EPA/822/B-92/004
             United States
             Environmental Protection
             Agency
Office of Science and Technology
Health and Ecological Criteria Division
Washington, DC 20460
May 1992
\Epy\    Interim Guidance on
             Interpretation and Implementation
   "         of Aquatic  Life Criteria for Metals
   V;

-------
         Interim Guidance on
Interpretation and Implementation of
   Aquatic Life Criteria for Metals
                 May 1992
        Health and Ecological Criteria Division
         Office of Science and Technology
        U.S. Environmental Protection Agency
             Washington, DC 20460

-------
     This document provides guidance for interpreting and im-
     plementing aquatic life criteria for metals in waters of the
     United States. It is issued in support of EPA regulations and
initiatives involving the application of water quality criteria and
standards. This document is agency guidance only. It does not es-
tablish or affect legal rights or obligations. It does not establish a
binding norm, or prohibit alternatives not included in the docu-
ment. It is not  finally determinative of the issues  addressed.
Agency decisions in any particular case will be made by applying
the law and regulations on the basis of specific facts when regula-
tions are promulgated or permits are issued.
   This document is  expected to be revised periodically to reflect
advances in this rapidly evolving area. Comments from readers
are welcomed. Send comments to Health and Ecological Criteria
Division (WH-586), U.S. EPA, 401 M  Street SW, Washington, DC
20460.
Foreword
                                                           iii

-------
     This guidance addresses the use of EPA (and corresponding
     State) metals criteria in water quality standards intended to
     protect aquatic life. This guidance also addresses the deriva-
tion of NPDES permit limits from such criteria. The main body of
the document presents recommendations on the best current ap-
proaches for implementing aquatic life criteria  for metals and
measuring attainment of such criteria. This guidance supersedes
past  criteria document statements expressing criteria in terms of
an acid soluble analytical method. Appendix  A presents a case
study illustrating derivation of site-specific criteria (item 3 below).
Appendix B presents recommendations on  the  derivation  of
NPDES permit limits from ambient metals criteria. As described in
Appendix B, it supersedes part of the Technical Support Docu-
ment [1] discussion of metals.
   The principal issue is the correlation between metals that are
measured  and  metals that  are  biologically  available.  The
bioavailability and toxitity of metals depend strongly on the exact
physical and chemical form of the metal, and  on the species af-
fected. The form of the metal, in turn, can vary depending on the
chemical characteristics of the surrounding water matrix. Because
of differences between various effluents and site waters, and be-
tween  laboratory  toxicity  test waters  and many site waters,
establishment  and implementation of  metals criteria  are  not
straight forward.  Consequently,  this  guidance presents  three
reasonable approaches that differ in their complexity.
   (1)  The simplest approach is to measure total recoverable
       metals in ambient waters, and to compare such meas-
       urements to national or state-wide criteria.

   (2)  A closer focus on biologically available metals can be
       obtained by measuring dissolved metals  in ambient
       waters, and comparing such measurements to criteria
       appropriate for dissolved metal. Since  effluent limits,
       for both technical and legal (40 CFR 122.45) reasons,
       are generally expressed in terms of total recoverable
       metal,  it is necessary to translate between the  total
       recoverable concentration in the effluent and the dis-
       solved concentration in the ambient  water.

   (3)  Because of the complexity of metal chemistry, there is
       no one chemical analytical method that can accurately
       determine  the metals that are bioavailable and toxic.
       For  implementing  metals  criteria  established  from
       laboratory toxicity tests, an adjustment of the criteria
       value can address this constraint. It involves measur-
Synopsis

-------
Introduction
                                  ing a pollutant's  water-effect ratio  in  the  receiving
                                  water covered by  the standard. The water-effect ratio
                                  compares the toxicity of a pollutant in the actual site
                                  water to its toxicity in laboratory water, for two or
                                  more aquatic species. Because the metal's toxicity in
                                  laboratory water is the basis for the national criterion,
                                  the water-effect ratio is used in an adjustment to ob-
                                  tain a site-specific value. Implemented in conjunction
                                  with either of the first two alternatives, this adjust-
                                  ment may  either  increase or decrease  the  numeric
                                  value of the criterion.
     The principal problem in relating discharges of toxic metals to
     environmental impacts is the different toxicities of various
     metal species in ambient waters, and the varying fractions of
such species with location and time. This results in the same metal
concentration exerting different toxicity from place to place and
from time to time. The chemical species involved include metals
dissolved in a variety of forms, and metals sorbed to or within
particulate matter. Metals may differ markedly from each other
with respect to speciation and bioavailability.
   Although metal toxicity may vary depending on the chemical
characteristics  of the water body, the national criteria have been
designed to protect all or almost all bodies of water. However, this
does not mean that the national criteria will always be overprotec-
tive. For  example, some untested locally important species might
be very sensitive to the material of concern, or the local aquatic or-
ganisms  might  have  increased  sensitivity due to  diseases,
parasites, other pollutants or water quality conditions, or extreme
flow or temperature conditions [2].
   Another problem involves metal speciation in effluents, and
the potential transformations that may occur in moving from the
chemical environment of the effluent to the chemical environment
of the  receiving water. Consequently, in  contrast to an ambient
measurement,  which should respond predictably to metal that is
actually bioavailable, an effluent measurement needs to respond
also to metal that may not be bioavailable  under effluent chemical
conditions, but would  possibly become bioavailable under am-
bient chemical conditions.
   Because  of the complexity of metal speciation and its effect on
toxicity, the relationship between measured  concentrations and
toxicity is not precise. Consequently, any chemical  analytical
method that could be recommended would not guarantee precise
comparability  between concentrations measured in the field and
concentrations  employed in the toxicity  tests  underlying the
criteria. However, the three approaches presented in this guidance
should provide acceptable approximations.

-------
      EPA has recognized four methods of sample preparation for
      metals analysis. These lead to measurement of: (a) total me-
      tals, (b) total recoverable metals, (c) acid soluble metals, and
(d) dissolved metals. Ordinarily, the four methods measure all of
the dissolved metal present at the time of sampling. They differ in
the amount of particulate metal that they measure.
   The total metals procedure, the total recoverable metals proce-
dure [3], and the acid soluble metals procedure [4, 5] measure
metals that  are dissolved in  water or  become  dissolved when
treated with acid. They differ in the concentration of acid and in
the temperature  used  during  the analytical  procedure,  both
decreasing in the order cited above.
   The dissolved procedure  [3]  measures metal that passes
through a 0.45 [am filter at the  time of sample collection.  The
results from this procedure are reported as "dissolved," although
it may include metal that was bound to micro-particles (<0.45 ^tm)
at the time of sample collection. More recent dissolved procedures
recommend positive-pressure, in-line  filtration through poly-
carbonate membrane filters having a uniform pore size selected
from  a range  of 0.1-0.4 jam [6],  and emphasize ultra-clean
laboratories, labware,  and reagents  [7, 13]. Measurements using
different filter sizes may, however, give different results.
   Metals criteria documents issued in 1980 recommended the
use of the total recoverable method. Beginning in 1984, although a
final acid  soluble method was not available, the  criteria docu-
ments have stated that an acid soluble method would be a better
way  of measuring attainment of  the  criteria.  Noting the un-
availability of a final method, they recommended the continued
use of the total recoverable method, which they acknowledge may
be overly protective.
   Because the acid soluble method uses a less rigorous digestion,
it was expected that it would generally measure less of the particu-
late metal than the total  recoverable method. It was  therefore
believed that the acid soluble method would more accurately
measure bioavailable metal. Recently available ambient and ef-
fluent  data suggest,  however,  that acid  soluble  results are
ordinarily nearly identical to total recoverable results, while being
somewhat different from dissolved results. Because an increased
understanding of the complexity of metals bioavailability indi-
cates that the acid soluble method will not significantly improve
the correlation between  measured metal and bioavailable metal,
this guidance is not recommending the use of this method.
      Bioavailability and toxicity vary with the form of the metal.
      Particulate  metal  is  generally  expected  to  have  less
      bioavailability  than  dissolved  metal.  Nevertheless,  the
toxicity of ambient particulate metal is not necessarily zero. For
example, some metal that is in the particulate phase in the ambient
Background
on Analytical
Methods
Bioavailability
and  Toxicity

-------
Dissolved
and  Total
Recoverable
Approaches
                          water environment may become dissolved  in the chemical en-
                          vironment associated with the gill or the gut.
                             In natural waters, some metals may exist in a variety of dis-
                          solved species that differ significantly in toxicity. For copper, the
                          divalent free cation and some inorganic complexes have substan-
                          tial toxicity, whereas dissolved organic complexes generally have
                          significantly less toxicity.  As a  result, the  same concentration of
                          dissolved copper may exert different toxicity in different waters.
                             Toxicity tests that form the basis for the criteria are usually per-
                          formed in an untreated or slightly treated natural water from an
                          uncontaminated source, or in water that has been first purified
                          and then reconstituted by the  addition of  appropriate mineral
                          salts.  Because such dilution water is generally lower in metal-
                          binding particulate matter and dissolved organic matter than most
                          ambient waters, these toxicity  tests may  overstate the ambient
                          toxicity of non-biomagnified metals that interact with particulate
                          matter or dissolved organic matter.
                             In most but not all toxicity tests underlying the criteria, the
                          percentage of metal in the particulate phase is fairly low. For am-
                          bient waters, on the other hand, recent data suggest that typically
                          30-80  percent of the copper, nickel, and zinc, and 90-95 percent of
                          the lead  may be  in a particulate phase measured by the total
                          recoverable method but not by the dissolved method.
                             In freshwater laboratory tests, organic carbon concentrations
                          of a few mg/L are typical, with chronic tests having higher con-
                          centrations than acute tests. In ambient waters,  organic carbon
                          concentrations are typically somewhat higher than this, and may
                          be substantially higher at the edge of small mixing zones.
                             Because of the greater fraction of particulate metal in ambient
                          waters, as well as the higher levels of dissolved organic binding
                          agents in ambient waters, the fraction of metal that is biologically
                          available may often be lower under ambient field conditions than
                          under laboratory conditions, particularly for freshwaters.
      Aquatic life criteria in ambient waters may be implemented
      either as total recoverable metal or as dissolved metal. Ef-
      fluent  limits  must  generally be  expressed  as  total
recoverable metal. For analyses of metals in the low ng/L range or
below, ultra-clean sample handling techniques [7, 13] should be
used.

Ambient Waters

When used for expressing ambient water quality criteria, the total
recoverable  method provides  greater  safety  than does  the
dissolved method. Nevertheless, when used for  ambient  waters,

-------
total recoverable measurements may result in overestimating the
toxicity. While toxicity testing has shown dissolved measurements
to  be  better  predictors  of  toxicity than total recoverable
measurements, there are also some potential concerns with this
approach, as discussed below.
    First, EPA water  quality criteria are generally based on the
reported total recoverable  concentrations in the toxicity tests.  If
used  for  dissolved standards, the criteria  values need  to  be
downwardly adjusted to account for the typical dissolved fraction
in test dilution water. For copper, approximately 86 percent of the
reported total concentration was  dissolved during freshwater
acute toxicity tests with the more sensitive species. Consequently,
the copper freshwater acute criterion should be adjusted to 86 per-
cent of its  total recoverable  criterion in order  to  serve  as  a
dissolved criterion, particularly in waters having low  concentra-
tions  of dissolved organic  binding agents. While the adjustment
may be small  for most metals, a few metals, such as aluminum,
may require much larger  adjustments to account  for the much
lower percentage  dissolved  typically  occurring  during toxicity
tests.  Chronic  criteria may require larger adjustments than acute
criteria, due to the higher particulate concentrations caused by the
addition of food during chronic tests.
    Except for copper in freshwater, the factors are not yet avail-
able for converting EPA's published criteria into dissolved criteria.
A re-examination of data underlying the metals criteria is now un-
derway to compile the dissolved concentrations measured during
toxicity tests.  While preliminary analysis does not indicate that
these dissolved adjustment factors are of sufficient magnitude to
be of great concern, they should be considered in any adoption of
dissolved metal standards subsequent to distribution of this infor-
mation.

    Second,  by measuring  comparatively  little of  the particulate
fraction, it may be possible that the dissolved method could oc-
casionally understate the toxicologically effective concentration.
Although toxicity data suggest that this is not ordinarily a prob-
lem, it is more likely to be a concern if the dissolved concentration
is only a very small percentage of the particulate concentration,
such as may occur with aluminum.
    In some situations the dissolved method may overstate the
toxicologically effective concentration.  When certain metals (such
as copper) become complexed with elevated concentrations of dis-
solved organic matter, a reduction in toxicity may occur, compared
to toxicity in  laboratory water, which is  low in  organic  matter.
Where dissolved organic  matter  is likely  to interact  with the
toxicant, the water-effect ratio approach is likely to be more ac-
curate and is currently the recommended solution.

    A review  of the limited number of  available site-specific
studies found that the water-effect ratio (site water LC50 versus

-------
Water-Effect
Ratio
Approach
                           lab  water LC50) was generally  significantly  larger than  the
                           measured  total recoverable  versus  dissolved  ratio [10]. These
                           limited freshwater data thus suggest that use of properly formu-
                           lated dissolved criteria would be at least as protective as criteria
                           derived from careful measurements of water-effect ratios.
                           Effluents

                           The dissolved method is generally not applied to effluents to
                           determine achievement  of  effluent quality  goals. Such use is
                           generally barred  by regulation  (40  CFR 122.45). Because  the
                           chemical conditions in ambient surface waters may differ sub-
                           stantially from those in the effluent, there is no assurance that
                           effluent particulate  metal would not  dissolve after discharge.
                           A common method of removing metals from wastewaters is to
                           chemically precipitate the metal and settle the resulting particles.
                           Expressing a metals limitation in terms  of dissolved metal would
                           imply little concern about the effectiveness of the settling process
                           or the fate of the discharged particulate metal.
                              Determining  the total  recoverable  effluent limitation cor-
                           responding to a dissolved criterion would involve specifying the
                           fraction of effluent total recoverable metal that would exist as dis-
                           solved metal under the chemical conditions of the receiving water.
                           In the absence of site information, any values assumed for this
                           fraction should be environmentally conservative.
                              Where greater accuracy is  desired,  the dissolved fraction of
                           total recoverable metal could be evaluated by direct measurement
                           of dissolved and total recoverable metal in the affected ambient
                           waters, or possibly by geochemical model calculations (as dis-
                           cussed in Appendix B). All of the techniques involve approx-
                           imations.
      Due to the complexity of metals speciation, and due to the
      varying degrees of bioavailability and toxicity of the many
      forms and complexes, there is no chemical method that can
assure that a unit of concentration measured in the field would al-
ways be lexicologically  equivalent  to  a  unit  of concentration
employed in the laboratory toxicity tests underlying the criteria.
   For metals criteria derived from laboratory toxicity tests, one
approach is to use a biological method to compare bioavailability
and toxicity in receiving waters versus laboratory test waters. This
involves running toxicity tests with  at least two species, measur-
ing acute (and possibly chronic) toxicity values for the pollutant
using (a) the local receiving water, and (b) laboratory toxicity test-
ing  water,  as  the  sources  of toxicity test dilution water.  A
water-effect  ratio is  the  acute (or chronic) value in site water

-------
divided by the acute (or chronic) value in laboratory waters. An
acute value is  an LC50 or EC50 from a 48-96 hour test, as ap-
propriate for the species. A chronic value is a concentration
resulting from  hypothesis testing or regression analysis of meas-
urements of survival, growth, or reproduction in life cycle, partial
life cycle, or early life stage tests with aquatic species.
   Because the metal's toxicity in laboratory water is the basis for
EPA's criterion, this water-effect ratio is used to adjust the national
criterion (or corresponding State criterion) to  a site-specific value.
This adjustment may either increase or decrease the criterion. Be-
cause the water-effect ratio reflects differences in water chemistry,
it is acceptable  to assume that a ratio derived  from acute LCSOs or
ECSOs may be applied to both acute and chronic criteria, provided
that  the water-effect ratio is determined with an acutely sensitive
species. Nevertheless, performing chronic tests is an option that
could produce a different water-effect ratio, due to  changes in
water chemistry caused by the addition of food during chronic
tests. While  this may somewhat improve  the accuracy of the
resulting criteria, it will substantially increase  the testing costs.
   The principle of criteria adjustment using a water-effect  ratio
was  set forth in previous guidance [8, 9]. The procedure applies to
criteria derived from laboratory toxicity data. As such, it does not
apply to the residue-based mercury chronic criteria, or the field-
based selenium freshwater criterion. The basic features of the
procedure  as recommended  herein  are  described  below.  Dis-
chargers or private entities wishing to perform such testing should
consult with the appropriate regulatory agency before proceeding.
   (1)  At least two sensitive species, including at least one in-
        vertebrate, should be tested through standard toxicity
        testing protocols, using site dilution water and using
        laboratory dilution water. Test organisms should be
        drawn  from the same population and  tested under
        identical  conditions  (except for water  source).  Test
        species  should ordinarily be  selected  from those
        species that were used for national criteria develop-
        ment in order to be able to  ascertain whether  the
        laboratory water results are comparable to the those in
        the criteria document.

   (2)  Site water samples used for testing  are to be repre-
        sentative of the receiving  water to  which the  site-
        specific criteria value is to apply. For flowing waters, it
        is generally recommended that  at least one sample
        correspond to a condition in which point or nonpoint
       pollutant contributions  are reasonably well mixed
        with the flow of the receiving water. For other types of
        waters, it is generally recommended that a sample cor-
       respond  to  a  dilution  situation well  outside  any
       regulatory mixing zone. These recommendations are

-------
         intended to yield a water-effect ratio appropriate for         1
         the affected receiving water as a whole. These recom-
         mendations supersede  those  of  the  previous  site-
         specific guidance  [8, 9],  which recommended  that
         pristine waters always be used.

     (3)  The laboratory dilution water should  be comparable
         to what was  used in  tests  underlying the national
         criteria. For any pollutant with a national or State
         criterion  calculated  from   site-specific  hardness,
         laboratory-water and site-water toxicity results should
         be computationally normalized to the same hardness,
         using the specified hardness slope.

     (4)  The toxic metal should be  added in the form of an in-
         organic salt having relatively high solubility. Nitrate
         salts are generally acceptable; chloride  and  sulfate
         salts  of many  metals  are  also acceptable.  Results
         should be based on measured or nominal initial con-
         centrations  if  static tests are performed, and on
         average measured concentrations if flow-through tests
         are performed.

     (5)  Water quality characteristics affecting bioavailability
         and toxicity should be monitored. Measurements or-          *
         dinarily should  include  both dissolved  and  total          "
         recoverable  metal  concentrations,  hardness,  pH,
         alkalinity, suspended solids, conductivity, dissolved
         solids or salinity, total organic carbon, dissolved or-
         ganic carbon, temperature, and specific metal binding
         ligands (where known to be important).

     (6)  The number of site water samples to be tested  may
         vary with the size of the affected water body  (or the
         size of the metal loading). Except in the smallest sys-
         tems, a  minimum of two site water samples should be
         collected in different seasons during times of relatively
         low flow or low dilution. In moderately sized and
         larger systems  (e.g., multiple m /sec or  double to
         triple digit cfs low flow), additional samples should be
         collected during other  times in the year and possibly
         at additional  locations appropriate for the segment
         under study.

     (7)  In studies involving continuous discharges, samples
         ordinarily  should not  be  taken from storm affected
         waters, which  may contain particulate  matter not
         present during  design flow conditions. On the other
         hand, in effluent dominated situations,  at least one
         sample should represent a higher dilution condition
         (less than 50 percent effluent) in order to portray am-
         bient conditions. In all situations, it should be recog-
8

-------
        nized that the water-effect ratio may be affected by
        constituents  contributed  by  point  and  nonpoint
        sources. Consequently, new measurements should be
        undertaken if newly implemented controls or other
        changes  substantially  affect  ambient  levels  of
        suspended solids, organic carbon, or pH.

    (8)  Additional testing should be performed before accept-
        ing  unusually or  inexplicably high values  for the
        water-effect ratio, based on experience with the par-
        ticular pollutant, and based on the chemical  charac-
        teristics  of  the  water.  Retesting  should  also  be
        performed for ratios having wide uncertainty ranges.
        EPA intends to compile additional  information to as-
        sist in judging water-effect ratios in this manner. These
        recommendations,  which focus concern on large and
        uncertain water-effect ratios, supersede the previous
        guidance [8,  9]  recommendations  that  encourage
        retesting or rejection of small water-effect ratios.

    (9)  Ordinarily,  the acute and chronic criteria for  the site
        are calculated by  multiplying the national  or  State
        criteria by the geometric mean water-effect ratio  for
        the two or more tested  species. The  previous site-
        specific guidance [8, 9] provides some additional sug-
        gestions for  situations  where  the  measured ratios
        differ significantly, and provides other alternatives for
        setting the chronic criterion.

    (10) As with other types of water quality-based control ac-
        tions for toxic pollutants, it is recommended that the
        chemical-specific approach be implemented  in con-
        junction with assessments of whole  effluent  toxicity
        and field ecology ("bioassessment") [1]. Nevertheless,
        in light of the stated limitations of these latter techni-
        ques with regard to identifying causative agents and
        predicting future changes [I], considerable caution is
        warranted  in  using such  information (particularly
        ecological data) to make inferences about the ade-
        quacy of particular numeric criteria.

    The water-effect ratio is affected not only by speciation among
the various dissolved and particulate forms, but also by additive,
synergistic, and  antagonistic  effects of other materials in  the af-
fected site waters. As such, the water-effect ratio is a much more
comprehensive measure than a ratio of total recoverable metal to
dissolved metal. Because  the basic technique involves adding
soluble metal salts to site water samples, it is most accurate where
rapid sorption or complexation processes are involved.

    Because  effluent  limits  are  generally expressed  as total
recoverable metal, simplicity would suggest deriving water-effect
ratios in terms of total recoverable measurements. Derivation in

-------
Relationship
with Sediment
Criteria
Current
Technical
Support and
Future
Research
                          terms of dissolved measurements is also acceptable, and may be
                          preferred in situations involving highly variable suspended solids
                          concentrations.
                             Data available from a limited number of site-specific studies
                          performed in rather clean freshwater suggest that copper, lead,
                          and cadmium often have substantial water-effect ratios, while
                          zinc, in situations where it is preponderantly dissolved, often does
                          not [10]. Much less information is available for such ratios in salt
                          water.
     For national or state-wide criteria expressed as dissolved or
     total recoverable metal, and for site-specific criteria derived
     from water-effect ratios, questions may be raised about the
adequacy of water column criteria for protecting sediment. The
issue is whether particulate metal settling from the water column
could contribute to sediment quality problems, even where  no
toxicity is manifested in the water column.
   Because  sediment toxicity is considered to be  determined
primarily by the concentrations of pollutant dissolved in the sedi-
ment  interstitial  water,  the  question  becomes  whether  the
pollutant would have a greater propensity to become dissolved or
bioavailable in the sediment than in  the water column.  While
available information does not suggest that  this is ordinarily the
case, the ongoing development of sediment criteria should resolve
this issue. Nevertheless, in those cases where the beneficial uses of
a receiving water are known to be impaired by the toxicity of me-
tals  in  sediments,  water quality-based control requirements
should be designed to abate any sources that would continue to
cause sediment toxicity.
     The Environmental Research Laboratories in Duluth and in
     Narragansett will continue to answer technical  questions
     about the possible problems in applying the above methods
to criteria for specific metals. The contact for freshwater  is Charles
Stephan (Duluth, Minnesota telephone (218) 720-5510). The con-
tact for salt water is Gary Chapman (Newport, Oregon  telephone
(503) 867-4027).
   EPA intends to undertake further work to facilitate the im-
plementation  of  metals   criteria   in  terms   of   dissolved
measurements. For metals such as copper, silver, zinc,  lead, and
cadmium, the dependency of toxicity on factors other than hard-
ness will  be considered for inclusion. Where appropriate and
feasible, EPA may develop equations relating  dissolved metal
criteria to hardness and organic matter concentration, and possib-
ly pH and other water quality characteristics.  EPA might also
consider other biological or chemical techniques for ascertaining
the effective concentration of bioavailable metals.
                        10

-------
                 APPENDIX A
                  CASE STUDY:
     Determination of the Water-Effect Ratio
             Using Indicator Species
                     Norwalk River
                Georgetown, Connecticut*
'Adapted from the 1983 Water Quality Standards Handbook [8] for
purposes of illustrating an application.

-------
       Connecticut's Upper Norwalk watershed, where this study
       was conducted, covers an area of 18.5 square miles and in-
       cludes the region extending from the headwaters of the
Norwalk River to its confluence with Comstock Brook.
   Two secondary treatment plants discharge a total of 0.44 mgd
of municipal wastewater to a reach 9-14 stream miles upstream of
the study  site. An area of failed  septic systems in the same
upstream vicinity also contributes to the pollutant loading of the
river.
   Although water quality is degraded somewhat in the immedi-
ate vicinity of these municipal pollutant sources, as the river flows
southward towards Long Island Sound, it recovers to support a
valuable recreational trout  fishery. There are no  industrial point
source discharges of metals upstream of the study area.
   Within the study area itself, the Gilbert and Bennett Manufac-
turing Company discharges treated process water to the Norwalk
River at a point below Factory Pond in Georgetown, Connecticut.
Gilbert and Bennett cleans, draws, and coats metal wire. Waste-
water is primarily generated during the wire cleaning process. The
wastewater  treatment system of the facility  consists  of  pH
neutralization and equalization followed by precipitation and
clarification of  the effluent before discharge to the  river. The
treated wastewater is discharged intermittently to the river.
   The Connecticut Department of Environmental Protection un-
dertook the study  of the Norwalk River site because the Gilbert
and Bennett metal loadings were calculated to result in excursions
of national water quality criteria for lead and zinc under design
flow conditions. In order to evaluate the effect of site water on the
toxicity of lead and zinc, EPA and the State used the indicator
species (water-effect ratio) protocol.
   By testing a sensitive invertebrate and a fish in both site and
reconstituted laboratory dilution water, the water-effect ratio pro-
cedure accounts for  differences in bioavailability and effective
toxicity of a pollutant in the two waters. The procedure responds
to the summation of all synergistic and antagonistic effects of site
water quality characteristics (including pH, hardness, particulate
matter, dissolved organic  matter, and other contaminants). The
procedure does not,  however, elucidate factors causing the  dif-
ference in toxicity.
   A water-effect ratio is the ratio of a species LC50 in site water
versus its LC50 in laboratory dilution water.  The 1983  Water
Quality Standards Handbook recommends that two relatively sen-
sitive indicator species be tested, and that the geometric mean of
the two results be used. The site-specific criterion would be calcu-
lated as the product of the  national criterion and the water-effect
ratio.
Introduction
                                                            13

-------
Analyses
Conducted
     The results from the testing of metals toxicity in laboratory
     and site water forms the primary basis for the site-specific
     criteria. To provide additional information, the monitoring
of ambient water chemistry, surveying of macroinvertebrates, and
testing of whole effluent toxicity were also performed.
                          Chemical and Ecological Monitoring

                          Concentrations of several metals were  measured in composite
                          samples taken at each of four ambient  stations  and in grab
                          samples of the Gilbert and Bennett main  effluent. One ambient
                          station was in the upstream control zone, two in the impact zone,
                          and one in the recovery zone.
                             Benthic populations were sampled at five locations to assess
                          the impact of the discharge on the stream community. One refer-
                          ence station was  located in  the upstream control  zone.  Three
                          stations were in the Bennett and Gilbert impact zone,  and one was
                          in the recovery zone. Physical substrate, stream velocity, and
                          water depth were similar at each location. Four Surber samples
                          were collected at each of the five locations. Organisms were sorted
                          in the field, preserved in 70 percent ethanol, and returned  to the
                          laboratory for identification and enumeration.
                          Toxicity Testing

                          Norwalk River water was withdrawn from a station upstream of
                          Gilbert and Bennett and transported  (along  with the  effluent
                          samples) back to the laboratory. Toxicity tests were conducted in
                          the sampled river water and in reconstituted water, with differing
                          amounts of either lead or zinc  added, in order to determine the
                          LC50. Whole toxicity  testing of one of the Gilbert and Bennett
                          effluents was also performed, using both upstream Norwalk River
                          water and reconstituted water for dilution.
                              Because  the  lead  and  zinc toxicity  tests were run  using
                          upstream water, they do not indicate the effects of synergism, an-
                          tagonism, or toxicant  additivity with constituents of the Gilbert
                          and Bennett effluent. Although this guidance recommends use of
                          downstream water for at least one sample, the case study predates
                          the guidance and does not follow this recommendation.
                              Ninety-six hour acute toxicity tests (static with measured con-
                          centrations of toxicant)  were conducted  with laboratory reared
                          rainbow trout (Oncorhynchus mykiss, formerly Salmo gairdneri) and
                          48-hour acute toxicity tests (static with measured concentrations)
                          were conducted with laboratory-reared Daphnia tnagna.
                        14

-------
Water Chemistry and Ecological Quality

Mean instream concentrations of lead, zinc, and cadmium were
higher in the impact and recovery zones than in the control zone.
Levels of cadmium and copper appeared to exceed  national acute
criteria at all sampling locations both upstream and downstream
of Gilbert and Bennett.
   It should be noted that the metals data were generated using
the sample handling and analytical protocols of the early 1980s,
rather than more recent protocols emphasizing ultra-clean tech-
niques. While  the link between chemical quality and ecological
quality is of great interest, it is not clear that these ambient metals
data are sufficiently reliable to be used in such comparisons. If the
ambient metals concentrations were reliably known, such infor-
mation would be most useful for comparing concentrations in the
control, impact, and recovery zones with the criteria derived from
the water-effect ratio.
   At the upstream reference location, 889 organisms from 44 taxa
were collected. Most of the species collected could be classified as
sensitive or facultative with respect to pollution tolerance. Species
diversity was high (Shannon Diversity index of 3.4), indicating ac-
ceptable water quality and aquatic habitat.
   At the three near downstream locations, the number of or-
ganisms, number of taxa, and diversity were reduced significantly.
At some of the impact zone stations, the number of organisms was
less than one-fourth, and number of taxa one-third  that of the ref-
erence site. The Shannon index registered as low as 1.0.
   At the recovery zone station (500 m downstream from the dis-
charge), a larger number of organisms were found  than at any of
the other stations, including the upstream reference site. Diversity
and  numbers of taxa, however, remained at levels more charac-
teristic of the impacted stations.
   The ecological assessment demonstrated that the ecology of
the Norwalk River was impaired, and  strongly suggested  that
some type of pollutant release from Gilbert and Bennett was in-
volved in the  impairment. However, as noted in  the  Technical
Support Document [1], ecological assessments cannot identify the
causative agents, and  generally do  not predict  the ecological
quality as a function of chemical-specific concentrations. Conse-
quently, these ecological data  do not indicate the appropriateness
of particular values for the water-effect ratio.
Findings
Toxicity Testing with Lead and Zinc

Table A-l shows the results of static toxicity tests with daphnids
and rainbow trout exposed to lead and zinc, in river water and in
laboratory water.
                                                           15

-------
 Table A-1.—Toxicity of lead and zinc in Norwalk River and in laboratory water.
DAPHNIAMAGNA

Lead
River water
Lab water
Zinc
River water
Lab water
48-hr LC50, HQ/L
(95% confld lim)

1300(950-1900)
320 (290-360)

900(740-1100)
400 (380-480)
Water-Effect
Ratio

4.1

2.3
RAINBOW TROUT
96-hr LC50, ng/L
(95% confld Urn)

9600(7500-12000)
2600(1900-3600)

1500(1200-1800)
1000(850-1200)
Water-Effect
Ratio

3.7

1.5
     The toxicity of both lead and zinc was lower in Norwalk River
 water than in laboratory water. For both metals, the more sensitive
 species, Daphnia magna, had the higher water-effect ratio. This is
 consistent with general tendencies observed in other studies [10].
  Whole Effluent Toxicity for One Effluent

  In unspiked,  static toxicity tests in which rainbow  trout were
  exposed to one of the Gilbert and Bennett effluents, the effluent
  was rendered nontoxic by relatively little dilution of the effluent.
  As effluent constituted the bulk (60-68 percent) of the test water at
  dilutions toxic to half the organisms, it was not surprising that the
  whole effluent toxicity tests could not discern differences between
  toxicity in laboratory and river water. That is, toxicity was much
  reduced before enough laboratory or river water could added to
  the effluent to discern differences between the added water.
     The monitored effluent was not sufficiently toxic to daphnids
  to allow calculation of the dilution lethal to half the organisms.
  However, to eliminate toxicity to all the  tested individuals, sig-
  nificantly more dilution was required with laboratory water than
  with river water. This suggests that this effluent may be less toxic
  in Norwalk River water than in laboratory water.
     If the lead and zinc concentrations had been measured during
  the whole effluent toxicity tests, it would be possible  to compare
  effect concentrations with the lead and zinc LC50 values shown in
  the  previous  section. In making such comparisons, however,  it
  must be recognized that the cause of toxicity of the effluent is not
  known.
     Finally, it should be noted that the tested effluent is only one of
  Gilbert and Bennett's releases to the Norwalk River.  There is no
  disparity between the observed significant instream impacts and
  the  relatively  low toxicity of the one monitored effluent. The
  ecological assessment suggested that the unmonitored release was
  more toxic than the monitored one.
16

-------
Calculation of the Site-Specific Criteria

The water-effect ratios for lead and zinc differed relatively little
between species. If the overall water-effect ratio for each metal
were  calculated  from  the  geometric  mean  of  the  species
water-effect ratios, then the water-effect ratio for lead would be
3.9, and that for zinc would be 1.9.

   It is assumed that the water-effect ratio would apply to both
the acute and the chronic criteria. As the national criteria for lead
and zinc are hardness dependent, for purposes of determining the
value of the site-specific criteria during the survey period, it is ap-
propriate to calculate the national criteria at the hardness of the
laboratory reconstituted water, if different from the site water. The
site-specific acute and chronic criteria for each metal would equal
the national (or state-wide) criteria multiplied by the water-effect
ratio for each metal.
   Because all of the above toxicity tests and water-effect ratios
were  based on total recoverable metal, the resulting site-specific
criteria would also be expressed as total recoverable metal.
   When the site-specific criteria were used to calculate effluent
limits, it was found that large reductions in current metals load-
ings would be required. This result is not surprising, considering
the ecological effects observed downstream.
                                                              17

-------
     APPENDIX B
Derivation of Effluent Limits
from Ambient Metals Criteria

-------
      The determination of the waste loads and effluent limits that
      allow attainment of water quality criteria is described in
      other EPA guidance  [1, 11, 12]. The Waste Load Allocation
Guidance  Manuals [11, 12] are  particularly suited to predicting
far-field dissolved and total recoverable concentrations, such as
would be  necessary for evaluating watersheds with multiple dis-
chargers.  The  Technical  Support  Document  (TSD)  [I]  has
additional guidance on the evaluation of mixing zones and the
derivation of permit limits. Nevertheless, some additional discus-
sion is provided  below.  This guidance supersedes the second
paragraph of Section 5.7.3 "Metals" of the TSD [1].
      Expressing  state-wide or  site-specific  criteria  as  total
      recoverable metal has the advantage of providing a simple
      basis for calculating effluent limits. All of the effluent total
recoverable  metal  would  contribute  to the  ambient  total
recoverable concentration.
   f criteria are expressed as dissolved metal, then it is necessary to
   establish what fraction of the effluent total recoverable metal
   contributes to the ambient dissolved metal.
   Three alternatives may be considered for relating the ambient
dissolved  criterion for a specific metal  to  the  effluent  total
recoverable  limits: (a)  directly measure  dissolved  and  total
recoverable metal in the receiving water, (b)  assign environmental-
ly conservative  default values  for the assumed ratio  between
dissolved and total recoverable metal, and (c) predict the percent-
age dissolved metal from a geochemical model such as MINTEQ.

   Regardless of which alternative is used, it must be recognized
that the goal is to set the effluent limit at a value such that the am-
bient  water quality standard  will  be attained.  In  addition,
compliance with regulatory requirements  for  technology-based
limits, antidegradation, and antibacksliding is necessary.
Using Site-Specific Measurements

The  concept is  to measure the dissolved-total  ratio for the
particular metal in the receiving water affected by the discharger.
Because the chemical properties of an effluent (particularly an
industrial effluent) may  be much different than the chemical
properties  of the  receiving water below the discharger, it  is
appropriate to measure the ratio in the receiving water rather than
in the effluent. As an approximation, it may be assumed that the
measured dissolved-total metal ratio in the affected waters reflects
the  fraction of effluent total recoverable metal that remains or
becomes dissolved in ambient water.
Total
Recoverable
Metal Criteria
Dissolved
Metal Criteria
                                                           21

-------
     Samples on which measurements are made should be repre-
  sentative of the bulk of the receiving water. It is recommended that
  sampling be performed over a period of time, with samples repre-
  senting the usual range of effluent and ambient quality, while
  emphasizing the season corresponding to the critical water quality
  conditions. Because the  control strategy assumes that the dis-
  solved  concentration  is   related  to   the   total   recoverable
  concentration, it would be appropriate to verify that the dissolved
  and total recoverable concentrations are in fact correlated.
     In freshwaters,  an alternative approach to downstream  sam-
  pling is to sample the effluent and the upstream waters ana mix
  samples at an appropriate dilution. The dilution and the seasons
  for sampling should be related to the critical conditions, although
  it may be appropriate to reduce the dilution if necessary to detect
  and quantify the metal.
     The most important constraint on the feasibility of carrying
  out site-specific  measurements is the  capability of  analytical
  laboratories to detect and accurately quantify both the dissolved
  and total recoverable metal. Graphite furnace (flameless) atomic
  absorption AA techniques are usually needed. Furthermore, great
  care is needed to prevent external contamination of samples. The
  EPA- and USGS-recommended sample  handling methods,  com-
  monly used,  may produce inaccurate results when judged against
  newer techniques that emphasize highly purified reagents, Teflon
  and polyethylene labware, and clean laboratory environments [7,
  13].
     The high degree of imprecision of metals measurements tends
  to result in overstatement of the true variability of the dissolved-
  total ratio. As a result, unless a mean or median observed ratio is
  used, it may be necessary to compensate for the effect of measure-
  ment  imprecision   (by   subtracting   out  the  measurement
  imprecision variance).
     In order  to provide some sense of the general magnitude  of
  typically observed dissolved-total metals ratios, data from several
  sources have been compiled in Table B-l. The ambient data under-
  lying the tabulated values are considered to be reasonably reliable.


  Using Environmentally Conservative Default Values

  This option is best applied as the first tier of a  tiered approach,
  where the second tier involves site-specific measurements. In this
  type of framework, the default (first-tier) percentage dissolved
  might be set  at a reasonable worst-case value.
     One possible worst-case assumption is that 100 percent of the
  effluent total recoverable metal will become dissolved in the
  receiving water. Such an assumption  may be particularly ap-
  propriate  for a  metal, such as mercury, for which there are
  substantial uncertainties regarding long-term processes convert-
22

-------
 Table B-1 .—Observed  average  fractions  of dissolved  metals in ambient
 waters.


METAL
Aluminum
Cadmium
Copper
Lead
Nickel
Silver
Zinc

FRESH WATER ~
EAST COAST [7]

0.40
0.62
0.10


0.20
SALT WATER
NY-NJ HARBOR AREA [13]

NEAR SURFACE

0.81b
0.50C
0.08C
0.73°
0.11°
0.60d

NEAR BOTTOM

0.56°
0.23d
0.03d
0.41d
0.08°
0.29d
FRESH
AND SALT
WATER
STORET"
0.07"

0.4(

0.6


'For STORET data, means were estimated from cumulative distributions of concentrations, 1984-
1990, in ambient streams, rivers, canals, lakes, reservoirs, and estuaries, for samples in which
both dissolved and total recoverable metal were analyzed. The fraction dissolved for STORET
data for cadmium, lead, silver, and zinc are not tabulated  because most of these data are
believed to be seriously compromised by external contamination of samples.
bDissolved and total recoverable concentrations well correlated.
cDissolved and total recoverable concentrations somewhat correlated.
"Dissolved and total recoverable concentrations not correlated.
"For any particular measurement of aluminum, the method of filtration may strongly affect the
result.
8For much of the STORET copper data, external contamination of samples is likely to somewhat
affect the absolute values of the measured concentrations, and may somewhat affect the dis-
solved-total ratio.
ing inorganic (including particulate) mercury into bioaccumula-
tive methyl mercury. Where the background metal concentration
is either negligible or entirely dissolved, and where the dissolved
criterion is less than the corresponding total recoverable criterion,
the assumption that all effluent total recoverable metal will be-
come   dissolved  yields  more  restrictive  limits  than  simply
implementing a total recoverable criterion.
    Other  environmentally conservative default values may also
be developed based on available  information. Such values  may
differ  in different parts of the country, due to variations in water
quality characteristics.
Using a Geochemical Model

The  equilibrium  metal  speciation  model MINTEQ  may  offer
assistance in understanding or predicting the fraction dissolved
[14].  Using  MINTEQ without obtaining site-specific  input data
may  not be feasible, however.  As the  effort in  obtaining the
appropriate input data  would  likely  be equivalent to  simply
measuring the site-specific fraction dissolved, the model may be
more useful for providing insight into the controlling factors and
predicting  the  effects  of  different environmental conditions.
MINTEQ may be particularly useful for predicting whether the
dissolved  concentration  will  respond to reductions in  the total
concentration or  whether  it  is  controlled  by factors such as
solubility.
                                                                  23

-------