&EPA
          United States
          Environmental Protection
          Agency
Office Of Water (WH-553)
Washington, IDC 20460
EPA-822-R-93-011
September 1993
          Technical Basis for Deriving
          Sediment Quality Criteria for
          Nonionic Organic Contaminants
          for the Protection of Benthic
          Organisms by Using Equilibrium
          Partitioning
                                    Printed on Recycled Paper

-------

-------
                  Sediment Quality Criteria Using Equilibrium Partitioning
This document presents the technical basis EPA has used in establishing the proposed method-
ology for deriving sediment quality criteria for the protection of benthic organisms from non-
ionic organic chemicals. It was issued in support of EPA regulations and policy initiatives
involving the application of biological and chemical assessment techniques to control toxic pol-
lution to surface waters and sediments. This document does not establish or affect legal rights
or obligations. It does not establish a binding norm and is not finally determinative of the issues
addressed. Agency decisions in any particular case will be made applying the law and regula-
tions on the basis of specific facts when permits are issued or regulations promulgated. This
document is expected to be revised periodically to reflect advances in this rapidly evolving area.

      This report has been reviewed by the Health and Ecological Criteria Division, Office of
Science  and  Technology, U.S. Environmental Protection Agency, as  well as other pertinent and
interested offices in the Agency, and approved for publication. Mention of trade names or com-
mercial products does not constitute endorsement or recommendations for use.

-------

-------
                Sediment JQuality Criteria Using Equilibrium Partitioning
CONTENTS
ABSTRACT
OVERVIEW   ...........  ...... • .  .  .....   3
   Toxidty and Bioavailabfility of Chemicals in Sediments ............. 3
   Partitioning of Nonionic Organic Chemicals  .  . .......  .  ..... .  . 5
   Effects Concentration   .......................  .... 7

BACKGROUND  ...................   ....   7
   Rationale for Selecting the EqP Method  ............  ....... 8
   Relationship to WQC Methodology ...... ...........  ..... 8
   Applications of SQC .....................  ....... 8

TOXICITY AND BlOAVAILABILITY OF CHEMICALS IN SEDIMENTS   .......   9
   Toxicity Experiments ... .....  .......... ....... ... 9
   Bioaccumulation ..... ' ........................  12
   Conclusion ...... . ........................  12

SORPTION OF NONIONIC ORGANIC CHEMICALS ...  .....  .....  14
   Partitioning in Particle Suspensions ....................  14
      Particle concentration effect  ...............  ;  ......  14
      Organic carbon fraction ........................  16
   Dissolved Organic Carbon (DOC) Complexing  ...............  17
   Phase Distribution in Sediments .......  ...............  17
   Unavailability of DOC Complexed Chemicals   ...............  18
   Field Observations of Partitioning in Sediments  ...............  19
      Organic carbon normalization  .....................  19
      Sediment /pore water partitioning  ...................  23
      Laboratory toxicity tests ............... .  ........  24
   Organic Carbon Normalization of Biological Responses ...........  26
      Toxicity and bioaccumulation experiments   . ...........  .  .......  26
      Bioaccumulation and organic carbon normalization  .....  .......  28
   Determination of the Route of Exposure  .....  . ............  31

APPLICABILITY OF WQC AS THE EFFECTS LEVELS FOR BENTHIC ORGANISMS ...  31
   Method-Relative Acute Sensitivity ....................  .  32
   Comparison of the Sensitivity of Benthic and Water Column Species   .....  32
      Most sensitive species ......................  ...  32
      All species ..... ......  .  ...............  ...  32

-------
                Sediment Quality Criteria Using Equilibrium P
    BentWc Community Colonization Experiments	36
    Water Quality Criteria (WQC) Concentration  Versus
      Colonization Experiments	      ..... t	36
    Conclusions.  .  .  .  ,	37

GENERATION OF SQC	  .  37
    Parameter Values	'.	37
    Measurement of JCow		38
       Literature Kow	40
       Estimated K0w	40
       KOW selection	42
       Koc determination	42
    Species Sensitivity	43
    Quantification of Uncertainty Associated with SQC	44
    Minimum Requirements to Compute SQC	46
       Laboratory octanol-water partition coefficient	  46
       Final chronic value	47
       Sediment toxicity test	47
       Analytical procedures	48
       Conclusion	48
    Example Calculations	48
    Field Data	49
       STORE! data	50
       National Status and Trends Program data	50
       Corps of Engineers data	51

CONCLUSIONS	55
    Research Needs	  57
REFERENCES
57

-------
        Sediment Quality Criteria Using Equilibrium Partitioning
                       Abstract
       The purpose of this report is to present the technical basis for es-
 tablishing sediment quality criteria for nonionic organic chemicals, us-
 ing equilibrium partitioning (EqP). Equilibrium partitioning is chosen
 because it addresses the two principal technical issues that must be re-
 solved: the varying bioavailability of chemicals in sediments and the
 choice of the appropriate biological effects concentration.
       The data that are used to examine the question of varying
 bioavailability across sediments are from toxicity and bioaccumulation
 experiments using the same chemical and test organism but different
 sediments. It has been found that if the different sediments in each ex-
 periment  are compared, there is essentially no relationship between
 sediment  chemical concentrations on a dry weight basis and biological
.effects. However, if the chemical concentrations in the pore water of the
 sediment  are used (for chemicals that are not highly hydrophobic) or if
 the sediment chemical concentrations on an organic carbon basis are
 used, then the biological effects occur at similar concentrations (typi-
 cally within a factor of two) for the different sediments. Most impor-
 tantly, the effects  concentrations are the  same as, or they can be
 predicted  from, the effects concentration determined in  water-only
 exposures.
       The EqP methodology rationalizes these results by assuming
 that the partitioning of the chemical between sediment-organic carbon
and pore water is at equilibrium. In each of these phases, the fugacity
or activity of the chemical is the same at equilibrium. As a consequence,
it is assumed that the organism receives an equivalent exposure from a
water only-exposure or from any equilibrated phase: either from pore
water via respiration; or from sediment carbon, via ingestion; or from a
mixture of the routes. Thus, the pathway of exposure is not significant.
The biological effect is produced by the chemical activity of the single
phase or the equilibrated system.
       Sediment quality criteria (SQC) for nonionic organic chemicals
are based  on the chemical concentration in sediment organic carbon.
For highly hydrophobic chemicals this is necessary because  the pore
water concentration is, for those chemicals, no longer a good estimate
of the chemical activity. The pore water concentration is the sum of the
free chemical concentration, which is bioavailable and represents the
chemical activity, and the concentration of chemical complexed to dis-
solved organic carbon, which is not bioavailable. Using the  chemical
concentration in sediment organic carbon eliminates this ambiguity.

-------
       Sediment Quality Criteria Using Equilibrium pltttitioning
       SQC also require that a chemical concentration be chosen that is
sufficiently protective of benthic organisms. The final chronic value
(FCV)  from the U.S. Environmental Protection Agency (EPA) water
quality criteria  is proposed._An analysis of the data compiled in the
water quality criteria documents demonstrates that benthic species, de-
fined as either.epibenthic or infaunal species, have a similar sensitivity
to water column species. This similarity is the case if the most sensitive
species are compared and if all species are compared. The results of
benthic colonization experiments also support the use of the FCV. Thus,
if effects concentrations in sediments can be accurately predicted using
the KOC and data from water-only tests, the SQC protecting benthic spe-
cies can be predicted using the KOC and FCV/
       Equilibrium partitioning cannot remove all .the variation in the
experimentally  observed sediment-effects concentration and the  con-
centration predicted from water-only exposures. A variation factor of
approximately four to five remains. Thus, a quantification of this uncer-
tainty should accompany the SQC.
       The derivation of SQC requires that a minimum database be
available. This includes: (1) the octanol/water partition coefficient of
the chemical, which should be measured with modern experimental
techniques, which appear to remove the large variation in reported val-
ues, (2) the derivation of the final chronic value, which should also be
updated to include the most recent toxicological information, and (3) an
SQC check test to establish variation of the EqP prediction. The SQC is
then the FCV x Koc with confidence limits based on SQC check tests.

-------
                         Sediment Quality Criteria Using Equilibrium Partitioning
 OVERVIEW

 This report presents the technical basis  for estab-
 lishing sediment quality criteria (S.QC) for nonionic
 .organic chemicals using the equilibrium partitioning
 (EqP) method. The term sediment quality  criteria, as
 used herein, refers to numerical concentrations for in-
 dividual chemicals that are  applicable  across  the
 range of sediments encountered in practice. Sediment
 quality criteria are intended to be predictive of bio-
 logical effects. As a consequence, they can be used in
 much the same way as final chronic values (FCV) are
 used in water quality criteria—as the concentration of
 a chemical that is protective ofbenthic aquatic life.
       The specific regulatory uses of SQC have not
 been established. However, the range of potential ap-
 plications is quite large because  the  need for the
 evaluation  of potentially  contaminated sediments
 arises in many contexts. SQC are meant to be used
 with direct toxicity testing of sediments as a method
 of evaluation.  They provide a chemical by chemical
 specification of  what sediment concentrations are
 protective of benthic aquatic life.
       This1 overview (Section 1) summarizes the evi-
 dence and  the major lines of reasoning of the  EqP
 methodology, with supporting references cited in the
 body of the report. Section 2 reviews the background
 that led to the  need for SQC and also the selection of
 the EqP methodology. Section 3 reviews the develop-
 ment of concentration-response curves  for pore-
 water concentrations and sediment  oirganic-carbon
 normalized concentrations to determine toxicity and
 bioavailability in contaminated sediments. It  also
 presents analyses of sediment toxicity and bioaccu-
 mulation experiments. Section 4 reviews  the parti-
 tioning of nonionic organic chemicals to sediments
 using laboratory and field studies. Section  5 reviews
 a comparison  of benthic and 'water column species
 using aquatic toxicity data contained in EPA's Water
 Quality Criteria (WQC) documents to show the ap-
 plicability of WQC as the effects levels for benthic or-
 ganisms. Section 6 reviews the computation of an
 SQC and presents an analysis for quantifying the un-
 certainty associated with SQC. This section also pre-
 sents minimum  data requirements  eind  example
calculations and compares  the  SQC computed for
 five chemicals  to field data.-Section 7 presents con-
clusions and further research  needs. Section 8 lists
the references used in this document.

Toxicity  and Bioavailabiiity of
 Chemicals  in Sediments

Establishing SQC requires a  determination of the ex-
tent of the bioavailability  of  sediment associated
chemicals. It has frequently been observed that simi-
lar concentrations of a chemical, in units of mass of
chemical per mass of sediment dry weight (e.g., mi-
 crograms chemical  per gram sediment [ug/g]) can
 exhibit a range in toxicity in different sediments. If the
 purpose of SQC is  to establish chemical concentra-
 tions that apply to sediments of differing types, it is
 essential that the reasons for this varying bioavailabil-
 ity be understood and explicitly included in the crite-
 ria. Otherwise the criteria cannot be presumed to be
 applicable across sediments of differing properties.
       The importance of this issue cannot be over-
 emphasized. For example, if 1 ug/g of Kepone is the
 LCso for an organism in one sediment and 35 ug/g is
 the LCso in another sediment, then unless the cause
 of this difference can be associated with some explicit
 sediment properties it is not possible to decide what
 would be the LCso of a third sediment without per-
 forming a toxicity test. The results of toxicity tests
 used to establish the toxicity of chemicals in sedi-
 ments would  not be generalizable to other sedi-
 ments. Imagine the situation if the results of toxicity
 tests  in water depended strongly  on the particular
 water source, for example, Lake Superior versus well
 water. Until the source of the differences was under-
 stood, it would  be fruitless to attempt to establish
 WQC. For this reason, bioavailability is a principal
 focus of this report.
       The key insight into the problem of quantify-
 ing the bioavailability of chemicals in sediments was
 that the concentration-response curve for the biologi-
 cal effect of concern can be correlated not to the total
 sediment-chemical concentration (micrograms chemi-
 cal per gram sediment), but to the interstitial water or
 pore water concentration (micrograms chemical per
 liter pore  water). In addition, the effects concentra-
 tion found for the pore water is essentially equal to
 that found in water-only exposures. Organism mor-
 tality, growth  rate,  and bioaccumulation  data are
 used to demonstrate this correlation, which is a criti-
 cal part of the  logic behind the EqP approach to de-
 veloping SQC. For nonionic organic chemicals, the
 concentration-response curves correlate equally well
 with the sediment-chemical concentration on a sedi-
 ment-organic carbon basis.
      These observations can be rationalized by as-
 suming that the pore water and sediment carbon are
 in equilibrium and that the concentrations are related
by a partition coefficient, Koc, as shown in Figure 1.
The name equilibrium partitioning (EqP) describes this
 assumption. The..rationalization for the equality of
water-only and sediment-exposure-effects concentra-
 tions on a pore water basis is that the sediment-pore
water equilibrium system (Fig. 1, right) provides the
same exposure as a water-only exposure (Fig. 1, left).
The chemical activity is the same in each system at
equilibrium. It should be pointed out that the EqP as-
sumptions are only approximately true; therefore,
predictions from the model have an inherent uncer-
tainty. The data presented below illustrate the degree
to which EqP can rationalize the observations.

-------
                          Sediment Quality Criteria Using Equilibrium Partitioning
              Water  Only
                Exposure
   Sediment  -  Pore Water
             Exposure
                   Biota
                                                Equilibrium  Partitioning
 Figure 1.—Diagram of the organism exposure routes for a water-only exposure (left and a sediment exposure (right). Equilib-
 rium partitioning refers to the assumption that an equilibrium exists between the chemical sorted to the particulate sedi-
 ment organic carbon and the pore water. K^ is the organic carbon partition coefficient.
       Figure 2 presents mortality data for various
 chemicals and sediments compared  to pore water
 concentrations when normalized on a toxic unit ba-
 sis. Pore water toxic units are the ratio  of the meas-
 ured pore water  concentration to the LCso from
 water-only toxicity tests. Three different sediments
 are tested for each chemical as indicated. The EqP
 model predicts that the pore water LCso will equal
 the water-only LCso which is obtained from a sepa-
 rate water-only exposure toxicity test. Define:
 pore water toxic unit
           _ (pore water concentration)
                 (water-only LCso)
(1)
       Therefore, a toxic unit of one occurs when the
pore water concentration equals the water-only LCso,
at which point it would be predicted that 50 percent
mortality would be observed. The correlation of ob-
served mortality to predicted pore water toxic units
in Figure 2 demonstrates (a) the efficacy of using
pore water  concentrations to remove sediment-to-
sediment differences and (b) the applicability of the
water-only effects concentration and, by implication,
the validity  of the EqP model. By contrast, the mor-
.tality versus sediment chemical concentration on a
dry weight basis varies dramatically from sediment
to sediment. This will be presented subsequently.
       The equality of the effects concentration on a
 pore water basis suggests that the route of exposure is
 via pore water. However, the equality of the effects con-
 centration on a sediment-organic carbon basis, which is
 demonstrated below, suggests that the ingestion of
 sediment-organic carbon is the primary route of expo-
 sure. It is important to realize that if the sediment and
 pore water are in equilibrium, then the effective expo-
 sure concentration is the same regardless of exposure
 route. Therefore, it is not possible to determine the pri-
 mary route of exposure from equilibrated experiments.
       Whatever the route of exposure, the correla-
 tion to pore water suggests that if it were possible to
 either measure  the pore water chemical concentra-
 tion, or predict it from the total sediment concentra-
 tion and the relevant sediment properties such as the
 sediment organic carbon concentration, then that
 concentration could be used to quantify the exposure
 concentration for an organism. Thus, the partitioning
 of chemicals between the solid and the  liquid phase
 in a sediment becomes a necessary component for es-
 tablishing SQC.
      In addition, if it were true that benthic organ-
 isms are as sensitive as water column  organisms—
 and the evidence to be presented appears to support
 this supposition—then SQC could be established us-
 ing the FCV from WQC documents as the effects con-
centration for  benthic organisms.  The apparent
equality between the effects concentration as meas-

-------
                           Sediment Quality Criteria Using Equilibrium Partitioning
                                Poire Wator Normalization
                                                                                                       .-AW
    100


•S   80

 >•   6O

5   40
 O

     20

      0

                                                                                • Kapono
                                                                                • PlMnanthrmm
                                                                                * Endrln
                                                                                • FkraranthaiM
                                                                                v Ae«naphth«n«
                                                                                ^ DOT
               0.01
                     0.10
                                            1.OO
                                                  10.00
                                                                        100.00
                                 Porn Water Toxic Units
 Figure 2.—Mortality versus predicted pore water toxic units for five chemicals and three sediments per chemical

 Sf ^fiSr^h?^g'e,hatChing {'°WeSt °rganiC Carb°n C0ntent)' «•**•«*« (intermedia^ organic       co
 SS'i?t2i-^"JS8 (h'gheSt>°rgani° Carb0n C0ntent)- See Tables 1 and 2 for data s°"«*s- Predicted pore water toxfc
 units are the ratio of the pore water concentration to the water-only LC50 (Eqn. 1).
 ured in pore water and in water-only exposures (Fig.
 2) supports using an effects concentration derived
 from water-only exposures.
       The  calculation  procedure for establishing
 SQC is as follows. If FCV (ug/L) is the final chronic
 WQC  for the chemical of interest, then  the SQC
 (u,g/g sediment) are computed using the partition
 coefficient Kp (L/kg sediment) defined as the ratio of
 chemical concentration in the sediment and in the
 pore water at equilibrium.
                                            cent by weight, the organic carbon appears to be the
                                            predominant phase for chemical sorption. The parti-
                                            tion coefficient, Kp, the ratio of sediment concentra-
                                            tion, Cs, to pore water concentration, Cd, is given by

                                                                                        (3)
              SQC-KpFCVxO.001
                                             (2)
                                g
This is the fundamental equation from which SQC
are generated. Its utility depends on the existence of
a methodology for quantifying partition coefficients.


Partitioning of Nonionic Organic
Chemicals

The partitioning of nonionic organic chemicals to soil
and sediment particles is reasonably well  under-
stood, and a standard model exists for describing the
process. The hydrophobicity of the chemical is quan-
tified by using the octanol/water  partition coeffi-
cient, Kow. The sorption capacity of the sediment is
determined by the mass fraction of organic  carbon
for the sediment,/oc. For sediments with/oc a 02 per-
                                            where Koc is the partition coefficient for sediment or-
                                            ganic carbon.
                                                  The  only other environmental variable that
                                            has a dramatic effect on partitioning appears to be
                                            the particle concentration in the suspension in which
                                            Kp is measured. There is considerable controversy re-
                                            garding the mechanism responsible for the particle
                                            concentration  effect, and a number of explanations
                                            have been offered. However, all the interpretations
                                            yield the same result for sediment/pore water parti-
                                            tioning, namely that Koc » Kow for sediments.
                                                  Using Equations 2 and 3, a SQC is calculated
                                            from
                                                           SQC./ocKocFCV.
(4)
                                           This equation is linear in the organic carbon fraction,
                                           foe- As a consequence, the relationship can be ex-
                                           pressed as
                                                            SQC
                                                                     /oc
                                                                 -KocFCV.
(5)

-------
                          Sediment Quality Criteria Using Equilibrium Partitioning
 If we define
                   SQCoc-
SQG
 foe
                                              (6)
'as the organic carbon-normalized SQC concentration
 (mioogram chemical per gram organic carbon), then
                 SQCoc-KocFCV.
                   (7)
 Thus, we arrive at the following important conclu-
 sion: For a specific chemical having a specific Koc, the
 organic  carbon-normalized sediment concentration,
 SQCoo is independent of sediment properties.
       Hydrophobic chemicals also tend to partition
 to colloidal-sized  organic  carbon  particles that are
 commonly referred to as dissolved organic carbon, or
 DOC. Although DOC affects the apparent pore water
 concentrations of highly hydrophobic chemicals, the
 DOC-bound fraction of the chemical appears not to be
 bioavailable and Equation 7 for SQCoc still applies.
                                                     Therefore, we expect that toxicity in sediment
                                               can be predicted from the water-only effects concen-
                                               tration and the KOC of the chemical. The utility of
                                               these ideas can be tested with the same mortality
                                               data as those in Figure 2 but restricted to nonionic or-
                                               ganic chemicals for which organic carbon normaliza-
                                               tion applies. The concept of sediment toxic units is!
                                               useful in this regard. These units are computed as the
                                               ratio of the organic carbon-normalized sediment con-
                                               centrations, Cs//oo and the predicted sediment LCso.
                                               using Koc and the water-only LCso. That is,
                                                       /predicted\          • ^
                                                       sediment	^	.
                                                       toxic unit   Koc (water-only LCso)
                                                                        (8)
                                 Figure 3 presents the percent mortality versus
                           predicted sediment toxic units.  The correlation is
                           similar to that obtained using the pore water concen-
                           trations  in Figure 2. The cadmium data are not in-
                           cluded because its partitioning is not determined by
                            Organic Carbon Normalization
          100

      ~    80
       X    60
      "5

      I
n
•:    40
            20

             0
                                            &>
                                            •   v
                                    «
                                         o
                                             1
                                                  ^ A A Dicldrin
                                                  O © • Kapon*
                                                  a D • Ph*nanthr*n*
                                                  .  -> * Endrln
                                                  O O * Ruoranfh*n*
                                                  T v T Ae*naphth*n*
                                                  [' > * DDT
             0.01
  O.10           1.OO          10.OO

 Predicted Sediment  Toxic Units
                                                                         100.00
Figure 3.—Mortality versus predicted sediment toxic units. Predicted sediment toxic units are the ratio of the organic carbon-
normalized sediment chemical concentration to the predicted sediment LCso (Eqn. 8). Sediment types are indicated by the
single hatching (lowest organic carbon content), cross-hatching (intermediate organic carbon content), and filled symbols
(highest organic carbon content). See Tables 1 and 2 for data sources. Koc values are computed from Kow for DDT (5.84), en-
drin (4.84), fluoranthene (5.00), dieldrin (5.25). phenanthrene (4.46), and acenaphthene (3.76) with Equation 11. Km, for
DDT is the log average of the reported values in the Log P database [75]. The kepone Koc is the log mean of the ratio of or-
ganic carbon-normalized kepone concentration to pore water-kepone concentration from the toxicity data set. ffcws for the re-
maining compounds were computed by the U.S. EPA, Environmental Research Laboratory, Athens, Georgia. Methods are
presented later in this document.

-------
                          Sediment Quality Criteria Using Equilibrium Partitioning
 sediment organic carbon.  The predicted  sediment
 toxic units for each chemical follow a similar concen-
 tration-response  curve  independent:  of  sediment
 type. The data demonstrate that 50-percent mortality
 occurs at about one sediment toxic, unit, independent
. of chemical, species of organism, or sediment type, as
 expected if the EqP assumptions are correct.
       If the assumptions of EqP were exactly true,
 and there were no experimental variability or meas-
 urement error, then all data in Figures 2 and 3 should
 predict 50 percent mortality at one toxic unit. There is
 an uncertainty factor of approximately four to five in
 the results. This variation reflects inherent variability
 in these experiments and phenomena  that have not
 been accounted for in the EqP model. It also appears
 to be the limit of the accuracy and precision that can
 be expected.


 Effects Concentration

 The development of SQC requires an effects concen-
 tration for benthic organisms. Because many of the
 organisms used to establish the WQC are benthic,
 perhaps the WQC are adequate estimates of the ef-
 fects concentrations for benthic organisms. To exam-
 ine this possibility, the acute toxiciry database, which
 is used to establish the WQC, is segregated  into ben-
 thic and water column species, and the relative sensi-
 tivities of each  group  are compared.  Figure  4
 compares the acute values for the most sensitive ben-
                      BENTWC
                 i	i
         A FRESH WATER
         • SALT WATER
                -1012346

                BENTHC ORQAMSM FAV (ug/l)
 Figure 4.-—A comparison of the final acute values (FAV) for
 water column versus benthic organisms. Each data point
 represents an FAV for a particular chemical in either a
 freshwater or a saltwater exposure. The data are from the
 WQC or draft criteria documents. See Table 4 for  data
 sources."
 thic (epibenthic and infaunal) species to the most
 sensitive water column species. The data are from the
 40 freshwater and 30 saltwater U.S. Environmental
 Protection Agency (EPA) criteria documents. Al-
 though there is considerable scatter, these results, a
 more detailed analysis of all the acute toxiciry data,
 and the results of benthic colonization experiments,
 support the contention of equal sensitivity.
 BACKGROUND

 Under the Clean Water Act (CWA), the EPA is respon-
 sible for protecting the chemical, physical, and bio-
 logical integrity of the nation's waters. In keeping
 with this responsibility, EPA published WQC in 1980
 for 64 of the 65 priority pollutants and pollutant cate-
 gories listed as toxic in the CWA. Additional water
 quality documents that update criteria for selected
 consent decree and new chemicals havexbeen pub-
 lished since 1980. These WQC are numerical concen-
 tration limits that are protective of human health and
 aquatic life. Although these criteria play an impor-
 tant role in assuring a healthy aquatic environment,
 they are not sufficient to ensure appropriate levels of
 environmental and human health protection.
       Toxic contaminants in bottom sediments of
 the nation's lakes, rivers, wetlands, and coastal, wa-
 ters create the potential for continued environmental
 degradation even where water column contaminant
 levels comply with established WQC. The absence of
 defensible SQC makes it difficult to assess the extent
 of sediment contamination, implement measures to
 limit or prevent additional contamination from oc-
 curring, or to identify and implement appropriate re-
 mediation as needed.
       As a result  of the need to assist  regulatory
 agencies in making decisions concerning contami-
 nated sediment, the EPA's Office of Science and Tech-
 nology,  Health and  Ecological  Criteria Division,
 established a research team to review alternative ap-
 proaches to assess sediment  contamination.  Sedi-
 ment contamination and related problems were the
 subject of a conference [I]. Alternative approaches to
 establishing SQC [2] and their merits and deficiencies
 were discussed [3]. Additional efforts were under-
 taken to identify the scope of national sediment con-
 tamination [4]  and to review proposed approaches
 for addressing  contaminated sediments  [5,6]. The
EqP  method was selected because  it provides the
most practical, scientifically defensible, and effective
regulatory tool  for addressing individual honionic
chemicals associated  with contaminated  sediments
on a national basis [7].

-------
                          Sediment Quality Criteria Using Equilibrium Partitioning
 Rationale for Selecting
 the EqP Method

 The principal reasons for the selection of the-EqP
(method include the following:

  1. The EqP method was most likely to yield sedi-
     ment criteria that are predictive'of biological ef-
     fects in the field and defensible when used in a
     regulatory context These criteria address the is-
     sue of bioavailability and are founded on the ex-
     tensive biological  effects database used to
     establish national WQC.
                        *
  2. Sediment criteria can be readily incorporated into
     existing regulatory operations because a unique
     numerical sediment-specific criterion can be es-
     tablished for any chemical and compared to field
     measurements to assess the likelihood of signifi-
     cant adverse effects.

  3. Sediment criteria provide a simple and cost-effec-
     tive means of screening sediment measurements
     to identify areas of concern and provide informa-
     tion to regulators in a short period of time.

  4. The method takes advantage of the data and exper-
     tise that led to the development of national WQC.

  5. The methodology can be used as a regulatory tool to
     ensure that uncontaminated sites are protected from
     attaining unacceptable levels of contamination.


 Relationship  to WQC Methodology

The first question to be answered is this: Why not use
the WQC procedure for the development of SQC? A
detailed methodology has been developed that pre-
sents the supporting logic, establishes the  required
minimum lexicological data set, and specifies the nu-
merical procedures to be used to calculate the criteria
values 18]. Further, WQC developed through this
methodology are routinely used in the regulation of
effluent discharges. Therefore, it is only natural to ex-
tend these methods directly to sediments.
       The WQC are based on total chemical concen-
tration, so the transition to using dissolved chemical
concentration for those chemicals that partition to a
significant extent would not be difficult. The experi-
ence with site-specific modifications of the national
WQC has demonstrated that the water-effect ratio,
the ratio of chemical concentrations in site water to
laboratory water that produces the same effect, has
averaged 35 [9,10]. The implication is that differences
of this magnitude result from variations in site-specific
water chemistry and are not an overwhelming impedi-
ment to nationally applicable numerical WQC.
        The WQC are based on using the total chemi-
  cal"concentration as a measure of bioavailable chemi-
  cal concentration. However, the use of total sediment
  chemical concentration as a measure of the bioavail-
  able—or  even  potentially  bioavailable—chemical
  concentration is not supported by the available data
  [11]. The results of recent experiments indicate that
  sediments can differ irutoxidty by factors of 100 or
  more for the same total chemical concentration. This
  difference is a significant obstacle. Without a quanti-
  tative estimate of the bioavailable chemical concen-
  tration in a sediment, it is impossible to predict a
  sediment's toxicdry on the basis of chemical measure-
  ments, regardless of the  method used to assess bio-
  logical impact—be it laboratory toxicity experiments
  or field data sets comprising benthic biological and
  chemical sampling [12-15].
        Without  a   unique  relationship  between
  chemical measurements  and biological end points
  that applies across the range  of sediment properties
  that affect bioavailability, the cause and effect linkage
  is not supportable. If the same total chemical concen-
  tration is 100 times more toxic in one sediment than it
  is in another, how can we set  universal SQC that de-
  pend  only on the total sediment chemical concentra-
  tion?  Any SQC  that are based on total sediment
  concentration have a potential uncertainty of this
  magnitude. Thus, bioavailability must be explicitly
  considered for any sediment evaluation methodol-
  ogy that depends on chemical measurements to es-
  tablish defensible SQC.

 Applications of SQC

 SQC that are reasonably accurate in their ability to
 predict the potential for biological impacts are useful
 in many activities [16]. Sediment quality criteria can
 play a significant role in  the identification, monitor-
 ing, and cleanup of contaminated sediment sites on a
 national basis and provide a basis to ensure that sites
 that are uncontaminated will remain so. In some
 cases,  sediment criteria alone are sufficient to iden-
 tify and establish cleanup  levels for contaminated
 sediments. In other cases, they  must be supple-
 mented with biological sampling and testing before
 decisions are made.
       In many ways, sediment criteria developed
 using  the  EqP methodology  are similar to WQC.
. However, their application may be quite different. In
 most  cases, contaminants exceeding WQC in the
 water  column need only be controlled at the source
 to eliminate unacceptable adverse impacts. Contami-
 nated  sediments have often been in place for quite
 some time, and controlling the source of that pollu-
 tion (if the source still exists) may not be sufficient to
 alleviate the problem. The difficulty is compounded
 because the safe removal and treatment or disposal
 of contaminated  sediments can be laborious and
 expensive.

-------
                         Sediment Quality Criteria Using Equilibrium Partitioning
       Sediment criteria can be used as a means for
 predicting or identifying the degree and extent of
 contaminated areas such that more informed regula-
 tory decisions can be made. Sediment criteria will be
 particularly valuable in monitoring applications in
. which sediment  contaminant  concentrations  are
 gradually approaching the criteria. The comparison
 of field measurements to sediment criteria will pro-
 vide reliable warning of potential problems. Such an
 early warning provides an opportunity to take cor-
 rective action before adverse impacts occur.


 TOXICITY AND BlOAVAILABILITY

 OF CHEMICALS IN  SEDIMENTS

 A key insight into the problem of quantifying the
bioavailability of chemicals in sediments was that the
concentration-response curve for the biological effect
of concern could be correlated, not to the total sedi-
ment chemical concentration (micrograms  chemical
per gram dry sediment) but to the pore water con-
centration  (micrograms  chemical per liter pore
water) [17]. However, these results do not necessarily
imply that pore water is the primary route of expo-
sure because all exposure pathways are  at equal
 chemical activity in an equilibrium experiment (see
 Fig. 1) and the route of exposure cannot be deter-
 mined. Nevertheless, this observation was the critical
 first step in understanding bioavailability of chemi-
 cals in sediments.


 Toxicity Experiments

 A substantial amount of data has been assembled
 that addresses the relationship between toxicity and
 pore water chemical concentrations. Table 1 lists the
 sources  and  characteristics  of  these experiments.
 Some of these data are presented in Figures 5 to 8.
 The remaining data are presented elewhere in this
 document. In Figures 5 to 8 the biological response—
 mortality or growth rate suppression—is plotted ver-
 sus the total sediment concentration in the top panel,
 and versus the measured pore water concentration in
 the bottom panel. Table 2 .summarizes the LCso and
 ECso estimates and 95 percent confidence limits for
 these data on a total sediment and pore water basis,
 as well as the water-only values.
      The results from kepone experiments (Fig. 5)
are illustrative of the general trends in these data [17,
 18]. For the low organic carbon sediment (foe = 0.09
percent), the 50th percentile total kepone concentra-
Table 1. — Sediment toxicity data and bioaccumulation data.
CHEMICAL
ACENAPHTHENE
ACENAPHTHENE
CYPERMETHRIN
DDT
DIELDRIN
DIELDRIN
DIELDRIN
ENDRIN
ENDRIN
FLUORANTHENE
FLUORANTHENE
KEPONE
KEPONE
KEPONE
PERMETHRIN
PHENANTHRENE
PHENANTHRENE
ORGANISM
^ i^^-^— — 	
Eohaustorius estuarius
Leptocherius plumulosus
Chironomus ientans
Hyalella azteca
Chironomus tentans
Hyalella azteca
Hyalella azteca
Diporeia sp.
Hyalella azteca
Rhepoxynius abronius
Rhepoxynius abronius

Chironomus tentans
Chironomus tentans
Chironomus tentans
Eohaustorius estuarius
Leptocherius plumulosus '
SEDIMENT SOURCE
South Beach. OR
McKinney Slough, OR
Eckman Slough, OR
South Beach, OR
McKinney Slough, OR
Eckman Slough, OR
River and pond
Soap Creek, Mercer Lake
Airport Pond, MN
West Bearskin Lake. MN
Pequaywan Lake, MN
Airport Pond, MN
Lake Michigan
Soap Creek, Mercer Lake
Amended Ona Beach, OR
Yaquina Bay, OR
Soil
Soil
Soil
River and pond
South Beach, OR
MeKinney Slouth; OR
Eckman Slough, OR
South Beach, OR
McKinney Slough, OR
EXPOSURE
DURATION
(DAYS)
10
10
1
10
10
10
10
10
10
10
10 .
14
14
14
1
10
10
BIOLOGICAL
END POINT
Mortality
Mortality
Body burden
Mortality
Mortality
Mortality
Mortality
Mortality
Mortality
Mortality
Mortality
Body burden
Growth
Mortality
Body burden
Mortality
Mortality
REFERENCE
[56]
[56]
[22]
[20,21]
[55]
[53]
[54]
[52]
[20,21]
[57]
[19]
[17,24] '
t!7]
[17]
[22]
[56]
[56]
FIGURE
23
23
8

26

24
23
7
23
6

5

8
23
23

-------
                            Sediment Quality Criteria Using Equilibrium Partitioning
                                        Dry  Weight  Normalization
      Kapona - Mortality
                                                                    K«pon*-- Growth
o.i       uo       IDA     100.0
   Dry Weight Concentration (ug/g)
                                                             O.1       1.0       1O.O     1OO.O
                                                             •,   Dry Weight Concentration (ug/e)
                                        Pore  Water  Normalization
                          Kapona - Mortality
                  1OO
                  80
                  eo
                  4O
                  80
                   O
                                                Kapona - Growth
                                       100

                                    H   80

                                    J   80
                                    3   4O
                                    «   20
                                         0
* -
 1        1O       1OO      1OOO
   Pore Water Concentration (ug/U
                                                              1        10       100     1000
                                                                fan W«t«r Conewrtntton (ug/U
Rgure 5.—Comparison of percent mortality (left) and growth rate reduction (right) of C. tentans to Kepone concentration in
bulk sediment (top) and pore water (bottom) for three sediments with varying organic carbon concentrations [17].
                                       Dry  Weight  Normalization
                            Fkioranth«n«
                                                                       Cadmium
                                                          100
                                                       2   *
                                                       *.   •<>
                                                       5   40
                                                       i   20
                                                            O
                                                                ao
                                                              *
                                                              e
                          5      1O     18     2O
                      Dry Weight Concentration (ug/g)
                                         0     2O     4O    6O     8O
                                           Dry Weight Concentration (i»«/g)
                                       Pore  Water Normalization

                            Fkioranthana                                Cadmium

                                                          100
                                                       3  «o
                                                          80
                                                       !  40
                                                          2O
                                                           0
                          20    4O     60     80
                      Pore Water Concentration (ug/U
                                         0       2OOO     4OOO     6OOO
                                           Pore Water Concentration (iKi/L)
Figure 6.r-Comparison of percent mortality of R. abronium to fluoranthene [19] concentrations in bulk sediment (top) and
pore water (bottom) for sediments with varying organic carbon concentrations..

-------
                          Sediment Quality Criteria Using Equilibrium Partitioning
  Tabte 2.—LCso and ECso for sediment dry weight and sediment-organic carbon normalization and for pore-water and
  water-only exposures.                                                                  <™awn anu

                                                        LCSO AND ECSO
CHEMICAL
(END POINT)
KEPONE
(MORTALITY)
KEPONE
(GROWTH)
FLUORANTHENE
(MORTALITY)
DOT
(MORTALITY)
DOT
(MORTALITY)
ENDRIN
• (MORTALITY)

ENDRIN
(MORTALITY)

fce(%)
0.90
1.50
12.0
0.09
1.5
12.0
0.2
0.3
0.5
3.0
7.2
10.5
3.0
3.0
11.0
3.0
6.1
11.2
3.0
11.0
11.0
TOTAL SEDIMENT («/«)
0.90 (0.73-1.10)
6.9(5.85-8.12)
35.2 (30.6-40.3)
0.46 (0.42-0.51)
9.93 (7.74-12.8)
37.3 (31.5-44.2)
3.2 (2.85-3.59)
6.4 (5.56-7.27)
10.7 (8.34-13.7)
10.3 (8.74-12.2)
17.5 (12.5-24.3)
44.9 (36.7-55.0)
1.54 (1.18-2.00)
4.16 (3.91-4.42)
10.95 (9.34-12.9)
3.39 (2.61-4.41)
5.07 (4.05-6.36)
5.91 (4.73-7.37)
4.76 (3.70-6.13)
18.9 (13.6-26.3)
10.5 (8.29-12.7)
PORE WATER (u*/D
29.9(25.3-35.6)
31.3 (25.7-38.1)
18.6 (15.7-21.9)
17.1 (15.7-18.7)
48.5 (34.6-67.8)
20.1 (16.7-24.1)
21.9 (19.6-24.4)
30.9 (27.0-35.4)
22.2 (17.5-29.3)
0.74 (0.67-0.82)
1.45 (1.20-1.75)
0.77 (0.67-0.89)



1.80 (1.44-2.24)
1.92 (1.55-2.36)
1.74 (1.37-2.20)
2.26 (1.67-3.05)
3.75 (2.72-5.19)
2.81 (2.44-3.23)
ORGANIC CARBON (pt/c)
1,000 (811-1.220)
460 (390-541)
293 (255-337)
511 (467-567)
662 (516-1,050)
"• 311 (262-368)
1,600 (1,430^1.800)
2,130 (1.850-2,420)
2.140 (1.670-2,740)
344 (291-405)
243 (174-338)
428 (350-524)
51.3 (39.3-66.7)
139 (130-147)
99.6 (84.9-117)
113 (87.0147)
83.1 (66.4-104)
52.8 (42.2-65.8)
159 (123-204)
172 (124-239)
95.8 (75.4-115)
WATER ONLY (pS/L)
26.4 (22.7-30.6)


16.2 (15.0-17.5)





0.45 (0.38-0.53)
0.48 (0.42-0.55)
0.52 (0.45-0.60)



4.81 (4.46-5.20)
3.39 (3.10-4.98)
3.71 (3.11-4.44)



REFERENCE
[17]


[171


• [19]


[20]


[21J


[20]


[21]


   The LCsos and ECsos and the 95% confidence limits in parentheses are computed by the modified Spearman-Karber method I123J
tion for both Chironomus tenians mortality (LCso) and
growth rate reduction from a life cycle test (ECso) are
<1 ug/g. By contrast, the 1.5 percent organic carbon
sediment ECso and LCso are approximately 7 and 10
ug/g, respectively. The high organic Cetrbon sediment
(12 percent) exhibits still higher LCso and ECso val-
ues on a total sediment kepone concentration basis
(35 and 37 ug/g, respectively).
       However, as shown in the bottom panels, es-
sentially all the mortality data collapse into a single
curve and the variation in growth rate is significantly
reduced when  the pore water concentrations are
used as the correlating concentrations. On  a pore
water basis, the biological responses as measured by
LCso or ECso vary by approximately a factor of two,
whereas when they are evaluated on a total sediment
kepone basis, they exhibit an almost 40-fold range in
kepone toxicity.
       The comparison between the pore water ef-
fects concentrations and the water-only results indi-
cates that they are similar. The pore water LCsos are
19 to 30 ug/L, and the water-only  exposure LCso is
26 ug/L. The pore water ECsos are 17 to 49 ug/L, and
the water-only ECso is 16 ug/L (Table 2).
        Laboratory experiments have also been per-
 formed to characterize the toxicity  of fluoranthene
 [19] to the sediment-dwelling marine amphipod Rhe-
 poxynius abronius. Figure 6 presents the R. abronius
 mortality data for the fluoranthene experiment. The
 results of the fluoranthene experiments parallel those
 for kepone. The sediment with the lowest organic
 carbon content (02 percent) exhibits  the lowest LCso
 on a total sediment concentration basis (3.2 ug/g)
 and as  the organic carbon concentration increases
 (0.3 and 0.5 percent) the LCso increases (6.4 and 10.7
 ug/g). On a pore water basis, the data again collapse
 to  a  single concentration-response  curve and the
 LCsos differ by less than 50 percent
       Figure 7 presents mortality data for DDT and
- endrin using the freshwater amphipod Hyalella azteca
 [20,21]. The responses  for DDT  [20] are similar to
 those  observed for kepone, cadmium,  and fluoran-
 thene. On a total sediment concentration basis the or-
 ganism  responses differ for the various sediments
 (LCsos are 10.3 to 45 ug/L), but on a pore water basis
 the responses are again similar (LCsos are 0.74 to 1.4
 Ug/L) and comparable to the water-only LCsos of ap-
 proximately 0.5 ug/L. The DDT  data reported by
 Schuytema et al. [21] are more variable. By contrast,

-------
                         Sediment Quality Criteria Using Equilibrium Partitioning
                                     Dry  Weight Normalization
                               DOT
                                                                    Endrln
                                       »OC (5-.)
                                       * *•*
                                       • 10*
                         80    10O   ISO   200
                     Dry Weight Concentration (ug/g)


„
£


I


100
80

80
4O
2O
O1
fee IX)
' • M _ II**
K
e 11JI

.*"•

I -I*'"5.'1

am-
.

•


•
     0.1      . 1.0
        Dry Weight
  1O.O .    100.0
ntratton 
• *

• •
• •
•>^/
8" •
• •
•


.
•
•
     0.1      1.O      1O.O     10O.O
        Pom Water Concentration (ug/U
Figure 7.—Comparison of percent mortality of H. azteca to DDT (left) and endrin (right) concentrations in bulk sediment (top)
and pore water (bottom) for sediments with varying organic carbon concentrations [20, 21].
the organism survival for endrin exposures on a dry
weight basis varies by a factor of almost six among
the six sediments. The LCsos are 3.4 to 18.9 ng/g. The
pore water LCsos were less variable, 1.7 to 3.8 u.g/L
and comparable to the water-only exposure LCso of
approximately 4 ug/L (Table 2).


Bioaccumulation

A direct measure of chemical bioavailability is the
amount of chemical retained in organism tissues.
Hence, tissue bioaccumulation data can be used to
examine the extent of chemical bioavailability. Chiro-
nomus tentans was exposed to two synthetic pyre-
throids, cypermethrin and permethrin, spiked into
three sediments, one of which was laboratory-grade
sand [22]. The bioaccumulation from the sand was
approximately an order of magnitude higher than it
was from the organic carbon-containing sediments
for both cypermethrin and permethrin (Fig.  8, top
panels).
      On a pore water basis, the bioaccumulation
appears to be approximately linear and independent
of sediment type (bottom panels). The mean bioaccu-
mulation factor  (BAF) for cypermethrin (and per-
methrin) varies from 6.2 to 0.6 (4.0 to 023) (ug/g or-
ganism/ug/g  sediment) as sediment foe increases
(Table 3). By contrast the mean BAFs on a pore water
basis vary by less than a factor of two.
       Bioaccumulation was  also . measured by
Adams et al. [17,23] and Adams [24] in the C. ten-
tons-kepone experiments presented previously (Fig-
ure 3). The body burden variation on a total sediment
basis is over two orders of magnitude (BAF = 600 to
3.3 ng/g organism /ng/g sediment), whereas the
pore water bioaccumulation factor is within a factor
of four (5,200 to 17,600 ng/kg organism/ng/L), with
the very low organic carbon sediment exhibiting the
largest deviation (Table 3).


Conclusion

These observations—that organism concentration re-
sponse and  bioaccumulation from different  sedi-
ments can be reduced to one curve if pore water is
considered as the concentration that quantifies expo-
sure—can be interpreted in a number of ways. How-
ever, these results do not necessarily imply that pore
water is the primary route of exposure because all ex-
posure pathways are at equal chemical activity in an

-------
                      Sediment [Qujality Criteria Using Equilibrium Partitioning
                                    Dry  Weight Normalization
   10000
 £ 1000
 i   100
 •
 I    .
                        Cyparmathrin

                                                     1000.0

                                                    2 100.0

                                                    I  10.0
                                                        1.O
                     10     100   1000   toooo
                Dry Weight CMcentnrttoa hg/g)
                                                        O.T
                                                                     Parmethrln
                                                             Dry Weight
                                                                      100      1000
                                                                          <"0/g)
                                  Pore  Water Normalization
  10000

N  1000
          100

           10
                      Cyparmatfirin
«»0j»
                                                    1OOO.O
9
8
            0.01    0.10   1.00   10.00  10000
                   Weter Concentration (ug/L)
9
-
-
o
                                                                    Parmathrin
                                                             foc(J«|
                                                             • "0.1

                                                             *"

                                                 <"»»      0.10      1.00      10.00
                                                         Weter Concentration (ug/L)
                                                                               versus concentration in bulk
                                                                                 122].
Table 3.—Bioaccumulation factors for C. tentans.
                                           BIOACCUMULATION FACTORS*
                                   'TOTAL SEDIMEN
                                    IK/K ORGANISM
                                    M/g SEDIMENT
                                                         PORE WATER
                                                       Hi/kg ORGANISM
                                    6.21 (4.41-8.01)
                                                              80.1 (73.5-86.7
                                    0.50 (0.30-0.71)
                                                              51.3 (43.8-58.8)
                                    0.60 (0.37-0.83)
                                                              92.9 (87.0-98.8)
                                    4.04 (2.89-5.20)
                                                              39.7 (25.0-54.3]
                                    0.38 (0.17-0.59)
                                                              52.5 (22.6-82.4)
                                    0.23 (0.18-0.28)
                                                              29.7 (15.6-43.7)
                                                          17.600 (6,540-28.600)
                                                           5,180 (1,970-8,3901
                                                           5,790 (2.890-8.7001
     95% confidence limits shown in parentheses.

-------
                         Sediment Quality Criteria Using Equilibrium Partitioning
 equilibrium experiment. The route of exposure can-
 not be determined, as we can see by comparing the
 concentration-response correlations  to  pore water
 and organic carbon-normalized sediment concentra-
 tions. That both are equally successful at correlating
•the data suggests that neither the pore water nor the
 sediment exposure pathway can be implicated as the
 primary exposure route.
       In order to relate pore water exposure to sedi-
 ment carbon exposure, it is necessary to establish the
 relationship  between  these two   concentrations.
 Therefore, an examination of the state of the art of
 predicting the partitioning of chemicals between the
 solid and the liquid phase is required. This examina-
 tion is described in the following section.
SORPTION OF  NONIONIC

ORGANIC CHEMICALS


Partitioning In Particle Suspensions

A number of empirical models have been suggested
to explain the sorption of nonionic hydrophobic or-
ganic chemicals to natural soils and sediment parti-
cles (see Karickhoff [25] for an excellent review). The
chemical property that indexes hydrophobicity is the
octanol/water partition coefficient, Kow- The impor-
tant particle property is the weight fraction of or-
ganic carbon, /oc. Another important environmental
variable appears to be the particle  concentration
itself.
       In many experiments using particle suspen-
sions, the partition coefficients have been observed to
decrease as the particle concentration used in the ex-
periment is increased [26]. Very  few experiments
have been done on settled or undisturbed sediments;
therefore, the  correct  interpretation of particle sus-
pension experiments is of critical importance. It is
not uncommon for the partition coefficient to de-
crease by two to three orders of magnitude at high
particle concentrations. If this partitioning behavior
is characteristic of bedded sediments, then quite low
partition coefficients would be appropriate, which
would result in lower sediment chemical concentra-
tions for SQC If, however, this phenomenon is an ar-
tifact or a result of a phenomenon that does not apply
to bedded sediments, then a quite different partition
coefficient would be used. The practical importance
of this issue requires a detailed discussion of the par-
ticle concentration effect.
       Particle concentration  effect. For the revers-
ible (or readily desorbable) component of sorption, a
particle interaction model (PIM) has teen proposed
that accounts for the particle concentration effect and
predicts the partition coefficient of nonionic hydro-
phobic chemicals over a range of nearly seven orders
of magnitude with a logio prediction standard error
of 038 [27]. The reversible component partition coef-
ficient, .Kp, is the ratio of reversibly bound chemical
concentration, Cs (ug/kg dry weight), to the dis-
solved chemical concentration, Cd (ug/L):
            Cs-Kp-Cd.

The PIM model for K p1, is

        ~ „      /ocKoc
                                           (9)
                                          (10)
                   1 + m/oc
                                                   where
    % * = reversible component partition coefficient
     P   (L/kg dry weight)

    Koc = particle organic caxbon partition
         coefficient (L/kg organic carbon)

   /oc  = particle organic caxfoon weight fraction (kg
         organic carbon/kg dry weight)

    m  = particle concentration in the suspension
         (kg dry weight/L)

    vx  = 1.4, an empirical constant (unitless).

The regression  of Koc to the octanol/water  coeffi-
cient, Kow, yields

       logioKoc = 0.00028 + 0.983 logioKow    (11)

which is that essentially K0c approximately  equals
Kow. Figure 9 presents the observed versus predicted
reversible component partition coefficients using this
model [27]. A substantial fraction of the  data in the
regression is at high particle concentrations (m/ocKow
> 10), where the partitioning is determined only by
the solids concentration and vx. The low particle con-
centration data (m/ocKow < 1) are presented in  Figure
10 for the conventional adsorption (left) and revers-
ible component  (right) partition coefficient, Kp, nor-
malized by /oc, that is Koc = Kp/foc. The relationship
Koc - Kow is demonstrated from the agreement be-
tween the line of perfect equality and the data. It is
important to note that while Equation 10 applies only
to the reversible component partition coefficient, Kp,
the equation: Kp -foe Kow applies to the conventional
adsorption partition coefficient as well (Fig. 10, left).
      A number of explanations have been offered
for the particle concentration effect. The most popu-
lar  is the existence of an additional third sorbing
phase or complexing component that is associated

-------
                         Sediment Quality Criteria Using Equilibrium Partitioning
                   Reversible Component  Partition Coefficient
                    O)
                   JC
                   o

                   o>
                   
-------
                          Sediment Quality Criteria Using Equilibrium Partitioning
 with the particles but is inadvertently measured as
 part of the dissolved chemical concentration because
 of experimental limitations. Colloidal particles that
 remain in solution after particle*separation [28,29]
 and dissolved ligands or macromolecules that desorb
.from the particles and remain in solution [30-33]
 have been suggested. It has also been suggested that
 increasing particle concentration increases the degree
 of particle aggregation, decreasing the surface area
 and hence the partition coefficient [34]. The effect has
 also been attributed to kinetic effects [25].
       Sorption by nonseparated  particles or com-
 plexing by dissolved organic carbon can produce an
 apparent decrease in  partition coefficient  with in-
 creasing  particle  concentration if the  operational
 method of measuring dissolved chemical concentration
 does not properly discriminate the truly dissolved or
 free chemical concentration from the complexed or col-
 loidally sorbed portion. However, the question is not
 whether improperly measured dissolved concentra-
 tions can lead to an apparent decrease in partition coef-
 ficient with  increasing particle concentrations. The
 question is whether these third-phase models explain
 all (or most) of the observed partition coefficient-parti-
 cle concentration relationships.
       An alternate possibility is that the particle
 concentration effect is a distinct phenomena that is a
 ubiquitous feature of  aqueous-phase particle  sorp-
 tion. A number of experiments have been designed
 explicitly to exclude possible third-phase  interfer-
 ences. Particle concentration effects are displayed in
 the resuspension experiment for polychlorinated
 biphenyls (PCBs) and metals [35-37], in which parti-
 cles are resuspended into a reduced volume of super-
 natant, and in the dilution experiment in which the
 particle suspension is diluted with supernatant from
 a parallel vessel [35]. It is difficult to see how third-
 phase models'can account  for these results because
 the concentration of the colloidal particles is constant
 while the concentration of the sediment particles var-
 ies substantially.
       The model (Eqn. 10) is based on the hypothe-
 sis that particle concentration  effects result from an
 additional desorption  reaction induced by  particle-
 particle interactions [27]. It has been suggested that
 actual particle collisions are responsible [38]. This in-
 terpretation relates vx to the collision efficiency for
 desorption and demonstrates that it is independent
 of the chemical and particle properties, a fact that has
 been experimentally observed [27,36].
       It is not necessary to decide which of these
 mechanisms is responsible for the effect if all the pos-
 sible interpretations yield the same result for sedi-
 ment/pore  water partitioning. Particle interaction
 models would predict that Koe = Kow because the par-
 tides are stationary in sediments. Third-phase mod-
 els would also  relate the  free (i.e., uncomplexed)
 dissolved chemical concentration to participate con-
 centration via the same equation. As for kinetic ef-
 fects, the equilibrium concentration is again given by
 the relationship Koc = Kow. Thus there is~ unanimity
 on the proper partition coefficient to be used in order
 to relate the free dissolved chemical concentration to
 the sediment concentration: Koc s Kow.
       Organic carbon fraction. The unifying para-
 meter that permits the development of SQC for non-
 ionic hydrophobic organic chemicals that are applic-
 able to a broad range of sediment types is the. organic
 carbon content of the sediments. This development
 can be shown as follows: The sediment/pore water
 partition coefficient, Kp, is given by

                                            (12)

 and the solid phase concentration is given by

                   Cs=/ocKocCd             (13)

 where Cs is the concentration on sediment particles.
 An important observation can be made that leads to
 the idea  of organic carbon-normalization. Equation
 12 indicates that the partition coefficient for any non-
 ionic organic chemical is linear in the organic carbon
 fraction, foe. The partitioning data examined in Fig-
 ure 10 can be used to examine the linearity of Kp to
 foe- Figure 11 compares Kp/Kow to/oc for both the ad-
 sorption and the reversible component partition coef-
 ficients.  The data are restricted to m/0c Kow < 1 to
 suppress particle effects. The line indicates the ex-
 pected linear relationship in Equation 12. These data
 and an analysis presented below appear to support
 the linearity of partitioning to a value of foe = 0.2 per-
 cent. This result and the toxicity experiments exam-
. ined below suggest that for foe > 0.2 percent, organic
 carbon-normalization is valid.
       As a consequence of the linear relationship of
 Cs and foe,  the relationship between sediment con-
 centration, Cs, and free dissolved concentration, Cd,
 can be expressed as
                   Joe
 If we define
                                            (14)
                                            (15)
 as the organic carbon normalized sediment concen-
 tration (jig chemical/kg organic carbon), then from
 Equation 14: -
                   Cs,oc ~
(16)

-------
                         Sediment Quality Criteria Using Equilibrium Partitioning
                           Partition  Coefficient  - m foe Kow  < 1
                            ;
                         Adsorption                       Reversible  Component
          I
          o
          ^»
-i
              -2
-3
• AMIetct
A CwkofwM
T Lbwraii
.<<
t> Cvkvyl
                                  Methyl PwatMm .
                                •$• PmtMon
                                •A- QMMHCH
                                  HP COT
 O.O1    O.1O    1.0O   1IO.OO  1OO.OO

             foe C%)
                                                              O.10    1.OO   .1O.OO  100.0O

                                                                   foe (X)
•Figure 11.—Comparison of the normalized partition coefficients for adsorption (left) and reversible component sorption
 (right) to sediment organic carbon. The data are restricted so that particle effects are not expected to be significant: m/bcKow
 < 1. The line represents perfect agreement [27].
Therefore, for a specific chemical with a specific KOCl
the organic carbon-normalized  total sediment con-
centration, Cs,oc, is proportional  to the dissolved free
concentration, Cd, for any sediment with/oc > 0.2 per-
cent. This latter qualification is  judged  to be neces-
sary because at/oc < 0.2 percent the other factors that
influence partitioning (e.g., particle size and sorption
to nonorganic mineral fractions) become relatively
more important [25]. Using the proportional relation-
ship given by Equation 16, the concentration of free
dissolved  chemical can be predicted from  the nor-
malized sediment concentration and Koc. The  free
concentration is of concern as it is.the  form that is
bioavailable. The evidence is discussed in  the next
section.


Dissolved Organic Carbon (DOC)
Complexing

In addition to  partitioning to  particulate organic
carbon (POC) associated with sediment particles, hy-
drophobic chemicals can also partition to the organic
carbon in colloidal sized particles. Because these par-
ticles are too small to be removed by conventional fil-
tration  or  centrifugation  they are  operationally
defined as DOC. Because sediment interstitial waters
frequently contain significant levels of DOC, it must
be considered in evaluating the phase distribution of
chemicals.
    ,„. .A distinction is made between the free dis-
solved chemical concentration,  Cd, and the  DOC-
                                      complexed chemical, CDOC- The partition coefficient
                                      for DOC, KDOO is analogous to Koc as it quantifies
                                      the ratio of DOC-bound chemical, CDOO to the free
                                      dissolved concentration, Cd:
                                                    CDOC =
                                                                             (17)
                                      where mooc is the DOC concentration. The magni-
                                      tude of KDOC and the DOC concentration determine
                                      the  extent of DOC complexation that takes place.
                                      Thus, it is important to have estimates of these quan-
                                      tities when calculating  the level of free dissolved
                                      chemicals in sediment pore waters.
                                            A recent compilation of KDOC together with
                                      additional  experimental determinations is available
                                      [39]. A summary that compares the partitioning of six
                                      chemicals to POC, natural DOC, and Aldrich humic
                                      acid (HA) is shown on Figure 12. The magnitude of
                                      the partition coefficients  follow the order: POC > HA
                                      > natural DOC. The upper bound on KDOC would ap-
                                      pear to be KDOC = Koc, the POC partition coefficient.


                                      Phase Distribution in Sediments

                                      Chemicals  in sediments are partitioned into three
                                      phases: free chemical; chemical sorbed to POC, and
                                      chemical sorbed to DOC. To evaluate the partitioning
                                      among these three phases, consider the mass balance
                                      for total concentration CT:
                                         CT = 0Cd +
                                                                + 0mDocKDocCd  (18)

-------
                          Sediment Quality Criteria Using Equilibrium Partitioning
o
O   '
o
o
                              POC  ~ ''
                              Hunle AaM
                              Pen W«t«r DOC
             BaP   DDT  HCBP MCBP  PYR  TCBP


                     Chemicals
      12.—Partition coefficients of chemicals to particu-
late organic carbon (POC), Aldrich humic acid, and natural
DOC.  Benzo[a]pyrene (BaP);  2,2',4,4',5,5' hexachloro-
biphenyl (HCBP); DDT; 2,2'.5.5'-tetrachlorobiphenyl (TCBP);
pyreoe (PYR); 4-monochlorobipheny!  (MCBP).  Data from
Eadie et al. [39].
        I
                                                          1.0000
        o
             O.1000
        s    0.0100
        5
                                                     O
                                                    U
                                                          O.OO10
             O.0001
                                                                                        S«dlm*nt
                                                                                        Por» Waur
                   0     10    20    30     40    50    60

                        DOC Concentration (mg/LJ

         Figure 13.—Partition coefficients of chemicals to particu-
         late organic carbon (POC), Aldrich humic acid, and natural
         DOC.  Benzo[a]pyrene  (BaP);  2,2',4,4',5,5'  hexachlorc-
        .biphenyl (HCBP); DDT; 2,2',5,5'-tetrachlorobiphenyl (TCBP);
         pyrene (PYR);  4-monochlorobiphenyl  (MCBP).  Data from
         Eadie et al. [39].
where 0 is the sediment porosity (volume of water/ vol-
ume of water plus solids) and m is the sediment solids
concentration (mass of solids /volume of water plus
solids). The three terms on the right side of the equa-
tion are the concentration of free chemical in the in-
terstitial water, and that sorbed to the POC and DOC,
respectively. Hence, from Equation 18 the free dis-
solved concentration can be expressed as
          Cd--
                         CT
               0 + m;
(19)
The concentration associated with the particle carbon
(Eqn. 16) and DOC (Eqn. 17) can then be calculated.
The total pore water concentration is the sum of the
free and DOC complexed chemical, so that

      Cpore = Cd + CDOC - Cd(l + mDoc^DOc)-   (20)

       Figure 13 illustrates the phase partitioning be-
havior of a  system for a unit concentration  of  a
chemical with the following properties: Koc = KDOC =
106 L/kg,/oc = 2.0 percent, m = 0.5 kg solids/L sedi-
ment, and TTJDOC varies from 0 to 50 mg/L, a reason-
able range  for  pore  waters  [40]. Witi^ no  DOC
present, the pore water concentration equals the free
concentration. As DOC increases, the pore water con-
centration increases because of the  increase in com-
plexed chemical, CDOO Accompanying this increase
in CDOC is a slight—in fact, insignificant—decrease in
Cd (Eqn.  19) and a proportional decrease in Cs
(Eqn. 16).
                                                            It is important to realize that the free chemical con-
                                                      centration, Cd, can be estimated directly from Cs,oc, the or-
                                                      ganic carbon-normalized sediment concentration, using
                                                      Equation 16, and that the estimate is independent of the
                                                      DOC concentration. However, to estimate Cdfrom the pore
                                                      water concentration requires that the DOC concentration
                                                      and KDOC be known. The assumption Cpore = Cd is clearly
                                                      not warranted for very hydrophobic chemicals. For these
                                                      cases Cs,oc gives a more direct estimate of the free dissolved
                                                      bioavailable concentration, Cd,  than does the pore water
                                                      concentration.

                                                      Bioavailability of DOC-Complexed
                                                      Chemicals

                                                      The proportion of a chemical in pore water that is
                                                      complexed  to DOC can be substantial (Fig.  13).
                                                      Hence, the question of bioavailability of DOC-com-
                                                      plexed chemical can be important in assessing toxic-
                                                      ity directly from measured pore  water concentra-
                                                      ions. A  significant quantity of data indicates that
                                                      DOC-complexed chemical is not bioavailable. Fish
                                                      [41]-and amphipod [42] uptake of polycydic aromatic
                                                      hydrocarbons (PAHs) are significantly reduced  by
                                                      adding DOC. An example is shown in Figure 14 for a
                                                      freshwater amphipod [42]. For a highly hydrophobic
                                                      chemical such as benzo[a]pyrene (BaP) the effect is
                                                      substantial, whereas for less hydrophobic chemicals
                                                      (e.g., phenanthrene) the reduction in uptake rate is
                                                      insignificant. This result was expected because, for a
                                                      fixed amount of DOC, the  quantity of DOC-com-
                                                      plexed chemical  decreases  with  decreasing KDOC
                                                      (Eqn. 17).

-------
                          Sediment Quality Criteria Using Equilibrium Partitioning
   I
   o»
   -i
   a
   oc
   0
   1C
   CO
   a
        300
200 -
100
                                                                 DOC  Partition  Coefficient
     >***
                       Chemical
 Figure 14.—Average uptake rate of chemicals by Pontopo-
 reia hoyi with (filled) and without (hatched) DOC present.
 Benzo[a]pyrene (BaP); 2,2',4.4'-tetrachlorobiphenyl (TCBP);
 pyrene, phenanthrene. Data from Landrum et al. [42].
                                                       o
                                                       o
I
I
(0
                                                       o
                                                       o
                                                       o
                                                           10000
                                                            1000 r
                                                     100 r
                                                            •  Humlc Add
                                                            A  Por* Water
                                                                       tOO
                                                                               1OOO
                                                                                      10OOO
                                                       Koc from Reverse Phase (L/g oc)

                                             Figure 15.—Comparison of the DOC partition coefficient
                                             calculated from the suppression of chemical uptake versus
                                             the Cis reversed-phase HPLC column estimate. Circles are
                                             Atdrich humic  acid; triangles are interstitial water DOC.
                                             Chemicals are listed in Rgure 14 caption (also anthracene
                                             and benzo[a]anthracene).
       The quantitative demonstration  that DOC-
complexed chemicals are not bioavailable requires an
independent  determination of the concentration of
complexed chemical. Landrum et al. [42] have devel-
oped a Cis reverse-phase HPLC column technique
that separates the complexed and free chemical. Thus
it is possible to compare  the measured  DOC-com-
plexed chemical to the quantity of complexed chemi-
cal inferred from the uptake experiments, assuming
that all the complexed chemical is not bioavailable
[42,43]. As shown on Figure 15, although the KDOC
inferred from uptake suppression is larger than that
inferred from the reverse-phase separation for  HA,
these data support  the assumption that the DOC-
complexed fraction,  CDOO is not bioavailable. Hence
the bioavailable form of dissolved chemical is Cd, the
free uncomplexed component. This is an important
observation because it is Cd that is in equilibrium
with Cs,oo the organic carbon-normalized sediment
concentration (Eqn. 15).


Field Observations of Partitioning in
Sediments

An enormous quantity of laboratory data exists for
partitioning in particle suspensions. However, pore
water and sediment data from field  samples are
scarce: Two types of data from field samples are ex-
                                            amined. The first is a direct test of the partitioning
                                            equation Cs,oc = Koc Cd, which is independent of the
                                            DOC concentration. The second examines the sedi-
                                            ment and pore water concentrations and accounts for
                                            the DOC that is present.
                                                   Organic carbon normalization. Consider a
                                            sediment sample that is segregated into various size
                                            classes after collection. The particles in each class
                                            were in contact with the pore water. If sorption equi-
                                            librium has been attained for each class, then, letting
                                            Cs(;) be the particle chemical concentration of the;th
                                            size class, it is true that
                                                           Cs(;) =/oc(;)KocCd
                                        (21)
                                            where/oc(;) is the organic carbon fraction for each
                                            size class /'. On an organic carbon-normalized basis
                                            this equation becomes
                                                           Cs,oc(;)  =
                                        (22)
                                            where Cs/oc(;) = Cs(;)//oc(;). This result indicates
                                            that the organic carbon-normalized sediment concen-
                                            tration of a chemical should be equal in each size
                                            class because K0c and Cd are the same for each size
                                            class. Thus a direct test of the validity of both organic
                                            carbon normalization and EqP would be to examine
                                            whether Cs/oc(;) is constant across size classes in a
                                            sediment sample.

-------
                          Sediment Quality Criteria Using Equilibrium Partitioning
        Data from three field  studies,  Prahl  [44],
 Evans et al. [45], and Delbeke et al. [46], can be used
 to test this prediction. In  Prahl's study,  sediment
 cores were collected at three stations near the Wash-
 ington State coast (Stations 4, 5 and 7). These were
•sieved into a silt-and-day sized fraction (<64 um),
 and a sand-sized fraction (>64 mm). This latter frac-
 tion was further separated into a low density fraction
 (<1.9 g/cm3) and the remaining higher density sand-
 sized particles.  The concentrations of  13 individual
 PAHs were measured in each size fraction.
       It is important to realize that these  size frac-
 tions  are not pure day, silt, or sand but  are natural
 particles in the size classes denoted by clay, silt, arid
 sand. The organic carbon fractions, shown on Figure
 16, range from 0.2 percent for the high-density sand-
 sized  fraction to greater than 30 percent for the low-
 density  fraction.  This  exceeds  two  orders   of
 magnitude  and essentially spans the range usually
 found in practice. For example, 90 percent of the es-
 tuarine and coastal sediments sampled for the Na-
tional Status and Trends program exceed  0.2 percent
 organic carbon [47].

            Organic  Carbon Fractions
 100.0
  10.0
   1.0
   0.1
                               • st«. 7
                               m st«. s
                               • SU. 4
              LOW     SAND   SILT/CLAY

                  S*dlm«nt Fraction
Figure 16.—-The organic carbon fractions (% dry weight) in
the low-density fraction .64 um, <1.9 g/cm3; the sand-
sized fraction >64 um, >1.9 g/cm3; the silt/clay-sized frac-
tion <64 urn. Numbered stations as indicated. Data from
[44].
       Figure 17 (top) compares the dry weight-nor-
malized day/silt sized fraction sediment PAH con-
centrations,  Cs(j),  to the  sand-sized  high- and
low-density PAH concentrations on a dry weight ba-
sis. The dry weight-normalized data have distinctly
different concentrations—the low-density high-or-
ganic carbon fraction is highly enriched, whereas the-
sand-sized fraction is substantially below the clay/silt
 fraction concentrations. Figure 17 (bottom) presents
 the same data but on an organic carbon-normalized
 basis, Cs/oc(;). In contrast to dry weight normaliza-
 tion, the PAH concentrations are essentially the same
 in each size dass, as predicted by Equation 22.
       In the field data from Evans et al. [45] sedi-
 ments were collected at five sites along the River
 Derwent, Derbyshire, United Kingdom, and sepa-
 rated into six sediment size classes. The size dasses
 were  representative  of day and silt  (<63 u,m). to
 course sand (1.0 to 2.0 mm). Organic carbon content
 and total PAH were measured in each sediment size
 class. Figure 18 presents the size classes and associ-
 ated organic carbon content. Evans et aL attribute the
 bimodal distribution  of foe to two types of organic
 matter. Organic matter in the 1.0 to 2.0 mm size dass
 may be  from fragmentary plant material while  the
 size classes less then 500 u,m organic carbon content
 is the result of aging humic material. The organic
 content in this study ranges from 2.0 to 40 percent.
       Figure 19 presents a comparison of PAH con-
 centration for different sediment classes for  dry
 weight normalization and organic carbon normaliza-
 tion. The top left panel compares PAH concentrations
 on sand (63 urn - 500 urn) and clay/silt (< 63 um) on a
 dry weight basis. The top right panel compares PAH
 concentrations on course sand (05 um - 2.0 um) and
 day/silt (< 63 um) on a dry weight basis. The data in-
 dicates that the PAH  concentration is higher in the
 course sand fraction of sediment. Recall from Figure
 18 that the day/silt and course sand fractions contain
 higher fraction organic carbon content. The bottom
 panels of Figure 19 present the organic carbon nor-
 malized  comparison of PAH concentrations by sedi-
 ment  class.  For both panels,  the organic carbon
 normalized  PAH concentrations are similar regard-
 less  of the  sediment size  class  as predicted  by
 Equation 22.
       Lastly, Delbeke et al. [46] collected sediments
 from seven sites in the Belgian continental shelf and
 the Scheldt  estuary. These sites were analyzed for
 eight PCB congeners and organic carbon in the bulk
 sediment and clay/silt (< 63 um) sediment fraction.
 In addition, analyses of .the samples were done to de-
 termine the percent of size  fractions ranging from
 500 um to 3 um which made up the sample. The PCB
 congeners tested for in this study were IUPAC num-
 bers 28,52,101,118,153,138,170 and 180.
       Using concentrations  reported for bulk sam-
 ples, concentrations for clay/silt samples, and per-
 cent size fractions of each sample, calculations were
 done to estimate concentrations on the greater than
 63 um portion of the sample. Similar calculations
 were done to determine organic carbon content on
 the >63 um portion of the sample. Organic content
varied from 0.01 percent to  10 percent inclusive of
both <63 uin and >63  um portions of the sediment.

-------
            • ,'f* '.. '- -j -S>*
                1OOO

            a
            »   too
              100000
            8  1OOOO
            X   1OOO
                 100
                   100
                          Sediment Quality Criteria Using Equilibrium Partitioning
                                         Dry  Weight Normalization
                         Send vs Cliy/SHt
                            Low Density vs Clay/Silt
                                                       10000
                                                     f
                     1OOO
                            10      100      1000
                           PAH 
c
Dz
e



t
ii
^
'/
i>

ii
/
*
El
• U










Sta
Sta










H
K





-
/



.











1

















r
t
t
t
j
t
t 1
/ 1
fl
1





"

Hx
P
wfc
mfc







-

x
/



















^
/
^
^
^
f



7
^
/
/
/
/
/
/

/
/
/
/





7

^
/
^














1


a
F.

^
2
/
?
j;
^


7



'
/
/

'

,
5
_

-

—
-
—




                         63 urn
Il3-126um  t2S-250um  26O-600W1  O6-1mm
                                                      SIZE CLASS
Figure 18.—The organic carbon fractions (% dry weight) in five sediment size classes ranging from clay and silt (<63 um) to
course sand (1.0 to 2.0 mm). Stations are indicated by hatch type. Data from [45].

-------
                          Sediment Quality Criteria Using Equilibrium Partitioning
                                     Dry  Weight Normalization
           1000
         p 100
         *  «
         I,
                     Sand va Clay/Sit
         i
            0.1
                                        1000

                                     a wo

                                     *  »
                                     i
                      110     wo
                     PAH (uo/Q dry wt)
                        woo
                                                          0.1
                                              Course Sand va Clay/Sit
                                          0.1      1      w     wo
                                                 PAH (ua/g dry wt)
                                 WOO
          WOOD


         ^ WOO


            wo
        I
                                    Organic Carbon Normalization
                     Sand va Clay/Sit                  	  Court* Sand v« day/Sit
O63-125un
A125-250um
V250-500um
  WOO
g
I «°
I"
       I  O0.5-1.0mm
         OlQ-2.0mm
                     w     wo    woo
                      PAH (ue/g OC)
                                                 W     WO    WOO

                                                  PAH Cuo/g OC)
                                WOOD
Figure 19.—Comparison of PAH concentrations of the sand-sized and course sand-sized fraction sediment particles.indi-
cated by symbols, ordinate, to the clay/silt fraction, abscissa (Stations C, Da, G, H, K). Top panels are for dry weight normali-
zation; bottom panels are for organic carbon normalization. Data from [45].
Figure 20 presents the percent organic carbon on the
<63 um portion of the sample (filled bar) and on the
>63 um portion of the sample (shaded bar). A com-
parison of the PCB congener concentrations on a dry
weight basis (top) and on an organic carbon basis
(bottom) is shown in Figure 21. Organic carbon con-
tent in the >63 um class size at stations 2 and 4 is 0.01
percent and 0.06 percent respectively, as indicated in
Figure 20. The data for these stations are shown on
Figure 21 using filled symbols. Though an foc > 02
percent has been presented as the value.for which or-
ganic carbon normalization applies, normalization at
these/oc values seems appropriate for this data set.
       The top panel of Figure 21 indicates no evi-
dent relationship between PCBs in the <63 um sam-
pl   -id PCBs in the >63 um sample on a dry weight
b.    When concentrations in  either class size are
nc,  alized to organic carbon content then the con-
centrations are similar for both class  sizes as shown
in the bottom panel. This indicates that PCB concen-
trations are similar across sediment class sizes which
supports organic carbon normalization.
                                                                 Organic  Carbon  Fractions
                                        wo
                                      O001
                                                          STATION
                                   Figure 20.—The organic carbon fractions (% dry weight) in
                                   two sediment size classes, <63 urn and >63 umi Seven
                                   stations are indicated. Data from [46].

-------
                          Sediment Quality Criteria Using Equilibrium Partitioning
         Dry Weight Normalized
 O001
    OOO1    O01    0.1      1      10

            PCB (ng/g}<63 urn
100
      Organic Carbon  Normalized
tcooo
                                        A
                                        •
                                        V
                                        •
                                        D
                                        O
                                     i   O -
         Sta 1
         Sta 2
         Sta 3
         Sta 4
         Sta 5
         Sta 6
         Sta 7
                             WOO   10000
         PCB (ng/o OCX 63  urn
Figure 21.—Comparison of eight PCB congener concentra-
tions of >63 urn sized particles, ordinate, to <63 urn sized
particles,. abscissa (Stations  1-7). Top panel  is for dry
weight normalization;  bottom panel is for  organic carbon
normalization. PCB congeners are IUPAC Nos 28, 52,101,
118,138,. 153,179 and 180. Data from [47].
       It can be concluded from the data of Prahl,
Evans et al., and Delbeke et al., that the organic car-
bon-normalized  PAH and PCB concentrations are
relatively independent of particle size class and that
organic carbon is the predominant controlling factor
in determining the partition coefficient of the differ-
ent sediment size particles in a sediment sample. The
organic carbon  concentration of the high-density
sand-sized fraction in Prahl's data (02 to 0.3 percent)
suggests  that organic carbon normalization is appro-
priate at  these low levels. The data from Evans et al.
suggests  that EqP can be applied to organic carbon
originating from more than one source, that is, frag-
mentary plant matter and aging humic: material.
                                                            Sediment/pore water partitioning. Normally
                                                     when measurements of sediment chemical concen-
                                                     tration, Cs, and total pore water chemical concentra-
                                                     tions, Cpore, are made, the value of the  apparent
                                                     partition coefficient is calculated directly from the ra-
                                                     tio  of these quantities. As a consequence of DOC
                                                     complexation, the apparent partition coefficient, Kp,
                                                     defined as
                                                                                                 (23)
                                                     is given by
                                               /ocKoc
                                            (24)
                                                                                1 •*• W
                As mooc increases, the quantity of DOC-complexed
                chemical increases and the apparent partition coeffi-
                cient approaches
                        /ooKoc
(25)
 which is just the ratio of sorbed to complexed chemi-
 cal. Because the solid-phase chemical concentration
 is proportional to the free dissolved portion of the
 pore water concentration, Cd, the actual partition co-
 efficient, Kp, should be calculated using the free dis-
 solved concentration. The free dissolved concentration
 will typically be lower than the total dissolved pore
 water chemical concentration in the presence of signifi-
 cant levels of pore water DOC (e.g., Fig. 13). As a result,
 the actual partition coefficient calculated with the free
 dissolved  concentration is  higher  than the apparent
 partition coefficient calculated with the total dissolved
 pore water concentration.
       Direct observations of pore water partition
 coefficients are restricted to the apparent partition co-
 efficient, Kp (Eqn. 23), because total concentrations
 in the pore water are reported and DOC complexing
 is expected to be significant at the DOC concentra-
 tions  found in  pore  waters. Data  reported by
 Brownawell and Farrington in 1986 [48] demonstrate
 the importance of DOC complexing in pore water.
 Figure 22 presents the apparent partition coefficient,
 measured for 10 PCB congeners at various depths in
 a sediment core, versus/oc Kow, the calculated parti-
 tion coefficient. The line corresponds to the relation-
 ship, Koc = Kow, which is the expected result if DOC
 complexing were not significant. Because DOC con-
 centrations were measured for these data, it is possi-
ble to estimate Cd with Equation 20 in the form:
                               Cd-
                                        -pore
                                           (26)
                                   1 +

-------
                          Sediment Quality Criteria Using Equilibrium Partitioning
 O
 \


 I
 c
 O
 O
 O
 CL
 IB
 O
               •  Apparent K'p
               e  Actual Kp
         4.0
5.O      6.0       7.0

    Log 10 foe  Kow
Figure 22.—Observed partition coefficient versus the prod-
uct of organic carbon fraction and octanol/water partition
coefficient The line represents equality. The partition coef-
ficients  are computed  by using total dissolved  PCB
(squares), and free PCB (circles) which  is computed with
Equation 26 with Kooc - Kow. Data from [48].
and to compute the actual partition coefficient: Kp =
Cs/Cd. The data indicate that if KDOC =? Kow is used,
the results, shown on Figure 22, agree with the ex-
pected partition equation, namely mat Kp = foe Kow.
A similar three-phase model  was  presented  by
Brownawell and Farrington in 1984 [49].
       Other data with sediment/pore water parti-
tion coefficients for which the DOC  concentrations
have not been reported [50,51] are available to assess
the significance of DOC partitioning on the apparent.
sediment  partition coefficient.  Figure 23 presents
these.apparent organic carbon-normalized partition
coefficients, that is KOC = Kp /-foe versus KOW The ex-
pected relationship for DOC concentrations of 0, 1,
10, and 100 mg/L is  also shown. Although there is
substantial scatter in  these data, reflecting the diffi-
culty of measuring pore water concentrations, the data
conform to DOC levels of 1.0 to 10 mg/L, which is well
within the observed range for pore waters [40,48].
Thus, these results do not refute the hypothesis that KOC
- Kow in sediments but show the need to account for
DOC complexing in  the analysis  of pore water
chemical concentrations.
      Laboratory toxicity tests. Another way to
verify Equation 22 is from data collected during sedi-
ment toxicity tests in the laboratory. These tests yield
sediment  (Cs/oc) and pore water (Cd) chemical 'con-
centrations at several dosages bounding an experi-
mentally estimated toxic  concentration for the test
organism. The organic content of the sediment must be
measured  also. Sediment toxicity tests are done under
quiescent conditions and sediment and pore water are
in equilibrium. The results of these tests can be used to
                  o
                  o
                  O»
                  X
                  O
                  O
                 o
                 I
                                   I	1

                              A  diver (V«tetis)

                              B  Spch. (PAH*)
                                                         100.0
                                                                         8
                                            Log 10 Kow
Rgufe 23.—Observed apparent partition coefficient to organic carbon versus the octanol/water partition coefficient. The
lines represent the expected relationship for DOC concentrations of 0,1,10, and 100 mg/L KDOC = AW Data from [51] for
PCB congeners and other chemicals and from [50] for phenanthrene, fluoranthene, and perylene.

-------
                          Sediment Quality Criteria Using Equilibrium Partitioning
 compute the organic carbon partition coefficient Koc.
 To verify Equation 22, estimates of Koc computed
 from Equation 11 using laboratory measurements of
 Kow are then compared to partitioning in the sedi-
 ment  toxicity test. Sediment toxicity tests and Kow
' measurements are available for five chemicals: en-
 drin  [20,21,52], dieldrin {54,55], acemaphthene [56],
 phenanthrene [56], and fluoranthene [19,57]. Sediment
 toxicity tests for these chemicals were performed as
 part of the development of SQC. Mortality results for
 these tests were presented in Figures 2 and 3..
       Figure 24 shows organic carbon normalized
 sorption isotherm for acenaphthene, endrin, phenan-
 threne and fluoranthene, where the sediment concen-
 tration (ug/g OC)  is  plotted versus pore  water
 concentration (ug/L). These  tests represent  fresh-
 water and marine sediments having a range of or-
 ganic carbon content of 0.07 to 11.0 percent. In each

                 ACENAPHTHENE
     10000
100

 «

  1
                        O - Swortz et oL. 1991 -
                                                    panel, the line corresponds to Equation 16 where Koc
                                                    is derived from Kow measurements in the laboratory.
                                                    A full discussion of laboratory K0w measurements is
                                                    presented subsequently. In each of the panels the toxic-
                                                    ity test data are in agreement with the line computed
                                                    from experimentally determined Koc. For these chemi-
                                                    cals DOC measurements are unavailable and partition-
                                                    ing to DOC in the pore water has not been considered.
                                                    The figure indicates, however, that DOC complexing in
                                                    these experiments appears to be negligible.
                                                           Partitioning in the dieldrin experiment indi-
                                                    cated-that DOC complexation may have been signifi-
                                                    cant. The partitioning isotherm for dieldrin, Figure
                                                    25),  represents organic carbon normalized sediment
                                                    concentration, versus total (top panel) and computed
                                                    dissolved (bottom  panel) pore water concentrations.
                                                    Dissolved pore water concentrations are  computed
                                                     100000
                                                                  PHENANTHRENE
                  100     ^000    10000    100000
                                                                         O - Swortz et ol. 1991 =
                                                                                          1QOOO
     10000
      1000
       100
                       ENDRIN
                                                                  FLUORANTHENE
8f    «
        0.1
                      O- Nebeker et oL, 1989
                      P- Schuytemo et d., 1989 :
                      V— SteMy,-1991         I
              i 11 mil  ill tiitil i i i inn!  i i i nmil ' i i i tun
         001     0.1
                             DOO    1000
                                                                         O- Swortz et ol., 1990
                                                                         O - De Witt et ol., 1992
                                                              t i i mid  i t i unit lit tnn  i i i Hud  til 11 in
           PORE WATER CONCENTRATION
                         (ug/L)
                                                                               100    1000  1OOOO
                                                            PORE WATER CONCENTRATION
                                                                          (ug/L)
 Figure 24.—Comparison of organic carbon partition coefficient (Koc) observed in toxicity tests (symbols) to Koc derived from
 laboratory Kow and Equation 11 (solid line). Symbols are sediment concentration, ordinate, versus pore water concentration,
 abscissa. Solid line is Cs.oc = Koc * Ca, where Logio Koc is 3.76 for acenaphthene, 4.84 for endrin, 4.46 for phenanthrene,
 and"5.00 for fluoranthene. These Logio Koc values are estimated from Logic KJW values measured at the U.S. EPA Environ-
 mental Research Laboratory at Athens, Georgia. Data sources as indicated.

-------
                             Sediment Quality Criteria Using Equilibrium Partitioning
                       DIELDRIN
             t I Hfltl  I I I IIIMt  I I I HUM  f I Illlltl  I  I I Illl
     TOTAL PORE WATER  CONCENTRATION (uQ/D
    10000
      "
_<     «
      at
         It I  1 1 1111(1
                     111 Illl  I I rii I ill  I ii 11 mi  I  iLJ*nii
                                  I  III ttllt  I III Mil
       am
                at
  CONFUTED PORE WATER CONCENTRATION (ug/U

   Figure 25. — Comparison of organic carbon partition coeffi-
   cient (Koc) observed in toxicity tests (symbols) to Koc de-
   rived from laboratory Kow and Equation 11 (solid line).
   Symbols are sediment concentration, ordinate, versus total
   (top panel) and free (bottom  panel) pore water concentra-
   tions, abscissa. Solid line is Cs,oe ~ Koc* Ca, where Logio
   Koc is 5.25 for dieldrin. The Logio Koc value is estimated
   from  Logio KOM value measured at the U.S. EPA Environ-
   mental Research  Laboratory at  Athens,  Georgia. Data
   source as indicated.
   using Equation 26, DOC measurements and an esti-
   mated log KDOC s 5.25. Log KDOC is estimated from
   log Koc - 525 for dieldrin. Figure 25 represents data
   from Hoke and Aukley [55] since Hoke [56] did not
   measure pore water. Adjusting for partitioning on to
   the DOC in the bottom panel results in better agree-
   ment with the experimentally determined Koc. These
   data represent one sediment with an organic carbon
content of 1.6 percent. It is important to note that
dieldrin has the highest Koc of the five chemicals
(LogioKoc dieldrin - 525, acenaphthene = 3.76, en-
drin = 4.84, phenanthrene = 4.46, fluoranthene = 5.00.
DOC complexing increases with an increasing parti-
tion coefficient which explains why DOC complexing
is significant for dieldrin.


Organic Carbon Normalization of
Biological Responses

The results discussed above suggest that if a concen-
tration-response curve correlates to pore water con-
centration, it should correlate equally well to organic
carbon-normalized total chemical concentration, in-
dependent of sediment properties. This is based on
the partitioning formula Cs,oc  = Ko<£d  (Eqn.16),
which relates the free dissolved concentration to the
organic carbon-normalized  particle concentration.
This applies only to nonionic hydrophobic organic
chemicals because the rationale is based on a parti-
tioning theory for this class of chemicals.
      Toxicity andbioaccumulation experiments. To
demonstrate this   relationship,  concentration-re-
sponse curves for the data presented in Figures 5 and
7 are used to compare results on a pore water-nor-
malized  and organic carbon-normalized  chemical
concentration basis. Figures 26 to 28 present these
comparisons for kepone, DDT, endrin, and fluoran-
thene. The mean and 95  percent confidence limits of
the LCso and ECso values for each set of data are
listed in Table 2. The top panels repeat the response-
pore water concentration plots shown previously in
Figures 5 to 7, while the lower panels present the re-
sponse versus the sediment concentration, which is
organic carbon-normalized (microgram chemical per
gram organic carbon).
      The general impression of these data is that
there is no reason to prefer pore water normalization
over  sediment  organic  carbon  normalization. In
some cases, pore water normalization is superior to
organic carbon normalization, for example, kepone-
mortality data (Fig. 26) whereas the converse some-
times occurs, for example kepone-growth rate  (Fig.
26). A more quantitative comparison can be made
with the LCsos and ECsos in Table 2. The variation of
organic carbon-normalized LCsos and ECsos between
sediments is less than a factor of two to three and is
comparable to the variation in pore water LCsos and
ECsos. A more comprehensive comparison has been
presented in Figures 2 and 3, which also examine the
use of the water-only LCso to predict the pore water
and sediment organic carbon LCsos.
      Bioaccumulation factors calculated on the basis
of organic carbon-normalized chemical concentrations
are listed in Table 3, for permethrin, cypermethrin, and

-------
                            Sediment Quality Criteria Using Equilibrium Partitioning
                100
             S  80
             ]f  •«>
             5  40
             i  20
                  0
                                      Pore Water  Normalization
                         Kapone - Mortality                         Kopon. - Growth
100
 80
 eo
 40
 20
  0
                           10       10O      1000
                        Water Concentration (ug/L)
   1        10       100     1OOO
    •Per* Water Concentration (ug/U
                                   Organic  Carbon Normalization
                        Kcpoiw - Mortality                         Kvpom - Growth


_

1




100
80

80
40
20
0
«oc(X)
' a O.O8 - _ A ./• — • •
• » 1.6 ' ' 1
e 12 ' ' 1

*


**

I
I

1O 100 1OOO 100OO
Organic Carbon Normalized (ug/g oc)

100
80

eo
40
20
0

M m m •
ff^

.'"''/







10 100 1000 10000
Organic Carbon Normalized (ug/g oc)
Figure 26.—Comparison of percent survival (left) and growth rate reduction (right) of C. tentans to kepone concentration in
pore water (top) and in bulk sediment, using organic carbon normalization (bottom) for three sediments with varying
carbon concentrations [17].                                                                          -e
                                     Pore Water Normalization
                              DDT
                                                                      Endrln
1OO
- 80
>. eo
I 4°
I .20
0
n<

f
*
•oc |SS)
t-* :«;
* 1OJ
100
~ 8O
1 80
1 40
| 20
o'
tec (X)
. • *.1 • "
e mi
m» *"
                    Pora Water Concentration (UJ/L)
 O.1       1.O       1O.O     100.O
    Pora Water Concentration (ug/L)
                                  Organic  Carbon Normalization
                              DDT


^

k,
••
i
a



100
80

80
4O
20
0


OB- «*• «•
•ll
O
*
• •
•
.* . • * •* »

                                                                      Endrln
                 1      1O     1OO   10OO  1000O
                  Organic Carbon Normalized {ug/g oc)

2
~

s
i

100
8O
eo

20
o
l>«
I

• •
jj
.*•!->• •-
•»••



.
•
  1      1O    1OO   1OOO  10000
  Organic Carbon Normalized (ug/g oc)
Rgure 27.—Comparison of percent survivalof H. azteca to DDT (left) and endrin (right) concentration in pore water (top) and
in bulk sediment, using organic carbon normalization (bottom) for three sediments with varying organic carbon concentra-
iions j

-------
                         Sediment Quality Criteria Using Equilibrium Partitioning
        Por« Wafer Normalization

                 FkK>r*nth«n« >
                = free dissolved chemical concentration
         o     20    40     ao     ao
           Per* Wafer ConoMrtnUon (•o/U

      Organic  Carbon Normalization

                 FluoranUwn*
         0    2000  4000   6000   8OOO
          Orjinlc Carbon Normalized (nf/fl oe)
Figure 28.—Comparison of percent survival of R. abronius
to fluoranthene concentration in pore water (top) and bulk
sediment, using organic carbon normalization (bottom) for
sediments with varying organic carbon concentrations [19].


kepone. Again, the variation of organic carbon nor-
malized BAFs between sediments is less than a factor
of two to three and is comparable to the variation in
pore water BAFs.
       Bioaccumulation and organic carbon nor-
malization. Laboratory and field data also exist for
which  no pore water  or -DOC measurements  are
available but for which sediment concentration, or-
ganic carbon fraction, and organism body burden
have been determined. These data can be used to test
organic carbon normalization for sediments and to
examine organism normalization as well. It is con-
ventional to use organism lipid fraction for this nor-
malization (see references in Chiou [58]). If Cb is the
chemical concentration per unit wet weight of the or-
ganism, then the partitioning equation is
                  Cb =
(27)
where
    KL  s lipid/water partition coefficient (L/kg
          lipid)

    ft  '= weight fraction of lipid (kg lipid /kg
          organism)
         The lipid-normalized organism concentration, Cb,L, is

                        r    Cb                   (28)
                        Q>,L - -z- - Kl-Cd.
                              fi,

         The lipid-normalized body burden and the organic
         carbon-normalized  sediment  concentration can be
         used to compute a bioaccumulation ratio, which can
         be termed the BSF [59]:
                      BSF-
                           Q>,L  KL   KL
                            s,oc
                                      KOW.
                                           (29)
The second equality results from using the partition-
ing Equations 16 and 28 and the third from the ap-
proximation that Koc = KOW. The BSF is the partition
coefficient between organism lipid and sediment or-
ganic carbon. If the equilibrium  assumptions are
valid for both organisms and sediment particles, the
BSF should be independent of both particle and or-
ganism properties. In addition, if lipid solubility of a
chemical is proportional to its octanol solubility, KL «
KOW, then the lipid normalized-organic carbon nor-
malized BSF should be a constant, independent of
particles, organisms, and chemical properties [59,60,
61]. This result can be tested directly.
      The representation of benthic organisms as
passive encapsulations of lipid that equilibrate with
external chemical concentrations is dearly only a first-
order approximation. Biomagnification effects, which
can occur via ingestion of contaminated food and the
dynamics of internal organic carbon metabolism, can
be included in a more comprehensive analysis [59]. It
is, nevertheless, an appropriate initialassumption be-
cause deviations from  the first-order representation
will point to necessary refinements, and for many
purposes this approximation may suffice.
      A comprehensive experiment involving four
benthic organisms—two species of deposit-feeding
marine polychaetes, Nereis and Nephtys, and  two
species of deposit-feeding marine clams, Yoldia and
Macoma—and five sediments has been performed by
Rubinstein et al. [62]. The uptake of various PCB con-
geners was monitored until steady-state body bur-
dens were  reached. Sediment organic carbon  and
organism lipid content were measured. Figures 29
and 30 present the log mean of the replicates for the
ratio of organism-to-sediment concentration for all
measured congeners versus K0w for each organism.
Dry weight normalization for  both organism  and
sediment (left panels), organic carbon normalization
for the sediment (center panels), and both organic

-------
                           Sediment Quality Criteria Using Equilibrium Partitioning
                      Dry Weight
             Ner/os
          Organic Carbon
           o
           i
           u
               3JO
               3.O
               1.0
o
D>
2   o.o
                               foe*(K)
                               4 «£
                               » 3.9 ,
                      6.5    7.5

                      LoglO Kow
     1.0


     0.0
                                                                        Organic Carbon, Lipid
                                      g  -1.0
                                      O

                                      jj  -2.0
                                                           3.0


                                                           2.0
                            I-
                            O
                            ?   0.0
                         8.S
                                                                     -i.<
                                                  Logic Kow
                                        8.5    7.6
                                        LogtO Kow
                                                    8.S
                     Dry Weight
            Nepfltys
         Organic Carbon
              3.0


              24


              1.0


              OJ>


             -1.0
    1.0
                                                                        Organic .Carbon, Lipld
8   0.0

5  -1.0
o"
   -2.O
           •.6    7.8     a8
           LoglO Kow
                                                 •JS     7.8
                                                 LoglO Kow
                                                           2.0
                                0.0
                                       U     7.8
                                       LoglO Kow
                                                    8.S
Figure 29.—Plots of the BSF (ratio of organism-to-sediment concentration) for three sediments for a series of PCB congeners
versus the logioKow for that congener. The dry weight normalization for both organism and sediment (left panels); organic
carbon normalization for the sediment (middle panels); and organic and lipid normalization (right panels) as indicated The or-
ganisms are Nereis (top) and Nephtys (bottom). Data from [62].
carbon  and lipid normalization (right panels) are
shown. The results for each sediment eire connected
by lines and separately identified.
       The BSFs based on dry weight normalization
are quite different for each of the sediments with the
low carbon sediment exhibiting the largest values.
Organic carbon normalization markedly reduces the
variability in the BSFs from sediment to sediment
(center panels). Lipid normalization usually further
reduces the variability. Note that the BSFs are reason-
ably constant for the polychaetes, although  some
suppression is evident at logioKow > 7.  The clams,
however, exhibit a marked declining relationship.
       Results of a similar though less extensive ex-
periment using one sediment and oligochaete worms
have been reported [52]. A plot of the organic carbon-
and lipid-normalized BSF versus JCOW  from this ex-
periment is shown on Figure 31, together with the
averaged polychaete data (Fig. 29). There appears to
be a systematic variation with respect to Kow, which
suggests that the simple lipid equilibration model
with a constant lipid-octanol solubility ratio is not
descriptive for all chemicals. This suggests  that a
more detailed model of benthic organism uptake is
required to describe chemical body burdens for all
                nonionic chemicals as a function of KOW [59]. How-
                ever, for a specific chemical and a specific organism,
                for example Nereis and any PCB congener (Fig. 29)
                organic carbon normalization reduces the effect of
                the varying sediments. This demonstrates the utility
                of organic carbon normalization and supports its use
                in generating SQC.
                       A further conclusion can be  reached from
                these results. It has been pointed out by Bierman [63]
                that the lipid- and  carbon-normalized BSF is in the
                range of 0.1 to 10 (Figs. 29 to 31) supports the conten-
                tion that the partition coefficient for sediments is Koc
                = Kow and  that the particle concentration effect does
                not appear to be affecting the free concentration in
                sediment pore water. The reason is that the lipid- and
                carbon-normalized BSF is the ratio of the solubilities
                of the chemical in lipid and in particle carbon (Eqn.
                29). Because the solubility of nonionic organic chemi-
                cals in various nonpolar solvents is similar [64], it
                would be expected that the lipid-organic carbon solu-
                bility ratio should be on the order of one. If this ratio
                is taken to be approximately one, then the conclusion
                from the BSF data is that Koc is approximately equal
                to Kow for sediments [63].

-------
1 30
9.0
' ! •"
o
«•
3 ox>
Sediment Quality Criteria Using Equilibrium Partitioning \ 1
Dry Weight .
' locifc)
* 82
• 3X1 .
• 1X1
1 6.6 6.3 7.6 ft,
LoglO Kow
Dry Weight
8 "
3 -ix>
-W,
feo(X)
1.0
o O.O
£
8 -ix)
o
S -2.0
s *•%.
1X>
8 O.O
f ,
g -ix)
e
^ -2.O
M»S ft* 7JS &£ ""s.
LoglO Kow
K^A//»
Organic Carbon
^;
* 9JS 7JS 8
LoglO Kow
Macoma
Organic Carbon
•Un
fc^.
Logic Kow
3.O
o
9. 2.0
0
* 1-%
(
2X>
I "
g 0.0
e
| -IX,
i -*°.
Organic Carbon, Llpld
•
.^^.

a ex* us 84
Logic Kow
Organic Carbon, Llpld
Tfe^Y-
1 U 7.B •.
LoglO Kow
t
Rgure 30. — Plots of the BSF (ratio of organism-to-sediment concentration) for three sediments for a series of PCB congeners
versus the logioKiw for that congener. The dry weight normalization for both organism and sediment (left panels); organic
carbon normalization for the sediment (middle panels); and organic and lipid normalization (right panels) as indicated. The or-
ganisms are Yoldia (top) and Macoma (bottom). Data from [62].
                                  Oligochaete - Polychaete BSF
                      100.O
                   ^  10.0
                   o
                   3

                   e   1>0
                   O
                   ja
                   u
                        0.1
                                 • CHIgoehaoto

                                 • Polychwt*
3.0      4.0     5.0     6.0      7.0     8.0

                     LoglO Kow
                                                                               9.0
Figure 31.—Plots of the BSF (ratio of organism lipid to sediment organic carbon concentration) for a series of PCB congeners
and other chemicals versus logio/W Data for oligochaetes [51] and polychaetes [62].

-------
                         Sediment Quality Criteria Using Equilibrium Partitioning
    ..  A final observation can be made. The data
 analyzed in-this section demonstrate that  organic
 carbon, normalization accounts 'for much of the re-
 ported differences in bioavailability of chemicals in
 sediments for deposit-feeding polychaetes, oligo-
' chaetes, and clams. The data presented in previous
 sections are for amphipods and midges. Hence these
 data  provide important additional siupport for  or-
 ganic carbon normalization  as a determinant of
 bioavailability for different classes of organisms.


 Determination of the Route of Exposure

 The exposure route by which organic chemicals are
 accumulated has been  examined in some detail  for
 water column organisms (e.g., by Thomann and Con-
 nolly [65]). It might be supposed that the toxicity and
 bioaccumulation data presented above can be used to
 examine the question of the route of exposure. The in-
 itial observations were  that biological effects appear
 to correlate to the interstitial water concentration, in-
 dependent of sediment type.  This has been inter-
 preted to mean that exposure  is primarily via pore
 water. However, the data correlate equally well with
 the organic carbon-normalized sediment concentra-
 tion (see Figs. 2 and 3). This observation suggests that
 sediment organic carbon is the route of exposure. In
 fact, neither conclusion follows necessarily from these
 data because an alternate explanation is available that
 is independent of the exposure pathwciy.
       Consider the hypothesis that the chemical po-
 tential or, as it is sometimes called, the fugacity [66]
 of a  chemical controls its biological  activity. The
 chemical potential, ud, of  the free concentration of
 chemical in pore water,  Cd, is
                                           (30)
where u,o is the standard state chemical potential and
RT is the product of the universal gas constant and
absolute temperature [67]. For a chemical dissolved
in organic carbon—assuming that particle organic
carbon can be  characterized as a  homogeneous
phase—its chemical potential is
                 -Ho + RTln(Cs,oc)
(31)
 where Cs,oc is the weight fraction of chemical in or-
 ganic carbon. If the pore water is in equilibrium with
 the sediment organic carbon then
                    ud-Hoc.
(32)
 The chemical potential that the organism experiences
 from either route of exposure  (pore water or sedi-
 ment) is the same. Hence, so long as the sediment is
 in equilibrium with the pore water, the route of expo-
sure is immaterial. Equilibrium experiments cannot
distinguish between different routes of exposure.
       The data analysis presented above, which
normalizes biological response to either pore water
or organic carbon-normalized sediment concentra-
tion, suggests that biological effects are proportional
to chemical potential or fugacity.
       The issue with~respect to  bioavailability is
this: In which phase is u, most  easily and reliably
measured? Pore water concentration is one option.
       However, it is necessary that the  chemical
complexed to DOC be a small fraction of the total
measured concentration or that the free concentra-
tion be directly measured, perhaps by the cis column
technique [42]. Total sediment concentration normal-
ized by sediment organic carbon fraction is a second
option. This measurement is not affected  by DOC
complexing. The only requirement is that sediment.
organic carbon be the only sediment phase  that con-
tains significant amounts of the  chemical.  This ap-
pears to be a reasonable assumption for most aquatic
sediments. Hence, SQC are based on organic carbon
normalization because pore water normalization is
complicated by DOC complexing for highly hydro-
phobic chemicals.


APPLICABILITY OF WQC AS

THE  EFFECTS  LEVELS FOR

BENTHIC  ORGANISMS

The EqP method for deriving SQC utilizes partition-
ing theory to relate the sediment concentration to the
equivalent free chemical concentration in pore water
and in sediment organic carbon. The pore water con-
centration for SQC should be the effects concentra-
tion for benthic species.
      This section examines the validity  of using
the  EPA WQC concentrations  to define the effects
concentration for benthic organisms. This use of
WQC assumes that (a) the  sensitivities of benthic
species and species tested to derive WQC, predomi-
nantly water column species, are similar, (b) the lev-
els of protection afforded by WQC are appropriate
for benthic organisms, and c) exposures are similar
regardless of feeding type or habitat. This section ex-
amines the assumption of similarity of sensitivity in
two ways. First, a comparative toxicological exami-
nation of the acute sensitivities of benthic and water
column species, using data compiled from the pub-
lished EPA WQC for nonionic  organic chemicals as
well as metals and ionic organic chemicals, is pre-
sented. Then a comparison of the FCVs  and the
chronic sensitivities of benthic saltwater species in a
series of sediment colonization experiments is made.

-------
                          Sediment Quality Criteria Using Equilibrium Partitioning
       •J!'
 Method-Relative Acute Sensitivity

 The relative acute sensitivities of benthic and water
 column species are examined by using LCsos for
 freshwater and saltwater species from draft or pub-
• lished WQC documents that contain minimum data-
 base requirements  for calculation  of  final acute
 values (Table 4). These data sets are selected because
 exposures -.were via water, durations were similar,
 and data and test conditions have been scrutinized
 by reviewing the original references.  For each of the
 2,887 tests conducted in fresh water, using 208 spe-
 cies with 40 chemicals, and the 1,046 tests conducted
 in salt water, using 118 species with 30 chemicals, the
 chemical, species, life stage, salinity,  hardness, tem-
 perature, pH, acute value, and test  condition (i.e.,
 static, renewal, flow-through, nominal, or measured)
 were entered into a database. If necessary, original
 references were consulted to determine the tested life
 stage and any other missing information. Each life
 stage of the tested species was classified according to
 habitat (Table 5). Habitats were based on degree of
 association with sediment. A life stage that occupied
 more than one habitat was assigned to both of the ap-
 propriate habitats.
       For  each chemical, if a life stage was tested
 more  than once or more than one  life stage was
 tested, data were systematically sorted in a three-step
 process to arrive at the acute value based on the most
 experimentally sound testing methodology and the
 most sensitive life stages. First, if a  life stage for a
 species was tested  more than once, flow-through
 tests with measured concentrations had precedence,
 and data from other tests were omitted. When there
 were no  flow-through tests with measured concen-
 trations, all acute values for that life stage were given
 equal weight. If the remaining acute values for that
 life stage differed by greater than a factor of four, the
 higher values were omitted and the geometric mean
 of the lower acute values was calculated to derive the
 acute value for that life stage. Second, life stages were
 classified as either "benthic" (infaunal species [habi-
 tats 1 and 2] or infaunal and epibenthic species [habi-
 tats 1,2,3, and 4]), or "water column" (habitats 5 to
 8). Third, if two or more life stages were classified as
 either benthic or water column and their acute values
 differed by a factor of four, the higher values were
 omitted and the geometric mean of the lower acute
 values was calculated to derive the  acute value  for
 that life stage of the benthic or water column species.
 This procedure is similar to that used for WQC [8].

 Comparison of the Sensitivity of
 Benthic and Water Column Species

 Most Sensitive Species. The relative acute sensitivi-
 ties of the most sensitive benthic and water column
 species were examined by comparing the final acute
values (FAV) for benthic and water column organ-
isms,  using acute LCso concentrations from the 40
freshwater and the 30 saltwater WQC documents.
When benthic species were defined as only infaunal
organisms (habitat types 1 and 2) and water column
species were defined as all others (habitat types 3 to
8), the water column species were typically the most
sensitive. The results are cross-plotted  on  Figure 32
(top). The line represents perfect agreement.
      Data on the sensitivities of benthic infaunal
species are limited. Of the 40  chemicals for which
WQC for freshwater organisms are available, two or
fewer infaunal species were tested with 28 (70 per-
cent) of the chemicals, and five or fewer species were
tested with 34 (85 percent) of the chemicals. Of the 30
chemicals for which WQC for saltwater organisms
are available, 2 or fewer infaunal species were tested
with 19 (63 percent) of the chemicals, and 5 or fewer
species were tested with 23 (77 percent) of the chemi-
cals. Of these chemicals only zinc  in salt water has
been tested using infaunal species from three or more
phyla and eight or more families, the minimum acute
toxitity database required for criteria derivation. As a
result, FAVs could not be computed for several of the
chemicals. Therefore, it is probably  premature to con-
clude from the existing data that infaunal species are
more tolerant than water column species.
       A similar examination  of the most sensitive
benthic and water column species,  where the defini-
tion of benthic includes both infaunal and epibenthic
species (habitat types 1 to 4), is based on more data
and suggests a similarity in sensitivity (Fig. 32, bot-
tom). In this comparison, the number of acute values
for freshwater benthic species for each chemical aver-
aged nine, with a range of 2  to 27; the number of
acute  values for saltwater benthic species for each
chemical substance averaged 11, with a range of 4 to
26. The variability of these data is high, suggesting
that for some chemicals, benthic and water column
species may differ in sensitivity and that additional
testing is desirable, or that this approach to examin-
ing species sensitivity is not sufficiently rigorous.
       Examination of individual criteria documents
in which benthic species were markedly less sensi-
tive than water column species suggests that the ma-
jor factor for this difference is that benthic species
phylogenetically related to sensitive water column
species have not been tested. Apparent  differences in
sensitivity, therefore, may reflect an absence of ap-
propriate data. Data that are available suggest that,
on the average, benthic and water column species are
similarly sensitive and support the use of WQC to
derive SQCfor the protection of infaunal and epiben-
thic species.
       All species. A more general  comparison of the
species sensitivities can be made if all the  LCso data
are used. One approach examines the  relative loca-

-------
                            Sediment Quality Criteria Using Equilibrium Partitioning
Table 4.—Draft of published WQC documents and number of infaunal (habitats 1 and 2)  eoibenthic (habitat-? i ann A\
and water column (habitats 5-8) species tested acutely for each substance             ep'bentn,c (habitats 3 and 4),

CHEMICAL
ACENAPTHENE
ACROLEIN
ADRIN
ALUMINUM
AMMONIA
ANTIMONY (III)
ARSENIC (III)
CADMIUM
CHLORDANE
CHLORINE
CHLORPYRIFOS
CHROMIUM (III)
CHROMIUM (VI)
COPPER
CYANIDE
DDT
DIELDRIN
2,4,-DIMETHYLPHENOL
ENDOSULFAN
ENDRIN
HEPTACHLOR
HEXACHLOROCYCLOHEXANE
.LEAD
MERCURY
NICKEL
PARATHION
PARATHION, METHYL-
PENTACHLOROPHENOL
PHENANTHRENE
PHENOL
SELENIUM (IV)
SELENIUM (VI)
SILVER
THALLIUM
TOXAPHENE
TRIBUTYLTIN
1,2,4-TRICHLOROBENZENE
2,4,5-TRlCHLOROPHENOL
ZINC

DATE OF
PUBUCATIO
9/87"
9/87"
1980
1988
1985:1989
9/87"
1985
1985
1980
1985
1986
1984
1985
1985
1985
1980
1980
6/88b
1980
1980
1980
1980
1985
1985
1986
1986
10/88"
-1986
9/87b
5/88"
1987
1987
9/87*
11/88"
1986
9/87"
9/88"
9/87"
1987
NO.OFS ITWATEH SPECIES NO. OF FRESHWATPP «Fr F*
TOTAL"
mH^H
10
. -
16
-
20
11
12
38
8
23
15
-
23
25
9
17
21
9
12
21
19
19
13
33
23
-
-
19
10
-
16
-
21
-
15
19
15
11
33
| INFAUNA
^^••••H
-
0
-
2
3
2
10
1
2
2
-
8
6
1
1
1
2
2
1
1 •
2
2
10
7
-
-
7
4
-
1
-
1
-
2
1
7
4
10
EPIBENTHIC
™^™^^^— ^
3
-
11
-
- 7
6
3
18
7
9
8
-
9
5
4
11
15
2
8
14
14
14
3
7
10
-
-
7
6
-
5
—
6
-
9
8
7
5
9
WATER
COLUM
^••i^MH
7
_
12
-
16
5
8
18
7
15
10
-
9
18
5
12
15
6
8
16
13
12
10
18
9
_

11
4
-
13
_
16
_
11
15
4
5
17
TOTAL
^^••Mi
12
21
15
48
9
16
56
14
33
18
17
33
57
17
42
19
12
10
28
18
22
14
30
21
37
36
41
9
32
23
12
19
8
37
9
14
10
45
" The total numbers of tested species may not be the same as the sum of the number of spec
cies may occupy more than one habitat.
b Draft aquatic life criteria document, U.S. Environmental Protection Agency. Office of Water Re
Standards division, Washington, D.C.
INFAUNAL

1
2

2
1
1
13
1
1
2
3
1
8
1
3
1
1
1
3

1

11

7
1
9
2
6
2
1
1
1
5
1
2
1
5
^"^"•^••i^™™^
ies from each habitat
gulations and Standa
EPIBENTHI


10

17

6
16
4
9
8

10
15
6
15
9
3
4
12
8
4
4
8

14
9
11
1
9
6
4
9
1
13
1


12
i^^n^K^^
type becaus
rds. Criteria
WATER
COLUMN

7
15
11
33

13
31
10


19

36

29

7

17
12
18
11
12
13
23

23
6
20
19
10
13
3
23
6

8
30
•MH^HM
e a spe-
and

-------
                           Sediment Quality Criteria Using Equilibrium Partitioning
                       INFAUNAL
             FRESH WATER
             SALT WATER
   -2-1012346

LOQ-K) MFAUHAL ORQAMSM FAV (ug/1)


               BENTHIC
          A  FRESH WATER
          •  SALT WATER
J3-3-2    -101     2346

         LOG*)  BENTHC ORQAMSM FAV (ug/0

figure 32.—Comparison of LCso or ECso acute values for
the most sensitive benthic (abscissa) and water column
(ordinate) species for chemicals listed in Table 5. Benthic
species are defined as infaunal  species (habitat types  1
and 2, left panel) or infaunal and epibenthic species (habi-
tat types 1-4); see Table 6.


tion of benthic species in the overall species sensitiv-
ity distribution. For each chemical in either  fresh or
salt water, one can examine the distribution of ben-
thic  species in a rank-ordering of all the  species'
LCsos. If benthic species were relatively insensitive,
then they would predominate in ranking among the
larger LCso concentrations. Equal sensitivity would
be indicated by a  uniform  distribution of species
within the overall ranking. Figure 33 presents the re-
sults for tests of nickel in  salt water. The LCsos are
plotted in rank order, and the benthic species are in-
dicated. Infaunal species are among the most tolerant
(left panel), whereas infaunal and epibenthic species
are uniformly  distributed  among the species (right
panel).
                                                         Table 5.—Habitat classification system for life stages of
                                                         organisms.
                                                         HABITAT
                                                          TYPE
                                                                                DESCRIPTION
                                                                 Life stages that usually live in the sediment and whose
                                                                 food consists mostly of sediment or organisms living in
                                                                 the sediment: infaunal nonfilter feeders.
                                                                 Life stages that usually live in the sediment and whose
                                                                 food consists mostly of plankton and/or suspended
                                                                 organic matter filtered from the water column: infaunal
                                                                 filter feeders.
                                                                 Life stages that usually live on the surface of sediment
                                                                 and whose food consists mostly of organic matter in
                                                                 sediments and/or organisms living in or on the sediment:
                                                                 epibenthic bottom feeders.
                                                                 Life stages that usually live on the surface of sediment
                                                                 and whose food is mostly from the water column,
                                                                 including suspended detritus, plankton, and larger prey:
                                                                 epibenthic water column-feeders.
5
6
7
8
Life stages that usually live in the water column and
whose food consists mostly of organisms that live on or
in the sediment.
Life stages that usually live in and obtain their food from,
the water column but have slight interaction with
sediment because they occasionally rest or sit on the
sediment and/or occasionally consume organisms that
live in or on the sediment.
Life stages that live in or on such inorganic substrates as
sand, rock, and gravel, but have negligible contact with
sediment containing organic carbon.
Life stages that have negligible interactions with
sediment because they spend essentially all their time in
the water column and rarely consume organisms in direct
contact with the sediment; that is fouling organisms on
pilings, ships, and so on, and zooplankton, pelagic fish,
and so on.
                                                      This comparison can be  done  chemical by
                                               chemical. However, to make the analysis more ro-
                                               bust, the LCso data for each chemical-water type can
                                               be normalized to zero log mean and unit log variance
                                               as follows:
                                                             LCson,ij -
                                                                      log(LC50ij) - u.
(33)
                                                                            Oi
                                               where i indexes the chemical-water type, u,j is the log
                                               mean and si is the log standard deviation, j indexes
                                               the LCsos within the ith class, and LCson,ij is the nor-
                                               malized LCso. This places all the LCsos from each set
                                               of chemical-water type on the same footing. Thus,
                                               the data can now be combined and the uniformity of
                                               representation of benthic species can be examined in
                                               the combined data set.
                                                      The comparison is made in Figure 34. If the
                                               sensitivity of benthic species is not unique, then a
                                               constant  percentage  of benthic species-normalized
                                               LCsos, indicated by the dashed line, should be repre-

-------
                            Sediment Quality Criteria Using Equilibrium Partitioning
                                 Species Sensitivity for Ni in  Sea water
                                 Infaunal
                                                                    Infaunal & Eplbenthlc

Q 100000
\
^ 10000
in
2
1000

100
o.
i — i 	 • — i 	 1 	
; © Watar Column
• • Banthlc © .
®«@
©
A
@®®
0
®
®
o :
3
	 j 	 1 	 1 	 1
0 O.2 0.4 O.6 O.8 1.
1000000
100000
10000
1000

100
0 O.
© Watar Column
- e Banthic •©*'
e****
' .."• ' ]
©
. ©
•• i
©
•






0 0,2 O.4 0.6 0.8 1.0
Rank Rank
Figure 33.—L.CSOS versus rank for nickel in seawater. Infaunal organisms (left) and infaunal and epibenthic (right) are identified by
the solid symbols. The plot illustrates the distribution of benthic organisms in the overall species sensitivity distribution.
                                                     Infaunal
                               Saltwater
                                                           i
                                                               so
                                                               30
                                                               20
1O
                                                                           Freshwater
                      8  16 28 38 48 88 OS 78 88 OS

                       MMH of Parcantlla Rang* {•/.)
     8 IS 28 38 48 88 88 78 88 SB

      Mean of Parcantlla Rang* (5{)

H 80
£ 60
£
5>
o «0
1 20
ffl
O

•
•


1





1





I










%
1








s

•',.











j-
1








I
%•
|;
i



"




n
%
£










>^
i
!



•




i
^
i
1

•

_ .
-
-
                                           Infaunal & Epibenthic
                                                                           Freshwater
                      8 18 28 38 48 65 
-------
                          Sediment Quality Criteria Using Equilibrium Partitioning
  sented in each 10-percentile (decile) interval of data
  for all species. That is, the 10 rectangles in each histo-
  gram should be  identical in 'height. The  infaunal
  species (top panel) display a tendency to be under-
  represented in the lowest deciles. However, the in-
*  faunal and epibenthic species (bottom panels) more
  closely follow this  idealized  distribution.  Infaunal
  and epibenthic freshwater species are nearly uni-
  formly distributed, whereas the saltwater benthic
  species  are  somewhat  underrepresented  in  the
  lowest ranks.
        Given the  limitations of these data,  they ap-
  pear to indicate that, except for possibly freshwater
  infaunal species,  benthic species are not uniquely
  sensitive or insensitive and that SQC derived by us-
  ing the FCV should protect benthic species.

  Benthic Community Colonization
  Experiments

  Toxicity tests that determine the effects of chemicals
  on the colonization of communities  of benthic salt-
  water species [68-74] appear to be particularly sensi-
  tive at measuring the impacts of chemicals on benthic
  organisms. This is probably because the experiment
  exposes the most  sensitive life stages of a wide vari-
  ety of benthic saltwater species, and they are exposed
  for a sufficient duration to maximize response. The
  test typically includes  three  concentrations  of  a
  chemical and a control, each with 6 to 10 replicates.
  The test chemical is added to inflowing  ambient
  seawater containing planktonic larvae and other life
  stages of marine organisms that can settle  on clean
  sand in each replicate aquarium. The test  typically
  lasts from  two to four months, and the number of
  species and  individuals in  aquaria receiving the
  chemical are enumerated and compared to controls.
        If this test is extremely sensitive and if con-
  centrations in interstitial water, overlying water, and
  the sediment particles reach equilibrium, then the ef-
  fect and no-effect concentrations from this test can be
  compared  with the FCV from  the saltwater WQC
  documents to examine the applicability of WQC to
  protect benthic organisms. An FCV is the concentra-
  tion, derived from  acute and chronic toxicity data,
  that is predicted  to protect organisms from chronic
  effects of a chemical [8]. In addition, similarities in
  sensitivities of taxa tested as individual species and
  in the colonization experiment can indicate whether
  the conclusion of similarity of sensitivities of benthic
  and water column species is reasonable.
        The benthic colonization experiment is con-
  sistent with the assumptions used to derive SQC. The
  initially clean sandy sediment will rapidly equili-
  brate with the inflowing overlying water  chemical
  concentration as the pore water concentrations reach
  the overlying water concentration. The production of
sedimentary organic matter should be slow enough
to permit its equilibration as well. As a consequence,
the organisms will be exposed to an equilibrium sys-
tem with a unique chemical potential. Thus, the as-
sumption  of  the EqP is met by this  design. In
addition, the  experimental design guarantees that
the interstitial water-sediment-overlying water is at
the chemical potential of the overlying water. Hence
there is a direct correspondence between the expo-
sure in the colonization experiment and the water-
only exposures  from which WQC are  derived,
namely, the overlying water chemical concentration.
This allows a direct comparison.

Water Quality Criteria (WQC)
Concentrations Versus Colonization
Experiments

Comparison of the concentrations of six chemicals
that had the lowest-observable-effect concentration
(LOEC) and the  no-observable-effect concentration
(NOEC) on benthic colonization with the FCVs either
published in saltwater portions of WQC documents
or estimated from available toxicity data  (Table 6)
suggests that the level of protection afforded by
WQC to benthic organisms is appropriate. The FCV
should be lower than the LOEC and larger than the
NOEC.
      The FCV from the WQC document for pen-
tachlorophenol of 7.9 ng/L is less than the LOEC for
colonization of 16.0 ug/L. The NOEC of 7.0 ug/L is
less than the FCV. Although no FCV is available for
Aroclor 1254, the lowest concentration causing no ef-
fects on the sheepshead minnow (Cyprinodon vari-
egatus)  and pink shrimp (Penaeus  duorarum) as
cited in the WQC document is about 0.1 ug/L. This
concentration is less than the LOEC of 0.6 ug/L and
is similar to the NOEC of 0.1 ug/L based on a nomi-
nal concentration in a colonization experiment. The
lowest concentration tested with  chlorpyrifos (0.1
ug/L) and fenvalerate (0.01 ug/L) affected coloniza-
tion of benthic species. Both values are greater than
either  the  FCV  estimated for chlorpyrifos (0.005
ug/L) or the FCV estimated from acute and chronic
effects  data for fenvalerate (0.002 ug/L). The draft
water  quality criteria document  for 1,2,4-trichlo-
robenzene  suggests that the FCV should be 50.0
ug/L. This value is slightly above the LOEC from a
colonization experiment (40.0 ug/L) suggesting that
the criterion might be somewhat underprotective for
benthic species. Finally,  a colonization  experiment
with toxaphene  provides the only evidence  from
these tests that the FCV might be overprotective for
benthic species;  the FCV is 02 ug/L  versus  the
NOEC for colonization of 0.8 ug/L.
       The taxa most sensitive to chemicals, as indi-
cated by their LCsos and the  results of colonization
experiments,  are generally  similar, although, as

-------
                         Sediment Quality Criteria Using Equilibrium Partitioning
       Table ,6.—Comparison of WQC FCVs and concentrations affecting (LOEC) and not affecting (NOEC)
       benthic colonization,                   i '-•>•
SUBSTANCE
PENTACHLOROPHENOL
AROCLOR1254
CHLORPYRIFOS
FENVALERATE
1,2,4-TRICHLOROBENZENEa
TOXAPHENE
COLONIZATION
VERSUS FCVa "'"
Colonization LOEC
FCV
Colonization NOEC
Colonization LOEC
Estimated FCV
Colonization NOEC
Colonization LOEC
FCV
Colonization NOEC
Colonization LOEC
Estimated FCV
Colonization NOEC
Estimated FCV
Colonization LOEC
Colonization NOEC
Colonization LOEC
Colonization NOEC
FCV
"CONC. M8/L
16.0
7.9
7.0
0.6
-0.1
0.1
0.1
0.005
-
0.01
-0.002
-
50.
40.
-
11.0
0.8
0.2
SENSITIVE TAXA
Molluscs, Abundance
Molluscs, Crustacea, Fish
-
Crustacea
Crustacea, Fish
-
Crustacea, Molluscs, species richness
Crustacea
-
Crustacea, Chordates
Crustacea
-
Crustacea, Fish
Molluscs, abundance
-
Crustacea, species richness
-
Crustacea, fish
REFERENCE
[69,70]


[71]
[68]

[72]


[73]


[74]


[68]


          "Six day exposure to established benthic community  '
might be expected, differences occur. Both the WQC
documents and the colonization experiments suggest
that Crustacea are most sensitive to Aroclor 1254,
chlorpyrifos, fenvalerate, and toxaphcne. Coloniza-
tion experiments indicated that molluscs are particu-
larly sensitive to three chemicals,  an observation
noted only for pentachlorophenol  in WQC  docu-
ments. Fish, which are not tested in  colonization ex-
periments, are particularly sensitive to four of the six
chemicals.

Conclusions

Comparative  toxicological  data  on the acute and
chronic sensitivities of freshwater and saltwater ben-
thic species in the ambient WQC documents are lim-
ited.  Acute  values  are   available for only  34
freshwater infaunal species from four phyla and only
28 saltwater infaunal species from five phyla. Only
seven freshwater infaunal species and 24 freshwater
epibenthic species have been tested with five or more
of the 40 WQC chemicals. Similarly, nine saltwater
infaunal species and 20 epibenthic species have been
tested with five or more  of the 30 substances for
which saltwater criteria are available.
      In spite of the paucity of acute toxicity data on
benthic species, available data suggest that benthic
species are not uniquely sensitive and that SQC can
be  derived from  WQC. The data suggest that the
most sensitive infaunal species are typically less sen-
sitive than the most sensitive water column (epiben-
thic and water column) species. When both infaunal
and epibenthic species are classed as benthic, the sen-
sitivities  of benthic and water column species are
similar, on average. Frequency distributions of the
sensitivities of all species to  all  chemicals indicate
that  infaunal species  may be relatively insensitive
but that infaunal and epibenthic species appear al-
most evenly distributed among both sensitive and in-
sensitive species overall.
       Finally, in experiments to determine the ef-
fects of chemicals on colonization of benthic saltwa-
ter organisms, concentrations affecting colonization
were generally greater, and concentrations not affect-
ing  colonization were generally lower, than  esti-
mated or actual saltwater WQC FCVs.


GENERATION OF SQC

Parameter Values
The equation from which SQC are calculated is
                                            (34)
(see Eqns. 2 to 7 and associated text). Hence, the SQC
concentration depends only on these two parame-
ters. The Koc of the chemical is calculated from the
Kow  of the chemical via the regression Equation .11.

-------
                          Sediment Quality Criteria Using Equilibrium Partitioning
 The reliability of SQCoc depends directly on the reli-
 ability of Kow. For most chemicals of interest,  the
 available KowS  (e.g., [75]) are highly variable—a
 range of two orders of magnitude is not unusual.
 Therefore the measurement methods and/or estima-
• tion  methodologies  used to obtain each estimate
 must be critically evaluated to ensure their validity.
 The technology for measuring K0w has improved in
 recent years. For example, the generator column [76]
 and  the slow stirring [73] method appear to give
 comparable  results,  whereas earlier methods pro-
 duced more variable results. Hence,  it  is recom-
 mended that literature values for KowS not be used
 unless they  have been measured using the newer
 techniques.


 Measurement of Kbw

 The Kow is defined as the ratio of the equilibrium
 concentrations of a dissolved substance in a system
 consisting of n-octanol and water and is ideally  de-
 pendent only on temperature and pressure:
                                           (35)
 where COCT is the concentration of the substance in
 n-octanol and Cw is the concentration of the sub-
 stance in water. The Kow is frequently reported in the
 form of its logarithm to base ten as log P.
       At the EPA Environmental Research Labora-
 tory (ERL) at Athens, Georgia, three methods were
 selected for measurement and one for estimation of
 Kow for the five chemicals for which SQC are being
 developed. The measurement methods were shake-
 centrifugation (SQ,  generator column  (GC),  and
 slow-stir-flask (SSF). The estimations were made us-
 ing the computer expert system SPARC. The discus-
 sion of these methods is adapted from ERL at Athens
 research protocols.
       The SC method [78] is routinely used to meas-
 ure the partitioning of compounds with Kow values
 on the order of 102 to 10. The method involves add-
 ing a layer of octanol containing the compound of in-
 terest onto the surface of the water contained in a
 centrifuge tube. Both phases were mutually presatu-
 rated before beginning the measurements. Equilibra-
 tion is  established by  gentle agitation  and  any
 emulsions formed are broken by centrifugation. The
 concentration in each phase is determined, usually
 by  a chromatographic method, and the Kow value
 calculated using Equation 35.
       The  original GC method, limited to com-
 pounds with Kovf values of less than 106, was modi-
 fied [76] and used to determine K0w values up to 108.
 Briefly, the method requires the packing of a 24-cm
 length of tubing with silanized Chromosorb  W. Oc-
 tanol, containing the chemical in a known concentra-
tion, is then pulled through the dry support by gentle
suction until the octanol appears at the exit of the col-
umn. Water is then pumped through the column at a
rate of less than 2 mL per minute to allow equilibra-
tion of the chemical between the octanol and water.
The first 100 mL are discarded followed by collection
of an amount of water sufficient to determine ths
chemical  concentration. The Kow is calculated using
Equation 35.
      The SSF method [77] achieves equilibrium of
the compound between octanol and water by a gen-
tle stirring of the phases contained in a six-liter flask.
One  liter aliquots of the aqueous phase are with-
drawn at two-day intervals and the concentration of
the chemical determined. Equilibrium  is considered
to be established when the concentration of the
chemical  is constant in successive samples (usually
after two to six days). The procedure is to set up three
six-liter flasks in a constant temperature room. Five
liters of water are added  to each flask and the water
is stirred  with teflon-coated magnetic stir bars over-
night to achieve temperature equilibration. The tem-
perature equilibrated octanol is added very gently
along the side  wall  to  avoid  mixing of the two
phases. At the time of sampling, a one-liter aqueous
sample is drained from a sampling port at the base of
the flask without disturbing the octanol layer. The
concentration in each phase is  determined, usually
by a chromatographic method, and the K0w value
calculated using Equation 35.
      When repetitive measures were made in the
Athens laboratory, a protocol was established to as-
sure  compatibility with  future experiments. These
protocols described the entire experimental scheme
including planning, sample requirements, experi-
mental set  up and chemical analysis, handling of
data, and quality assurance. Only established ana-
lytical  methods for solute concentration measure-
ment were applied and the purity and identity of the
chemical  was determined by spectroscopic means.
The name on the label of the chemical's container
was not proof of identity.
      Standard reference compounds (SRCs) were
tested with each experiment. SRCs  are compounds
that are used as quality assurance standards and as
references in inter-laboratory generation of data. The
value of the process constant(s) has been established
by repetitive measurements for an SRC and serves as
baseline information for evaluating all experimental
techniques and all aspects of quality assurance. Be-
cause the SRC is taken through the entire experimen-
tal  scheme,  its  acceptable  result  assures  the
experimenter that equipment and measurement
methods are functioning satisfactorily. Table 7 shows
the logioKow values  for endrin,  dieldrin, acenap-
thene,  phenantherene, and fluoranthene  and  the

-------
                         Sediment Quality Criteria Using Equilibrium Partitioning
          Table 7.—LogioKow values measured by shake-centrifugation (SC), Generator column (GC), and
          slbw-stir-flask (SSF) for Endrin, Dieldrin, Acenapthene, Phenanthrene, Fluoranthene and
          Concurrently analyzed standard reference compounds.
CHEMICAL
ENDRIN
DIELDRIN
ACENAPHTHENE
PHENANTHRENE
FLUORANTHENE
BIPHENYL
PYRENE
SC
4.65
4.91
4.79
4.76
4.84
4.83
4.84
4.83




5.04
5.00
5.04
5.03
5.04
4.88
4.99
5.04
3.82
3.84
3.88
3.84
4.29
4.25
4.33
4.33
4.99
5.00
5.01



4.06
5.17
GC
4.67
5.01
4.73
4.62
5.09
5.28






4.89
4.88
5.18
5.15
5.26
5.38
5.67

4.18
4.17
4.16
4.17
4.47
4.41
4.46
4.24
5.19
5.35
s 5.47
5.48




SSF
4.86
4.59
4.97
4.95
5.02
4.82
5.04
4.91
5.07
4.93
4.96
4.78
" 5.33
5.43
5.38
5.33
5.43
5.08
5.28

3.81
3.84
3.84

4.57
4.53
4.50

4.98
5.02
5.02
5.10
5.14
5.23


              Source: U.S. EPA Environmental Research Laboratory, Athens Georgia.
SRCs, biphenyl and pyrene, measured at the Athens
laboratory by the SC methods. The SRCs were not
measured by the GC or SSF methods.
       The logio of the average of eight previous
measurements of K0w by the shake-centrifugation
method for biphenyl is 4.06. The logio of the average
of 13 previous measurements of K0w by the shake-
centrifugation method for pyrene is 5.17. These aver-
age KowS are in good agreement with the SQC shake-
centrifugation  measurements  for  biphenyl  and
pyrene made concurrently with the measurements
for the five chemicals providing quality assurance for
the experimental techniques (see Table 7).

-------
                          Sediment Quality Criteria Using Equilibrium Partitioning
       Literature Kow An extensive literature search
 was performed for the five compounds and two
 standard reference compounds, biphenyl and  py-
 rene. Generally, problems encountered in compiling
 and reporting fate constants from published data and
• from databases during a several years have ranged
 from retrieval of misquoted numbers to solution of
 nested citations [79]. Some citations were three or
 more  authors removed from the original work or
 contained data referenced as unpublished data or as
 personal communication. The same problems were
 experienced during a ERL, Athens literature search.
 The largest difference in misquoting numbers was six
 orders of magnitude. For these reasons, ERL ob-
 tained data from the primary sources and released
 values only from  these primary  sources. Unpub-
 lished data or data that originated through personal
 communication were rejected as well as  data that
 were insufficiently documented to determine their
 credibility and applicability or reliability.
       Tables 8 and 9 show the measured and esti-
 mated logioKow values, respectively, retrieved by this
 literature search. Each of the measured values was
 experimentally  determined by  the researcher using
 one of several laboratory  methods. The individual
 experimental methods are not identified here. The es-
 timated literature values were  computed by the re-
 searchers by one of several published  techniques.
 The individual computational techniques also are not
 identified here.
       Estimated Kow A promising new computa-
 tional method for  predicting reactivity  is the com-
 puter  expert  system  SPARC (SPARC  Performs
 Automated Reasoning in  Chemistry) being devel-
 oped by ERL's Samuel W. Karickhoff and scientists at
 the University of Georgia [106]. The system has the
 capability of crossing chemical boundaries to cover
 all organic chemicals and uses algorithms based on
  mdamental chemical structural theory to estimate
 "parameters. Organic chemists have in the past estab-
 lished the types of structural groups or atomic arrays
 that impart certain types of reactivity and have de-
 scribed, in "mechanistic" terms, the effects on reac-
 tivity of the structural constituents appended to the
 site of reaction.
       To encode this  knowledge base, Karickhoff
 and his associates developed a classification scheme
 that defines the role of structural constituents in af-
 fecting or modifying reactivity. SPARC quantifies re-
 activity by classifying molecular structures and
 selecting appropriate "mechanistic" models. It uses
 an approach that combines principles of quantitative
 structure-activity relationships, linear free energy the-
 ory (LFET), and perturbed molecular orbital (PMO)
 or quantum chemistry theory. In general, SPARC uses
 LFET to compute thermal properties and PMO theory
 to describe quantum effects such as delocalization en-
 ergies or polarizabilities of pi electrons.
   Table 8.—Measured logioKow values found in the
   literature.
CHEMICAL
ENDRIN
DIELDRIN
ACENAPTHENE
PHENANTHRENE
FLUORANTHENE
BIPHENYL
PYRENE
LOGIOKOW VALUE
4.40
4.92
5.01
5.195
4.09
4.54
4.65
5.401
6.2
3.92
4.28
4.46
. 4.562
4.57
4.63
5.155
3.16
3.63
3.75
3.76
3.79
3.89
4.008
4.01
4.04
4.09
4.10
4.96
5.05
5.09
5.18
5.22
5.52
REFERENCE
[80]
[81]
[82]
[77]
[81]
[83]
[84]
[77]
[85]
[86]
[87]
[88]
[77]
[78]
[89]
[77]
[90]
[84]
[91] -
[92]
[80]
[76]
[77]
[82]
[86]
[84]
[89]
[80]
[84]
[93]
[78]
[89]
[94]
      SPARC computes K0w values from activity co-
efficients in the octanol (~lo) and water (~lw) phases
using Equation 36:

     LogioKow = logio(-lw/~lo) + logio(Mo/Mw)  (36)

where Mo and Mw are solvent molecularities of oc-
tanol and water, respectively. SPARC computes activ-
ity coefficients for any solvent/solute pair for which
the structure parser can process the structure codes.
Ultimately, any solvent/solute combination can be
addressed. NJew solvents can be added as easily as
solutes by simply providing a Simplified Molecular

-------
                           Sedimetrt Quality Criteria Using Equilibrium Partitioning
   Table 9.—Estimated logioKow values found in the
   literature.
' CHEMICAL
ENDRIN
DIELDRIN
ACENAPTHENE
PHENANTHRENE
FLUORANTHENE
BIPHENYL
PYRENE






LOGlOKOW VALUE
3.54
5.6
3.54
3.70 -
3.92
3.98
4.03
4.15
4.22
4.33
4.43
4.44
4.45
4.63
4.64
4.90
4.95
5.22
5.29
5.33
3.79
3.95
3.98
4.14
4.25
4.42
4.50
4.85
4.88
4.'90
5.12
5.22
5.32
REFERENCE
[95]
[96]
[95]
[96]
[97]
[95]
[98]
[99]
[100]
[101]
[102]
[100]
[95]
[99]
[96]
[95]
U.S. EPA, Graphical
Exposure Modeling '
System [GEMS]*
[96]
[99]
[101]
[96]
[97]
[100]
[99]
[102]
[103]
[104]
[100]
[105]
[95]
[99]
[96]
[101]
    The Graphical Exposure Modeling System (GEMS) is an inter-
    active computer system located on the VAX Cluster in the Na-
    tional Computer Center in Research Triangle Park, North
    Carolina, under management of EPA's Office of Toxic Sub-
    stances. PC GEMS is the version for personal computers.
Interactive Linear Entry System (SMILES) string [107,
108]. Activity coefficients for either solvent or solute
are computed by solvation models that are built from
structural constituents requiring no data besides the
structures.                 —
       A goal for SPARC is to compute a value that is
as accurate as a value obtained experimentally for a
fraction of the cost required to measure it. Because
SPARC does not depend .on laboratory  operations
conducted on compounds with structures closely re-
lated to that of the solute of interest, it does not have
the inherent problems of phase separation encoun-
tered in measuring highly hydrophobic compounds
(logioKow > 5). For these compounds, SPARCs com-
puted value should, therefore, be more reliable than
a measured one.  Reliable  experimental  data with
good documentation are still necessary, however, for
further testing and validation of SPARC.
       CLOGP [109] is a computerized program that
estimates the logioKow, based  on Leo's Fragment
Constant Method [105]. CLOGP provides an estimate
of logioKow using fragment constants  (ft)  and struc-
tural factors (Fj) that have been empirically derived
for many molecular  groups. The estimated logioKow
is obtained from the  sum of constants and factors for
each  of  the molecular  subgroups comprising the
molecule using Equation 37.
                                                                               n
                                             (37)
                                                      The method assumes that logioKow is a linear addi-
                                                      tive function of the structure of the solute and its con-
                                                      stituent parts and that the most important structural
                                                      effects are described by available factors. The struc-
                                                      ture of the compound is specified  using the SMILES
                                                      notation.  The CLOGP algorithm is included in the
                                                      database QSAR (see Table 10) located at EPA's Envi-
                                                      ronmental Research Laboratory at Duluth,  Minne-
                                                      sota. All CLOGP values reported here were obtained
                                                      through   QSAR.   Table  10  shows  the  estimated
                                                      logioKow  values  that were  computed with  SPARC
                                                      and CLOGP.
                                                       Table 10.—QSAR -"obtained LogioKow values estimated
                                                       by SPARC and CLOGP
CHEMICAL
ENDRIN
DIELDRIN
ACENAPTHENE
PHENANTHRENE
FLUORANTHENE gan
BIPHENYL
PYRENE
SPARC
5.40
5.40
3.88
4.58
5.21
4.25
5.13
CLOGP
-
-
4.07
4.49
4.95
4.03
4.95
  •Quantitative Structure^Activity Relationships (QSAR) is an in-
   teractive chemical database and hazard assessment system
   designed to provide basic information for the evaluation of
   the fate and effects of chemicals in the environment. QSAR
   was developed jointly by the U.S. EPA Environmental Re-
   search Laboratory, Duluth, Minnesota, Montana State Univer-
   sity Center for Data System and Analysis, and the Pomona
   College Medicinal Chemistry Project.

-------
                         Sediment Quality Criteria Using Equilibrium Partitioning
            Selection. Investigators selected SSF de-
 rived Kow  values to derive the KoC to calculate SQC
 concentrations because SSF is tiie superior method
 for chemicals with low and high Kow values, has the
 least statL ,cal bias, and is highly.reproducible. Use
• of values  from one  method provides consistency
 across chemicals. This choice was made after an
 analysis of the Kow values generated by the three
 measurement methods discussed above (GC, SC, and
 SSF) and the SPARC estimation method for the five
 chemicals for which SQC are currently being devel-
 oped (acenaphthene, dieldrin, endrin, fluoranthene,
 and phenanthrene). Kows were measured with repeat
 experiments and a mean Kow was computed for each
 method, for each chemical (Table 11).
       The mean measured Kows and the SPARC esti-
 mation method provide similar Kow estimates. To se-
 lect a final Kow for  computing the SQC, the four
 methods were compared and the bias of each method
 were quantified. (Bias is defined as the mean differ-
 ence between the best-fit  estimate of  Kow using all
 four methods and the estimates from each method.)
 Figure 35 presents the mean measured K0ws for each
 chemical and the range of values. SGtends to estimate
 lower values while the GC method estimates higher
 values. GC values exhibit greater variability than the
 SC and SSF values. SSF estimates of Kows were gener-
 ally within the range of the SC and GC methods.
       A  .stistical analysis of the three  measure-
 ment me;. ;ds and SPARC method was performed.
 The following linear model was used to compute es-
 timates of Kow for each chemical (represented by El,
 E2, E3, E4, E5) and the biases contributed by each of
 the  estimation methods (represented by Bl, B2, B3,
 B4). The regression model is as follows:

         logioKow -  El * Endrin +
                   E2 * Dieldrin +
                   E3 * Acenaphthene +
                   E5 * Fiuoranthene +

                   Bl * Shake Centrifugation +
                   B2 * Generator Column +
                   B3 * Slow-Stir-Rask
                   B4 * SPARC
       To compute logirjKow the variables, ENDRIN,
... SPARC, are set to either 0 or 1 and the appropriate
coefficient for the chemical and method  that corre-
sponds to the particular Kow measurement or esti-
mate is selected. Table 12 presents the model Kow
results and bias contributed by each method. Shake
centrifugation and SPARC  estimates provide the
greatest  bias, followed by  the  generator  column
method. Slow-stir-flask provides the least bias. Slow-
stir-flask was chosen as the method to use to deter-
mine Kow for use in computing SQC since it appears
to have the least bias. In addition, it exhibits similar
variability to the shake centrifugation method and
less variability than the generator column method.
  Table 12.—Model results to determine method bias
COEFFICIENT
El
E2
E3
E4
E5

Bl
B2
B3
B4
ESTIMATE OF Kow OR BIAS FOR
Endrin
Dieldrin
Acenaphthene
Phenanthrene
Fluoranthene

Shake Centrifugation
Generator Column
Slow-stir Flask
SPARC
VALUE
4.88
5.17
3.94
4.40
5.13"

•0.115
0.091
0.040
0.193
           Determination. The previous section dis-
 cusses selecting the method for measuring Kows for
' use in computing SQC. It is widely accepted that KocS
 can be estimated from K0w. The K0c used to calculate
 the sediment quality criteria is based on the regres-
 sion of logioKoc to logioKow, Equation 11.
     Table ±L—Kg* as measured by the EPA Environmental Laboratory at Athens, GA.

                                    LogioKow (NUMBER OF DETERMINATIONS)
CHEMICAL
ENDRIN
DIELDRIN
ACENAPHTHENE
PHENANTHRENE
FLUORANTHENE
SHAKE CENTRIFICATION
4.80 (8)
5.01 (8)
3.85 (4)
4.30 (4)
5.00 (3)
GENERATOR COLUMN
4.97 (6)
5.16 (7)
4.17 (4)
4.40 (4)
5.39 (4)
SLOW STIR FLASK
4.92 (12)
5.34(7)
3.83 (3)
4.54 (3)
5.09 (6)

-------
                          Sediment Quality Criteria Using Equilibrium Partitioning
 o
§
                    ACENAPHTHENE    ErCWi
PtdANTHRENE
FLUORANTHENE
Figure 35.—Laboratory Kbw values for five chemicals using three experimental methods with replication. Ka* values were
measured at the Environmental Research Laboratory, Athens, GA. For each chemical the average of the methods is indicated
by 0 for shake centrifugation, D for slow stir flask and O for generator column. Ranges are indicated by I.
       This equation is based on any analysis of an
extensive body of experimental data for a wide range
of compound types and experimental  conditions,
thus encompassing a wide range of Kows and/ocs.
       Sediment  toxicity tests provide a favorable
environment for  measuring Kow.  Figures 24 and 25
presented plots of the organic carbon-normalized sorp-
tion isotherm from sediment toxicity tests for the five
chemicals where the sediment concentration (ug/g oc)
is plotted versus pore water concentration (ug/L).
Also included in  each  panel is the line to the parti-
tion, Equation 16, where Koc is computed from the
slow-stir flask Kow values. These plots can be used to
compare the Koc computed from laboratory deter-
mined Kow and the regression equation with the par-
titioning behavior of the  chemical in the sediment
toxicity tests. For each of the chemicals the Koc line is
in agreement with the data demonstrating the valid-
ity of the use of the slow-stir flask Kow  in the -SQC
computation.

Species Sensitivity

The FCV is used as the appropriate end point for the
protection of benthic organisms. Therefore, its appli-
cability to benthic species for each chemical should
be verified. The previous work has indicated that this
is a reasonable assumption across all criteria chemi-
cals. To test this assumption for a particular chemical
a statistical method known as Approximate Ran-
domization [110] is applied to each chemical.  The
idea is to test whether the difference between the fi-
nal acute value (FAV) derived from considering only
    benthic organisms is statistically different from the
    FAV contained in the Water Quality Criteria (WQ'C).
           The  Approximate Randomization  method
    tests the significance level of the test statistic by com-
    paring it to the distribution of statistics generated
    from many random reorderings of the LCso values.
    For example, the test statistic in this case is the differ-
    ence between the WQC FAV, computed from the
    WQC LCso values, and the benthic  FAV, computed
    from the benthic organism LCso values. Note that the
    benthic organism  LCso values are a subset of the
    WQC LCso values. In the Approximate Randomiza-
    tion method for this test, the number of data points
    coinciding with the number of benthic organisms are
    selected from the WQC data set. A "benthic" FAV is
    computed. The original WQC FAV and the "benthic"
    FAV are then used to compute the difference statis-
    tics. This is done many times and the distribution
    that results is representative of the population of FAV
    difference statistics. The test statistic is compared to
    this distribution to determine its level of significance.
           For each chemical, an initial test of the differ-
    ence between the freshwater and saltwater FAVs for
    all species (water column and benthic) is performed.
    The probability distribution of the FAV differences
    for fluoranthene are shown in the top panel of Figure
    37. The horizontal line that crosses the distribution is
    the test statistic computed  from the original WQC
    and benthic FAVS. For fluoranthene,  the test statistic
    falls at the 78th percentile. Since  the probability is
    less than 95 percent, the hypothesis of no significant
    difference in sensitivity is accepted.

-------
                           Sediment Quality Criteria Using Equilibrium Partitioning
        Since freshwater and saltwater species show
 similar sensitivity, a test of difference in sensitivity
 for benthic and WQC organism^ combining freshwa-
 ter and saltwater species can be made.  The bottom
t panel of Figure 36 represents the bootstrap analysis
 to test the hypothesis of no difference in sensitivity
 between benthic and WQC organisms  for fluoran-
 thene.  The test statistic for this analysis falls at the
 74th percentile and the hypothesis of no difference in
 sensitivity is accepted.
        Table 13 presents the Approximate Randomi-
 zation  analysis for five chemicals for  which SQC
 documents are being developed. Four chemicals,
 (acenaphthene,  phenanthrene,  fluoranthene  and
 dieldrin) indicate that there is no difference in sensi-
 tivity for freshwater and saltwater species. The test
 for endrin fails at the 99 percentile which indicates
 that FAVs for freshwater and saltwater are different.
 Therefore, separate analyses for the freshwater and
 saltwater organisms are performed.
        Table 14 presents the results of the statistical
 analysis for each chemical for benthic organisms and
 WQC organisms. In all cases the hypothesis of no differ-
 ence in sensitivity is accepted. Therefore, for each indi-
 vidual chemical the WQC is accepted as the appropriate
 effects concentrations for benthic organisms.

 Quantification of Uncertainty
 Associated with SQC

 The uncertainty in the SQC can be estimated from
 the degree to which the equilibrium  partitioning
 model, which is the basis for the criteria, can rational-
 ize the available sediment toxicity data. The EqP
 model  asserts that (1) the bioavailability of nonionic
 organic chemicals from sediments is equal on an or-
 ganic carbon basis; and (2) that the effects concentra-
 tion in sediment can be estimated  from  the product
 of the  effects concentration from  water-only expo-
 sures and the partition coefficient K0c-  The  uncer-
 tainty associated with the sediment quality criteria
 can be obtained from a quantitative estimate of the
 degree to which  the available data support these
 assertions.
             FRESHWATER V8 SALTWATER
           11limn  i I limn	1—i i  i i  i—i  mmi. i ..|.|m ,
                      PROHABOTY
Figure 36.—Probability distributions of randomly generated
differences between saltwater FAVs and freshwater FAVs
(top panel) and randomly generated differences between
WQC FAVs and benthic FAVs (bottom  panel) using the Ap-
proximate Randomization method. Horizontal line in both
panels indicates the test statistic which is the FAV differ-
ence from original LCso data sets.
       The data used in the uncertainty analysis are
the water-only and sediment toxicity tests that have
been conducted in support of the sediment criteria
development effort. A listing of the data sources used
in the EqP uncertainty analysis is presented in Table
15. These freshwater and saltwater tests span a range
of chemicals and organisms; they include both water-
only and sediment exposures, and they are replicated
within  each  chemical-organism-exposure  media
       Table 13.—Approximate randomization analysis freshwater versus saltwater organisms.
NUMBER
CHEMICAL
ENDRIN
DIELDRIN
ACENAPHTHENE
PHENANTHRENE
FLUORANTHENE
SALTWATER
19
21
10
11
8
FRESH WATER
35
19
10
8
12
FINAL ACUTE VALUE (FAV)
SALT WATER
0.033
0.662
139.05
16.6
16.13
FRESH WATER
0.189
0.377
80.01
59.63
33.58
DIFF
0.156
-0.305 :
-59.04
43.02
17.45
%
99
31
33
73
78

-------
                          Sediment Quality Criteria Using Equilibrium Partitioning
    -Table 14.—Approximate randomization analysis benthic versus WQC organisms.
-: •' -• 	 ' 	
CHEMICAL
ENDRIN
ENDRIN
DIELDRIN
ACENAPHTHENE
PHENANTHRENE
FLUORANTHENE
NUMBER ... 	
WATER TYPE
Fresh
Salt
Combined
Combined
Combined
Combined
WATER COL +
BENTHIC
,- 35 ,
19
40
20
19
20
BENTHIC
21
12
26
10
13
14
FINAL ACUTE VALUE (FAV)
WATER COL +
BENTHIC
0.189
0.033
0.621
132.6
26.62
37.91
BENTHIC
.234
0.023
0.532
173.9
19.27
34.27
DIFF
•O.045
0.010
0.090
-38.34
7.35
3.64
%
7
66
72
31
80
74
 Table 15.—Data for uncertainty analysis number of replicates//bc.
ENDRIN
Hy
3(3/-)
2/11%
4/11%
3/3.0%
3/6.1%
3/11.%
DIELD
Hy
V-
2/1.7%
4/2.9%
4/8.7%


FLUOR
Re
l/-
I2/.18%
2/.30X
!2/.48% '


ACENAP
Le
L 4/--
2/1.6%
2/2.5%
2/3.6%


ACENAP
Eo
4/-
2/1.2%
2/2.5%
2/3.6%


PHEN
Le
4/-
2/1.9%
2/2.5%
2/3.6%


PHEN
Eo
4/-
2/1.0% '
2/2.5%
2/3.3%


 Hy B Hyalella	Re = Rhapoxynlut	Le = Loptoehelru*    Eo = Eohaustorius
  Data sources: endrin [20, 21, 52], dieldrin [53] fluoranthene [19], acenaphthene [56] phenanthrene [56].
treatment. These data are analyzed using an analysis
of variance (ANOVA) to estimate the uncertainty
(i.e., the variance) associated with varying the expo-
sure media and the uncertainty associated with ex-
perimental error. If the EqP model were perfect, then
there would be only  experimental  error. Therefore,
the uncertainty associated with the use of EqP is the
variance associated with varying exposure media.
       Sediment and water only LCsos are computed
from the sediment and water-only toxicity tests. The
EqP model can be used to normalize the data in order
to put it on a common basis. The LCso from water-
only exposures, LCsow (ug/L), is related to the LCso
for sediment  on  an organic carbon basis, LCsos,oc
(ug/goc) via the partitioning equation:
               LCsOs,oc = Koc LCsOw
(38)
       The EqP model asserts that the toxicity of sedi-
ments expressed on an organic carbon basis equals toxic-
ity in water-only tests multiplied by the Koc. Therefore,
either LCsos,oc (ug/goc), from sediment toxicity ex-
periments or Koc x LCsow.(ug/goc), are estimates of
the true LCso for this chemical-organism pair.
       In this analysis, the accuracy of Koc is not
treated separately. Any error associated with Koc will
be reflected in the uncertainty attributed to varying
the exposure media.
       In order to perform an analysis of variance, a
model of the random variations is required. As dis-
cussed  above,  experiments  that  seek to validate
Equation 38 are subject to various sources of random
variations. A number of  chemicals  and organisms
have been tested. Each chemical-organism pair was
tested in water-only exposures and in different sedi-
ments. Let a represent the random variation due to
the varying exposure media. Also, each experiment
was replicated. Let e represent the  random variation
due to replication. If the model were perfect,,there
would be no random variation other than that result-
ing from the experimental error which is reflected in
the replications. Thus, a represents the uncertainty
due to the approximations inherent in the model, and
e represents the experimental error. Let oa2 and oe2
be the variances of these random variables. Let i in-
dex a specific chemical-organism pair. Let j index the

-------
                          Sediment Quality Criteria Using Equilibrium Partitioning
 exposure media, water-only, or the individual sedi-
 ments. Let k-index the replication of the experiment.
 Then the equation that describes this relationship is
            Ln(LCsoi,j,k) - Hi + ay + £i,j/k
(39)
 where   Ln(LCso)i,j,k,   are  either   In(LCsow)   or
 In(LCsopw) corresponding to a water-only or sedi-
 ment exposure; urs are the population of In(LCso) for
 chemical-organism pair i. The error structure is as-
 sumed to be lognormal which corresponds to assum-
 ing that the errors are proportional to the means, for
 example, 20 percent; rather than absolute quantities,
 for example, 1 mg/L. The statistical problem is to es-
 timate us and the variances of the model error, aa2,
 and the measurement error, aE2. The maximum likeli-
 hood method is used  to make these estimates [109].
 The results are shown in Table 16.
  Tablo 16.—Analysis of variance for derivation of
  criteria confidence limits.
SOURCE OF UNCERTAINTY
Exposure media
Replication
Sediment Quality Criteria
PARAMETER
Oa
Ot
OSQC
VALUE
(M/Coc)
" 0.39
0.21
0.39
  Note that OSQC • oa the variability due to EqP
       The last line of Table 16 is the uncertainty as-
sociated with the SQC, that is, the variance associ-
ated  with the  exposure  media  variability.  The
confidence limits for the SQC are computed using
this uncertainty for SQC. For the 95 percent confi-
 dence interval limits, the significance level is 1.96 for
 normally distributed errors. Thus

       ln(SQCoc)upper - ln(SQCoC) + 1.96oSQc   (40)

       ln(SQCoc)lower - In(SQCoc) - 1.96osQC   (41)

 The confidence limits are given in Table 17.


 Minimum Requirements to
 Compute SQC

 It has been demonstrated that the computation of
 sediment quality criteria for a particular chemical re-
 quires key parameter values as well as evidence that
 EqP is applicable for a particular chemical. Minimum
 requirements for these parameters are warranted so
 that they provide the level-of protection intended by
 SQC and that are within the limits of uncertainty set
 forth  in this document. This section  outlines mini-
 mum data  requirements and guidance for deriving
 them. This is a necessary step to develop reliable pa-
 rameters to be used in computing SQC. The mini-
 mum requirements for an EqP based  SC are as
 follows.

      • Octanol-Water Partition Coefficient (Kow)

      • Final chronic value (FCV)

      • Sediment Toxicity Tests

Procedures to ensure that these data  meet assump-
tions of the EqP approach will also be addressed.
      Laboratory octanol-water partition coefficient
Kow data developed by the slow stir flask measure-
ment  technique is  required. This method has been
shown to provide the least amount of variability and
TaWe 17.—Sediment quality criteria confidence limits for five chemicals.
                                               SEDIMENT QUALITY CRITERIA 95% CONFIDENCE LIMITS ((lg/goc)
CHEMICAL
ACENAPTHENE
FLUORANTHENE
PHENANTHRENE
ENDRIN
DIELDRIN
TYPE OF WATER BODY
Fresh Water
Salt Water
Fresh Water
Salt Water
Fresh Water
Saltwater
Fresh Water
Saltwater
Fresh Water
Salt Water
SQCoe
ug/goc
132
232
616
296
182
238
4.22
0.76
11.1
20.4
LOWER LIMIT
61.5
103
290
140
85
111
1.96
0.354
5.17
9.50
UPPER LIMIT
283
498
1,300
640
391
511
9.06
1.63
23.8
43.8

-------
                          Sediment Quality Criteria Using Equilibrium Partitioning
  the least bias when comparisons of Kow estimation
  techniques were done for Kows derived for the five
  chemicals for which SQC have been developed. A
  minimum of three K0w values are required. These
  values may be taken from the literature provided
»  that methods followed yield a degree of confidence
  similar to that provided by the methodology used by
  EPA to derive Kow values.
       If slow stirk flask Kow values do not exist then
  laboratory experiments must be  conducted. Meas-
  urements of Kow done at the EPA ERL, Athens, Geor-
  gia were presented. At a minimum, these procedures
  are recommended. EPA laboratory procedures in-
  clude a quality assurance and control plan. The plan
 includes testing the  compound  by  spectroscopic
 means to ensure its identity and purity as well as
 running concurrent Kow measurements of reference
 compounds  which  have  KowS  that  have  been
 verified.
       Final chronic value. The FCV is computed as
 part of the derivation of the water quality criteria for
 .a compound, and is defined as the quotient of the Fi-
 nal Acute Value (FAV) and the Final Acute-Chronic
 Ratio  (8). The data required to compute the WQC
 FCV are water-only toxicity tests for a variety of or-
 ganisms meeting minimum data base requirements.
 The FCV computation and  minimum database re-
 quirements are presented  in  the EPA document
 which describes methods to be used in deriving na-
 tional ambient WQC (8).
       ^WQC are based on an assessment of a com-
 pound's acute and chronic toxiciry for organisms rep-
 resenting a range of sensitivities, most importantly
 most sensitive organisms. This is appropriate since
 the objective of WQC is to set limits based on the best
 estimate  of organism sensitivity. The toxiciry data
base should therefore include all available data that
meets  requirements. That is, a complete search, re-
trieval and review for any applicable data must be
conducted, to locate all preexisting toxicity data. For
some compounds a WQC FCV may exist which
would provide a significant amount of toxicity data.
Literature searches are recommended to locate other
sources of toxiciry data.
       A reevaluation of an already existing FCV is
warranted because data post dating publication of
the national FCV can be incorporated into the FCV
value.  Also minimum  database  requirements  have
changed since some WQC have been published. For
those compounds for which WQC FCVs do not exist,
compiled toxicity data are evaluated to see if mini-
mum data requirements as put forth by EPA (8) are
met. If so an FCV could then be computed. If there is
not enough water only toxicity data to compute an
FCV additional water only tests will be conducted so
that there is enough data to satisfy minimum data-
base requirements.
        Sediment toxicity test. Verification of applica-
 bility of EqP theory is required for each compound.
 Sediment toxicity tests can be used for this. These
 tests provide a sediment based LCso. Comparison of
 the EqP predicted LCso with the sediment LCso con-
 centration is direct confirmation of the EqP approach.
 The validity of EqP is confirmed when the toxiciry
 test results fall within the limits of uncertainty deter-
 mined in this document.
        Guidelines for conducting sediment toxicity
 tests ensure that the tests  are uniform and are de-
 signed to incorporae the assumptions of EqP. These
 tests must represent a range of organic carbon con-
 tent and include organisms that exhibit sensitivity to
 the chemical in question. The range of organic carbon
 must be no less than a factor of 3 and a factor of 10 is
 recommended. Organic carbon content  shuld be no
 less than 0.2 percent. Replicated toxicity tests for at
 least two sediments are required. Organisms to be
 used in the sediment toxicity tests are  benthic ani-
 mals which are  most sensitive to the compound in
 question. Guidelines on appropriare selection of ben-
 thic organisms is given  in the American Society for
 Testing and Materials annual handbook [111].
       Several studies are required as part of sedi-
 ment toxicity testing. A water-only flow through test
 is required. Water-only tests are run for five concen-
 trations of the compound in question and a control.
 The endpoint of interest is the 10-day mortality of the
 test species. This value will  be compared to the pore
 water and sediment mortality  from the sediment
 spiking tests discussed next.
       Two sediment spiking tests are required. The
 first test is for the purpose of identifying sediment
 spiking concentrations so that pore water concentra-
 tions in spiked  sediments bracket the LCso  deter-
 mined in the water-only test. In addition, this test is
 done to determine the  time-to-equilibrium  of the
 compound between the pore water and sediment.
 Sorption equilibrium, and assumption of EqP theory,
 is  essential  for  valid  porewater  and sediment
 concentrations.
       Three spiking treatments are recommended
 for this first test: low, medium  and high concentra-
 tion. The amount of compound to add to each treat-
 ment is calculated using the initial chemical weight,
 the % total organic carbon (TOC), % dry weight and
 total volume of spiked sediment. The results are sedi-
 ment concentrations that bracket the predicted LCsos
 estimated from the water-only LCso (ug/L) andKoc.
 Samples for chemical analyses in bulk sediment and
 pore water are collected at various time intervals.
 Nominal sediment spike concentrations, measured
 sediment  TOC and  measured and  EqP-predicted
 compound concentrations in sediments and pore wa-
 ters are obtained  for each sample period to establish
 time-to-equilibrium and  to verify that spiking pro-
 duces the appropriate concentrations in the pore
water.

-------
                         Sediment Quality Criteria Using Equilibrium Partitioning
       In the second sediment spiking tests three
 sediments representing a range of organic  carbon
 content are spiked to yield five estimated sediment
 concentrations to bracket the  predicted sediment
 LCso. The amount of compound for spiking is based
* on similar computations as in the first sediment spik-
 ing experiment. Each treatment (sediment by concen-
 tration) is held for the appropriate time based on
 time to equilibrium established in the first spiking
 test. Day 0 samples are taken for sediment and pore
 water analyses. Then organisms are placed in repli-
 cated beakers and 10-day sediment toxicity tests with
 the equilibrated  spiked sediments are conducted.
 Eight replicates for each treatment are required. Four
 replicates are used for day 10 sediment and pore
 water analyses while the remaining four replicates
 are used to assess organism mortality.
       These experiments provide data to compute
 pore water toxic units  and sediment toxic units
 (Equations 1 and 8). The results of these equations
 serve as direct comparisons of the predicted toxicity
 (Equation 1 and 8 numerator) to the observed toxic-
 ity (Equation 1 and 8-denominator). That is,  the va-
 lidity of EqP for a chemical is confirmed when the
 pore water and sediment toxic units fall within the
 limits of uncertainty determined in this document.
       Analytical procedures. The purpose of these
 procedures is to verify that:

       • the WQC FCV applies to benthic organisms

       • the Koc from the slow stir flask Kow is an
         accurate estimate of Kow

       A test that the WQC FCV, which is applicable
 to the most sensitive water column organisms, is ap-
 plicable  to the most sensitive benthic  orgnaisms is
 needed for each chemical. In computing SQC for en-
 drin/  dieldrin, acenapthene,  phenanthrene  and
 fluoranthene,  the  Approximate Randomization test
 was applied. This is a statistical test to compare the
 WQC toxicity database to benthic organism toxicity.
 The  methodology is  presented previously.  If  it is
 found that benthic organisms exhibit similar or less
 sensitivity to a chemical than those organisms used
 to compute WQC, then the WQC FCV can be applied
 in computing an SQC. If benthic organisms exhibit a
 greater sensitivity than the WQC orgnaisms then tox-
 icity experiments for benthic organisms are required.
       A check on the laboratory Koc must be done
 by comparing it to the Koc computed from sediment
 toxicity  tests. Pore water and sediment concentra-
 tions from  the sediment toxicity test provide  data
 necessary to compute Koc- This Koc is then compared
 to the Koc from the slow stir flask K0w.
       Lastly, when a site's sediments are being stud-
 ied, a check to show that the SQC applies to the site is
 needed. National SQC may be under or over  protec-
tive if 1) the species at the site are more or less sensi-
tive than those included in the data set used to derive
SQC or 2) the sediment quality characteristics of the
site alter the bioavailability predicted by EqP and, ul-
timately, the predicted  toxicity of the  sediment
bound chemical. Therefore, it is appropriare that site-
specific guidelines procedures address each of these.
conditions separately, as  well as jointly. Methods to
determine the applicability of national SQC to a site
and to determine site specific SQC if needed are pre-
sented in the EPA guidelines document for deriving
site specific sediment criteria [112].
       Conclusion. Minimum database and analyti-
cal requirements must be set when deriving national
sediment criteria. The reasons for  this is twofold.
First, the requirements provide that a level of protec-
tion intended by the criteria are met. Secondly, the re-
quirements provide that parameters used to compute
the  criteria satisfy assumptions underlying the EqP
theory. The key required parameters are Kow using
the  slow stir flask measurement method, the WQC
FCV and sediment toxicity tests. Procedures to verify
that  these values are appropriate to use in the SQC
computation are also required. It must be shown that
the  FCV is protective of benthic organisms. Confi-
dence in the Koc must also be established by compar-
ing the Kow to the observed K0c in sediment toxicity
tests. Individual sites may exhibit greater or lesser
toxicity to a chemical than that predicted by SQC to
an individual site. EPA procedures to test this as well
as to compute site specific SQC are available.


Example Calculations

Equation 34 can be used to compute SQCoc for a range
of KowS and FCVs. The results for several chemicals are
shown in Figure 38 in the form, of a nomograph. The di-
agonal lines are for constant FCVs as indicated. The ab-
scissa is iogioKow- For example, if a  chemical has an
FCV of 1.0 ug/L and a IogioKow of 4, so that K0w =
104, the logic SQCoc is approximately 1 and the SQC
= 10 = 10.0 ng chemical/g organic carbon.
       As can be seen,  the  relationships between
SQCoc and the parameters that determine its magni-
tude, Kow and FCV, are essentially linear on a log-log
basis. For a constant FCV, a 10-fold increase in Kow
(one log unit) increases the SQCoc by approximately
10-fold (one log unit) because Koc also increases ap-
proximately 10-fold.  Thus,  chemicals with similar
FCVs will have larger SQCocS if their KowS are larger.
       The chemicals listed  in Figure 37 have been
chosen to ilustrate the SQCoc concentrations that re-
sult from applying the EqP method. The water qual-
ity concentrations are the FCVs (not the final residue
values) computed as part of the development of SQC
for  acenaphthene,  endrin,  phenanthrene, dieldrin
and  fluoranthene or from  draft or published  EPA

-------
                          Sediment Quality Criteria Using Equilibrium Partitioning
                                 Sediment Quality Criteria
                     o
                     o
                     o»


                    JO

                     9

                    O
                     X
                     0
                     E
                    •a
                     o
                    u>
                     D)
                     O
 6


 5


 4


 3
                                                       1000   100
-1
-2
                           -3
FCV (ug/L)

10


1


O.I  *"


O.O1
                                           0.001
   M»thyl Parathion
   Toxaphan*
   Chlordana
   Parathion
   EndoaOlfan
   Acanaphthwi*
   Endrln
   Ph«nanthr«n*
   Chlorpyrifoa
   DMdrln
   Fluoranth*n»
                                    3 .    4
                            6
                                           Log10 Kow
Figure 37.—Logio SQC versus logio Kow. The diagonal lines indicate the FCV values. The criteria are computed from Equa-
tion 34. KM is obtained from Kow with Equation 11. The symbols indicate SQCOC for the freshwater (filled) and saltwater
(hatched) criteria for the listed chemicals. The vertical line connects symbols for the same chemical. The FCVs for methyl
parathion, toxaphene, chlordane, parathion are from WQC or draft documents, see Table 4. The FCVs for acenapthene, en-
drin, phenanthrene, fluoranthene and dieldrin are computed as part of the development of SQC. The octanol/water partition
coefficients for methyl parathion, toxahene, chlordane, parathion, endosulfan and chlorpyrifos are the log mean of the values
reported in the Log P database [75]. The KS>WS for acenaphthene, endrin, phenanthrene and dieldrin are those measured
from the slow stir flask method.
WQC documents  (see Table 4) for the remaining
chemicals plotted. Measurement of Kows for ace-
naphthene,  endrin,  phenanthrene,  dieldrin  and
fluoranthene are from the slow stir flask method as
previously presented. The Kows for the remaining
chemicals are the log averages of the values reported
in the Log P database [75]. While the SQCs for ace-
napthene, endrin, phenanthrene, fluoranthene and
dieldrin meet the  minimum database requirements
presented in the previous section, SQCs for the re-
maining chemicals are for illustrative purposes only
and should not be considered final SQC values. Final
SQC, when published, should reflect the best current
information for both FCV and Kow.
       The FCVs that are available for nonionic or-
ganic insecticides range  from approximately 0.01
Ug/L to 0.3 ug/L, a factor of 30. The SQCocs range
from approximately 0.01 ug/g organic carbon to in
excess of 10 ug/g organic carbon,  a factor of over
1,000. This increased range in values occurs because
the KowS of these chemicals span over two orders of
                           magnitude. The most stringent SQCoc in this exam-
                           ple is for chlordane, a chemical with one of the low-
                           est KowS among the chemicals with an FCV of
                           approximately 0.01 ug/L.
                                 By contrast, the PAHs included in this exam-
                           ple have a range of FCVs and Kows of approximately
                           an order of magnitude. But these values vary in-
                           versely: The chemical with the larger FCV has  a
                           smaller K0w. The result is that the SQCocs are ap-
                           proximately the same,  240 ug/g  organic carbon.
                           Classes of chemicals for which the effects concentra-
                           tions decrease logarithmically with increasing KowS,
                           for example, chemicals that are narcotics [113], will
                           have SQC that are more nearly constant.

                           Field  Data

                           Information on actual levels of the criteria chemicals
                           in the environment was assembled in order to pro-
                           vide an indication of the relationship between the

-------
                         Sediment Quality Criteria Using Equilibrium Partitioning
SQC concentrations and the actual concentration lev-
els observed in the sediments of U.S. surface water
bodies. Three separate databases were examined:

      • EPA's STORE! database .[114],

      • NOAA's National Status and Trends data-
        base, which focused on water bodies in
        coastal areas [47], and the

      • Corps of Engineers  database for San Fran-
        cisco Bay [116].

      The data that were retrieved have been sum-
marized on probability graphs that are presented in
the subsections  that follow for each of the data
sources. A large proportion of the observations are
below detection limit values and indicate only that
the actual concentration is unknown, but less than
the concentration  plotted. These data are  plotted
with a "less than" symbol. As a result, the probability
plots should not be interpreted as representations of
.the actual probability distribution of the monitored
samples. They do, however, provide a useful visual
indication of the range  of concentration levels of the
study chemicals in natural sediments.
       A suggestion of the probable extent to which
problem sediments might be encountered  is  pro-
vided by the plot overlay showing the SQC concen-
tration developed by this research. In the case of the
STORET data, the SQC is shown as a band because
the foe is not reported. The lines represent the SQC
for between an foe =  1  to 10  percent. The other two
data bases provide  the necessary information on
sediment organic carbon levels, and the results have
been properly normalized.
       Some salient  features of the available field
data displayed by the plots are summarized in Table
18. The SQC concentration is listed for each of the
five criteria chemicals,  together with the number of
samples and the approximate percentage of the sam-
ples that exceed the SQC. The table also lists the sedi-
ment concentrations that are exceeded by 10,5, and 1
percent of the measurements.
       We recognize that the tabulated information
represents only  approximate estimates, because of
the presence of large numbers of detection limit val-
ues. Nevertheless, it provides what we consider to be
a reliable expectation that only a small percentage of
sediment sites in the databases, less than 5 percent,
will have concentrations that exceed the SQC levels.
       We did not attempt a more rigorous analysis
to provide a more definitive characterization of the
spatial and temporal features of the database. Some
of the recorded data  dearly represent multiple sam-
ples at a particular site. The very high observed con-
centrations are relatively few in absolute number and
may reflect multiple samples at one or a few particu-
 larly contaminated sites. Some of the probability
 plots also show a discontinuity at the .high end. This
 is particularly true of the Corps of Engineers data
 that pool results from a limited number of stations in
 San Francisco Bay. Until a more detailed analysis is
 performed, the results of the preliminary screening
 should be considered approximate, upper bound es-
 timates of the probable prevalence of sediment sites
 that may exceed the SQC.
        STORET data. A STORET data retrieval was
 performed to obtain a preliminary assessment of the
 concentrations of the criteria chemicals in the sedi-
 ments of the nation's waterbodies. The data retrieved
 was restricted to samples  measured in the period
 1986 to 1990. The selection of this recent period elimi-
 nated much of the older data with the higher detec-
 tion limits to provide a more accurate  indication of
 current conditions. Log probability plots concentra-
 tions are shown in Figures.38 and 39. Concentrations
 are shown on a dry weight basis, because sediment
 organic carbon is not reported. The SQCs are com-
 puted on the basis of a sediment organic carbon con-
- tent (foe) of 1 percent and  10  percent,  which is the
 typical range for inland sediments. The STORET data
 distinguishes between the type of waterbody, and
 separate displays  are  provided  for  stations on
 streams, lakes, and estuaries.
        The PAH data are shown in Figure 38. The to-
 tal number of samples, and the number of detected
 samples  are indicated on  the  figures. The  plotted
 points are restricted to a subset of the total number of
 samples, so that the plots are legible. A few samples
 with detected concentrations, the solid symbols, ex-
 ceed the SQC  for foc  = 1 percent, and fewer exceed
 the SQC for/oc = 10 percent. The nondetected data,
 plotted at the detection limit with "<", are below the
 value indicated on the plot. In fact with nondetected
 data included in the probability plot, the actual plot-
 ting positions of the detected data is uncertain, since
 the nondetected data may in fact occupy plotting po-
 sitions further to the left, at lower probabilities. Thus
 the exceedence probabilities for the detected data are
 at least as large as is indicated on the plots. Approxi-
 mately 5 percent or less of the detected samples ex-
 ceed the/oc = 1 percent SQC.
        The data for endrin and dieldrin are shown in
 Figure 39. Similar results are obtained. Less than 3
 percent of the detected dieldrin and endrin samples
 exceed the lower SQC.
        National  Status and Trends Program data.
 NOAA's National Status and Trends Program devel-
 oped a database on the quality of marine sediments
 focusing on estuarine and coastal sites that are not in
 close proximity to known sources of contamination
 [116]. Figure 40 displays the distribution of sediment
 concentrations from the National Status and Trends
 Program sites for four of the five criteria chemicals

-------
                         Sediment Quality Criteria Using Equilibrium Partitioning
 Table 18.—Observed quality of natural sediments.

SALTWATER
CORPS OF
ENGINEERS
NOAA
NATIONAL STATUS
AND TRENDS
PROGRAM
EPA STORE!
ESTUARIES
FRESH WATER
EPA STORET
STREAMS
EPA STORET LAKES

SQC (|ig/g oc)
NO. OF SAMPLES
% that exceed SQC
10% exceed - conoc
5% exceed - conoc
1% exceed - cone
NO. OF SAMPLES
% THAT EXCEED SQC
10% exceed - cone
5% exceed - cone
1% exceed - cone
SQC(ng/g)for
sediment OC: 1-10%
NO. OF SAMPLES
% THAT EXCEED SQC
10% exceed - cone
5% exceed - cone
1% exceed - cone
SQC (ng/g) for
sediment OC: 1-10%
NO. OF SAMPLES
% THAT EXCEED SQC
10% exceed - cone
5% exceed - cone
1% exceed - cone
NO. OF SAMPLES
% THAT EXCEED SQC
10% exceed - cone
5 exceed - cone
1% exceed - cone
FIJUORANTHENE
293
231
3%
40
300
40,000
797
0.2%
4
7
40
3.0-30
88
1%
4
7
40
6.2-62
786
2%
4
7
40
57
5%
4
7
40
PHENANTHRENE
238
231
4%
60
300
25,000
736
0.1%
25
40
90
2.4-24
87
<1%
0.1
0.8
1.0
1.8-18
584
7%
1
2
? .
50
<0%
K?)
2(?)
10(?)
ACENAPTHENE
232
130
5%
30
200
3,000
245
0%
4
7
40
2.3-23
74
0%
0.3
0.3
0.5
1.3-13
681
<4%
1
1.5
40
56
2%
0.7
1
3
ENDRIN
0.744
260
0%
60
150
700
-
-
-
-
-
.007-0.07
^
150
<10%
all
data are

-------
52 Sediment Quality Criteria Using Equilibrium Partitioning i 1
f&
h
m***
*o5
!§
IS
••S'
»
n
10
10
V
*>
»
n
»
•n
«
X
•i
:
-i
o.
4
I
I
1
0
.<
J
-1
-4
*^t ^
STREAM
r~
r
r
i 1
5 LWE ^
r
r
r" '
f— '
r
r
r
TOTAL SAJbFLES: 881
I£ASU£D SA1FLES: 53
' lf~
r*~~
TO 2O SO SO SO M
TOTAL' SAiFLES: 58 '
MEASURED SAVPLES: 5

_•*««*<
««««*
•
  SO

                         PROBABLJTY
- ESTUART TOTAL SAllpUS: 8?'
f MEASURED SMFLES: 28
L-l -- -- -1

r
r ^««««
\
r
r. _«
	 - 	 3
-•
2
3
-- «r< .-- *
1
i
1
20     fo    »o eo

PROBABLTTY
                                 n
                                 10
                                 n
                                 «
                                 10
                                 10
                                 n
                                 •n
' STREAM
r

r
r
!"«««««

L1 1
' LAKE

r
r
r
r
r
TOTAL SAMPLES: 7B£-
fcCASURED SAUUS: V.»


' ^Pk«««^*<«^«<<^^^
.-/

10 20 00 MM
TOTAL SAMPLES: 5?'
MEASURED SAVPLES: 28

^f^

--r-.ir... .^
•"1
.^ T
i
i
i
i
1
M M.
' ' ' ' 	 !
I

:
1
n
1
V)
n
V
n
•n
10
«
n
4
a
t


.1
-4
J

- ESTUARY TOTAL SAMR£S: 88
T MEASURED SAIFLES: 32
c 	
r .. ..
r
r ^f*""™
r , ««««<«,«<^^
r
r

i
j
• -\

1
i
i
, ' 	 T
                                                   PROBABLTTY
Figure 38.—Comparison of sediment quality criteria for sediments containing 1 percent and 10 percent organic carbon to
the distributions of three sediment PAH concentrations from the U.S. EPA STORE! database from 1986 to 1990. Samples
above the detection limit (filled symbols) and samples less than the detection limit (less than symbols) are shown. Data from
U.S. EPA [114].

-------
10'
10°
io-'
io •'
10 ":

10 "!
10"'
         •«#
  0.1
10 '
10°  •
10"'[
10'2 'r
10
10
10

  0.1
    '3
                              Sediment Quality Criteria Using Equilibrium Partitioning
                TOTAL SAMPLES: 2677
               . MEASURED SAMPLES: 67
                   10  20
                                  60  90
                                              99
                                                   99,9
LAKE TOTAL SAMPLES: 478
r MEASURED SAMPLES: 12
. "
i <^
o  ._
   10-'
   10"
   10"
   10"
   10
               TOTAL SAMPLES:  ISO
               MEASURED SAMPLES: 0
                 <
             ,<«»;
                 10  20    SO     80
                   PROBABILITY
                                                                 JO
                                                                 10"
                                                                        ESTUARY
                                                                     0.1
                                                                                 TOTAL SAMPLES:  160
                                                                                 MEASURED SAMPLES: 3
                                                                                      PROBABILITY
                                                                                                           « <   i
Figure 39.—Comparison of sediment quality criteria for sediments containing 1 percent and 10 percent organic carbon to
the distributions of two sediment pesticide concentrations from the U.S. EPA STORE! Database.  Data are from 1986 to
1990. Samples above the detection limit (filled symbols) and samples less than the detection limit (less than symbols) are
shown. Data from U.S. EPA [114].
        Table 19.—San Francisco Bay sediment samples.
LOCATION
Port of San Francisco: Piers 27-29, 35, 38. 48, 70, 80 and 94
Rsherman's Wharf and Islais Creek
Suisun Channel
West Richmond
Pinole Shoal
Carquinez Strait
Mare Island Strait
Richmond Harbor Channel
Santa Fe Channel
Outer and Inner Richmond Harbor Channel
Port of Oakland Tier II: Berths 20-23, 2!>, 26-30, 31, 35-38, 60-63 and 82-84
Port of Oakland Outer and Inner Harbor
Treasure Island . -
San Leandro Bay
San Pablo Bay
NO. OF SAMPLES
PAHt ft PESTICIDES
21
2
6
11
44
10
6
48
6
6
40
27
5 composites
1 composite
6
YEARS
1988 and 1990
1990
1991
1990
1990
1990
1990
1990
1990
1991
1989-1990
1990-1991
1990
1990
1989-1990
   Note: PAHs = Fluoranthene, Phenanthrene, Acenapthene; Pesticides = Dieldrin, Endrin.

-------
                          Sediment Quality Criteria Using Equilibrium Partitioning
    10000
     1000
      100
=8    «

if     1
J3    ai
      aoi
     aooi
                    i IIHIIIJ - 1 — i i |  i i — | - |iinn i i

                   SQCoc
                            O - all f o,. (N=288)
                            •- foe > °-2« (N=245) I
               I lltl«l  I I Illllll
 100

  10



  0.1

 aoi

aooi
                       10 20   60   8080
                                                          ftOOOl
                                        oo oao
    0.1
10 20   SO   8090
                                                                                             00  OOA
                                                          10000

                       10 20   60   8OOO

                        PROBABUTY
                                                                     10 20   60   8000

                                                                      PROBABUTY
                                  00  OOJO
Figure 40.—Comparison of sediment quality criteria to the distributions of acenaphthene, fluoranthene, dieldrin, and phen-
anthrene organic carbon normalized sediment concentrations from NOAA's National Status and Trends Program. Data are
from 1984 to 1989. Samples with organic carbon greater than 0.2 percent (filled symbols) and samples for all organic carb-
on contents (open symbols) are shown. Data from NOAA [47].
the qualify of dredged sediments in order to deter-
mine their suitability for open water disposal. The
database did not indicate what determinations were
made concerning their acceptability for this purpose.
       Investigators compared the frequency of occur-
rence (in individual samples, not dredge sites) with the
SQC criteria  developed using the EqP methodology.
The major portion (93 percent) of the samples analyzed
had organic carbon fractions greater than 02 percent,
for which the SQC concentrations are  applicable. The
concentrations of  each chemical  measured in  these
sediments was normalized by the organic carbon con-
tent and the results are displayed below as probability
plots to illustrate the frequency at which different levels
are observed. Results are presented for the five criteria
chemicals. A horizontal line at the  concentration value
of the SQC provides a reference that indicates the rela-
tionship between observed range of quality and the
SQC for each chemical.
       PAH  results  are  summarized in Figure 41.
Less than 5 percent of the individual samples con-
                                                  tained concentrations  in excess of  the  sediment
                                                  quality. It is informative to note that the small set of
                                                  very high concentrations are nearly all from one sam-
                                                  ple site (Treasure Island). These samples are respon-
                                                  sible for the discontinuous pattern of the  frequency
                                                  distribution.
                                                        Figure 42 presents the monitoring program
                                                  results for the two pesticides in the same format. In
                                                  this case, virtually all of the samples were less than
                                                  the varying detection limits  of the analytical tests.
                                                  Each of the samples for which actual measurements
                                                  were obtained were at least an order of magnitude
                                                  lower than the SQC. An estimate of the possible fre-
                                                  quency distribution  of sediment concentrations of
                                                  dieldrin and endrin was developed by the applica-
                                                  tion of an analysis technique that accounts for the
                                                  varying detection limits and the presence  of nonde-
                                                  tected observations [117]. The results are illustrated
                                                  by the straight line, which suggests that no apprecia-
                                                  ble number,of exceedences is expected. However, the
                                                  virtual absence of detected concentrations makes the

-------
                          Sediment Quality Criteria Using Equilibrium Partitioning
                    if
10'

10 "

10 3

10 '

10 t

10°

10"

10"
1 	 '
F
r SQCOC
r
r
r •
r "
• 1 1 ,,i, ni i.i. i mi
1 	 ' 	 ' 	 i
,. 	 «•• • ^
• 1
1
• ° •*°° 1
• all foc (N-112) :
0 foc > 0.2X (N-105) 1
i , , i , , i i, 	 , |MI 	 •
                            0.1
                                           10  20
                                                      50
                                                            80  90
                                                                        99
                                                                             99.9
                    ^^
                     §


                    if
                          10'
                          10
                          10-

                          10'

                          102

                          10l

                          10 J

                          10 C

                          10"

                          10"
                                                               all foc  (N-207)
                                                               foc > 0.2X (N-190) 1
                                           10  20
                                                      50
                                                            80  90
                                                                        99
                                                                             99.9
                                    all foc (N-207)
                                    foc > 0.2X  (N-190)
                             0.1
                                           10  20
                                                      SO
                                                            80  90
                                                                        99
                                                                             99.9
                                               PROBABUTY
Rgure 41.—Comparison of sediment quality criteria to the distributions of acenapthene, fluoranthene, and phenantnrene or-
ganic carbon normalized sediment concentrations from the U.S. Army Corps of Engineers monitoring program of San Fran-
cisco Bay. Samples with organic carbon greater than 0.2 percent (filled symbols) and samples for all organic carbon contents
(open symbols) are shown. See Table 19 for description of location, number of samples, and sample period. Data from U.S.
Army COE [115].
distribution estimates unreliable. They are presented
only to suggest the probable relationship between
the levels of these two pesticides in relation to sedi-
ment quality criteria.
CONCLUSIONS
The technical basis and data that support the use of
the EqP method to generate SQC have been pre-
sented for nonionic organic chemicals. The use of or-
ganic carbon  normalization is equivalent to  using
pore water normalization as a means of accounting
                           for varying bioavailability (Figs. 2,3, 5-8,26-28). The
                           variation in organism body burden across sediments
                           can also be significantly reduced if organic carbon
                           and lipid normalization are used (Figs. 29-31). For
                           contaminated sediments, particle size effects are re-
                           moved if organic carbon-normalized concentrations
                           are compared (Figs. 17,19,21). The reason is that or-
                           ganic carbon is the proper  normalization for parti-
                           tioning  between   free  dissolved  chemical  and
                           sediment-bound chemical (Fig. 11).
                                  Using  pore  water normalization for highly
                           hydrophobic chemicals is complicated  by chemical
                           complexing to DOC (Fig. 13). Partitioning between

-------
                          Sediment Quality Criteria Using Equilibrium Partitioning
                    10
                        0.1
         01*09
         5s
                        0.1
10   20      60       80  00

     PROBABILITY
09     99*
Figure 42.—Comparison of sediment quality criteria to the distributions of endrin and dieldrin organic carbon normalized
sediment concentrations from the U.S. Army Corps of Engineers monitoring program of San Francisco Bay. See Table 19 for
description of location, number of samples and sample period. Samples with organic carbon greater than 0.2 percent (filled
symbols) and samples less than the detection limit (less than symbols) are shown. Also shown is an estimate of the distribu-
tion developed by accounting for nondetected observations (solid line). Data from U.S.  Army COE [115].
pore water and sediment organic carbon from field-
collected sediments can be rationalized if DOC com-
plexing is taken  into  account (Figs. 22  and 23).
However, the complexed chemical appears not to be
bioavailable (Fig. 15).
       These observations are consistent  with the
EqP model, which assumes the equivalence of water-
only exposure  and  the exposure from pore water
and/or sediment  organic carbon. Sediment quality
criteria are based on organic carbon normalization
because pore water normalization is complicated by
DOC complexing for highly hydrophobic chemicals.
       The justification for using the FCV from the
WQC to define the effects level for benthic organisms
has also been discussed. Water column and benthic
organisms appear to have similar sensitivities for
both the most sensitive species tested (Fig. 32) and all
tested  species (Fig. 34). Benthic colonization experi-
           ments also demonstrate that WQC can be used to
           predict effects concentrations for benthic organisms.
           A direct statistical test of the equality of the distribu-
           tions can be used to confirm or refute this assump-
           tion for individual chemicals (Fig. 36).
                 Equilibrium partitioning cannot remove all of
           the observed variation from sediment to sediment. It
           does reduce the much larger sediment-to-sediment
           variation that exists if no corrections for bioavailabil-
           ity are made (Figs. 5-8). A variation factor of approxi-
           mately two to three  remains (Figs. 2 and  3),  which
           includes measurement and other sources of variabil-
           ity. This is not unexpected as EqP is an idealization of
           the actual  situation. Other factors that are  not
           considered  in the model play roles in determining
           biological effects. Hence, it is recognized that a quan-
           tification of the uncertainty should accompany the
           SQC that reflect these additional sources of variation.

-------
                           Sediment Quality Criteria Using Equilibrium Partitioning
 Research Needs

 The final validation of SQC will come from field
 studies that are designed to evaluate the extent to
 which biological effects can be predicted from SQC.
» The colonization experiments (Table 6) are a labora-
 tory simulation of a field validation. Sediment qual-
 ity criteria can possibly be validated more easily than
 WQC because determining the organism exposure is
 more straightforward. The benthic population expo-
 sure is quantified by the organic carbon-normalized
 sediment concentration.
        It has been suggested that the kinetics of PAH
 desorption from sediments control the chemical body
 burden of a benthic amphipod  [118]. The extent to
 which kinetics can be important in fit'ld situations is
 unknown at present, and field studies would be an
 important component in examining this question. In
 addition, more laboratory  sediment toxicity tests,
 particularly  chronic tests involving multiple  sedi-
 ments, would also be helpful. In a typical practical
 application of SQC, mixtures of  chemicals are  in-
 volved. The extension of EqP methodology to mix-
 tures  would be  of  great  practical value. Initial
 experiments indicate that it should be possible [119].
       The EqP method  is presently restricted to •
 computing effects-based-criteria for the protection of
benthic organisms. The direct extension of this meth-
odology for computing sediment  criteria that are
protective of human health, wildlife, and marketabil-
ity of fish and shellfish requires that the equilibrium
assumption be extended to the water column and to
water column organisms. This assumption is, in gen-
eral, untenable. Water column concentrations can be
much lower than pore water concentrations if suffi-
cient dilution flow is present. Conversely, upper-tro-
phic-level  organisms are at concentrations  well
above equilibrium values [120]. Hence, the applica-
tion of the final residue values from the WQC for the
computation of SQC, as was done for certain interim
criteria [121], is not technically justifiable. At present,
organism lipid-to-sediment organic carbon ratios,
that is, BSFs (Eqn. 29), might be useful in estimating
the concentration of contaminants in benthic species,
for which the assumption of equilibrium is reason-
able. However, a site-specific investigation (e.g., Con-
nolly [122]) appears to be the only available method
for performing an evaluation of the effect of contami-
nated sediments on the body burdens of upper-tro-
phic-level organisms.         3
REFERENCES

  1. Dickson, K.L., A.W. Maki, and W.A. Bmngs, eds. 1987.
    Fate  and Effects of Sediment-Bound Chemicals  in
    Aquatic Systems. Proceedings of Sixth Pellston Work-
    shop, Pergamon Press, Elmsf ord, NY.
  2. Pavlou, SP., and D.P. Weston. 1983. Initial evaluation of
    alternatives for development of sediment related crite-
    ria for  toxic  contaminants in  marine waters, Puget
    Sound.  Phase I. Development of conceptual frame-
    work. Technical Report. JRB Associates, Bellevue, WA.

  3. Chapman,  G.A. 1987. Establishing Sediment Criteria
    for Chemicals - Regulatory Perspective. Pages 355-77 in
    K.L. Dickson, A.W. Maki, and W.A. Brungs, eds., Fate
    and Effects of Sediment-Bound Chemicals in Aquatic
    Systems. Pergamon Press, Elmsf ord, NY.

  4. Bolton, S H., R.J. Breteler, B.W. Vigon, J.A. Scanlon, and
    S,L. dark.  1985. National Perspective  On Sediment
    Quality. EPA 68-01-6986. U.S. Environmental Protection
    Agency, Office of Water Regulations and Standards,
    Criteria and Standards Division, Washington, D.C.

  5. Pavlou,  SP., and D.P. Weston. 1983. Initial evaluation of
   alternatives for development of sediment related crite-
   ria for  toxic contaminants in marine waters (Puget
 .  Sound) - Phase I and Phase H. EPA 68-01-6388. U.S. En-
   vironmentalProtection Agency, Office of Water Regula-
   tions and Standards, Criteria and Standards Division.
   Washington, DC.

  6. JRB Associates.  1984. Background and review docu-
   ment on the development of sediment criteria. EPA 68-
   01-6388. U.S. Environmental Protection Agency, Office
   of Water Regulations  and Standards,  Criteria  and
   Standards Division, Washington, DC.

  7.Battelle. 1984.  Sediment quality criteria  development
   workshop. U.S. Environmental Protection Agency, Of-
   fice of Water Regulations and Standards, Criteria  and
   Standards Division, Washington, DC.

 8. Stephan, C.E., D.I. Mount, D.J.  Hansen, LH. Gentile,
   G.A. Chapman, and W.A. Brungs. 1985. Guidelines for
   deriving numerical national water quality criteria for
   the protection of aquatic organisms and their uses.
   PB85-227049. National Technical Information Service.

 9. Spehar, R.L., and A.R. Carlson. 1984. Derivation of site-
   specific water quality criteria for cadmium and the St.
   Louis River basin, Duluth, Minnesota. Environ. Toxicol.
   Chem. 3:651-55.

10. Carlson,  A.R., H. Nelson, and D. Hammermeister. 1986.
   Development  and validation of  site-specific water
   quality criteria for copper. Environ. Toxicol. Chem.
   5:997-1012.

ll.Luoma, S.N., and W. Bryan. 1981.  A statistical assess-
   ment of the form  of trace metals in oxidized sediments
   employing  chemical extractants. Sci.  Total Environ.
   17:165-%.

12. Chapman, EM., and E.R. Long. 1983. The use of bioas-
   says as part of a comprehensive approach to marine
   pollution assessment. Mar. Pollut. Bull. 14:81-84.

13. Long, E.R., and P.M. Chapman. 1985. A sediment qual-
   ity triad:  Measures of sediment contamination, toxicity
   and infaunal community composition in Puget Sound.
   Mar. Pollut. Bull. 16:405-15.

-------
                           Sediment Quality Criteria Using Equilibrium Partitioning
14.Barrick, R.C, D.S. Becker, DP. Weston, and T.C. Ginn.
   1985. Commencement Bay nearshore/tideflats reme-
   dial investigation. Final Report.  EPA 910/9-85-134b.
   Prepared by Tetra Tech, for the Washington Depart-
   ment of Ecology and the U.S. Environmental Protection
   Agency. Washington, DC.

15. Long, EJt, and L.G. Morgan. 1990. The potential for
   biological effects of sediment  — sorbed contaminants
   tested in  the National Status and Trends Program.
   NOAA Technical Memorandum NOS OMA 52. Office
   of Oceanography and Marine Assessment, Seattle, WA.

16. Cowan, C.E., and CS. Zarba. June  1987. Regulatory ap-
   plications of sediment quality criteria - Final Report.
   EPA 68-01-6986. U.S. Environmental Protection Agency,
   Office of Water Regulations and Standards, Criteria
   and Standards Division, Washington, DC.

17. Adams, W.J., R-A. Kimerle, and  R.G. Mosher. 1985.
   Aquatic safety assessment of chemicals sorbed to sedi-
   ments. Pages 429-53 in R.D. Cardwell, R. Purdy, and
   R.C. Banner, eds., Aquatic Toxicology and Hazard As-
   sessment: Seventh Symposium. STP 854. American So-
   ciety for Testing and Materials, Philadelphia, PA.

18.Ziegenfuss, P.S., W.J.  Renaudette, and  W.J.  Adams.
   1986. Methodology for assessing  the acute toxicity of
   chemicals sorbed to sediments: Testing the equilibrium
   partitioning theory. Pages 479-93 in T.J. Poston, and R.
   Purdy, eds., Aquatic Toxicology  and Environmental
   Fate. 9th Volume. STP 921. American Society for Testing
   and Materials, Philadelphia, PA.

19.Swartz,  R.C,  D.W.  Schults,  T.H.  DeWitt,  G.R.
   Ditsworth, and  J.O.  Lamberson. 1990. Toxicity of
   fluoranthene in sediment to marine amphipods: A test
   of the equilibrium partitioning approach to sediment
   quality criteria. Environ. Toxicol. Chem. 9:1071-80.

20.Nebeker, A.V. G.S. Schuytema, W.L. Griffis, J.A. Bar-
   bitta, and LA. Carey. 1989. Effect of sediment organic
   carbon on survival of Hyalella azteca exposed to DDT
   and endrin. Environ. Toxicol. Chem. 8:705-18.

21. Schuytema, G.S.,  A.V.  Nebeker, W.L. Griffis, and C.E.
   Miller. 1989. Effects of freezing on  toxicity of sediments
   contaminated with DDT and Endrin. Environ. Toxicol.
   Chem. 8:883-91.

22.Muir, D.C.G., GP. Rawn, B.E. Townsend, W.L. Lock-
   hart,  and R. Greenhalgh. 1985.  Bioconcentration of
   cypermethrin, deltamethrin,  fenvalerate,  and  per-
   methrin by Chironomus tentans larvae in sediment and
   water. Environ. Toxicol. Chem. 451-61.

23. Adams, W.J., R.A.  Kimerle, and R.G. Mosher. 1983.
   Aquatic safety assessment of chemicals sorbed to sedi-
   ments. Special Study Report No. ES-EAG-83-1. Mon-
   santo Company,  Environmental  Sciences Center, St.
   Louis, MO.

24. Adams, W.J. 1987. Bioavailability of neutral lipophilic
   organic chemicals contained in sediments: A review.
   Pages 219-44 in K.L. Dickson, A.W. Maki, and W.A.
   Brungs,  eds., Fate and  Effects  of Sediment-Bound
   Chemicals  in  Aquatic  Systems.  Pergamon  Press,
   Elmsford, NY.

25. Karickhoff, S.W.  1984. Organic pollutant sorption in
   aquatic systems. J. Hydraul. Div. ASCE 110:707-35.

26. O'Connor, D.J., and J. Connolly. 1980. The effect of con-
   centration of adsorbing solids on the partition coeffi-
   cient. Water Resour. 14:1517-23.

27. Di Toro, D.M. 1985. A particle interaction model of re-
   versible  organic  chemical  sorption.  Chemosphere
   14:1503-38.

28. Benes, P., and V. Majer. 1980. Trace Chemistry of Aque-
   ous Solutions. Elsevier, New York, NY.

29. Gschwend, P.M. and S. Wu. 1985. On the constancy of
   sediment-water partition coefficients of hydrophobic
   organic pollutants. Environ. Sci. Technol. 19:90-96.

30. Carter, C.W.,  and I.H. Suffett, 1983. Interactions be-
   tween dissolved humic and fulvic acids and pollutants
   in aquatic environments. Pages 215-30 in R.L. Swann
   and A. Eschenroeder, eds., Fate of Chemicals in the En-
   vironment. ACS Symposium Series 225.  American
   Chemical Society, Washington, DC.

31. Voice, T.C, CP. Rice, and W.J. Weber,  Jr. 1983. Effect of
   solids concentration on the sorptive partitioning of hy-
   drophobic pollutants in aquatic systems. Environ. Sci.
   Technol. 17:513-18.

32. Curl, R.L., and G.A. Keolelan. 1984. Implicit-adsorbate
   model for apparent anomalies with organic adsorption
   on natural adsorbents.  Environ. Sci. Technol. 18:916-22.

33. Nelson, DM.,  W.R.  Penrose, J.O.  Karttunen,  and
   Mehlhaff. 1985. Effects of dissolved organic carbon on
   the adsorption properties of plutonium in natural wa-
   ters. Environ. Sci. Technol. 19:127-31.

34. Karickhoff, S.W., and K.R.  Morris. 1985. Sorption dy-
   namics of hydrophobic pollutants in sediment suspen-
   sions. Environ. Toxicol. Chem. 4:469-79.

35. Di Toro, D.M., and L. Horzempa. 1983. Reversible and
   resistant component model of hexachlorobiphenyl ad-
   sorption-desorption: Resuspension and dilution. Pages
   89-114 in D. Mackay, S. Paterson, S.J. Eisenreich, and
   M.S. Simmons, eds., Physical Behavior  of PCBs in the
   Great Lakes. Ann Arbor Science, Ann Arbor, MI.

36. Di Toro, D.M., J.D. Mahony, P.R. Kirchgraber, A.L.
   O'Byrne, L.R. Pasquale, and D.C. Piccirilli. 1986. Effects
   of nonreversibility, particle concentration, and ionic
   strength on heavy metal sorption. Environ. Sci. Tech-
   nol. 20:55.

37. Mdlroy, L.M., J.V. DePinto, T.C. Young, and S.C. Mar-
   tin. 1986.  Partitioning of heavy metals to suspended
   solids in the Flint River, Michigan. Environ. Toxicol.
   Chem. 5:609-23.

38. Mackay, D., and B. Powers.  1987. Sorption of hydro-
   phobic chemicals from water: A hypothesis for the
   mechanism of the particle concentration effect. Che-
   mosphere 16:745-57.

-------
Sediment Quality Criteria Using Equilibrium Partitioning
 39. Eadie,  B.J., N.R. Morehead, and P.E Landrum.  1990.
    Three-phase partitioning of hydrophobic organic com-
    pounds in Great Lakes waters. Chemosphere 20:161-78.

 40. Thurman, EM. 1985. Organic Geochemistry of Natural
    Waters. Martinus Nijhoff/Dr. W. Junk Publ., Dordrecht,
    The Netherlands.

 41. McCarthy, J.F., and B.D. Jimenez. 1985. Reduction in
    bioavailability to bluegills of polycydic aromatic hy-
    drocarbons bound to dissolved humic material. Envi-
    ron. Toxicol. Chem. 4:511-21.

 42. Landrum, P.P., S.R. Nihart, B.J. Eadie, and L.R. Herche.
    1987. Reduction in bioavailability of organic contami-
    nants to the amphipod Pontoporeia hoyi by dissolved or-
    ganic matter of sediment interstitial waters. Environ.
    Toxicol. Chem. 6:11-20.

 43. Landrum, P.P., M.D. Reinhold, SJl. Nihart, and B.J.
    Eadie. 1985. Predicting  the bioavailability of organic
    xenobiotics to Pontoporeia hoyi in the presence of humic
    and fulvic materials and natural dissolved organic mat-
    ter. Environ. Toxicol. Chem. 4:459-67.

,44.Prahl, EG. 1982. The geochemistry of polycydic aro-
    matic hydrocarbons in Columbia River and Washing-
    ton coastal sediments. Ph.D. thesis. Washington State
    University, Pullman, WA.

 45. Evans, KM., R.A. GUI, and P.W.J. Robotham. 1990. The
    PAH and organic content of sediment particle size frac-
    tions. Water Air and SoU Pollut. 51:13-31.

 46. Delbeke, K, R.J. Claude, and M. Bossicart. 1990. Or-
    ganochlorides in different fractions of sediments and
    different planktonic compartments of the Belgian Con-
    tinental Shelf and Schelt Estuary. Environmental Pollu-
    tion. 66:325-49.

 47. National  Oceanic  and Atmospheric Administration.
    1991. Second summary of data on chemical contami-
    nants in sediments from the National Status and Trends
    Program.  NOAA Technical Memorandum NOS OMA
    59. Office of Oceanography and Marine Assessment,
    Rockville, MD.

 48. BrownaweU, B.J., and J.W. Farrington. 1986. Biogeo-
    chemistry of PCBs in interstitial waters; of a coastal ma-
    rine sediment. Geochim. Cosmochim. Acta 50:157-69.

 49. BrownaweU, B.J., and J.W. Farrington. 1985. Partition-
    ing of PCBs in marine sediments. Pages 97-119 in A.C.
    Sigleo, and A. Hattori, eds., Marine and Estuarine Geo-
    chemistry. Lewis Publishers, Chelsea, MI.

 50. Socha, S.B., and R. Carpenter. 1987. Factors affecting
    pore water hydrocarbon concentrations in Puget Sound
    sediments. Geochim. Cosmochim. Acta 51:1237-84.

 51. Oliver, B.C. 1987. Biouptake of chlorinated hydrocar-
    bons from laboratory-spiked and field sediments by
    oligochaete worms. Environ. Sci. Technol. 21:785-90.

 52. Stehly, G.R. 1991. Memorandum to W.J. Berry, April 25,
    1993.
                              53. Hoke, R., and G. Ankley. 1991. Results of dieldrin sedi-
                                 ment spiking study conducted in support of USEPA de-
                                 velopment of sediment quality criteria. Memorandum
                                 to D. Hansen and D. Di Toro. June 18,1991.

                              54. Hoke, R., and G. Ankley. 1992. Results of dieldrin sedi-
                                 ment spiking study, Airport Pond retest, conducted in
                                 support of USEPA development of sediment quality
                                 criteria. Memorandum to D. Hansen and D.DiToro.
                                 January 27,1992.

                              55. Hoke, R. 1992.  Results of dieldrin sediment  spiking
                                 study conducted in support of USEPA development of
                                 sediment quality criteria. Memorandum to D. Hansen,
                                 D. Di Toro, and G. Ankley. December 2,1992.

                              56.Swartz, R.C.  1991. Acenaphthene and phenanthrene
                                 files. Memorandum to David Hansen, June 26,1991.

                              57.DeWitt, T.H., R.J. Ozretich, R.C. Swartz, J.O. Lamber-
                                 son,  D.W. Schults, G.R. Ditsworth,  JJCP. Jones,  L.
                                 Hoselton, and LM. Smith. 1992. The influence of or-
                                 ganic matter quality on the toxicity and partitioning of
                                 sediment  association  fluoranthene.   Environ.  Tox
                                 Chem. 11:197-208.

                              58. Chiou, C.T. 1985. Partition coefficients of organic com-
                                 pounds in lipid-water systems and correlations with
                                 fish bioconcentration factors. Environ. Sci. Technol
                                 1957-62.

                              59.Thomann, R.V., J.P Connolly, and TJ. Parkerton. 1991.
                                 An equilibrium model of organic chemical accumula-
                                 tion in aquatic foodwebs with sediment interaction. En-
                                 viron. Toxicol. Chem. (in press).

                              60. McFarland, V.A. 1984. Activity-based evaluation of po-
                                 tential bioaccumulation for Sediments. Pages 461-67 in
                                 R.L. Montgomery, and J.W. Leach, eds., Dredging and
                                 Dredged Material Disposal. Vol. 1. American Society of
                                 Civil Engineers, New York, NY.

                              61. Lake, J.L., N. Rubinstein, and S. Pavignano. 1987. Pre-
                                 dicting bioaccumulation: Development of a simple par-
                                 titioning model  for  use  as a  screening tool for
                                 regulating ocean disposal of wastes. Pages 151-66 in
                                 K.L. Dickson, A.W. Maki, and W.A. Brungs, eds.. Fate
                                 and Effects of Sediment-Bound Chemicals in Aquatic
                                 Systems. Pergamon Press, Elmsford, NY.

                              62. Rubinstein, N.I., J.L. Lake, R.J. Pruell, H. Lee II, B. Ta-
                                plin, J. Heltshe, R. Bowen, and S. Pavignano. 1988. Pre-
                                dicting bioaccumulation  of   Sediment-Associated
                                Organic Contaminants: Development of a Regulatory
                                Tool for Dredged Material Evaluation. Technical Report
                                D-87-1, U.S. Environmental Protection Agency (Narra-
                                gansett, RI) for the U.S. Army Corps of Engineers Wa-
                                terway Experiment Station, Vicksburg, MS.

                              63. Bierman, V.J., Jr.,  1990. Equilibrium and biomagnifica-
                                tion partitioning  of organic chemicals in benthic ani-
                                mals. Environ. Sci. Technol 24:1407-12.

                              64. Leo, A.J. 1972. Relationships between partitioning sol-
                                vent systems. In R.F. Gould, ed., Biological Correlations
                                - The Hansch Approach. Advances in Chemistry Series
                                114. American Chemical Society, Washington, DC.

-------
                            Sediment Quality Criteria Using Equilibrium Partitioning
 eS.Thomann, R.V., and JP. Connolly. 1984. Model of PCB
    in the Lake Michigan lake trout food chain. Environ.
    ScLTechnol. 18:65-71.
                                :
 66.Mackay, D. 1979. Finding fugacity feasible. Environ.
    ScLTechnol. 13:1218.

 67. Stunun, W., and J.J. Morgan. 1970. Aquatic Chemistry.
    Wiley-Interscience, New York, NY.

 68.Hansen, D.JV and M.E. Tagatz. 1980. A laboratory test
    for assessing the impacts of substances on developing
    communities of benthic estuarine organisms. Pages 40-
    57 in J.G. Eaton, P.R. Parrish, and A.C. Hendricks, eds.
    Aquatic Toxicology  (Third Symposium). STP 707.
    American Society for Testing and Materials, Philadel-
    phia, PA.

 69. Tagat^ M.E 1977. Effects of pentachlorophenol on the
    development of estuarine  communities. J. Toxicol. En-
    viron. Health 3501-6.

 70. Tagatz, M.E., CJt Deans, G.R. Plaia, and J.D. Pool.
    1983. Impact on and recovery of experimental mac-
    robenthic communities exposed to pentachlorophenol.
    Northeast Gulf Science 6:131-36.

 71.Hansen, D.J. 1974. Aroclor 1254: effect on composition
    of developing  estuarine animal communities in  the
    laboratory. Contrib. Mar. Sci. 18:19-33.

 72. Tagatz, M.E, N.R. Gregory, and G.R. Plaia. 1982. Effects
    of chlorpyrifos on field- and laboratory-developed es-
    tuarine benthic  communities. J. Toxicol. Environ.
    Health 10:411-21.

73. Tagatz, M.E., and JM. Ivey. 1981. Effects of fenvalerate
    on field- and laboratory-developed  estuarine benthic
    communities. Bull. Environ. Contam. Toxicol. 27:256-
    67.

74. Tagatz, M.E., G.R. Plaia, and C.H. Deans. 1985. Effects
    of 1,2,4-trichlorobenzene  on estuarine macrobenthic
    communities exposed via water and sediment. Ecotoxi-
    col. Environ. Saf. 10351-60.

75. Leo, A. and C. Hansch, eds. 1986. Log(P) Database and
   Related Parameters. Pomona College, Qaremont, CA.

76. Woodbum, KB., W.J. Doucette, and A.W. Andren. 1984.
    Generator Column Determination of Octanol/Water
   Partition  Coefficients  for Selected Polychlorinated
   Biphenyl Congeners. Environ. Sci. Technol. 18:457-59.

77. De Bruijn, J., F. Busser, W. Seinen, and J. Hermens. 1989.
   Determination of octanol/water partition  coefficients
   for hydrophobic organic chemicals with the "slow-stir-
   ring" method. Environ. Toxicol. Chem. 8:499-512.

78.Karickhoff, S.W., andD.S. Brown. 1979. Determination
    of octanol/water distribution coefficients, water solu-
   bilities, and  sediment/water partition coefficients for
   hydrophobic  organic  compounds. EPA 600/4-79-032.
   U.S. Environmental Protection  Agency, Environmental
   Research Laboratory, Athens, GA.

79.Kollig, HP. 1988. Criteria for evaluating the reliability
   of literature  data on environmental  processes con-
    stants. Toxicol. Environ. Chem. 17:287-311. Gordon and
    Breach, Science Publishers, Inc., Great Britain.

 SO.Rapaport, R.A., and  S.J.  Eisenriech.  1984.  Chroma-
    tographic determination of octanol-water partition co-
    efficients  (Kow's)  for 58  polychlorinated  biphenyl
    congeners. Environ. Sci. Technol. 18(3):163-70.

 81. Ellington, J.J., and F.E. Stancil, Jr. 1988. Octanol/water
    partition coefficients for evaluation of hazardous waste
    land disposal: Selected chemicals. EPA/600/M-88/010.
    U.S. Environmental Protection Agency, Environmental
    Research Laboratory, Athens, GA.

 82. Eadsforth, C.V., 1986. Application of reverse-phase
    HPLC for the determination of partition coefficients.
    Pest. Sci. 17:311-25.

 83. Brooke, D.N., A.J. Dobbs,  and N. Williams. 1986. Oc-
    tanol: Water partition coefficients (P): Measurement, es-
    timation and interpretation, particularly for chemicals
    with P1035. Ecotoxicol. Environ. Saf. 11251-60.

 84. De Kock,  A.C., and D.A. Lord. 1987. A simple proce-
    dure  for  determining octanol-water partition coeffi-
    cients  using  reverse phase high  performance liquid
    chromatography (RPHPLQ. Chemosphere 16(1):133-
    42.

 85.Briggs, G.G. 1981. Theoretical and experimental rela-
    tionships between soil adsorption, octanol-water parti-
    tion coefficients,  water solubilities, bioconcentration
    factors, and the parachor. J. Agric. Food Chem. 29:1050-
    59.

 86.Banerjee, S., SJH. Yalkowsky, and S.C.  Valvani. 1980.
    Water  solubility and  octanol/water partition coeffi-
    cients of organics: Limitations of the solubility-parti-
    tion coefficient  correlation.  Environ.  Sci.  Technol.
    14(10):1227-29.

 87. Haky, J.E., and A.M. Young. 1984. Evaluation of a sim-
    ple HPLC correlation method for the estimation of the
    Octanol-water partition coefficients of organic com-
   pounds. J. Liq. Chromatogr. 7(4):675-89.

 88. Hansch, C., and T. Fujita. 1964. A method for the corre-
   lation of biological activity and chemical  structure. J.
   Am. Chem. Soc 86:1616-26.

89. Bruggeman, W. A., J. Van der Steen, and O. Hutzinger.
    1982.  Reversed-phase  thin-layer chromatography of
   polynudear  aromatic  hydrocarbons and  chlorinated
   biphenyls: Relationship with hydrophobicity as meas-
   ured by aqueous solubility and octanol-water partition
   coefficient. J. Chromatogr. 238:335-46.

90. Rogers, K.S., and A.  Cammarata. 1969. Superdelo-
   calizability and charge  density: a correlation with parti-
   tion coefficients. J. Med. Chem. 12:692-93.

91. Veith, G.D., N.M. Austin, and R.T. Morris. 1979. A rapid
   method for estimating log P for organic chemicals.
   Water Res. 13:43-47.

92. Miller, M.M., S. Ghodbane,  SP. Wasik, Y.B. Tewari, and
   D.E. Martire. 1984. Aqueous solubilities, octanol/water

-------
Sediment Quality Criteria Using Equilibrium Partitioning
     partition coefficients, and entropies of melting of chlo-
     rinated benzenes and biphenyls. J. Chem. Enp. Data
     29(2):184-90.

  93. Means, J.C., S.G. Wood, J.J. Hasset, and W.L. Banwart.
     1980. Sorption of polynuclear aromatic hydrocarbons
     by  sediments  and  soils.  Environ.  Sci.  Technol.
     14(12):1524-28.

  94.Burkhard,  L.P., D.W. Kuehl, and G.D.  Veith. 1985.
     Evaluation of  reverse  phase  liquid chromatogra-
     phy/mass  spectrometry for  estimation  of  n-Oc-
     tanol/water   partition   coefficients  for   organic
     chemicals. Chemosphere 14(10):1551-60.

  95. Mabey, WR.,  J.H. Smith, R.T. Podoll, H.L. Johnson, T.
     Mill, T.W. Chou, J. Gates, I. Partridge, Waight, H. Jaber,
     and D. Vandenberg. 1982. Aquatic fate process data for
     organic priority pollutants. EPA-440/4-81-014. U.S. En-
     vironmental Protection Agency, Office of Water Regula-
     tions and Standards, Washington, DC.

  96. Yalkowsky, S H., S.C. Valvani, and D. Mackay. 1983. Es-
    timation  of the aqueous solubility of some aromatic
     compounds. Residue Rev. 85:43-55.

  97. Miller, MM.,  SP.  Wasik, G. Huang, W. Shiu, and D.
    Mackay.  1985. Relationships between octanol-water
    partition coefficient and aqueous solubility. Environ.
    Sci. Technol. 19(6)522-29.

 98. Yalkowsky, S.HV and S.C. Valvani. 1979. Solubilities
    and partitioning 2: Relationships between aqueous
    solubilities, partition coefficients, and molecular sur-
    face areas of rigid aromatic hydrocarbons. J.  Chem.
    Eng.Data24(2):127-29.

 99. Mackay, D., A. Bobra, and W.Y. Shui. 1980. Relation-
    ships between aqueous  solubility  and octanol-water
    partition coefficients. Chemosphere 9:701-11.

100. Kamlet, M.J. RM. Doherty, P.W. Carr, D. Mackay, M.H.
    Abraham, and R.W. Taft.  1988. Linear solvation energy
    relationships: Parameter estimation rules that allow ac-
    curate prediction  of octanol/water  partition  coeffi-
    cients and  other  solubility and toxic properties  .of
    polychlorinated biphenyls and polycyclic aromatic hy-
    drocarbons. Environ. Sci. Technol. 22(5):504-9.

101. Callahan, M.A., M.W. Slimak, N.W. Gabel, I.P. May, C.F.
    Fowler, J.R. Freed, P. Jennings, R.L. Durfee, F.C. Whit-
    more, B. Maestri, W.R. Mabey, B.R. Holt, and C. Gould.
    1979. Water-related environmental fate of  129 priority
    pollutants. Volume 2: Halogenated aliphatic hydrocar-
    bons, halogenated ethers, monocylic aromatics, phtha-
    late   esters,   polycyclic  aromatic   hydrocarbons,
    nitrosamines,  and  miscellaneous compounds.  EPA-r
    440/4-79-029b. U.S. Environmental Protection Agency,
    Office of Water Planning and Standards,  Office  of
    Water and Waste Management, Washington, DC.

102. Arbuckle, W.B. 1983. Estimating activity coefficients for
    use in calculating environmental parameters. Environ.
    Sci. Technol. 17(9)537-42.

103. Doucette, W.J., and A.W.  Andren. 1987. Correlation of
    octanol/water partition coefficients and total molecu-
                                  lar surface area for highly hydrophobic aromatic com-
                                  pounds. Environ. Sci. Technol. 21(8):821-24.

                              104.D'Amboise, M., and T. Hanai. 1982. Hydrophobicity
                                  and  retention in reversed phase liquid chromatogra-
                                  phy. J. Liq. Chromatogr. 5(2):229-44.

                              105. Lyman, W.J., W.F. Rheel, and D.H. Rosenblatt. 1982. Ta-
                                  ble 1-4 in Handbook of Chemical Property Estimation
                                  Methods: Environmental Behavior  of  Organic Com-
                                  pounds. McGraw-Hill Inc., New York, NY

                              106. Karickhoff,  S.W.,  L.A.  Carreira,   C  Melton,  V.K
                                  McDaniel, A.N. Vellino, and D.E. Nute. 1989. Computer
                                  prediction of chemical reactivity — The ultimate SAR.
                                  EPA/600/M-89/017. U.S.  Environmental Protection
                                  Agency, Environmental Research Laboratory, Athens,
                                  GA.

                              107. Anderson, E., G.D. Veith, and  D. Weininger.  1987.
                                  SMILES: a line notation and computerized interpreter
                                  for chemical structures. EPA/600/M-87-021. U.S. Envi-
                                  ronmental Protection Agency, Duluth, MN.

                              108. Weininger, D. 1988. SMILES, a chemical language and
                                  information system. 1. Introduction to methodology
                                  and encoding rules J. Chem. Inf. Comput. Sci. 2831-36.

                              109. Chou, J.T., and P.C. Jurs. 1979. Computer-assisted com-
                                 putation of partition coefficients from molecular struc-
                                 tures using fragment constants. J. Chem. Inf. Comput.
                                 Sci. 19(3):172-78.

                              110. Noreen, E.W. 1989. Computer Intensive Methods for
                                 Testing Hypotheses: An Introduction. John Wiley &
                                 Sons, Inc, New York, NY.

                              111. American Society for Testing and Materials, 1992. "An-
                                 nual  Book of ASTM Standards".

                              112. U.S.  Environmental  Protection Agency. 1993. Guide-
                                 lines  for the derivation of Site-Specific Sediment Qual-
                                 ity Criteria. (In Review).

                              113. Abernethy, S., and D. Mackay. 1987. A discussion of cor-
                                 relations for narcosis in aquatic species. Page 1-16 in
                                 K.L. Kaiser, ed. QSAR in Environmental Toxicology n.
                                 D. Reidel Publishing Co., Dordrecht, The Netherlands.

                              114. U.S.  Environmental  Protection Agency.  1989.  Hand-
                                 book: Water Quality Control Information Systems,
                                 STORET. Washington, DC

                             115. U.S. Army Corps of Engineers. 1991. Monitoring Pro-
                                 gram for San Francisco Bay Sediments. 1988 to 1990.

                             116. O'Connor, T.P. 1991. Concentrations  of  Organic Con-
                                 taminants in Mollusks and Sediments at NOAA Na-
                                 tional Status  and Trend Sites in  the  Coastal and
                                 Estuarine United States. Environ. Health Perspectives
                                 90:69-73.

                             117. El-Shaarawi, A.H., and DM. Dolan. 1989. Maximum
                                 Likelihood Estimation of Water Quality Concentrations
                                 from  Censored Data. Can. J. Fish. Aquat. Sci. 46:  pp.
                                 1033-39.

-------
                             Sediment Quality Criteria Using Equilibrium Partitioning
     *"  .a     -*•
  118. Landram, RE 1989. Bioavailabffity and toxicokinetics
     of polycydic aromatic hydrocarbons sorbed to sedi-
     ments for the amphipod Pontororeia hoyi. Environ. Sci.
     Technol. 23:588-95.

 119. Swartz, R.C., EF. Kemp. D.W. Schults, and J.O. Lamber-
.    son. 1988. Effects of mixtures of sediment contaminants
     on  the  marine  infaunal  amphipod,  Rhepoxynius
     abronhts. Environ. Toxicol. Chem. 7:1013-20.

 120. Connolly, J.P., and C.J. Pederson. 1988. A thermody-
     namic-based evaluation of organic chemical accumula-
     tion in aquatic organisms. Environ.  Sci.  Technol.
     22:99-103.
    U.S. Environmental Protection Agency, Office of Water
    Regulations and Standards, Criteria and Standards Di-
    vision, Washington, DC.

122. Connolly, J.P. 1991. Application of a food chain model
    to polychlorinated biphenyl contamination of the Lob-
    ster and Winter Flounder food chains in tfew Bedford
    Harbor. Environ. Sci. Technol. 25(4):760-70.

123. Hamilton, M.A., R.C. Russo, and R.V. Thurston. 1977.
    Trimmed Spearman-Karber method of estimating me-
    dian lethal concentrations in toxicity b'ioassays. Envi-
    ron. Sci. Technol. 11:714-19. Correction 12:417 (1978)
 121. Cowan, D.E., and D.M. Di Toro. 1988. Interim sediment
     criteria values for nonpoplar hydrophobic compounds.

-------