Battelle Sediment Conference Jacksonville, Florida February 4, 2009

1. Critical Evaluation of the Applicability of SedimentEffect Concentrations forPCBs in Site-Specific Ecological Risk Assessments Battelle Sediment Conference Jacksonville, Florida February 4, 2009 S. Kane Driscoll, Ph.D. N. W. Gard, Ph.D. T. C. Ginn, Ph.D.

2. Sediment Effect Concentrations • Threshold Effect Concentration (TEC) = 0.04 mg/kg • < TEC Adverse effects unlikely • Moderate Effect Concentration (MEC) = 0.40 mg/kg • > MEC Adverse effects frequently occur • Extreme Effect Concentration (EEC) = 1.7 mg/kg • > EEC Adverse effects usually or always occur SEC Source: MacDonald et al. (2000)

3. Developers’ Intended Uses • Reconcile existing sediment quality guidelines (SQGs) • Predict sediment toxicity • Demonstrate causal effects • Evaluate need for sediment remediation • Determine injury to benthic organisms

4. Derivation • PCB sediment effect concentrations (SEC) developed from 30 SQGs • SQGs grouped by narrative intent to develop 3 SECs • SECs then evaluated by: • Determination of predictive ability • Comparison to dose-response data from spiked-sediment toxicity tests • Comparison with SQGs developed via EqP approach

5. SEC Limiting Factors • Validity of underlying SQGs • Classification of SQGs • Effects of co-occurring chemicals • Representativeness of datasets • Predictive ability • Cause-effect relationships • Concurrence with site-specific data

6. Freshwater SQGs (mg/kg dry wt) Wide range of values within groupings Redundant use of same data in more than one SQG Limited data sets for EEC Misclassification of some values (e.g., NEC and PAET-WA are comparable to EECs based on developers’ definitions)

7. SQG Limitations • Small sample sizes • e.g., only 5 samples with toxic effects for Great Lakes SQGs • For some SQGs, low prevalence of toxicity above stated probable effect concentrations • Limited predictive value • Databases skewed toward reference sites and slightly-contaminated areas

8. Application of SQGs • Developers generally recognized limitations • Low predictive ability • Based on correlation, not cause-effect • Recommended use primarily as screening values • Predictions need to be confirmed by site-specific evaluation

9. Testing Tools

10. Sediment ToxicityTests

11. Co-occurring Chemicals • Most SQG data is from sites with multiple contaminants • Chemicals co-vary with each other • Many studies that are sources of SQG were unable to: • Correlate effects with any specific chemical • Attribute effects specifically to PCBs

12. Unrepresentative Datasets • For 40% of SQGs used in calculation of SECs, most data (70%) is < 0.5 mg/kg • For validation data, 69% (sw) and 89% (fw) of data < 0.5 mg/kg • Only 2% saltwater and 20% of freshwater validation data > EEC (1.7 mg/kg) • Potential toxicity of PCBs not well characterized at sediment concentrations > EEC

13. Causation • MacDonald et al. note that MEC and EEC are comparable to or higher than chronic thresholds calculated via EqP approach • Cited EqPs were developed for protection of wildlife and humans from consumption of contaminated fish • EqP-derived SQGs for wildlife have no relevance for SEC validation or determination of causation

14. Site-specific Conditions • Most sites have multiple contaminants • Two validation sets from sites with PCBs as the primary contaminant • Fox River-Green Bay • Waukegan Harbor • These sites can provide a check on the predictability of the SECs

15. Fox River-Green Bay • 0.09–6.8 mg/kg PCBs • 7 of 13 stations exceed EEC, 11 exceed MEC • However, only 1 station was toxic in tests with Chironomus ripariusandHexagenialimbata • SECs result in erroneous identification of numerous stations as toxic

16. Relationship between Sediment Toxicity Endpoints

17. Waukegan Harbor • 0.9–8.9 mg/kg PCBs • 17 of 18 stations exceed EEC, all exceed MEC • Lowest rate of amphipod survival was 66 percent • EPA study concluded sediments not generally lethal to amphipods • Negative effects on growth ambiguous, possibly due to sediment and test conditions of negative controls used for comparison

18. Relationship between Sediment Toxicity Endpoints

19. Waukegan Harbor Data • Samples account for 52 % of toxic samples >EEC in SEC validation data set • Predictive ability of the EEC is very sensitive to the single data set for Waukegan Harbor

20. Relevance of Site-Specific Studies • Application of SECs produces highly erroneous conclusions • Predictive ability of MEC and EEC < 15 percent for Fox River-Green Bay • No consistent negative relationship between toxicity and PCB concentrations • Some decline in percent survival at 6–8 mg/kg • However, some of the highest growth and survival rates also seen at 7–8 mg/kg

21. Conclusions • SECs are simple mathematical constructs • Share technical faults of underlying SQGs • Methodological issues (redundancy, misclassification, misinterpretation) influence final values • Biased by influence of samples with < 0.5 mg/kg PCB and presence of co-occurring chemicals • Site-specific application indicates low predictive ability of MEC and EEC

22. SEC Recommendations • Use in screening-level evaluations only • Should not be used to predict presence of sediment toxicity or attribution of causality to PCBs • Site-specific assessments of biological effects are required • Should not be used to determine the extent of injury to sediment-dwelling organisms