The test media can be prepared in a number of ways, with each method having advantages and disadvantages. In the following literature review, water-soluble fractions (WSF) are typically created using a generator column or gravel column. In such preparations, water is pushed through an oil-coated matrix, with the soluble fractions “dosing” the seawater. Recently, SINTEF developed a droplet generator that allows for the formation of physical dispersions with neutrally buoyant droplets of a diameter similar to chemically dispersed oil. This allows for a more direct comparison of physically and chemically dispersed oil. 

Water-soluble fractions (WSF) represent the soluble fractions of oil and are typically created using a generator column or gravel column. In such preparations, water is pushed through an oil-coated matrix, with the soluble fractions “dosing” the seawater. Oil is coated on to granular materials (e.g. gravel, or glass beads) and contained in a column having an entry port and exit port to accommodate the flow of filtered seawater through the column. Generally, fresh oil is added initially and naturally weathers over time. Adult female Calanus glacialis and Calanus finmarchicus were exposed to water soluble fraction (WSF) crude oil in a continuous flow experiment; hatching success of eggs and production of fecal pellets were assessed (Jensen and Carroll 2010). The WSF solutions were prepared by pumping filtered seawater through columns containing glass beads coated with crude oil (Duesterloh et al. 2002; Camus and Olsen 2008). Chemical analyses of WSF for both experiments included polynuclear aromatic hydrocarbons (PAH; based on USEPAs 16 priority compounds). Authors noted that for the C. glacialis experiment only 4 PAHs of the 16 analytes measured exceeded detection limits in the low and high dose treatments (naphthalene, fluorene, phenanthrene, and acenaphthylene). The low dose (3.6 μg/L PAH) treatment had a much lower naphthalene concentration relative to the high dose treatment. Total concentrations of 16 PAHs gradually decreased over time (naphthalene decreased more rapidly whereas heavier compounds were more stable). Results from C. glacialis tests showed no significant differences between highest treatment (10.4 μg/L PAH) and controls for cumulative egg or fecal pellet production; however, hatching success was significantly reduced. Exposure of C. finmarchicus to 7.0 μg/L PAH reduced feeding efficiency of adults. 

Water-accommodated fractions (WAF) are a preparation method where oil is placed over a volume of seawater and mixed with a magnetic stirrer, generally for 18 to 24-hours. The amount of mixing energy may vary, but is typically adjusted such that visible oil droplets do not become entrained in the seawater. The mixing period is then followed by a “resting” period which allows larger oil droplets to rise to the surface. The test media are removed via a port at the bottom on the mixing container. WAF preparations are intended to include the dissolved fraction and the smaller, neutrally buoyant oil droplets. A dispersant can be added to the WAF preparation to create a dispersed WAF or Chemically-Enhanced WAF (CEWAF). Methods for preparing petroleum-water mixtures for laboratory toxicity testing were first developed by Anderson et al. (1974) and have since been modified by several investigators to create more consistent petroleum-water and petroleum-dispersant mixtures that are more representative of field conditions (Singer et al. 2000, Nordtug and Oistein 2007, Aurand and Coelho 2005; Barron 2003, Clark et al. 2001). Many of the studies conducted with Arctic species have been conducted with WAF or CEWAF preparations. 

An additional method has been developed to create oil-in-water preparations that may better approximate the distribution of smaller, neutrally buoyant oil droplets in the physically dispersed oil. Nordtug et al. (2008) present a method for continuous and predictable in-line production of oil dispersions with defined size distribution of oil droplets based on theoretical models for droplet formation. Test solutions enter the experimental chamber via multiple nozzles which create turbulent dispersions of oil droplets and water soluble components. The mean droplet size and the droplet size distribution was determined by a combination of energy input (flow rate) and the number of nozzles used. The system enables simultaneous comparison between the effects of different concentrations of the dispersion and their corresponding water soluble fractions and the net effects of the oil droplet fractions. Flow-through into each individual exposure chamber was controlled by pumping exposure solution out near the surface, causing fresh solution to flow passively by entrainment into the chamber at a corresponding rate. Chemical measurements have been used to verify concentrations and quality of the dispersions and water soluble phases. The hydrocarbon profiles of the oil and the dispersions were very similar, whereas only a fraction of the components are in the dissolved phase. Tests with early life stage cod larvae (Gadus morhua) and copepods (C. finmarchicus) have been conducted with OWD preparations, allowing for evaluations of both the toxicity associated the dissolved fractions as well as adhesion of particulate oil (Olsvik et al. 2010 and Melbye et al. 2006). 

The type of oil used in test media preparation may affect toxicity, as well as factors like fresh or weathered oil is used. After oil is released into the environment, its chemical and physical characteristics change through processes of: evaporation, dispersion, emulsification, dissolution, oxidation, and biodegradation. The weathering process is described in more detail in other sections of this report. For laboratory toxicity studies, typically weathering is simulated by evaporation causing a loss of the most volatile hydrocarbons and a reduction in the oil volume. The volatile hydrocarbons generally represent the more acutely toxic fractions of oil. It is important to note that the evaporative weathering only accounts for one portion of the weathering process. Many studies are conducted with fresh oil as it represents a worst-case scenario, particularly for acute toxicity estimates. There have been several toxicity studies with Arctic species that have used weathered oil in test media. 

Photo 6-2: Preparing for bioassay test under Arctic conditions (William Gardiner)

The manner in which exposure concentrations are presented can also affect the usefulness and comparability of data. Measured effects concentrations may be reported as a percentage of the original test media (nominal concentrations) or as measured concentrations. Toxicity reported as nominal concentrations represent a percentage of a particular loading rate and are most closely tied to the laboratory method. Nominal concentrations are perhaps the most difficult data to relate to field data or to use in models, unless coupled with measured concentrations of hydrocarbons. Measured concentrations can include several analytical methods for expressing petroleum hydrocarbons. TPH are the most commonly reported measure for both laboratory and field studies. As such, they allow for comparisons between lab and field exposures. However, TPH includes many petroleum compounds that do not contribute to toxicity. This can be a significant source of variability and error when using toxicity data in models. Total petroleum aromatic hydrocarbons (PAHs) are believed to be more specific to the toxic fractions; however, this measure also includes non-toxic fractions. Finally, toxicity may be expressed in terms of a specific compound, such as naphthalene or 2-methylnaphthalene. While this removes the concern of including too many compounds, it also may not account for toxicity associated with secondary hydrocarbons. In many cases, it is important to understand how data are being applied and their limitations. 

Historically, the toxicity of oil, dispersed oil and single compounds was evaluated using continuous exposures, with test concentrations remaining more or less constant over the test period. In some cases, such as dispersed oil in open water, this may overestimate exposure. In such cases, spiked, or declining exposures have been used to better simulate field conditions (Photo 6-2). Each exposure method may be appropriate depending on the application of the data (e.g. spiked exposures for dispersed oil and continuous exposures for oil in ice). This topic is dealt with later in this chapter. 

Where possible, test species that are included in toxicity studies are valuable ecosystem components (VECs) that represent the relevant environmental compartments potentially exposed to oil or treated oil. The VECs that were selected for the Arctic are discussed in Chapter 2. However, species that are VECs do not necessarily lend themselves to toxicity studies. Characteristics for suitable test organisms include the following: 

• A sensitivity to oil and treated oil; 
• Relative abundance and an availability for collection or culture; 
• The ability to withstand laboratory handling; 
• Meaningful and measurable endpoints over the time period of the study; 
• Native to the site-specific conditions (e.g. cold water). 

There were toxicity data for a number of Arctic VECs including copepods, Arctic and Atlantic cod, sculpin, capelin, ice amphipods, benthic amphipods, crab, shrimp, and echinoderms. While there were toxicity data for a number of the VECs, in many cases the data are limited and endpoints. The suitability of data for conducting ecosystem-level evaluation of different OSR alternatives is considerably more limited, with copepods and Arctic cod having the most complete data sets. 

Endpoints, or measures of effect, are dependent upon the test organism and the study goal. In general, toxicity is expressed as a dose-response to a certain concentration of test media. The most common endpoint is mortality, with the median Lethal Concentration (or LC50) being that concentration of test media that will cause 50 % mortality to the exposed organisms. Sub-lethal effects such as growth or reproduction are usually expressed as a median Effect Concentration (EC50) representing the concentration at which 50 % of the exposed organisms have a significant change in those end-points. These indicators are often extrapolated to an assessment endpoint depending on the purpose of the assessment. Effect concentrations can for example be used to assess environmental threshold values like a Predicted Environmental No-effects Concentration (PNEC) or to construct a species sensitivity distribution used to assess the fraction of species that is at risk. Other common measures of toxicity include No-observed Effect Concentrations (NOEC) and Lowest-Observed Effects Concentrations (LOEC). 

In order to evaluate population or ecosystem level effects, there are a variety of extrapolation models that utilize the toxicity endpoints (e.g. LC50) as input. Population models coupled with field-collected data on the distribution, abundance, natural mortality and fecundity are used to predict population level effects associated with modeled exposure scenarios. Simulated exposures using physical fates models such as the Integrated Oil Spill Impact Model System (SIMAP) allow for different spill and response scenarios to be evaluated. Gallaway et al. (2007; 2013, in prep) used toxicity data from laboratory exposures of Arctic cod to WAF and CEWAF, field collected data from the western Beaufort Sea, and the SIMAP model simulations to predict potential population effects from physically and chemically treated oil in open water. 

Another approach to population models is to incorporate many ecological components of an ecosystem to predict the potential cascade effects and repercussion in the trophic chain over time (Peterson et al. (2003). An example of such a model is the SYMBIOSES model that is under development for the Barents Sea (Carroll and Smit 2011; DeLaender et al. 2010). The main objective of such an impact analysis tool is to examine extent of effects on a VEC (e.g. larval Arctic cod) that result in significant effects on fish recruitment and stock size. While this is similar to the goal of the model used by Gallaway et al. (2007; in prep), this ecosystem approach considers indirect effects of a discharge that originates from ecological interactions. Data used to quantify effects of oil associated compounds on the Arctic trophic linkages include many food web links from the planktonic community to fish larvae and adult fish. 

The SYMBIOSES model integrates food web analyses of the structure and functions of the whole microbial and plankton community of the Barents Sea (bacteria, protozoa, phytoplankton, zooplankton, fish eggs and fish larvae) in terms of response to oil; for instance using an in situ mesocosm experiment. Today, such mesocosms exist and can be used in boreal or Arctic ecosystems to address plankton community responses to petroleum discharges (Hjorth et al. 2008). The mesocosm could be deployed in the Arctic waters during the ice free season especially at the peak of the phytoplankton bloom when the ice break-ups occur. While mesocosms represent an increase in the level of complexity of exposure design, they are subject to similar artifacts as laboratory studies. Interactions observed in the laboratory should be confirmed with field data when possible. 

Empirical toxicity data may also be used as part of predictive environmental fate and effect models that aim to integrate the chemical relationship of petrochemicals with prevailing biological processes (Figure 6-1). To form a holistic ecosystem perspective, complex relationships have been simplified according to basic assumptions, such as: 1) Hydrocarbon block and Quantitative Structure Activity Relationship (QSAR) models, i.e. hydrocarbons sharing similar structure and physiochemical properties will behave similarly in the environment; 2) Partitioning theory, i.e. chemicals belonging to similar structural groups will partition to organic carbon, water, and biological membranes in a similar manner and further, most hydrocarbons partition to lipids within the organisms (e.g. Target Lipid Model; McGrath et al. 2004)); 3) Critical body residues (CBR) coupled with median effect levels will provide a good estimation of impact or injury at the species level of investigation (threshold response levels have been summarized for many species groups); 4) Sensitivity of early life stages – incorporation of data representing early life stage endpoints result in the most conservative estimation of population disruption (e.g. OECD guidelines have been set for fish embryo toxicity; LC50 values greater than 100 mg/L are considered non-toxic). 

Hydrocarbon models, such as PETROTOX, organize the numerous hydrocarbon compounds in solution into structurally based ‘blocks’. If the behavior and toxicity of each block is known, the model can represent the relative predicted toxic activity level of the combined compounds present in different environmental compartments and estimate the resulting toxicity. Such an approach relies on laboratory estimates of toxicity for a variety of hydrocarbon constituents. This model assumes additivity of the toxicity for each block which may not necessarily be true. At this point in time, the quantity and quality of data with specific hydrocarbons for Arctic species are limited, however, if temperate data are also used, this approach may be feasible. 

The CBR approach couples toxicity data with tissue residue levels to predict effects. Such an approach is based on the premise that the toxicity of oil and treated oil is related to those hydrocarbons taken into the tissues. While this may have merit for exposures that occur over a long enough period to allow for uptake, it does not account for acute effects from short term exposures or physical effects due to adhesion. This method also relies on tissue residue data which are limited for Arctic species.