Joy McGrath and Dominic Di Toro

Abstract
A method is presented for developing scientifically defensible numeric guidelines for oil-related constituents, specifically monocyclic aromatic hydrocarbons (MAH) and polycyclic aromatic hydrocarbons (PAHs), in the water column and sediment. The guidelines are equivalent to a HC5, a hazard concentration value that protects 95 percent of the test species. The model of toxicity used in this evaluation is the target lipid model (TLM) that was developed for assessing the toxicity of Type I narcotic chemicals (Di Toro et al. 2000). Structurally the aromatic components of oil should exert an acute narcotic mode of action. This research focused on validating the TLM for its appropriateness in assessing the toxicity of these oil components, both on an acute and chronic basis. The methodology was determined to be effective at predicting the toxicity of MAHs and PAHs. The resulting HC5 guidelines were found to be protective of sub-lethal effects commonly associated with blue sac disease resulting from early life stage exposure to PAHs. The use of toxic units (TUs) as the metric for expressing the toxicity of mixtures of oil-related components, or mixtures of hydrocarbons in general, is demonstrated to be an effective means of normalizing toxicity data across different sources and different species. The toxicity of the mixture depends on which hydrocarbons are present because the toxicity of the individual components in the mixture varies. A concentration of 1 TU from the components in the mixture implies toxicity. A concentration of 1 µg/L of total measured compounds in a mixture may or may not be toxic and depends on which components are present, the concentration of those components and the species exposed. The use of TUs eliminates this confusion and normalizes the data across sources. The methodology presented can be used by the oil spill community, which includes the regulators and the regulated industries, to compare residual concentrations of MAHs and PAHs against defensible numeric guidelines to assess potential ecological impacts.

Keywords: Polycyclic Aromatic Hydrocarbons, Target Lipid Model, Model Validation, Oil Toxicity, Guidelines

Acknowledgements
Funding for this project was provided by the NOAA/UNH Coastal Response Research Center (Grant number NA170Z2607). John Sondey at HydroQual, Inc. was responsible for producing final graphical displays.

Table of Contents

1.0	Introduction	1
2.0	Objectives	1
3.0	Methods	4
3.1	Aqueous Acute Toxicity Prediction - Target Lipid Model	4
3.2	Aqueous Chronic Toxicity Prediction	9
3.3	Sediment Toxicity Prediction	9
3.4	Uncertainty in Toxicity Predictions	9
3.5	Application to Mixtures	11
3.6	Computing Total PAH Toxic Units in Sediment	13
3.7	Literature Review	14
3.8	Physical Chemical Properties	15
4.0	Results	15
4.1	Literature Review	15
4.2	TLM Validation - Water Column	16
4.3	TLM Validation - Sediment	27
5.0	Discussion and Importance to Oil Spill Response/Restoration	34
6.0	Technology Transfer	37
7.0	Achievement and Dissemination	37
References		38

List of Figures and Tables

Figure 1	Schematic Diagram of log(LC50) versus log(KOW). At log(KOW) equal to 0, the KOW is equal to 1 meaning that chemical concentration in the water is equal to the chemical concentration in octanol. (Adopted from Di Toro et al. 2000)	7
Figure 2.	Log(LC50) Versus log(KOW) for the Indicated Species. (Redrawn from Di Toro et al., 2000).	8
Figure 3.	Diagram of how acute toxicity data from aqueous and sediment tests using mixtures of chemicals will be displayed. Percent effect as a function of predicted total TUs. Dashed lines represent HC5 and HC95 values	13
Figure 4.	Acute Exposures- Single Compounds - TLM predicted acute aqueous LC50 versus observed LC50 for MAHs and PAHs. Solid line represents 1:1 relationship. Dashed lines represent 90% confidence interval	17
Figure 5.	Acute Exposures - Mixtures - Percent mortality as a function of total measured concentration (mg/L) (top panels) and predicted aqueous TUs (bottom panels). Data on the right is for Pimephales promelas exposure to WSFs prepared from neat and weathered Exxon Valdez crude oil (ENSR, 2000). Data on the left is for Oithona davisae exposure to WSF prepared from a mixture of 9 PAHs (Barata et al. 2005). Solid lines represent 50% mortality (horizontal) and 1 TU (vertical). Dashed vertical lines represent HC5 and HC95 for each species	19
Figure 6.	Chronic Effects – Single Compounds - Distribution of acute to chronic ratios (ACRs) for aliphatic hydrocarbons (A), PAHs (B), BTEX (C) and the combined data set (D).	21
Figure 7	Chronic Effects – Mixtures – Observed mortality as a function of total measured concentration in WSF (mg/L) (left panel) and predicted aqueous TUs (right panel). Data are for Cyprinodon variegates embryos exposed to WSF prepared from No. 2 fuel oil (Anderson et al. 1977). Dashed vertical lines represent 5th and 95th percentiles based on variations in CTLBB and ACR.	22
Figure 8	Sub-lethal effects – Single Compounds – Comparison of OEC/LOEC/NOEC observed from early life stage fish exposures to single compounds to TLM chronic effect concentrations. The effects are those associated with sub-lethal endpoints such as abnormal larvae development and blue sac disease-like symptoms. The symbols represent the TLM chronic endpoint. The lines associated with the TLM chronic endpoints represent the 5th and 95th percentiles based on variations in CTLBB and ACR. The number located above the reported effect concentration is the assigned ranking number. See Table 4 for references.	25
Figure 9	Sub-lethal effects – Mixtures – Comparison of 18-d NOEC and LOEC from early life stage toxicity tests exposing Oryzias latipes to three prepared mixtures of PAHs to TLM chronic endpoints. The circles represent the TLM effect concentration. The bars represent the 5th and 95th percentiles based on variations in CTLBB and ACR. Data are from Rhodes et al. 2005.	26
Figure 10	Acute Sediment Exposures – Single Compound - Comparison of observed and TLM predicted sediment effect concentrations (see Table 5 for data). Solid line is 1:1 relationship. The dashed lines are 90% confidence intervals	28
Figure 11	Acute Sediment Exposures – Mixtures – Percent mortality of R. abronius as a function of PAH concentraton (mg/kg) (top panel) and normalized to total PAH sediment toxic units (bottom panel). All data are 10-d exposures. Solid lines at a toxic unit of 1.0 and 50 % mortality are shown for guidance. Dashed lines represent 5th and 95th percentiles based on variation in CTLBB. (• Swartz et al. 1997; Boese et al. 1999; Tay et al. 1992; Techratech, 1982; Swartz et al. unpublished; Swartz et al. 1989; • Ozetich et al. 2000; Page et al. 2000)	30
Figure 12	Acute Sediment Exposures – Mixtures – Percent mortality as a function of PAH concentraton (mg/kg) (top panel) and normalized to total PAH sediment TUs (bottom panel). See Table 6 for exposure information. Solid vertical and horizontal lines at a TU of 1.0 and 50 % mortality, respectively, are shown for guidance.	31
Figure 13	Chronic Sediment Exposures – Single Compounds - Comparison of reported OEC/LOEC/NOEC and TLM effect concentrations from long-term exposures of fluoranthene and phenanthrene to the marine copepods Coullana sp. and Schizopera knabeni. See Table 8 for references.	33
Figure 14	Chronic HC5 concentration for PAHs versus log (KOW). The HC5 concentration is based on the 5th percentile CTLBB for all species in the database. The observed NOECs for PAHs are denoted with a 0 (Table 10).	36
Table 1	Chemicals and their properties (at 25C) measured in various data sets.
Table 2	Observed water-only acute LC50/EC50 values for PAHs and TLM predictions
Table 3	Acute and chronic values used in the development of ACRs. Chronic endpoints are those effecting growth, reproduction or mortality of an organism.
Table 4	Comparison of target lipid model chronic endpoints to NOECs, LOECs, OECs (sub-lethal effects) from early life stage tests for single PAH exposures.
Table 5	Observed sediment acute LC50 values for PAHs and TLM predictions.
Table 6	Summary of sediment data sets for acute toxicity from PAH mixtures.
Table 7	Example demonstrating benefit of using adjustment factor to convert TU from 13PAH to Total PAH TU. Data are from Swartz et al. unpublished (Table D5).
Table 8	Chronic effects from single PAH sediment exposures.
Table 9	Summary of CTLBBs for different chemical classes.
Table 10	Chronic HC5 values for MAHs and PAHs for aqueous and sediment toxicity.
Table 11	PAH NOEC toxicity data for various species (measured data considered only).

1.0 Introduction
This research project is directed at evaluating the impacts of low levels of residual oils based on assessment of the toxicity of oil-related constituents and their exposure risks. The primary compounds of interest in this research are the monocyclic aromatic and polycyclic aromatic hydrocarbons (MAHs and PAHs). Structurally, these compounds are Type I narcotic chemicals and their acute toxic mode of action is narcosis. The toxicity of such compounds is additive and therefore a methodology is needed that accounts for exposure to a mixture of hydrocarbons. The target lipid model (TLM) and the concept of toxic units (TUs) are used to assess the toxicity of oil-related chemicals. The TLM is a method of computing water criteria for Type I narcotic chemicals (Di Toro et al. 2000). Equivalent sediment criteria were derived using the equilibrium partitioning (EqP) model (Di Toro and McGrath, 2000). Aqueous and sediment criteria were derived using methodology based on narcotic chemicals in general, rather than MAHs and PAHs specifically. The United States Environmental Protection Agency (U.S. EPA) developed sediment benchmarks (water-only benchmarks were not presented) for PAH mixtures based on the TLM (U.S. EPA, 2003). These benchmarks were derived with the objective of protecting sediment organisms from long-term effects of mortality, growth and reproduction. Recent literature suggests that exposure to PAHs during an organism’s early life stage results in various sub-lethal effects similar to those of blue-sac disease (Carls et. 1999; Heintz et al. 1999; Incardona et al. 2004). These sub-lethal effects were not addressed in the EPA’s sediment benchmarks for PAH mixtures and therefore they may not be protective of these types of effects.

For this research, the appropriateness of the TLM for predicting the toxicity of MAHs and PAHs as single chemical exposures and in mixtures is investigated. The TLM is validated for predicting the toxicity of these chemicals in water-column and sediment exposures, both on an acute and chronic basis. Additionally, the TLM is evaluated to determine if computed endpoints are protective of sub-lethal endpoints such as yolk sac edema, hemorrhaging and larvae abnormalities. Once it has been demonstrated that the TLM methodology can be used to assess whether effects are expected at a certain concentration of a chemical or from a mixture of chemicals, the methodology is used to derive defensible chemical-specific guidelines. HC5 values, the ecological hazard concentrations that protect 95% of species, are presented on an aqueous basis and a sediment basis as a function of log octanol-water partitioning coefficient, KOW.

The beneficiaries of this research include regulators and the regulated communities because they will have an innovative scientifically defensible method to estimate toxicity of oil components in water and sediment.
Moving a research oriented model like CDOG into operational use requires more than adding a Graphical User Interface (GUI). Particularly for emergency response, the user has a task to accomplish as part of the response. That “task” is not running the model, though the task may involve running the model. For example, NOAA/HAZMAT provides trajectory analysis during a spill response. The GNOME model may be run, but the GNOME output is post-processed to add additional information from the trajectory forecaster to the Federal On Scene Coordinator (FOSC).


2.0 Objectives
The overall objective of this project is to evaluate the impacts of low levels of residual oils based on the toxicity assessment of oil-related constituents and their exposure risks. Generally, the short-term toxicity of petroleum is attributed to the water-soluble MAHs (i.e., benzene, toluene, ethylbenzene and xylene, commonly referred to as BTEX) and PAHs. BTEX are not expected to persist in the environment because they are volatile compounds and evaporate quickly and are fairly biodegradable. PAHs are less volatile and the heavier PAHs (those with 4 or 5 rings) are known to persist and they can potentially have a long-term impact on the aquatic environment. Specific objectives and how they were met include:

• Identify key components of residual oil that contribute to toxicity – The literature was reviewed for toxicity data resulting from exposure to single chemicals and mixtures of chemicals expected to be present in oils (i.e., MAHs and PAHs) both in the water column and in sediment. Toxicity endpoints reflected both short-term (acute) and long-term (chronic) effects.
• Provide a synthesized and reviewed effects database to use as a screening tool to evaluate risk to aquatic resources from petrogenic sources of MAHs and PAHs.
• Validate that the TLM of toxicity can predict the aqueous toxicity of oil related compounds – Toxicity predictions based on the TLM methodology were performed on the data sets identified in the literature review. Predictions were made for single chemical exposures for both acute and chronic effects. The predicted toxicity was compared to the observed toxicity.
• Establish that the TU can be used as a universal endpoint for expressing the toxicity from mixtures of compounds such as those resulting from different oil sources – The toxicity of different mixtures of compounds that result from various oil sources were normalized to TU and compared to mass based (i.e., mg/L) endpoints.
• Demonstrate that EqP theory is appropriate for converting the TLM-predicted aqueous effect concentrations for oil-related compounds to equivalent sediment effect concentrations – Sediment toxicity predictions based on the TLM and EqP methodologies were performed on the data sets identified in the literature. Predictions were made for single chemical exposures and mixtures of chemicals for both acute and chronic effects. Sediment toxicity predictions were compared to observed effects data.
• The TLM in conjunction with EqP theory was used to derive guidelines for oil-related compounds that may be protective of aquatic and benthic species from long-term sub-lethal effects.

These project objectives are directly related to the work of the oil spill community. A scientifically defensible methodology is presented for computing guidelines to judge what levels of oil-related compounds should not have an adverse impact on the aquatic community. The methodology is simple to use, only the log octanol-water partitioning efficients (KOW) of the chemicals in the mixture are needed. The method is versatile in that it allows the user to compute protective guidelines for a particular species or guidelines that protect 95% of all species considered in the method development (see Section 3.0 Methods).

In addition, the use of the TU as the toxicity metric for mixtures normalizes differences in toxicity among chemicals. This is of critical importance to the oil spill community because it allows data from one location/test to be directly comparable to data for another location/test. Historically, the measured concentrations of chemicals in the water phase have been summed to assess the aqueous toxicity of crude oils. Water-soluble fractions (WSFs) are prepared from different sources of oil and although the compositions of the WSFs vary and different compounds are measured, the interpretation of the toxicity is based on summed concentrations. The effects are often expressed as total petroleum hydrocarbons on a mass per volume basis (i.e., mg TPH/L) (Anderson et al. 1974; Rossi and Anderson, 1976; Moles et al. 1979; Brodersen, 1987) or total PAH concentrations (i.e., mg TPAH/L) (Anderson, 1977; Neff and Stubblefield, 1995; Carls et al.1999; Heintz et al. 1999). The fundamental problem with reporting toxicities of hydrocarbon mixtures on a total concentration basis is that it assumes all of the compounds have equal toxicities. The toxicity of hydrocarbons varies widely and has been related to the log KOW (Konemann, 1981; Veith et al. 1983). The toxicities of individual chemicals in a mixture are normalized by expressing the toxicity of each chemical in terms of a TU (Hermens, 1989; Peterson, 1994). Once the toxicities are normalized, comparisons of different oils can be made.

This research effort for determining impacts of low levels of residual oil components focused on meeting the objectives described above. It is impractical and probably impossible to develop guidelines that address every situation that may occur in the environment. The following is a listing of conditions and assumptions of the TLM that can impact the toxicity of oil-related components, which were NOT addressed in this research, but are important to consider when determining the effects from oil.

• The TLM assumes that octanol is a good surrogate for organism lipid and that the KOW describes the partitioning of the chemical between the water and the organism lipid – Other solvents or physical-chemical properties may be better descriptors for the partitioning.
• Some data suggest that a toxicity cut-off exists for compounds with log KOW greater than approximately 5.5 (i.e., compounds with log KOW values of 5.5 or greater have similar toxicity). The TLM ignores a cut-off and relies on the KOW to be the descriptor for the toxicity. The KMW, membrane-water partition coefficient, has been shown to be a better surrogate than KOW for describing the partitioning of a chemical into the organism lipid membrane. (Gobas et al. 1988; Vaes et al 1998; Verbruggern 2004). The log(KMW)-log(KOW) relationship is linear until a log KOW of approximately 5.5 where the log KMW no longer increases linearly with log KOW (Verbruggern 2004). However, the KOW is used in the TLM because a cut-off point for PAHs has not been confirmed or identified and toxicity from PAHs with log KOW values greater than 5.5 has been observed. The TLM would not have correctly predicted the toxicity if a cut-off or the KMW was used in the analysis.
• The TLM assumes that metabolites have the same mode of action as the parent compound – Recent literature suggests that some metabolites may be more toxic than parent compounds.
• Effects of photoactivation of PAHs – Some PAHs exert photoenhanced toxicity after accumulation in the tissues of organisms and exposure to ultraviolet radiation. Photoinduced toxicity can be orders of magnitude greater than baseline toxicity (i.e., narcosis) (Ankley et al. 1994; Spehar et al. 1999; Lyons et al. 2002).
• Other water-soluble components of oil – This research focused on MAHs and PAHs, the most commonly measured oil components, as being the causative agents of oil toxicity. There may be other water-soluble components that exert toxicity. Note that this methodology can be applied to any chemical, but the toxicity computed will be the chemical’s “baseline” or minimum toxicity.
• Time-variable concentrations – This methodology assumes that the exposure concentrations are constant with time or that equilibrium conditions have been achieved, particularly for sediment exposures. Time-variable concentrations are not addressed and it is difficult to assign observed effects to a concentration if the concentration is varying with time.
• Non-standard conditions – The physicochemical properties used in this research were determined at 25 ° C. These parameters vary with temperature. For example, organic compounds are usually less soluble in colder temperatures. Therefore, the computed guidelines would be conservative for temperature conditions less than 25 ° C.
• Other sediment partitioning phases – Equilibrium partitioning theory assumes organic carbon is the only partitioning phase, described by KOC, for non-polar organic contaminants. Some field data sets indicate that other partitioning phases may be present and include soot carbon or coal. PAHs seem to have a higher affinity for these other carbon phases than for organic carbon as described by KOC. Therefore, the PAHs will be less bioavailable than predicted using KOC as the sole partitioning coefficient. Ignoring other possible partitioning phases is conservative since the maximum toxicity will be predicted.



3.0 Methods
3.1 Aqueous Acute Toxicity Prediction - Target Lipid Model

Based on chemical structure, the MAHs and PAHs should exhibit a narcotic mode of action (Verhaar et al. 1992). The TLM (Di Toro et al. 2000) will be used to assess the toxicity of MAHs and PAHs resulting from exposure to low levels of oil in the water column. The TLM is for Type I narcotic chemicals, which are non-ionic organic chemicals with a similar mode of toxic action, namely narcosis. Narcosis is a reversible state of arrested activity. The toxic mode of action is non-specific and is caused by a variety of chemicals, such as gases, aliphatic and aromatic hydrocarbons, alcohols, chlorinated compounds, ethers and ketones (Veith et al. 1983; Verhaar et al. 1992). With narcosis, it is assumed that an equilibrium exists between the chemical concentration in the organism and the chemical concentration in the external media phase, such as the water phase.

The TLM model is based on the inverse relationship between toxicity in water-only exposures, as defined by the LC50, and the K_OW. Numerous quantitative-structure activity relationships (QSARs) of log(LC50) versus log(KOW) have been developed for narcotic chemicals (Konemann 1981; Veith et al., 1983; Van Leeuwen et al., 1992). The TLM is an extension of these models. It is based on a single universal slope for the log (LC50)-log (K_OW) relationship, independent of species. It includes corrections for chemical classes that are slightly more potent than baseline narcotics in particular MAHs and PAHs. And it interprets the y-intercepts as the species-specific critical body burdens for a particular endpoint. A complete presentation of the narcosis TLM is available (Di Toro et al. 2000). A brief description is provided below.

This result can be understood by examining Figure 1. For a chemical with a log(K_OW) equal to zero or K_OW equal to 1.0, the y-intercept is the LC50. At a K_OW equal to one, the chemical concentration in the water is equal to the chemical concentration in the octanol. Assuming octanol is a good surrogate for lipid, then the chemical concentration in the water is equal to the chemical concentration in the lipid (target tissue) of the organism. Therefore, the y-intercept is equal to the target lipid concentration producing an effect (i.e., 50% mortality, LC50) for that organism. In the TLM, the y-intercept is the CTLBB and has units of µmol/g octanol = µmol/g lipid). The CTLBB varies for each species. The slope of the line is a chemical property of the lipid and should be the same regardless of the species tested. The slope of line is referred to as the universal narcosis slope. Figure 2 shows the relationship of log(LC50)-log(K_OW) for several species in the database. The symbols represent many chemicals from different chemical classes. The line drawn through the data is the universal narcosis slope (i.e., it is the same for all species).

Figure 1. Schematic Diagram of log(LC50) versus log(K_ow) equal to 0, the (K_ow) is equal to 1 meaning that chemical concentration in the water is equal to the chemical concentration in octanol. (adopted from Di Toro et al. 2000).

The TLM has been extensively validated using a large database to predict the aquatic toxicity of classic Type I narcotic chemicals. CTLBBs for 33 aquatic species (Di Toro et al. 2000) and 5 algae (McGrath et al. 2004) were computed with the TLM using the K_OW values determined from a 1997 version of SPARC, a U.S. EPA sponsored computer program that estimates numerous physicochemical properties for chemicals based on chemical structure (Karickoff et al. 1991). Since 1997, SPARC has undergone enhancements in its programming. Therefore, K_OW values computed using the 1997 version may be different than those values computed using the enhanced on-line version. To take advantage of the enhancements to SPARC, new K_OW values were determined for all chemicals in the narcosis toxicity database and revised coefficients for the TLM were computed (see Appendix A for coefficients)

Figure 2 Log (LC50) Versus log (k_ow) for the Indicated Species. (Redrawn from Di Toro et al., 2000).

There are four data sets that have acute mortality data for species other than R. abronius. Comparisons of percent mortality as a function of measured PAH (mg/kg) and total PAH TU are shown in Figure 12. DeWitt et al. (1992b) determined the mortality to two species (E. estuaries and L. plumulosus). Since the CTLBB for these species are very similar (41.4 and 43.1 mmol/g octanol), the computed total PAH TUs are almost identical. Therefore, the observed range of mortality for the two species is shown as a function of the sediment concentration (Figure 12, panels E and F). Due to limited data available to compute CTLBB for sediment organisms, HC5 and HC95 values could not computed for E. estuaries, Chironomus riparius, Schizopera knabeni and Ampelisca abdita. For all four data sets, the dose-response is as expected with low total PAH TUs corresponding to low mortality and around 1 TU corresponding to about 50 % mortality.





Figure 12. Acute Sediment Exposures – Mixtures – Percent mortality as a function of PAH concentraton (mg/kg) (top panel) and normalized to total PAH sediment TUs (bottom panel). See Table 6 for exposure information. Solid vertical and horizontal lines at a TU of 1.0 and 50 % mortality, respectively, are shown for guidance.


It should be noted that the measured concentrations in the field datasets represent different numbers of PAHs measured ranging from 13 to 40. For example, a value of 10 mg/kg in a sediment sample contaminated from the Exxon Valdez oil spill (Page et al. 2002) that was computed as the sum of 40 PAHs is not necessarily equivalent to a value of 10 mg/kg in a sediment sample collected in Eagle Harbor, WA (Swartz et al. 1989) that was computed as the sum of 13 PAHs. In addition, the bioavailability of the sediment-associated PAHs is a function of the organic carbon concentration of the sediment (Di Toro et al. 1991). Two sediment samples that have 10 mg/kg of the same 40 PAHs, may have different toxicities if the organic carbon concentrations are different. Therefore, concentrations of measured PAHs in different sources should not be compared nor should effects be assigned based on mass based concentrations. Using the summed TU approach for sediments (i.e., converting individual sediment PAH concentrations to individual PAH TU, summing the individual PAH TU and then applying a factor to convert to total PAH TU) considers the toxicity from unmeasured PAHs and considers the difference in organic carbon concentration and therefore reduces the uncertainty associated with PAHs measurements across different sources and allows for comparisons to be made.

The importance of applying the adjustment factor to account for unmeasured PAHs is demonstrated using a San Diego Bay sediment dataset (Swartz et al. unpublished). In this dataset, 13 PAHs were measured in 54 sediment samples. The test organism was R. abronius and the endpoint was 10-d mortality. Table 7 presents a comparison of the TUs computed from the measured 13 PAH and from Total PAH as they relate to the observed mortality. A correction prediction was one where the observed mortality was greater than or equal to 50% and the computed TUs were greater than or equal to 0.4 (the uncertainty based on the ratio of the HC5 and the TLM endpoint for R. abronius) or where the observed mortality was less than 50% and the computed TUs were less than 0.4. When the TUs were computed from the 13 PAH only, there were 42 correct predictions. When the adjustment for unmeasured PAHs was considered the number of correct predictions increased to 45. In addition to the number of correct predictions, one can compare the number of times that the criterion was protective, meaning high mortality was observed and correctly predicted. The observed mortality was greater than or equal to 50% in 13 samples. Based on TUs from 13 PAHs, one sample would have been predicted to be toxic (station STA28 100% with TU of 0.608 from 13 PAH). In comparison, the TUs from Total PAHs were greater than 0.4 in eight of the samples with high mortality. This analysis demonstrates the importance of adjusting for the toxicity from unmeasured PAHs. The overall number of correct predictions (toxic and non-toxic) increased. Furthermore, the number of correct toxic predictions increased from 1 to 8 with the incorporation of the adjustment factor.

4.3.3 Chronic Effects – Single Compound Exposures
There are several data sets available that report chronic effects from exposure to sediment spiked with single PAHs. The chronic effects are growth and reproduction endpoints. The tests species are Coullana sp. and S. knabeni (Figure 13). The CTLBBs for these species were measured on a lipid basis (Lotufo, 1998). The effect of fluoranthene on grazing and reproduction for Coullana sp. was reported (Lotufo, 1998). The 10-d LOECs ranged from 3133 to 8800 mg/goc for grazing and reproduction, respectively. The 10-d NOECs ranged from 1200 to 3111 mg/goc. The TLM predicted chronic effect concentration of 1410 mg/goc was below the observed LOECs indicating that the TLM correctly predicted the effects. The chronic effects of phenanthrene and fluoranthene on grazing and reproduction of S. knabeni were determined (Lotufo, 1997; Fleeger and Lotufo, 1999). Phenanthrene had a reported 14-d IC25 (inhibition concentration - concentration causing 25% reduction of measured endpoint in relation to control) of 1730 µg/g_oc for reproduction effects. The reported 10-d NOEC and LOEC were 1470 and 3000 µg/g_oc, respectively, for hatching success. The TLM predicted chronic effect concentration of 3560 µg/g_oc was slightly higher than the reported OEC/LOEC. Fluoranthene had reported 10-d LOECs of 1200 and 3133 µg/g_oc for grazing and reproduction, respectively (Lotufo, 1998). The TLM predicted chronic effect concentration of 4130 µg/g_oc was slightly higher than the reported LOEC. These data are summarized in Table 8 and Figure 13.


The CTLBBs for S. knabeni and Coullana are measured values based on a single compound (fluoranthene) and the uncertainty associated with them could not be determined. In addition, Lotufo (1998) stated that the lipid content was based on an initial value, which could have changed during the exposure time. The variability in the reported CTLBB for these species is unknown. Therefore, these data are presented for informational purposes only and the results of the analysis should not be used to assess the performance of the TLM.




Figure 13. Chronic Sediment Exposures – Single Compounds - Comparison of reported OEC/LOEC/NOEC and TLM effect concentrations from long-term exposures of fluoranthene and phenanthrene to the marine copepods Coullana sp. and Schizopera knabeni. See Table 8 for references.


4.3.4 Chronic Effects – PAH Mixtures
Fleeger and Lotufo (1999) investigated the chronic effects on S. knabeni from exposure to a sediment spiked with diesel fuel. They measured the concentrations of 20 PAHs and TOC in the sediment. The sediment was spiked with the diesel fuel and dilutions were prepared for use in the bioassay. The resulting total PAH concentrations in the sediment were 19, 45, 93, 130, 185 and 370 mg/kg. The LOEC and NOEC for reproductive effects were 93 and 45 mg/kg, respectively. The equivalent TLM Total PAH TU for the LOEC and NOEC were 1.7 and 0.8, respectively (Table 10D). Based on these equivalent TUs, effects should have been have observed at 45 mg/kg. While the TLM was overprotective (i.e., effects were predicted at concentrations where no effects were observed), these data cannot be used to assess the performance of the TLM because of the uncertainty in the CTLBB (See Section 4.3.3).



5.0 Discussion and Importance to Oil Spill Response/Restoration
This research focused on developing a methodology that can be used to assess the toxicity of low levels of residual oil, focusing on the MAH and PAH components. The methodology has to be able to predict the toxicity of these components in the aqueous phase and in the sediment, on acute and chronic bases. The methodology had to include the derivation of a metric that enabled the toxicity from one location to be comparable to the toxicity at another.

It has been demonstrated that the TLM can be used to assess the acute and chronic toxicity of oil-related components on an acute basis in the water column and sediments. It has been shown that the use of an ACR to relate the acute toxicity to an equivalent chronic toxicity is appropriate without consideration of the toxicity mechanism. Within the uncertainty of the model, the TLM has been shown to be protective of sub-lethal effects that result from exposure to low levels of PAHs during an organism’s early life stages. The TLM coupled with the theory of additivity via the concept of TUs can be used to predict the toxicity of mixtures of chemicals present in oils. TUs are a means of normalizing the toxicity of different chemicals. A total TU of 1.0, implies toxicity. A total concentration of 1µg/L is not necessarily toxic. Rather, its effect is dependent on the specific chemical(s) to which the organism is exposed and the sensitivity of the organism. It should be noted, that the available toxicity data for mixtures that met the criteria for inclusion, both on an acute and chronic basis, are limited. There is research need to determine the toxicity of mixtures of chemical resulting from exposure to oil for species where a CTLBB is available. These data would further validate that the TLM and TU concept are appropriate for predicting the toxicity.

For sediment assessment, the TLM aqueous effect concentrations are converted to sediment effect concentrations via EqP methodology. In the sediment, the chemical concentration can be related to the bioavailability by normalizing the chemical concentration to the organic carbon concentration. The TU metric allows for the consideration of the toxicity from unmeasured PAHs by incorporating the U.S.EPA adjustment factors, which cannot be done on a mass basis (i.e., mg/kg). Organic carbon normalization and TU conversion allow for the generation of a toxicity metric that is applicable across a broad range of sediment types and representative of toxicity from Total PAH.

In addition to being used as a tool that can assess whether effects are expected at a certain concentration of a chemical or from a mixture of chemicals, the methodology can be used to derive defensible chemical-specific guideline values. When the statistics (geometric mean and variance) of the CTLBB of all test species are used in Equation 18, the computed HC5 is the aqueous concentration that protects 95% of the species. These HC5s can be used as decision-making values to determine the concentration of a particular chemical that will not likely have an adverse impact on the aquatic community. The geometric mean CTLBB of all tests species for baseline chemicals (i.e., aliphatic hydrocarbons, alcohols, see Appendix A) is 123 µmol/g octanol. The geometric mean CTLBB for specific chemical classes is computed as

	C*L(geometicmean, Chemicalclassi) = C*L(geometicmean,baseline)XCi	(23)
whereC is the chemical class correction factor for chemical class i (see Appendix A). The CTLBBs for the various chemical classes are provided in Table 9. The resulting equation to compute the chronic HC5 values for PAHs is:
		(24)
		(25)

The 95% confidence sample size-dependent extrapolation factor (k_z = 2.3) was based on the number of ACRs (29) rather than the number of CTLBBs (47) to ensure that the calculations were conservative. A summary of chronic HC5 values for MAHs and PAHs is provided in Table 10. EqP theory was used to convert the aquatic HC5s to sediment HC5 values (Equation 15). The values are presented on a molar basis (i.e., µmol/L, mmol/KgOC) and a mass basis (i.e., µg/L, mg/KgOC). Details of the calculations are provided in Appendices E and F.

To assess the management implications of adopting chronic HC5s derived via the TLM for decision-making, a comparison was made of the aqueous HC5 values to NOECs values for PAHs. The NOECs values used in this analysis are presented in Table 11 and Figure 14. A total of 27 NOECs were compiled for 7 different PAHs. Two NOECs fell below the HC5 line. Both of these NOECs were for phenanthrene and ranged from 5 to 5.5 µg/L, which were considerably lower than NOECs for other compounds. The excursion of two data points below the HC5 is consistent with the 95% protection level goal (i.e., 92.6% (25/27)). The HC5 derived from the TLM is very close to the expected level of protection. Therefore, these values are appropriate for use as numeric chemical-specific benchmarks and can be used to assess the ecological risk of contaminated sediments and/or establish safe levels for cleanup activities. Of course, one needs to consider the assumptions of the TLM and the conditions that can impact the toxicity of MAHs and PAHs (refer to Section 2.0) to determine the appropriateness for applying these guidelines to a specific location.




Figure 14. Chronic HC5 concentration for PAHs versus log (KOW). The HC5 concentration is based on the 5th percentile CTLBB for all species in the database. The observed NOECs for PAHs are denoted with a 0 (Table 10).

6.0 Technology Transfer
The results of this research effort on the application of the TLM to predict acute and chronic effects on oil-related compounds and the derivation of HC5 will be submitted to Environmental Chemistry and Toxicology. An abstract focusing on the application of TLM to be protective of sub-lethal effects from exposure to PAHs has been submitted to the Society of Environmental Chemistry and Toxicology (SETAC) for presentation at the Fall 2006 North America meeting. Intermediate results of this research effort were presented at a March 2005 workshop entitled “Emerging Research in Oil Spill Response and Restoration”, at a September 2006 summit for the PAH Toxicology Working Group, at a September 2006 workshop entitled “Innovative Coastal Modeling: Integrating Physical, Biological and Toxicological Models”, all sponsored by CRRC, and at the 2005 SETAC North America meeting.


7.0 Achievement and Dissemination
Workshops and Conferences

Workshop on Emerging Research in Oil Spill Response and Restoration. 2005. “ Impacts of Low Levels of Residual Oil: Predicting the Acute and Chronic Toxicity of Residual Oil”. Office of Response and Restoration of NOAA. Silver Spring, Maryland. March 2005.

26th Annual SETAC North America Meeting. 2005. “Predicting the Acute and Chronic Effects of PAHs using the Target Lipid Model of Narcotic Toxicity: Is a New Paradigm Needed? Baltimore, Maryland. November 2005.

PAH Toxicology Working Group Summit. 2006. “ Impacts of Low Levels of Residual Oil: Predicting the Acute and Chronic Toxicity of MAHs and PAHs”. CRRC. Seattle, Washington. August 2006.

Workshop on Innovative Coastal Modeling: Integrating Physical, Biological and Toxicological Models. 2006. “ Impacts of Low Levels of Residual Oil: Predicting the Acute and Chronic Toxicity of MAHs and PAHs”. CRRC. Durham, New Hampshire. September 2006.

27th Annual SETAC North America Meeting. 2006. “Predicting the Chronic Effects of PAHs using the Target Lipid Model of Narcotic Toxicity”. Montreal, Canada. November 2005.
7.0 Achievement and Dissemination

Table 1. Chemicals and their properties (at 25 o C) measured in various data sets
		Molecular Weight (g/mol)		Subcooled Liquid Solubility ( m g/L) b,d	Key PAHs e
	Log K ow a		Solid Solubility ( m g/L) b
ethane	1.730	30.00	1350000	1350000
benzene	1.943	78.11	1780000	1780000
propane	2.370	44.09	248000	248000
toluene	2.438	92.14	515000	515000
butane	2.868	58.12	61400	61400
isobutane	2.869	58.12	48900	48900
o-xylene	2.946	106.17	220000	220000
cyclopentane	2.991	70.00	156000	156000
ethylbenzene	3.006	106.17	152000	152000
m-xylene	3.032	106.17	160000	160000
p-xylene	3.051	106.17	215000	215000
9,10-Anthracenedione	3.080	208.22	116000	130000
naphthalene	3.256	128.19	31000	110000	A,B,C
isopentane	3.335	72.15	13800	13800
acenaphthylene	3.436	152.20	16100	75000	A,B,C
C3-benzenes c	3.455	120.00	19700	19700