Hatakeyama S, Yasuno M. 1982. Accumulation and effects of cadmium on guppy (Poecilia reticulata)
fed cadmium-dosed cladocera (Moina macrocopa). Bull Environ Contam Toxicol 29:159-166.
70
-------
4.0 New Information.
Reviewer
Comments
EPA Response
Hatakeyama S, Yasuno M. 1987. Chronic effects of Cd on the reproduction of the guppy (Poecilia
reticulata) through Cd-accumulated midge larvae (Chironomus yoshimatsui). Ecotoxicol Environ
Saf 14:191-207.
Hermanutz RO, Allen KN, Detenbeck NE, Stephan CE. 1996. Exposure of bluegill (Lepomis macrochirus)
to selenium in outdoor experimental streams. Duluth, MN, USA. USEPA Mid-Continent Ecology
Division.
Hilton JW, Bettger WJ. 1988. Dietary vanadium toxicity in juvenile rainbow trout: a preliminary study.
Aquat Toxicol 12:63-71.
Hodson PV, Blunt BR, Spry DJ. 1978. Chronic toxicity of water-borne and dietary lead to rainbow trout
(Salmo gairdneri) in Lake Ontario water. Water Res 12:869-878.
Hook SE, Fisher NS. 2001a. Sublethal effects of silver in zooplankton: Importance of exposure pathways
and implications for toxicity testing. Environ Toxicol Chem 20:568-574.
Hook SE, Fisher NS. 2001b. Reproductive toxicity of metals in calanoid copepods. Mar Biol 138:1131-
1140.
Hook SE, Fisher NS. 2002. Relating the reproductive toxicity of five ingested metals in calanoid
copepods with sulfur affinity. Mar Environ Res 53:161-174.
Hoyle 1, Shaw BJ, Handy RD. 2007. Dietary copper exposure in the African walking catfish, Clarias
gariepinus: Transient osmoregulatory disturbances and oxidative stress. Aquat Toxicol 83:62-72.
Huang X, Guo F, Wang W-X. 2010. Responses of abalone Haliotis diversicolor to sublethal exposure of
waterborne and dietary silver and cadmium. Ecotoxicol Environ Saf 73:1130-1137.
Irving EC, Baird DJ, Culp JM. 2003. Ecotoxicological responses of the mayfly Baetis tricaudatus to
dietary and waterborne cadmium: Implications for toxicity testing. Environ Toxicol Chem
22:1058-1064.
Jeng SS, Sun LT. 1981. Effects of dietary zinc levels on zinc concentrations in tissues of common carp. J
Nutr 111:134-140.
71
-------
4.0 New Information.
Reviewer
Comments
EPA Response
Julshamn K, Andersen K-J, Ringdal 0, Brenna J. 1988. Effect of dietary copper on the hepatic
concentration and subcellular distribution of copper and zinc in the rainbow trout (Salmo
gairdneri). Aquacult 73:143-155.
Kamunde CN, Grosell M, Lott JNA, Wood CM. 2001. Copper metabolism and gut morphology in
rainbow trout (Oncorhynchus mykiss) during chronic sublethal dietary copper exposure. Can J
Fish Aquat Sci 58:293-305.
Kamunde C, Grosell M, Higgs D, Wood CM. 2002. Copper metabolism in actively growing rainbow trout
(Oncorhynchus mykiss): interactions between dietary and waterborne copper uptake. J Exp Biol
205:279-290.
Kamunde C, Wood CM. 2003. The influence of ration size on copper homeostasis during sublethal
dietary copper exposure in juvenile rainbow trout, Oncorhynchus mykiss. Aquat Toxicol 62:235-
254.
Kang J-C, Kim S-G, Jang SW. 2005. Growth and hematological changes of rockfish, Sebastes schlegeli
(Hilgendorf) exposed to dietary Cu and Cd. J World Aquacult Soc 36:188-195.
Kim S-G, Kang J-C. 2004. Effect of dietary copper exposure on accumulation, growth and hematological
parameters of the juvenile rockfish, Sebastes schlegeli. Mar Environ Res 58:65-82.
Kim S-G, Kim J-W, Kang J-C. 2004. Effect of dietary cadmium on growth and hematological parameters
of juvenile rockfish, Sebastes schlegeli (Hilgendorf). Aquacult Res 35:80-86.
Kjoss VA, Grosell M, Wood CM. 2005. The influence of dietary Na on Cu accumulation in juvenile
rainbow trout exposed to combined dietary and waterborne Cu in soft water. Arch Environ
Contam Toxicol 49:520-527.
Kjoss VA, Wood CM, McDonald DG. 2006. Effects of different ligands on the bioaccumulation and
subsequent depuration of dietary Cu and Zn in juvenile rainbow trout (Oncorhynchus mykiss).
Can J Aquat Sci 63:412-422.
Knox D, Cowey CB, Adron JW. 1982. Effects of dietary copper and copper:zinc ratio on rainbow trout
Salmo gairdneri. Aquacult 27:111-119.
72
-------
4.0 New Information.
Reviewer
Comments
EPA Response
Knox D, Cowey CB, Adron JW. 1984. Effects of dietary zinc intake upon copper metabolism in rainbow
trout (Salmo gairdneri). Aquacult 40:199-207.
Kock G, Bucher F. 1997. Accumulation of zinc in rainbow trout (Oncorhynchus mykiss) after waterborne
and dietary exposure. Bull Environ Contam Toxicol 58:305-310.
Kolts JM, Boese C, Meyer JS. 2009. Effects of dietborne copper and silver on reproduction by
Ceriodaphnia dubia. Environ Toxicol Chem 28:71-85.
Kumada H, Kimura S, Yokote M. 1980. Accumulation and biological effects of cadmium in rainbow
trout. Bull Jap Soc Sci Fish 46:97-103.
Lanno RP, Slinger SJ, Hilton JW. 1985a. Maximum tolerable and toxicity levels of dietary copper in
rainbow trout (Salmo gairdneri Richardson). Aquacult 49:257-268.
Lanno RP, Slinger SJ, Hilton JW. 1985b. Effect of ascorbic acid on dietary copper toxicity in rainbow
trout (Salmo gairdneri Richardson). Aquacult 49:269-287.
Lapointe D, Gentes S, Ponton DE, Hare L, Couture P. 2009. Influence of prey type on nickel and thallium
assimilation, subcellular distribution and effects in juvenile fathead minnows (Pimephales
promelas). Environ Sci Technol 43:8665-8670.
Lee SC, Lee J-H, Bai SC. 2008. A preliminary study on effects of different dietary selenium (Se) levels on
growth performance and toxicity in juvenile black seabrearn, Acathopagrus schlegeli (Bleeker).
Asian-Aust J Anim Sci 21:1794-1799.
Lee RF, Noone T. 1995. Effect of reproductive toxicants on lipovitellin in female blue crabs, Callinectes
sapidus. Mar Environ Res 39:151-154.
Lin YH, Shiau SY. 2007. The effects of dietary selenium on the oxidative stress of grouper, Epinephelus
malabaricus, fed high copper. Aquacult 267:38-43.
Lin YH, Shie YY, Shiau SY. 2008. Dietary copper requirements of juvenile grouper, Epinephelus
malabaricus. Aquacult 274:161-165.
73
-------
4.0 New Information.
Reviewer
Comments
EPA Response
Lorentzen M, Maage A, Julshamn K. 1998. Supplementing copper to a fishmeal-based diet fed to
Atlantic salmon parr affects liver copper and selenium concentrations. Aquacult Nutr 4:67-72.
Lundebye A-K, Berntssen MHG, Wendelaar Bonga SE, Maage A. 1999. Biochemical and physiological
responses in Atlantic salmon (Salmo salar) following dietary exposure to copper and cadmium.
Mar Pollut Bull 39:137-144.
Maage A. 1990. Comparison of cadmium concentrations in Atlantic salmon (Salmo salar) fry fed
different commercial feeds. Bull Environ Contam Toxicol 44:770-775.
Maage A, Julshamn K. 1993. Assessment of zinc status in juvenile Atlantic salmon (Salmo salar) by
measurement of whole body and tissue levels of zinc. Aquacult 117:179-191.
Mai K, Li H, Ai Q, Duan Q, Xu W, Zhang C, Zhang L, Tan B, Liufu Z. 2006. Effects of dietary squid viscera
meal on growth and cadmium accumulation in tissues of Japanese seabass. Aquacult Res
37:1063-1069.
Malchow DE, Knight AW, Maier KJ. 1995. Bioaccumulation and toxicity of selenium in Chironomus
decorus larvae fed a diet of seleniferous Selenastrum capricornutum. Arch Environ Contam
Toxicol 29:104-109.
Metzler D. 2003. Effect of food-borne cadmium on the reproduction of Ceriodaphnia dubia [MS thesis],
Newark, DE, USA: University of Delaware. 151 p.
Miller PA, Lanno RP, McMaster ME, Dixon DG. 1993. Relative contributions of dietary and waterborne
copper to tissue copper burdens and waterborne-copper tolerance in rainbow trout
(Oncorhynchus mykiss). Can J Fish Aquat Sci 5:1683-1689.
Mount DR, Barth AK, Garrison TD, Barten KA, Hockett JR. 1994. Dietary and waterborne exposure of
rainbow trout (Oncorhynchus mykiss) to copper, cadmium, lead and zinc using a live diet.
Environ Toxicol Chem 13:2031-2041.
Munger C, Hare L. 1997. Relative importance of water and food as cadmium sources to an aquatic
insect (Chaoborus punctipennis): Implications for predicting Cd bioaccumulation in nature.
Environ Sci Technol 31:891-895.
74
-------
4.0 New Information.
Reviewer
Comments
EPA Response
Murai T, Andrews JW, Smith RG. 1981. Effects of dietary copper on channel catfish. Aquacult 22:353-
357.
Ng TY-T, Wood CM. 2008. Trophic transfer and dietarytoxicity of Cd from the oligochaete to the
rainbow trout. Aquat Toxicol 87:1-8.
Ng TY, Klinck JS, Wood CM. 2009. Does dietary Ca protect against toxicity of a low dietborne Cd
exposure to the rainbow trout? Aquat Toxicol 91:75-86.
Ogino C, Yang G-Y. 1978. Requirement of rainbow trout for dietary zinc. Bull Jpn Soc Sci Fish 44:1015-
1018.
Ogle RS, Knight AW. 1989. Effects of elevated foodborne selenium on growth and reproduction of the
fathead minnow (Pimephales promelas). Arch Environ Contam Toxicol 18:795-803.
Overnell J, Mcintosh R. 1988. The effect of supplementary dietary copper on copper and
metallothionein levels in the fish, dab, Limanda limanda. Mar Environ Res 26:237-247.
Overnell J, Fletcher TC, Mcintosh R. 1988a. Factors affecting hepatic metallothionein levels in marine
flatfish. Mar Environ Res 24:155-158.
Overnell J, Fletcher TC, Mcintosh R. 1988b. The apparent lack of effect of supplementary dietary zinc
on zinc metabolism and metallothionein concentrations in the turbot, Scophthalmus maximus
(Linnaeus). J Fish Biol 33:563-570.
Pinho GLL, Pedroso MS, Rodrigues SC, De Souza SS, Bianchini A. 2007. Physiological effects of copper in
the euryhaline copepod Acartia tonsa: Waterborne versus waterborne plus dietborne exposure.
Aquat Toxicol 84:62-70.
Powell ML, Jones WT, Gibbs VK, Hammer HS, Lawrence JM, Fox J, Lawrence AL, Watts SA. 2010. Dietary
copper affects survival, growth, and reproduction in the sea urchin Lytechinus variegates. J
Shellfish Res 29:1043-1049.
75
-------
4.0 New Information.
Reviewer
Comments
EPA Response
Rainbow PS, Poirier L, Smith BD, Brix KV, Luoma SN. 2006. Trophic transfer of trace metals from the
polychaete worm Nereis diversicolor to the polychaete N. virens and the decapod crustacean
Palaemonetes varians. Mar Ecol Prog Ser 321:167-181.
Reynders H, Campenhout KV, Bervoets L, De Coen WM, Blust R. 2006. Dynamics of cadmium
accumulation and effects in common carp (Cyprinus carpio) during simultaneous exposure to
water and food (Tubifex tubifex). Environ Toxicol Chem 25:1558-1567.
Richardson NL, Higgs DA, Beames RM, McBrideJR. 1985. Influence of dietary calcium, phosphorus, zinc
and sodium phytate level on cataract incidence, growth and histopathology in juvenile Chinook
salmon (Oncorhynchus tshawytscha). J Nutr 115:553-567.
Roberts DA, Poore AGB, Johnston EL. 2006. Ecological consequences of copper contamination in
macroalgae: Effects on epifauna and associated herbivores. Environ Toxicol Chem 25:2470-
2479.
Rodgher S, Lombardi AT, Melao MGG, Tonietto AE. 2008. Change in life cycle parameters and feeding
rate of Ceriodaphnia silvestrii Daday (Crustacea, Cladocera) exposure to dietary copper.
Ecotoxicol 17:826-833.
Rodgher S, Lombardi AT, Melao MGG. 2009. Evaluation onto life cycle parameters of Ceriodaphnia
silvestrii submitted to 36 days dietary copper exposure. Ecotoxicol Environ Saf 72:1748-1753.
Rowe CL, Heyes A, Hopkins W. 2009. Effects of dietary vanadium on growth and lipid storage in a larval
Anuran: Results from studies employing ad libitum and rationed feeding. Aquat Toxicol 91:179-
186.
Rozon-Ramilo LD, Dube MG, Squires AJ, Niyogi S. 2011. Examining waterborne and dietborne routes of
exposure and their contribution to biological response patterns in fathead minnow (Pimephales
promelas). Aquat Toxicol 105:466-481.
Ruangsomboon S, Wongrat L. 2006. Bioaccumulation of cadmium in an experimental aquatic food
chain involving phytoplankton (Chlorella vulgaris), zooplankton (Moina macrocopa), and the
predatory catfish Clarias macrocephalus X C. gariepinus. Aquat Toxicol 78:15-20.
76
-------
4.0 New Information.
Reviewer
Comments
EPA Response
Sabatini SE, Chaufan G, Juarez Ab, Coalova 1, Bianchi L, Eppis MR, del Carmen Rfos de Molina M. 2009.
Dietary copper effects in the estuarine crab, Neohelice (Chasmagnathus) granulata, maintained
at two different salinities. Comp Biochem Physiol Part C 150:521-527.
Serra R, Isani G, Cattani 0, Carpene E. 1996. Effects of different levels of dietary zinc on the guilthead,
Sparus aurata during the growing season. Biol Trace Elem Res 51:107-116.
Shaw BJ, Handy RD. 2006. Dietary copper exposure and recovery in Nile tilapia, Oreochromis niloticus.
Aquat Toxicol 76:111-121.
Shim KF, Lee TL. 1993. Zinc requirement of the guppy (Poecilia reticulate Peters). J Aqua Trop 8:81-90.
Sofyan A, Shaw JR, Birge WJ. 2006. Metal trophic transfer from algae to cladocerans and the relative
importance of dietary metal exposure. Environ Toxicol Chem 25:1034-1041.
Sofyan A, Rosita G, Price DJ, Birge WJ. 2007. Cadmium uptake by Ceriodaphnia dubia from different
exposures: Relevance to body burdens and toxicity. Environ Toxicol Chem 26:470-477.
Spry DJ, Hodson PV, Wood CM. 1988. Relative contributions of dietary and waterborne zinc in the
rainbow trout, Salmo gairdneri. Can J Fish Aquat Sci 45:32-41.
Szczerbik P, Mikolajczyk T, Sokolowska-Mikolajczyk M, Socha M, Chyb J, Epler P. 2006. Influence of
long-term exposure to dietary cadmium on growth, maturation and reproduction of goldfish
(subspecies: Prussian carp Carrassium auratus gibelio B.). Aquat Toxicol 77:126-135.
Takeda H, Shimma Y. 1977. Effects of toxic amounts of dietary zinc on the growth and body
components of rainbow trout at two levels of calcium. Bull Freshw Fish Res 27:103-109.
Tashjian DH, Teh SJ, Sogomonyan A, Hung SSO. 2006. Bioaccumulation and chronic toxicity of dietary 1-
selenomethionine in juvenile white sturgeon (Acipenser transmontanus). Aquat Toxicol 79:401-
409.
Teh SJ, Deng X, Deng D-F, Teh F-C, Hung SSO, Fan TW-M, Liu J, Higashi RM. 2004. Chronic effects of
dietary selenium on juvenile Sacramento splittail (Pogonichthys macrolepidotus). Environ Sci
Technol 38:6085-6093.
77
-------
4.0 New Information.
Reviewer
Comments
EPA Response
Thomas JK, Janz DM. 2011. Dietary selenomethionine exposure in adult zebrafish alters swimming
performance, energetics and the physiological stress response. Aquat Toxicol 102:79-86.
Vidal D, Bay SM, Schlenk D. 2005. Effects of dietary selenomethionine on larval rainbow trout
(Oncorhynchus mykiss). Arch Environ Contam Toxicol 49:71-75.
Wallace WG, Brouwer TMH, Brouwer M, Lopez GR. 2000. Alterations in prey capture and induction of
metallothioneins in grass shrimp fed cadmium-contaminated prey. Environ Toxicol Chem
19:962-971.
Wekell JC, Shearer KD, Houle CR. 1983. High zinc supplementation of rainbow trout diets. Prog Fish-
Cult 45:144-147.
Wekell JC, Shearer KD, Gauglitz EJ. 1986. Zinc supplementation of trout diets: tissue indicators of body
zinc status. Prog Fish-Cult 48:205-212.
Woock SE, Garrett WR, Partin WE, Bryson WT. 1987. Decreased survival and teratogenesis during
laboratory selenium exposures to bluegill, Lepomis macrochirus. Bull Environ Contam Toxicol
39:998-1005.
Woodward DF, Brumbaugh WG, DeLonay AJ, Little EE, Smith CE. 1994. Effects on rainbow trout fry of a
metals-contaminated diet of benthic invertebrates from the Clark Fork River, Montana. Trans
Am Fish Soc 123:51-62.
Woodward DF, Farag AM, Bergman HL, DeLonay AJ, Little EE, Smith CE, Barrows FT. 1995. Metals-
contaminated benthic invertebrates in the Clark Fork River, Montana: Effects on age-0 brown
trout and rainbow trout. Can J Fish Aquat Sci 52:1994-2004.
Xie L, Funk DH, Buchwalter DB. 2010. Trophic transfer of Cd from natural periphyton to the grazing
mayfly Centroptilum triangulifer in a life cycle test. Environ Pollut 158:272-277.
Yoshimatsu T, Higuchi T, Zhang D, Fortes NR, Tanaka K, Yoshimura K. 2006. Effect of dietary cobalt
supplementation on the population growth of rotifer Brachionus rotundiformis. Fish Sci 72:214-
216.
78
-------
4.0 New Information.
Reviewer
Comments
EPA Response
Physiological and Other Endpoints
Alves LC, Wood CM. 2006. The chronic effects of dietary lead in freshwater juvenile rainbow trout
(Oncorhynchus mykiss) fed elevated calcium diets. Aquat Toxicol 78:217-232.
Baker RTM, Handy RD, Davies SJ, Snook JC. 1998. Chronic dietary exposure to copper affects growth,
tissue lipid peroxidation, and metal composition of the grey mullet, Chelon labrosus. Mar
Environ Res 45:357-365.
Baldisserotto B, Chowdhury MJ, Wood CM. 2005. Effects of dietary calcium and cadmium on cadmium
accumulation, calcium and cadmium uptake from the water, and their interactions in juvenile
rainbow trout. Aquat Toxicol 72:99-117.
Barata C, Markich SJ, Baird DJ, Soares AMVM. 2002. The relative importance of water and food as
cadmium sources to Daphnia magna Straus. Aquat Toxicol 61:143-154.
Berntssen MHG, Aspholm 00, Hylland K, Wendelaar Bonga SE, Lundebye A-K. 2001. Tissue
metallothionein, apoptosis and cell proliferation responses in Atlantic salmon (Salmo salar L.)
parr fed elevated dietary cadmium. Comp Biochem Physiol Part C 128:299-310.
Berntssen MHG, Hylland K, Wendelaar Bonga SE, Maage A. 1999. Toxic levels of dietary copper in
Atlantic salmon (Salmo salar L.) parr. Aquat Toxicol 46:87-99.
Berntssen MHG, Lundebye A-K. 2001. Energetics in Atlantic salmon (Salmo salar L.) parr fed elevated
dietary cadmium. Comp Biochem Physiol Part C 128:311-323.
Brafield AE, Koodie AV. 1991. Effect of dietary zinc on the assimilation efficiency of carp (Cyprinus
carpio L.). J Fish Biol 39:893-895.
Campbell HA, Handy RD, Sims DW. 2002. Increased metabolic cost of swimming and consequent
alterations to circadian activity in rainbow trout (Oncorhynchus mykiss) exposed to dietary
copper. Can J Fish Aquat Sci 59:768-777.
79
-------
4.0 New Information.
Reviewer
Comments
EPA Response
Campbell HA, Handy RD, Sims DW. 2005. Shifts in a fish's resource holding power during a contact
paired interaction: The influence of a copper-contaminated diet in rainbow trout. Physiol
Biochem Zool 78:706-714.
Cockell KA, Hilton JW, BettgerWJ. 1992. Hepatobiliary and hematological effects of dietary disodium
arsenate heptahydrate in juvenile rainbow trout (Oncorhynchus mykiss). Comp Biochem Physiol
Part C Toxicol Pharmacol 103:453-458.
Galvez F, Hogstrand C, McGeer JC, Wood CM. 2001. The physiological effects of a biologically
incorporated silver diet in rainbow trout (Oncorhynchus mykiss). Aquat Toxicol 55:95-112.
Handy RD, Sims DW, Giles A, Campbell HA, Musonda MM. 1999. Metabolic trade-off between
locomotion and detoxification for maintenance of blood chemistry and growth parameters by
rainbow trout (Oncorhynchus mykiss) during chronic dietary exposure to copper. Aquat Toxicol
47:23-41.
Handy RD. 1993b. The accumulation of dietary aluminium by rainbow trout, Oncorhynchus mykiss, at
high exposure concentrations. J Fish Biol 42:603-606.
Kamunde C, MacPhail R. 2011. Metal-metal interactions of dietary cadmium, copper and zinc in
rainbow trout, Oncorhynchus mykiss. Ecotoxicol Environ Saf 74:658-667.
Lanno RP, Hicks B, Hilton JW. 1987. Histological observations on intrahepatocytic copper-containing
granules in rainbow trout reared on diets containing elevated levels of copper. Aquat Toxicol
10:251-263.
Overnell J, Fletcher TC, Mcintosh R. 1988. The apparent lack of effect of supplementary dietary zinc on
zinc metabolism and metallothionein concentration intheturbot, Scophthalmus maximus
(Linnaeus). J Fish Biol 33:563-570.
Park CW, Shimizu C. 1989. Quantitative requirements of aluminium and iron in the formulated diets
and its interrelation with other minerals in young eel. Nippon Suisan Gakkashi 55:111-116.
Poston HA. 1991. Effects of dietary aluminium on growth and composition of young Atlantic salmon.
Progressive Fish-Culturist 53:7-10.
80
-------
4.0 New Information.
Reviewer
Comments
EPA Response
Pratap HB, Fu H, Lock RAC, Wendelaar Bonga SE. 1989. Effect of waterborne and dietary cadmium on
plasma ions of the teleost Oreochromis mossambicus in relation to water calcium levels. Arch
Environ Contam Toxicol 18:568-575.
Pratap HB, Wendelaar Bonga SE. 1993. Effect of ambient and dietary cadmium on pavement cells,
chloride cells, and Na+/K+-ATPase activity in the gills of the freshwater teleost Oreochromis
mossambicus at normal and high calcium levels in the ambient water. Aquat Toxicol 26:133-150.
Szebedinszky C, McGeer JC, McDonald DG, Wood CM. 2001. Effects of chronic Cd exposure via the diet
or water on internal organ-specific distribution and subsequent gill Cd uptake kinetics in juvenile
rainbow trout (Oncorhynchus mykiss). Environ Toxicol Chem 20:597-607.
Zhang L, Wang X-L. 2005. Effects of Zn pre-exposure on Cd and Zn bioaccumulation and
metallothionein levels in two species of marine fish. Aquat Toxicol 73:353-369.
Zhang L, Wang X-L. 2006. Alteration of dissolved cadmium and zinc uptake kinetics by metal pre-
exposure in the black sea bream (Acanthopagrus schlegeli). Environ Toxicol Chem 25:1312-1321.
Reviewer
4
References cited:
Campbell PGC. 1995. Interactions between trace metals and aquatic organisms: A critique of the free-
ion activity model. In: Metal speciation and bioavailability in aquatic systems, (Tessier A, Turner
DR, eds). New York, NY, USA:John Wiley & Sons, 45-102.
Meyer JS. 2002. The utility of the terms "bioavailability" and "bioavailable fraction" for metals. Marine
Environmental Research 53:417-423.
Reviewer
5
References cited in the review
Balistrieri, L.S., D.A. Nimick, and C.A. Mebane. 2012. Assessing time-integrated dissolved
concentrations and predicting toxicity of metals during diel cycling in streams. Science of the
Total Environment. 425: 155-168. https://doi.Org/10.1016/j.scitotenv.2012.03.008
81
-------
4.0 New Information.
Reviewer
Comments
EPA Response
Bervoets, L. and R. Blust. 2000. Effects of pH on cadmium and zinc uptake by the midge larvae
Chironomus riparius. Aquatic Toxicology. 49(1): 145-157. https://doi.org/10.1016/S0166-
445X(99)00066-1
Borgmann, U., Y. Couillard, P. Doyle, and D.G. Dixon. 2005. Toxicity of sixty-three metals and metalloids
to Hyalella azteca at two levels of water hardness. Environmental Toxicology and Chemistry.
24(3): 641-652
Bradley, R.W. and J.B. Sprague. 1985. The influence of pH, water hardness, and alkalinity on the acute
lethality of zinc to rainbow trout (Salmo gairdneri). Canadian Journal of Fisheries and Aquatic
Sciences. 42: 731-736. https://doi.org/10.1139/f85-094
Bringolf, R.B., B.A. Morris, C.J. Boese, R.C. Santore, H.E. Allen, and J.S. Meyer. 2006. Influence of
dissolved organic matter on acute toxicity of zinc to larval fathead minnows (Pimephales
promelas). Archives of Environmental Contamination and Toxicology. 51(3): 438-444.
https://doi.org/10.1007/s00244-005-0088-6
Brix, K.V., D.K. DeForest, L.M. Tear, M. Grosell, and W.J. Adams. 2017. Use of multiple linear regression
models for setting water quality criteria for copper: A complementary approach to the Biotic
Ligand Model. Environmental Science & Technology. 51(9): 5182-5192.
https://doi.org/10.1021/acs.est.6b05533
Buchwalter, D.B., D.J. Cain, C.A. Martin, L. Xie, S.N. Luoma, and T. Garland, Jr,. 2008. Aquatic insect
ecophysiological traits reveal phylogenetically based differences in dissolved cadmium
susceptibility. Proceedings of the National Academy of Sciences of the United States of America.
105(24): 8321-8326. https://doi.org/10.1073/pnas.0801686105
CCME. 2018. Canadian Water Quality Guidelines: Zinc. Scientific Criteria Document. Canadian Council
of Ministers of the Environment, ISBN 978-1-77202-043-4 PDF, Winnipeg. 127 pp.
Chapman, G.A., S. Ota, and F. Recht. 1980. Effects of water hardness on the toxicity of metals to
Daphnia magna. U.S. EPA, Office of Research and Development, Corvallis, Oreg. 23 pp.
82
-------
4.0 New Information.
Reviewer
Comments
EPA Response
Clifford, M. and J.C. McGeer. 2009. Development of a biotic ligand model for the acute toxicity of zinc
to Daphnia pulex in soft waters. Aquatic Toxicology. 91(1): 26-32
https://doi.Org/10.1016/j.aquatox.2008.09.016
Clifford, M. and J.C. McGeer. 2010. Development of a biotic ligand model to predict the acute toxicity
of cadmium to Daphnia pulex. Aquatic Toxicology. 98(1): 1-7.
https://doi.Org/10.1016/j.aquatox.2010.01.001
Croteau, K., A.C. Ryan, R. Santore, D. DeForest, C. Schlekat, E. Middleton, and E. Garman. 2021.
Comparison of Multiple Linear Regression and Biotic Ligand Models to Predict the Toxicity of
Nickel to Aquatic Freshwater Organisms. Environmental Toxicology and Chemistry. n/a(n/a).
https://doi.Org/https://doi.org/10.1002/etc.5063
Cusimano, R.F., D.F. Brakke, and G.A. Chapman. 1986. Effects of pH on the toxicities of cadmium,
copper, and zinc to steelhead trout (Salmo gairdneri). Canadian Journal of Fisheries and Aquatic
Sciences. 43(8): 1497-1503. https://doi.org/10.1139/f86-187
De Schamphelaere, K.A.C. and C.R. Janssen. 2002. A biotic ligand model predicting acute copper
toxicity for Daphnia magna: the effects of calcium, magnesium, sodium, potassium, and pH.
Environmental Science and Technology. 36(1): 48 -54.
https://doi.org/10.1021/es000253sS0013-936X(00)00253-4
De Schamphelaere, K.A.C. and C.R. Janssen. 2004a. Bioavailability and chronic toxicity of zinc to
juvenile rainbow trout (Oncorhynchus mykiss): comparison with other fish species and
development of a Biotic Ligand Model. Environmental Science and Technology. 38(23): 6201 -
6209. https://doi.org/10.1021/es049720m
De Schamphelaere, K.A.C. and C.R. Janssen. 2004b. Effects of dissolved organic carbon concentration
and source, pH, and water hardness on chronic toxicity of copper to Daphnia magna.
Environmental Toxicology and Chemistry. 23(5): 1115-1122. https://doi.org/10.1897/02-593
DeForest, D.K., K.V. Brix, L.M. Tear, and W.J. Adams. 2018. Multiple linear regression models for
predicting chronic aluminum toxicity to freshwater aquatic organisms and developing water
83
-------
4.0 New Information.
Reviewer
Comments
EPA Response
quality guidelines. Environmental Toxicology and Chemistry. 37(1): 80-90.
https://doi.org/10.1002/etc.3922
DeForest, D.K., R.C. Santore, A.C. Ryan, B.G. Church, M.J. Chowdhury, and K.V. Brix. 2017. Development
of biotic ligand model-based freshwater aquatic life criteria for lead following US Environmental
Protection Agency guidelines. Environmental Toxicology and Chemistry. 36(11): 2965-2973.
https://doi.org/10.1002/etc.3861
Diamond, J.M., E.L. Winchester, D.G. Mackler, W.J. Rasnake, J.K. Fanelli, and D. Gruber. 1992. Toxicity
of cobalt to freshwater indicator species as a function of water hardness. Aquatic Toxicology.
22: 163-180. https://doi.org/10.1016/0166-445X(92)90038-0
Erickson, R.J., D.A. Benoit, V.R. Mattson, H.P. Nelson, and E.N. Leonard. 1996. The effects of water
chemistry on the toxicity of copper to fathead minnows. Environmental Toxicology and
Chemistry. 15(2): 181-193. https://doi.org/10.1002/etc.5620150217
Grosell, M., J. Blanchard, K.V. Brix, and R. Gerdes. 2007. Physiology is pivotal for interactions between
salinity and acute copper toxicity to fish and invertebrates. Aquatic Toxicology. 84(2): 162-172.
https://doi.Org/https://doi.org/10.1016/j.aquatox.2007.03.026
Hall, L.W., R.D. Anderson, B.L. Lewis, and W.R. Arnold. 2008. The Influence of Salinity and Dissolved
Organic Carbon on the Toxicity of Copper to the Estuarine Copepod, Eurytemora affinis.
Archives of Environmental Contamination and Toxicology. 54(1): 44-56.
https://doi.org/10.1007/s00244-007-9010-8
Hansen, J.A., P.G. Welsh, J. Lipton, D. Cacela, and A.D. Dailey. 2002. Relative sensitivity of bull trout
(Salvelinus confluentus) and rainbow trout (Oncorhynchus mykiss) to acute exposures of
cadmium and zinc. Environmental Toxicology and Chemistry. 21(1): 67-75.
https://doi.org/10.1002/etc.5620210110
Heijerick, D.G., K.A.C. De Schamphelaere, and C.R. Janssen. 2002. Predicting acute zinc toxicity for
Daphnia magna as a function of key water chemistry characteristics: development and
validation of a biotic ligand model. Environmental Toxicology and Chemistry. 21(6): 1309-1315
84
-------
4.0 New Information.
Reviewer
Comments
EPA Response
Heijerick, D.G., K.A.C. De Schamphelaere, P.A. Van Sprang, and C.R. Janssen. 2005. Development of a
chronic zinc biotic ligand model for Daphnia magna. Ecotoxicology and Environmental Safety.
62(1): 1-10. https://doi.Org/10.1016/j.ecoenv.2005.03.020
Heijerick, D.G., C.R. Janssen, and W.M. De Coen. 2003. The combined effects of hardness, pH, and
dissolved organic carbon on the chronic toxicity of Zn to D. magna : development of a surface
response model. Archives of Environmental Contamination and Toxicology. 44(2): 210-217.
https://doi.org/10.1007/s00244-002-2010-9
Hyne, R.V., F. Pablo, M. Julli, and S.J. Markich. 2005. Influence of water chemistry on the acute toxicity
of copper and zinc to the cladoceran Ceriodaphnia cf dubia. Environmental Toxicology and
Chemistry. (7): 1667-1675. https://doi.Org/10.1897/04-497R.l
Ivey, C.D., J.M. Besser, J.A. Steevens, M.J. Walther, and V.D. Melton. 2019. Influence of Dissolved
Organic Carbon on the Acute Toxicity of Copper and Zinc to White Sturgeon (Acipenser
transmontanus) and a Cladoceran (Ceriodaphnia dubia). Environmental Toxicology and
Chemistry. 38(12): 2682-2687. https://doi.org/10.1002/etc.4592
Markich, S.J., G.E. Batley, J.L. Stauber, N.J. Rogers, S.C. Apte, R.V. Hyne, K.C. Bowles, K.L. Wilde, and
N.M. Creighton. 2005. Hardness corrections for copper are inappropriate for protecting
sensitive freshwater biota. Chemosphere. 60(1): 1-8. https://doi.org/
Mebane, C.A. 2006. Cadmium risks to freshwater life: derivation and validation of low-effect criteria
values using laboratory and field studies. U.S. Geological Survey Scientific Investigation Report
2006-5245 (2010 rev.). 130 pp. https://doi.org/10.3133/sir20065245.
Mebane, C.A., M.J. Chowdhury, K.A.C. De Schamphelaere, S. Lofts, P.R. Paquin, R.C. Santore, and C.M.
Wood. 2020a. Metal bioavailability models: current status, lessons learned, considerations for
regulatory use, and the path forward. Environmental Toxicology and Chemistry. 39(1): 60-84.
https://doi.org/10.1002/etc.4560
Mebane, C.A., F.S. Dillon, and D.P. Hennessy. 2012. Acute toxicity of cadmium, lead, zinc, and their
mixtures to stream-resident fish and invertebrates. Environmental Toxicology and Chemistry.
31(6): 1334-1348. https://doi.org/10.1002/etc.1820
85
-------
4.0 New Information.
Reviewer
Comments
EPA Response
Mebane, C.A., T.S. Schmidt, J.L. Miller, and L.S. Balistrieri. 2020b. Bioaccumulation and toxicity of
cadmium, copper, nickel, and zinc to aquatic insect communities. Environmental Toxicology and
Chemistry. 39(4): 812-833. https://doi.org/10.1002/etc.4663
Meyer, J.S., S.J. Clearwater, T.A. Doser, M.J. Rogaczewski, and J.A. Hansen. 2007. Effects of Water
Chemistry on Bioavailability and Toxicity of Waterborne Cadmium, Copper, Nickel, Lead, and
Zinc to Freshwater Organisms. SETAC, Pensacola, Fla. 352 pp
Naddy, R.B., W.A. Stubblefield, R.A. Bell, K.B. Wu, R.C. Santore, and P.R. Paquin. 2018. Influence of
Varying Water Quality Parameters on the Acute Toxicity of Silver to the Freshwater Cladoceran,
Ceriodaphnia dubia. Bulletin of Environmental Contamination and Toxicology. 100(1): 69-75.
https://doi.org/10.1007/s00128-017-2260-x
Naddy, R.B., W.A. Stubblefield, J.R. May, S.A. Tucker, and J.R. Hockett. 2002. The effect of calcium and
magnesium ratios on the toxicity of copper to five aquatic species in freshwater. Environmental
Toxicology and Chemistry. 21(2): 347-352. https://doi.org/10.1002/etc.5620210217
Nimick, D.A., J.R. Gammon, and S.R. Parker. 2011. Diel biogeochemical processes and their effect on
the aqueous chemistry of streams: A review. Chemical Geology. 283(1-2): 3-17.
https://doi.Org/10.1016/j.chemgeo.2010.08.017
Niyogi, S., R. Kent, and C.M. Wood. 2008. Effects of water chemistry variables on gill binding and acute
toxicity of cadmium in rainbow trout (Oncorhynchus mykiss): A biotic ligand model (BLM)
approach. Comparative Biochemistry and Physiology Part C: Toxicology & Pharmacology. 148(4):
305-314. https://doi.Org/10.1016/j.cbpc.2008.05.015
Poteat, M.D. and D.B. Buchwalter. 2014. Four reasons why traditional metal toxicity testing with
aquatic insects is irrelevant. Environmental Science & Technology. 48(2): 887-888.
https://doi.org/10.1021/es405529n
Richards, J.G. and R.C. Playle. 1998. Cobalt binding to gills of rainbow trout (Oncorhynchus mykiss): an
equilibrium model. Comparative Biochemistry and Physiology Part C: Pharmacology, Toxicology
and Endocrinology. 119(2): 185-197. https://doi.org/10.1016/S0742-8413(97)00206-5
86
-------
4.0 New Information.
Reviewer
Comments
EPA Response
Richards, J.G. and R.C. Playle. 1999. Protective effects of calcium against the physiological effects of
exposure to a combination of cadmium and copper in rainbow trout (Oncorhynchus mykiss).
Canadian Journal of Zoology. 77(7): 1035-1047. https://doi.Org/doi:10.1139/cjz-77-7-1035
Santore, R.C., K. Croteau, A.C. Ryan, C. Schlekat, E. Middleton, and E. Garman. 2021. A Review of Water
Quality Factors that Affect Nickel Bioavailability to Aquatic Organisms: Refinement of the Biotic
Ligand Model for Nickel in Acute and Chronic Exposures. Environmental Toxicology and
Chemistry. n/a(n/a). https://doi.Org/https://doi.org/10.1002/etc.5109
Santore, R.C., A.C. Ryan, F. Kroglund, H.C. Teien, P.H. Rodriguez, W.A. Stubblefield, A.S. Cardwell, W.J.
Adams, and E. Nordheim. 2018. Development and application of a Biotic Ligand Model for
predicting the toxicity of dissolved and precipitated aluminum. Environmental Toxicology and
Chemistry. 37(1): 70-79. https://doi.org/10.1002/etc.4020
Schubauer-Berigan, M.K., J.R. Dierkes, P.D. Monson, and G.T. Ankley. 1993. pH-dependent toxicity of
Cd, Cu, Ni, Pb, and Zn to Ceriodaphnia dubia, Pimephales promelas, Hyalella azteca, and
Lumbriculus variegatus. Environmental Toxicology and Chemistry. 12(7): 1261-1266.
https://doi.org/10.1002/etc.5620120715
Tan, Q.-G. and W.-X. Wang. 2011. Acute Toxicity of Cadmium in Daphnia magna under Different
Calcium and pH Conditions: Importance of Influx Rate. Environmental Science & Technology.
45(5): 1970-1976. https://doi.org/10.1021/esl02453b
USEPA. 2016a. Aquatic Life Ambient Water Quality Criterion for Cadmium - 2016. EPA-820-R-16-002.
721 pp.
USEPA. 2016b. Draft Technical Support Document: Recommended Estimates for Missing Water Quality
Parameters for Biotic Ligand Model. EPA 820-R-15-106.
Van Genderen, E.J., R.W. Gensemer, C. Smith, R.C. Santore, and A.C. Ryan. 2007. Evaluation of the
Biotic Ligand Model relative to other site-specific criteria derivation methods for copper in
surface waters with elevated hardness Aquatic Toxicology. 84(2): 279-291.
https://doi.Org/10.1016/j.aquatox.2007.02.024
87
-------
4.0 New Information.
Reviewer
Comments
EPA Response
Wang, N., C.A. Mebane, J.L. Kunz, C.G. Ingersoll, W.G. Brumbaugh, R.C. Santore, J.W. Gorsuch, and
W.R. Arnold. 2011. Influence of dissolved organic carbon on toxicity of copper to a unionid
mussel (Villosa iris) and a cladoceran (Ceriodaphnia dubia) in acute and chronic water
exposures. Environmental Toxicology and Chemistry. 30(9): 2115-2125.
https://doi.org/10.1002/etc.596
Wang, N., C.A. Mebane, J.L. Kunz, C.G. Ingersoll, T.W. May, W.R. Arnold, R.C. Santore, T. Augspurger,
F.J. Dwyer, and M.C. Barnhart. 2009. Evaluation of acute copper toxicity to juvenile freshwater
mussels (fatmucket, Lampsilis siliquoidea) in natural and reconstituted waters. Environmental
Toxicology and Chemistry. 28(11): 2367-2377. https://doi.Org/10.1897/08-655.l
Welsh, P.G., J. Lipton, C.A. Mebane, and J.C.A. Marr. 2008. Influence of flow-through and renewal
exposures on the toxicity of copper to rainbow trout. Ecotoxicology and Environmental Safety.
69(2): 199-208. https://doi.Org/10.1016/j.ecoenv.2007.04.003
Welsh, P.G., J. Lipton, T.L. Podrabsky, and G.A. Chapman. 2000. Relative importance of calcium and
magnesium in hardness-based modification of copper toxicity. Environmental Toxicology and
Chemistry. 19(6): 1624-1631. https://doi.org/10.1002/etc.5620190620
Wood, C.M. 2012. An introduction to metals in fish physiology and toxicology: Basic principles. Fish
Physiology: Homeostasis and Toxicology of Essential Metals. 31(PART A): 1-51.
https://doi.org/10.1016/S1546-5098(11)31001-1
Wood, C.M., H.A. Al-Reasi, and D.S. Smith. 2011. The two faces of DOC. Aquatic Toxicology. 105S(3-4):
3-8. https://doi.Org/10.1016/j.aquatox.2011.03.007
Yoro, S.C., C. Panagiotopoulos, and R. Sempere. 1999. Dissolved organic carbon contamination induced
by filters and storage bottles. Water Research. 33(8): 1956-1959.
https://doi.org/10.1016/S0043-1354(98)00407-2
88
-------
References cited in EPA's Response
Alsop D, Ng TY-T, Chowdhury MJ, Wood CM. 2016. Interactions of waterborne and dietborne Pb in
rainbow trout, Oncorhynchus mykiss: bioaccumulation, physiological responses, and chronic
toxicity. Aquat Toxicol 177:343-354.
Amiard JC, Amiard-Triquet C, Barka S, Pellerin J, Rainbow PS. 2006. Metallothioneins in aquatic
invertebrates: their role in metal detoxification and their use as biomarkers. Aquat Toxicol
76(2): 160-202.
Besser JM, Ivey CD, Steevens J A, Cleveland D, Soucek D, Dickinson A, Van Genderen EJ, Ryan AC,
Schlekat CE, Garman E, Middleton E, Santo re R. 2021. Modeling the Bioavailability of Nickel
and Zinc to Ceriodaphnia dubia and Neocloeon triangulifer in Toxicity Tests with Natural
Waters. Environ Toxicol Chem. 40(ll):3049-3062. DOI: 10.1002/etc.5178.
Brix KV, DeForest DK, Tear L, Grosell M, Adams WJ. 2017. Use of multiple linear regression models
for setting water quality criteria for copper: A complementary approach to the biotic ligand
model. Environ Sci Technol 51:5182-5192.
Brix, K. V., Tear, L., Santore, R. C., Croteau, K., & DeForest, D. K. 2021. Comparative Performance of
Multiple Linear Regression and Biotic Ligand Models for Estimating the Bioavailability of
Copper in Freshwater. Environ Toxicol Chem. 40(6): 1649-1661. DOI: 10.1002/etc.5012.
Croteau K, Ryan A, Santore R, DeForest D, Schlekat C, Middleton E, Garman, E. 2021. Comparison of
Multiple Linear Regression and Biotic Ligand Models to Predict the Toxicity of Nickel to
Aquatic Freshwater Organisms. Environ Toxicol Chem. 40(8):2189-2205. DOI:
10.1002/etc.5063
DeForest, DK, Brix KV, Tear LM, Cardwell AS, Stubblefield WA, Nordheim E, Adams WJ. 2020.
Updated multiple linear regression models for predicting chronic aluminum toxicity to
freshwater aquatic organisms and developing water quality guidelines. Environ Toxicol
Chem. 39(9), 1724-1736.
Deleebeeck NME, De Schamphelaere KAC, Janssen CR. 2008. A novel method for predicting chronic
nickel bioavailability and toxicity to Daphnia magna in artificial and natural waters. Environ
Toxicol Chem. 27:2097-2107.
De Schamphelaere KA, Nys C, Janssen CR. 2014. Toxicity of lead (Pb) to freshwater green algae:
development and validation of a bioavailability model and inter-species sensitivity
comparison. Aquat Toxicol. 155:348-59.
De Schamphelaere K, Canli M, Van Lierde V, Forrez I, Vanhaecke F, Janssen C. 2004. Reproductive
toxicity of dietary zinc to Daphnia magna. Aquat Toxicol. 70:233-244.
De Schamphelaere K, Forrez I, Dierckens K, Sorgeloos P, Janssen C. 2007. Chronic toxicity of dietary
copper to Daphnia magna. Aquat Toxicol. 81:409-418.
89
-------
Evens R, De Schamphelaere KA, Janssen CR. 2009. The effects of dietary nickel exposure on growth
and reproduction of Daphnia magna. Aquat Toxicol. 94:138-144
doi:10.1016/j.aquatox.2009.06.011
Farley KJ, Meyer JS, Balistrieri LS, Iwasaki Y, Kamo M, Lofts S, Mebane CA, Naito W, Ryan AC,
Santore RC, Tipping E. 2015. Metal mixtures modeling evaluation: 2. Comparison of four
modeling approaches. Environ Toxicol Chem. 34:741-753.
Mebane CA, Dillon FS, Hennessy DP. 2012. Acute toxicity of cadmium, lead, zinc, and their mixtures
to stream-resident fish and invertebrates. Environ Toxicol Chem. 31(6): 1334-1348.
https://doi.org/10.1002/etc.1820
Mebane CA, Chowdhury MJ, De Schamphelaere KAC, Lofts S, Paquin PR, Santore RC, Wood CM.
2020. Metal bioavailability models: Current status, lessons learned, considerations for
regulatory use, and the path forward. Environ Toxicol Chem. 39:60-84.
Mebane CA, Schmidt TS, Miller JL, Balistrieri LS. 2020. Bioaccumulation and toxicity of cadmium,
copper, nickel, and zinc to aquatic insect communities. Environ Toxicol Chem. 39:812-833.
Meyer JS, Traudt EM, Ranville JF. 2018. Is the Factor-of-2 Rule Broadly Applicable for Evaluating the
Prediction Accuracy of Metal-Toxicity Models? Bull Environ Contam Toxicol. 100:64-68.
Niyogi S,Pyle GG, Wood CM. 2007. Branchial versus intestinal zinc uptake in wild yellow perch
(Perca flavescens) from reference and metal-contaminated aquatic ecosystems. Can. J. Fish.
Aquat. Sci. 64: 1605-1613
Nys C, Janssen C, De Schamphelaere KAC. 2017. Validating the chronic Pb algae bioavailability
model at high pH: Single-species evaluation. Prepared for International Lead Association,
Durham, NC, U.S.A.
Nys C, Janssen C, De Schamphelaere KAC. 2013. An investigation of the potential toxicity of dietary
Pb to Ceriodaphnia dubia. Durham, NC, U.S.A.
Roussel H, Joachim S, Lamothe S, Palluel O, Gauthier L, Bonzom J-M. 2007. A long-term copper
exposure on freshwater ecosystem using lotic mesocosms: individual and population
responses of three-spined sticklebacks (Gasterosteus aculeatus). Aquat Toxicol 82:272-280.
Schlekat CE, Van Genderen E, De Schamphelaere KA, Antunes PM, Rogevich EC, Stubblefield WA.
2010. Cross-species extrapolation of chronic nickel Biotic Ligand Models. Sci Total Environ
408:6148-6157.
U.S. Environmental Protection Agency. 2003. The Biotic Ligand Model: Technical Support Document
for Its Application to the Evaluation of Water Quality Criteria for Copper. EPA 822-R-03-027.
Washington, DC.
90
-------
US Environmental Protection Agency. 2007. Aquatic life ambient freshwater quality criteria-
Copper. EPA 822-R-07-001. Washington, DC.
US Environmental Protection Agency. 2016. Final Aquatic Life Ambient Water Quality Criterion for
Selenium - Freshwater 2016. EPA 822-R-21-006. Washington, DC.
US Environmental Protection Agency. 2018. Final aquatic life ambient water quality criteria for
aluminum. EPA-822-R-18-001. Washington, DC.
Versteeg DJ, Belanger SE, Carr GJ. 1999. Understanding single-species and model ecosystem
sensitivity: Data-based comparison. Environ Toxicol Chem. 18:1329-1346.
Wood CM, Al-Reasi HA, Smith DS. 2011. The two faces of DOC. Aquat Toxicol. 105S(3-4): 3-8.
https://doi.Org/10.1016/j.aquatox.2011.03.007
------- |