Aaron D. Redman

PETROTOX: An aquatic toxicity model for petroleum substances

A spreadsheet model (PETROTOX) is described that predicts the aquatic toxicity of complex petroleum substances from petroleum substance composition. Substance composition is characterized by specifying mass fractions in constituent hydrocarbon blocks (HBs) based on available analytical information. The HBs are defined by their mass fractions within a defined carbon number range or boiling point interval. Physicochemical properties of the HBs are approximated by assigning representative hydrocarbons from a database of individual hydrocarbons with associated physicochemical properties. A three-phase fate model is used to simulate the distribution of each structure among the water-, air-, and oil-phase liquid in the laboratory test system. Toxicity is then computed based on the predicted aqueous concentrations and aquatic toxicity of each structure and the target lipid model. The toxicity of the complex substance is computed assuming additivity of the contribution of the individual assigned hydrocarbons. Model performance was evaluated by using direct comparisons with measured toxicity data for petroleum substances with sufficient analytical characterization to run the model. Indirect evaluations were made by comparing predicted toxicity distributions using analytical data on petroleum substances from different product categories with independent, empirical distributions of toxicity data available for the same categories. Predictions compared favorably with measured aquatic toxicity data across different petroleum substance categories. These findings demonstrate the utility of PETROTOX for assessing environmental hazards of petroleum substances given knowledge of substance composition.

Guidance for improving comparability and relevance of oil toxicity tests

The complex nature and limited aqueous solubility of petroleum substances pose challenges for consistently characterizing exposures in aquatic life hazard assessments. This paper reviews important considerations for the design, conduct and interpretation of laboratory toxicity tests with physically and chemically dispersed oils based on an understanding of the behavior and toxicity of the hydrocarbons that comprise these substances. Guiding principles are provided that emphasize the critical need to understand and, when possible, characterize dissolved hydrocarbon exposures that dictate observed toxicity in these tests. These principles provide a consistent framework for interpreting toxicity studies performed using different substances and test methods by allowing varying dissolved exposures to be expressed in terms of a common metric based on toxic units (TUs). The use of passive sampling methods is also advocated since such analyses provide an analytical surrogate for TUs. The proposed guidance is translated into a series of questions that can be used in evaluating existing data and in guiding design of future studies. Application of these questions to a number of recent publications indicates such considerations are often ignored, thus perpetuating the difficulty of interpreting and comparing results between studies and limiting data use in objective hazard assessment. Greater attention to these principles will increase the comparability and utility of oil toxicity data in decision-making.

PETRORISK: A risk assessment framework for petroleum substances

PETRORISK is a modeling framework used to evaluate environmental risk of petroleum substances and human exposure through these routes due to emissions under typical use conditions as required by the European regulation for the Registration, Evaluation, Authorization and Restriction of Chemicals (REACH). Petroleum substances are often complex substances comprised of hundreds to thousands of individual hydrocarbons. The physicochemical, fate, and effects properties of the individual constituents within a petroleum substance can vary over several orders of magnitude, complicating risk assessment. PETRORISK combines the risk assessment strategies used on single chemicals with the hydrocarbon block approach to model complex substances. Blocks are usually defined by available analytical characterization data on substances that are expressed in terms of mass fractions for different structural chemical classes that are specified as a function of C number or boiling point range. The physicochemical and degradation properties of the blocks are determined by the properties of representative constituents in that block. Emissions and predicted exposure concentrations (PEC) are then modeled using mass-weighted individual representative constituents. Overall risk for various environmental compartments at the regional and local level is evaluated by comparing the PECs for individual representative constituents to corresponding predicted no-effect concentrations (PNEC) derived using the Target Lipid Model. Risks to human health are evaluated using the overall predicted human dose resulting from multimedia environmental exposure to a substance-specific derived no-effect level (DNEL). A case study is provided to illustrate how this modeling approach has been applied to assess the risks of kerosene manufacture and use as a fuel.

Use of passive samplers for improving oil toxicity and spill effects assessment.

Methods that quantify dissolved hydrocarbons are needed to link oil exposures to toxicity. Solid phase microextraction (SPME) fibers can serve this purpose. If fibers are equilibrated with oiled water, dissolved hydrocarbons partition to and are concentrated on the fiber. The absorbed concentration (Cpolymer) can be quantified by thermal desorption using GC/FID. Further, given that the site of toxic action is hypothesized as biota lipid and partitioning of hydrocarbons to lipid and fibers is well correlated, Cpolymer is hypothesized to be a surrogate for toxicity prediction. To test this method, toxicity data for physically and chemically dispersed oils were generated for shrimp, Americamysis bahia, and compared to test exposures characterized by Cpolymer. Results indicated that Cpolymer reliably predicted toxicity across oils and dispersions. To illustrate field application, SPME results are reported for oil spills at the Ohmsett facility. SPME fibers provide a practical tool to improve characterization of oil exposures and predict effects in future lab and field studies.

EXTENSION AND VALIDATION OF THE TARGET LIPID MODEL FOR DERIVING PREDICTED NO-EFFECT CONCENTRATIONS FOR SOILS AND SEDIMENTS

Substance risk assessments require estimation of predicted no-effect concentrations (PNECs) in soil and sediment. The present study applies the target lipid model (TLM) and equilibrium partitioning (EqP) model to toxicity data to evaluate the extrapolation of the TLM-derived aquatic PNECs to these compartments. This extrapolation assumes that the sensitivity of aquatic species is similar to that of terrestrial and benthic species. The acute species sensitivity distribution, expressed in terms of species-specific critical target lipid body burdens, was computed using the TLM-EqP framework and found to span a similar range as the aquatic organism species sensitivity distribution but with a slightly lower median value (less than 2 times). The species sensitivity distribution for acute-to-chronic ratios also exhibited a similar range and distribution across species, suggesting similar mechanisms of action. This hypothesis was further tested by comparing empirical soil/sediment chronic effect levels to the calculated PNEC derived using TLM-EqP. The results showed that 95% of the compiled chronic effects data fell above the PNEC, confirming an adequate protection level. These findings support the conclusion that TLM-derived aquatic PNECs can be successfully extrapolated to derive credible PNECs for soil and sediment compartments.

Chronic toxicity of selected polycyclic aromatic hydrocarbons to algae and crustaceans using passive dosing.

Because of the large number of possible aromatic hydrocarbon structures, predictive toxicity models are needed to support substance hazard and risk assessments. Calibration and evaluation of such models requires toxicity data with well-defined exposures. The present study has applied a passive dosing method to generate reliable chronic effects data for 8 polycyclic aromatic hydrocarbons (PAHs) on the green algae Pseudokirchneriella subcapitata and the crustacean Ceriodaphnia dubia. The observed toxicity of these substances on algal growth rate and neonate production were then compared with available literature toxicity data for these species, as well as target lipid model and chemical activity-based model predictions. The use of passive dosing provided well-controlled exposures that yielded more consistent data sets than attained by past literature studies. Results from the present study, which were designed to exclude the complicating influence of ultraviolet light, were found to be well described by both target lipid model and chemical activity effect models. The present study also found that the lack of chronic effects for high molecular weight PAHs was consistent with the limited chemical activity that could be achieved for these compounds in the aqueous test media. Findings from this analysis highlight that variability in past literature toxicity data for PAHs may be complicated by both poorly controlled exposures and photochemical processes that can modulate both exposure and toxicity. Environ Toxicol Chem 2016;35:2948-2957.

A re‐evaluation of PETROTOX for predicting acute and chronic toxicity of petroleum substances

The PETROTOX model was developed to perform aquatic hazard assessment of petroleum substances based on substance composition. The model relies on the hydrocarbon block method, which is widely used for conducting petroleum substance risk assessments providing further justification for evaluating model performance. Previous work described this model and provided a preliminary calibration and validation using acute toxicity data for limited petroleum substance. The objective of the present study was to re-evaluate PETROTOX using expanded data covering both acute and chronic toxicity endpoints on invertebrates, algae, and fish for a wider range of petroleum substances. The results indicated that recalibration of 2 model parameters was required, namely, the algal critical target lipid body burden and the log octanol-water partition coefficient (KOW ) limit, used to account for reduced bioavailability of hydrophobic constituents. Acute predictions from the updated model were compared with observed toxicity data and found to generally be within a factor of 3 for algae and invertebrates but overestimated fish toxicity. Chronic predictions were generally within a factor of 5 of empirical data. Furthermore, PETROTOX predicted acute and chronic hazard classifications that were consistent or conservative in 93 and 84% of comparisons, respectively. The PETROTOX model is considered suitable for the purpose of characterizing petroleum substance hazard in substance classification and risk assessments. Environ Toxicol Chem 2017;36:2245-2252.

Evaluating persistence of petroleum hydrocarbons in aerobic aqueous media.

Evaluating chemical persistence is essential in performing environmental risk assessments with biodegradation data being a key requirement of many national, regional and local regulatory bodies. The results from such assessments will, together with information relating to toxicity and bioaccumulation, determine classification and labeling under global (Global Harmonized System; GHS, Winder et al., 2005) and/or regional (Classification, Labelling and Packaging; CLP, Lahl and Hawxwell, 2006) regulations. The physical and chemical variability found in petroleum hydrocarbons however, present unique challenges in assessing

Role of entrained droplet oil on the bioavailability of petroleum substances in aqueous exposures

Bioavailability of petroleum substances is a complex issue that is affected by substance composition, the physicochemical properties of the individual constituents, and the exposure preparation system. The present study applies mechanistic fate and effects models to characterize the role of droplet oil on dissolved exposure and predicted effects from both neat and weathered crude oils, and refined fuel oils. The main effect from droplet oil is input of additional dissolved hydrocarbons to the exposure system following preparation of the initial stock solution. Toxicity was characterized using toxic units (TU) and shows that replenishment of bioavailable hydrocarbons by droplets in toxicity tests with low droplet content (e.g., <1mg/L) is negligible, consistent with typical exposure conditions following open ocean oil spills. Further, the use of volumetric exposure metrics (e.g., mg/L) introduces considerable variability and the bioavailability-based metrics (e.g., TUs) provide a more consistent basis for understanding oil toxicity data.

Investigating the role of dissolved and droplet oil in aquatic toxicity using dispersed and passive dosing systems.

Characterization of the aquatic toxicity of oil is needed to support hazard assessment and inform spill response. Natural processes and mitigation strategies involving dispersant use can result in exposures to both dissolved and droplet oil that are not typically differentiated when oil exposures are characterized in toxicity tests. Thus, the impact of droplets on aquatic toxicity is largely uncharacterized. To improve the understanding of the role of droplets, acute toxicity tests with Daphnia magna and Americamysis bahia were performed with Endicott crude oil in low-energy mixing systems with and without Corexit 9500 dispersant. Exposures were also prepared by placing crude oil in silicone tubing and passively dosing test media to provide dissolved oil exposures without droplets. A framework is described for characterizing dissolved phase exposures using both mechanistic modeling and passive sampling measurements. The approach is then illustrated by application to data from the present study. Expression of toxicity in terms of toxic units calculated from modeled dissolved oil concentrations or passive sampling measurements showed similar dose responses between exposure systems and organisms, despite the gradient in droplet oil. These results indicate that droplets do not appreciably contribute to toxicity for the 2 species investigated and further support hazard evaluation of dispersed oil on the basis of dissolved exposure metrics. Environ Toxicol Chem 2017;36:1020-1028.