Contradictory Conclusions Surrounding the Effects of Chemical Dispersants on Oil Biodegradation

The frequency and volume of oil spills from tankers has decreased significantly in the past 40 years (ITOPF, 2018). Knowledge gained from the 1989 Exxon Valdez oil spill has resulted in critical advances in both mechanical safety standards (e.g. double hulled tankers) and science-based oil spill response technology. Even though decades of research supports the overall environmental benefit of applying dispersants to surface oil spills, concerns surrounding the use of chemical dispersants persist as researchers are tasked with summarizing the knowledge gained from the Deepwater Horizon oil spill. During that tragedy, more than 500 million liters of light crude oil (Barbier, 2015) were discharged at the well head over 84 days in 1500 m of seawater 77 km offshore. The Deepwater Horizon oil spill response included mechanical recovery, aerial delivery of chemical dispersants and in situ burning. In a novel development, subsea dispersant injection (SSDI) was implemented to disperse the escaping oil at depth, hindering recoalescence into larger droplets and reducing volatile oil from surfacing near vessels and emergency response personnel (Rorick et al., 2012).

While dispersants were initially aimed at reducing bird mortality and minimizing ‘black tides’ arriving at shorelines (Canevari, 1974; Hayes, 1999; Lessard and DeMarco, 2000), it is now apparent that the dramatic dilution associated with dispersion and the radically increased surface area available for microbes to degrade the oil are major benefits. Nevertheless, even though dispersants were proven to be quite effective during the Deepwater Horizon spill (Bejarano et al., 2013), some members of the scientific community remain unconvinced of the benefits of dispersants (Kleindienst et al., 2015a).

Dispersants are carefully crafted mixtures of surfactant compounds and a hydrocarbon carrier solvent (Canevari, 1974; Word et al., 2015). When dispersants are mixed with oil they reduce the interfacial tension between the oil and the water, allowing the oil to be mixed into the water as tiny droplets with much less energy than would be required in the absence of dispersants. Because the effect is to ‘make the oil disappear’, there has been concern that dispersants mask a problem rather than deal with it (Bradshaw, 2014). Transforming an oil slick into tiny droplets in the water column is indeed the main function of dispersants, but this also has the important effect of increasing the surface area of the oil available to oil-degrading microorganisms and thus increasing its biodegradation (Brakstad et al., 2015). A typical dispersant application aims for one part dispersant for every 20 parts of oil, a dispersant to oil ratio (DOR) of 1:20.

‘Oil spill dispersants’ were first used on a large scale following the Torrey Canyon spill off the southwest coast of England in 1967, and there were anecdotal reports that they had significant adverse environmental impacts (Southward and Southward, 1978). But extrapolating environmental impacts from the reported response activities is difficult, at best. First, the measured effects were seen on shorelines that may or may not have received treatment with chemicals and/or flame throwers (Cowan, 1968). Second, the products used were principally commercial degreasers, designed to be used in factories and not the open environment. Nevertheless, the apparent negative effects ignited public concern surrounding the use of dispersants and motivated scientists to formulate environmentally safe yet effective products that are intended to be applied some distance from shorelines.

Current chemical dispersants on the United States (US) EPA National Contingency Plan Subpart J list (US EPA, 2020) have undergone substantial laboratory testing to ensure their environmental safety. Although these dispersants are pre-approved for use, they are used quite sparingly: NOAA has only 27 records of dispersant use in U.S.A. waters since 1968 (NOAA, 2019). There are 19 dispersants on the National Contingency Plan Product List (US EPA, 2020), but to date the most widely used in the USA is Corexit^®, with Corexit^® 9527 and 9500 being used in the response to the Deepwater Horizon blowout. The formulations of Corexit^® 9527 and Corexit^® 9500 are quite similar, as they both contain the ionic surfactant bis-(2-ethylhexyl)sulfosuccinate (DOSS), the non-ionic surfactants sorbitan monooleate (Span 80), sorbitan monooleatepolyethoxylate (Tween 80), and sorbitan trioleatepolyethoxylate (Tween 85), with hydrotreated light petroleum distillate and dipropylene glycol n-butyl ether (Corexit^® 9500) or 2-butoxyethanol (Corexit^® 9527) as solvents. All these components are Generally Regarded as Safe (GRAS) by the Food and Drug Administration when used as a food additive (Burdock and Carabin, 2004). The intrinsic toxicities of the Corexit^® formulations have been extensively studied and Corexit^® 9500 falls into the moderately toxic category for mysid shrimp and silverside fish (US EPA, 2010; Gardiner et al., 2013). Surfactants are the predominant ingredients in many household products as well as dispersants, and the acute toxicity of Corexit^® 9500 to marine aquatic organisms is either similar to or less toxic than common household cleaning agents (Word et al., 2015). Even though these dispersants have high dispersant effectiveness with minimal associated toxicity (Hemmer et al., 2011; Gardiner et al., 2013; Stroski et al., 2019), there is still concern from the public about the environmental impact of dispersants (Franklin and Warner, 2011).

Many laboratory and field studies have reported the fate and effects of oil, dispersed oil, and Corexit^® alone. Unfortunately, a subset of publications have fueled concern against dispersants by incorporating environmentally unrealistic methods that are known to produce questionable results and reporting unsupported conclusions. These publications have become a source of confusion among spill responders. In 2015, a study by Kleindienst et al. gained notoriety for claiming that dispersants exert a negative effect on microbial hydrocarbon degradation rates, but the article actually presented data showing rapid biodegradation in the presence of dispersant that would not have been seen with undispersed oil. Nonetheless, this finding was obscured by that unsubstantiated claim and Kleindienst et al. (2015b) is still widely cited despite more than a dozen papers disputing its conclusion (Prince et al., 2016a; Techtmann et al., 2017; McFarlin et al., 2018; Gofstein et al., 2020; Brakstad et al. 2018).

The objective of this paper is to identify common experimental variables in biodegradation experiments that are known to influence the results and impact the conclusions surrounding the use of dispersants. This review can assist oil spill response decision makers in determining if a study is likely to mimic the fate of oil during a real spill event. We aim to highlight the biodegradation rates of oil in different marine environments while explaining that contradictory conclusions often depend on what is being compared. We believe that when the biodegradation of dispersed oil is compared to the biodegradation of undispersed oil it is clear that effective dispersion dramatically decreases the lifetime of spilled oil in the environment.

Tables 1 and 2 include half-lives of total detectable hydrocarbons, and total alkanes, respectively, from studies that attempted to mimic environmental conditions during a surface oil spill. Studies that (1) utilized indigenous microorganisms as the inoculum, (2) were conducted at temperatures and with nutrient concentrations representative of the inoculum's environment, and (3) contained concentrations of oil or dispersant that represented those measured during an oil spill are listed in Tables 1 and 2. For comparison, Olson et al. (2017) and Zhuang et al. (2016) are included in the table as they were conducted with high oil and nutrient concentrations, with the later also including an artificial inoculum.

The rate at which oil degrades depends upon environmental conditions as well as physical attributes of the oil, such as the amount and type of oil. The main environmental variables that influence the fate of oil include its concentration, nutrient concentrations, oxygen concentration, temperature, microbial abundance and the presence of oil degrading genes in the indigenous microbiota. The following section outlines how these physical and environmental variables can be manipulated in laboratory studies to influence the fate of oil.

Depending upon the environment of interest, the method of introducing oil or dispersant into experimental incubations can impact bioavailability and thus biodegradation. Only a few compounds in a spilled oil have significant solubility (Prince et al., 2017), and these are lost from oil droplets quite near the source (Murray et al., 2019). Nevertheless, Kleindienst et al. (2015b) performed biodegradation incubations using a Chemically-Enhanced Water Accommodated Fraction (CE-WAF), a method originating from aquatic toxicity standard operating protocols (Singer et al., 2000). A CE-WAF is typically made with filtered seawater, oil (0.01 – 25 g/L) with a DOR of 1:20 (using Corexit^® 9500 or 9527), moderate mixing energy for 18 hours followed by a 6 hour settling time. CE-WAFs are the preferred systems for aquatic toxicity tests because the long settling time is a defined technique to minimize droplets of oil in the aspirated sample (since oil droplets rise to the surface). This subsequently limits the physical effects of droplets on the test species and allows the results to indicate chemical effects as intended for aquatic toxicity tests. Nevertheless, CE-WAFs were not designed for use in biodegradation experiments of whole oil because the defined removal of oil droplets creates an unrealistic system that limits the opportunity for certain oil-degraders to attach to the surface of droplets and biodegrade particulate oil. In fact most of the oil (typically >90%) added to the vessels to generate WAF and CE-WAF floats to the surface during the quiescent settling time, so any experiments on the fate of the accommodated fraction are ignoring the substantial majority of the oil. Furthermore, by removing most droplets, insoluble hydrocarbons are removed from the CEWAF sample. This not only creates a test system that neglects the influence of the majority of hydrocarbons within the oil, it also prevents the use of conserved internal markers, such as hopane, as these compounds remain with the particulate oil. Hopane is very resistant to biodegradation and serves as a conserved internal standard within the oil for assessing the biodegradation of oil compounds (Prince et al., 1994). This means that when using CE-WAFs, estimates of biodegradation have to rely on absolute amounts of measured analytes rather than on the ratio of biodegradable to non-biodegraded hydrocarbons, which in turn substantially increases the noise in the experimental data.

Conclusion: biodegradation incubations based on water-accommodated fractions neglect the impact of particulate oil droplets, which are actually the major fraction of dispersed oil in the sea. Experiments aimed at understanding the biodegradation of spilled oil should not use ‘water-accommodated fractions’ because these only represent a small fraction of the whole oil.

The inoculum (i.e. the quantity and diversity of microbe(s) added to the incubation) in biodegradation experiments can vary drastically and the type of inoculum has a substantial effect on the rate of oil biodegradation. Different research groups have used an abundance of a single microbial species, a selection of different species, an enriched natural community of microorganisms selected from a specific environment, or a fresh sample from a natural seawater. Diverse microbial communities are known to support biodegradation across the widest range of hydrocarbons, and soils are known to have a higher abundance of microorganisms than fresh water and seawater, sequentially. Of course experiments with freshly collected seawater are generally more environmentally relevant than those with laboratory-made seawater (e.g. Instant Ocean); however, lab-made test media allow control over environmental variables, such as nutrient concentrations and microbial structure.

Conclusion: Experiments should use natural fresh seawater from as close to the potentially impacted site as possible, or use water from a similar marine environment.

Crude oil is an unusual substrate for microbial growth in that while it provides a rich source of carbon and energy, it lacks the other elements essential for microbial growth, such as biologically available nitrogen, phosphorus and iron. Fortunately for oil-degrading microbes, these nutrients are available at low levels in seawater (e.g. Levitus, 1993), and as dispersed oil dilutes in the hours and days after dispersion, the microbes degrading the oil are provided the nutrients they need at appropriate levels.

If oil gets to a shoreline it becomes more concentrated as a coating on beach solids. Biodegradation can then become limited by available nutrients, even though they are naturally provided by tidal exchange in seawater environments. In such situations, biodegradation could be increased by the careful application of fertilizers, as demonstrated during the Exxon Valdez oil spill response (Bragg et al., 1994). But it would not be advisable to add nutrients to open water environments, such as lakes, rivers, or oceans, where many trophic levels (i.e. algae, fish and marine mammals) are sensitive to and negatively impacted by high nutrient concentrations.

When nutrients are added to oil biodegradation experiments with high oil concentrations (> 20 ppm), microorganisms are no longer limited by indigenous nutrient concentrations and are thus able to biodegrade oil at faster rates than incubations without added nutrients. When determining the environmental relevance of experimental biodegradation results, it is important to know the concentration of added nutrients and how this concentration reflects the environment of interest. As expected, the fastest biodegradation rates in Table 1 are from experiments with added nutrients. Both Olson et al. (2017) and Zhuang et al. (2016) incubated high oil concentrations, 67 and 700 ppm, respectively, with high nutrient concentrations and reported half-lives of ~ 3 days for total alkanes (Table 2). Such unrealistically high nutrient concentrations are commonly added to biodegradation experiments to support the biodegradation of high concentrations of oil, but it is important to recognize that such high concentrations of oil only occur on shorelines, and have little relevance to open water, where substantial dilution occurs (Lee et al., 2013).

Conclusion: Oil added to experimental incubations should be added at concentrations as environmentally realistic as possible – a few ppm or less for open sea environments, to ensure indigenous nutrient concentrations will sustain the full extent of biodegradation and prevent nutrient limitation. If laboratory experiments are incubated for long periods of time (e.g. > 28 days) in a closed system, it may be necessary to add a small amount of nutrients to maintain indigenous nutrient concentrations, as natural nutrient renewal is prevented in sealed containers. For soil and shoreline biodegradation experiments, the advantages of adding nutrients are well documented (Bragg et al., 1994) but concentrations should be monitored as high nutrient concentrations can be toxic to some microorganisms (Chaineau et al., 2005).

The widespread use of refrigeration to preserve meats (Daughtry et al., 1997) has led to the general expectation that low temperatures dramatically slow microbial growth, with the corollary that oil biodegradation will be slow in cold water. But in fact it is now clear that the indigenous microbes of any particular environment are adapted to the temperatures of that environment, and various community members possess reasonably similar growth rates at their natural temperatures. Furthermore, all show similar dependence around their natural conditions – they slow a little if it gets colder, and speed up if it gets a little warmer (Ratkowsky et al., 1982).

The key finding is that the rate of metabolism (i.e. growth and biodegradation) at the ‘normal’ temperature of growth for a microbial community is remarkably similar in Arctic and temperate climates (Rivkin et al., 1996). For example, McFarlin et al. (2014), Brakstad et al. (2015), and Prince et al. (2013) conducted experiments at −1 °C, 5 °C and 8 °C and reported half-lives of 37, 26, and 14 days, respectively, for total measurable oil (Table 1). Alkanes which make up approximately 3 – 10% of a crude oil, were shown to degrade at an estimated rate of 0.23 day⁻¹ (half-life = 3) at −1.7°C in microcosms containing sea ice and seawater collected from Canada's Northwest Passage and in the absence of added nutrients (Garneau et al., 2016). The Arctic and sub-Arctic experiments mimicked by McFarlin et al. (2014) and Brakstad et al. (2015) produced slightly longer half-lives compared to the temperate experiment mimicked by Prince et al. (2013), but the difference likely includes temperature effects on the viscosity of oil. Similarly, Rubicic et al. (2018) measured biodegradation rates of dispersed oil over 64 days at different temperatures and reported that total alkanes had a slight decrease in half-lives at 13 °C (5–7 days) compared to 5 °C (10–17 days) (Table 1), but again the overall biodegradation rate of total dispersed oil was quite similar at 5 °C and 13 °C as shown by the residual chemical profile at day 64. This indicates extrapolates from one environment at its natural temperature to another by using a relationship such as Q₁₀ (the change in rate over a 10 °C range) may be misleading (Bagi et al., 2013).

Alas what may limit biodegradation is the physical properties of the spilled oil, in particular the pour point; waxy crudes rich in larger alkanes gel at temperatures <0 °C, and pose a challenge to both mechanical recovery and dispersion (Strøm-Kristiansen et al., 1997). Most oils, however, remain liquid while floating at sea and are amenable to dispersion with the aid of dispersants during the window of opportunity (Lessard and DeMarco, 2000). Once dispersed, even at polar temperatures, crude oils are likely to undergo significant biodegradation (McFarlin et al., 2014; Garneau et al., 2016).

Conclusion: Extrapolating rates from one temperature to another is unlikely to be accurate – far better to do experiments at the temperature of interest.

The type of crude oil and its concentration has a significant effect on the rate of biodegradation of both total hydrocarbons and individual hydrocarbons. It is well known that light crude oils biodegrade faster than heavier crudes, as heavier crudes contain a relatively larger quantity of higher molecular weight hydrocarbons, resins and asphaltenes that are known to degrade more slowly (Prince et al. 2013). Oils that have been biodegraded in their initial reservoir are not likely to degrade promptly if spilled at sea (King et al., 2014).

As Prince and Butler (2014) pointed out, to determine the effects of dispersant application on the biodegradation of a surface oil spill, one should compare the dispersed oil plume to a contiguous surface oil slick, not to a physically-dispersed oil. Prince and Butler (2014) utilized floating glass booms to hold oil as a slick during an incubation experiment. Their treatment with Corexit dispersed the oil immediately into the flask and 84% of the oil was lost in 40 days, while the treatment with oil-alone stayed as a slick and only 33% of the oil was lost in 40 days. This emphasizes that if spill responders choose not to apply a dispersant, they must recognize that it is likely that the oil will remain a surface slick of concentrated oil that would impact surface dwelling organisms and continue to float until it hits a shoreline. Therefore, if researchers aim to determine the effects of dispersant application on the fate of a surface oil spill, a control treatment mimicking a surface oil slick should be included in experimental designs.

Various methods have been utilized to add oil to biodegradation experiments, with the majority directly adding oil and dispersant to the water's surface. Methods for creating the dispersed treatments are more variable with some incorporating direct addition of oil and dispersant separately, premixing oil and dispersant together, or utilizing a droplet generator. Unfortunately, many experiments have utilized ‘water accommodated fractions’, without (WAF) or with dispersants (Chemically Enhanced Water Accommodated Fraction – CEWAF). These follow the methodology of Singer et al. (2000), which was aimed at reproducible methods for determining the toxicity of dispersed oil. Physical fouling with oil droplets complicates efforts to determine the worst-case ‘toxicity’ of oil to macrobiota (small fish and shrimp) rather than its general lethality, due to challenges separating physical effects from chemical effects. Singer et al. (2000) developed methods that minimized the formation of droplets while encouraging the dissolution of soluble components, but in allowing most of the oil to float on the surface of their initial bottles, and drawing the ‘accommodated fractions’ from the bottom of those vessels, they typically ignored the vast majority, typically >90%, of the oil. Experiments that add low concentrations of oil directly to experimental systems more closely model marine oil spills.

Dispersant application during an oil spill response is not always necessary, especially if low volumes of oil are spilled. If enough wave action (or other mixing energy) is present, low oil concentrations disperse naturally, without the aid of a chemical dispersant. McFarlin et al. (2014) conducted incubations with natural seawater and 2.5 ppm of crude oil with and without a small amount of dispersant (0.167 ppm), and showed that physically dispersed oil biodegraded at a similar rate as chemically dispersed oil at low oil concentrations (Table 1). Prince et al., 2016b found no significant differences between three different dispersants, and Brakstad et al. (2018) showed that droplet size has a greater impact on biodegradation rate than the type and amount of dispersant applied. In fact oil biodegradation is most strongly influenced by the surface area of oil, with faster rates observed for smaller oil droplets regardless of the presence of dispersant (Brakstad et al., 2015).

Concentrations of dispersed oil under successfully dispersed slicks or in subsurface ‘plumes’ are very low (<1 ppm within hours; Lee et al., 2013; Bejarano et al., 2013), yet most experiments are conducted with oil at concentrations more than a magnitude higher because of the detection limits of analytical instruments. High oil concentrations in static biodegradation experiments decrease oil biodegradation rates to unrealistically low levels by creating a scenario that is never encountered in the real world. The higher the oil concentration in a closed experimental system, the higher the likelihood that oil will be in larger droplets, with smaller surface to volume ratios, and that nutrients will limit biodegradation. Microbes in experiments with lower oil concentrations are able to rely on the indigenous nutrients in the seawater for metabolic functions. Higher oil concentrations in dispersed oil incubations also increases the frequency that oil droplets encounter each other, promoting agglomeration to form larger droplets that eventually coalesce and float or adhere to the glass walls of incubation containers. Larger droplets (and the floating oil slick) likely hinder the diffusion of soluble aromatics to the droplet-water interface. Prince et al. (2017) showed that these effects began to be significant at 100 ppm oil in New Jersey seawater, and therefore recommended doing experiments at lower concentrations than that.

Conclusion: Dispersed oil is present at very low concentrations in the real world, so experiments aimed at understanding its fate must do experiments at reasonably similar concentrations. Experiments aimed at understanding the effects of dispersants on oil degradation must include a control where the oil will not disperse.

If conclusions from oil biodegradation experiments are to inform spill responders on the fate of oil and the efficiency of dispersants, then those experiments must include experimental variables (e.g. inoculum, temperature, and concentration of nutrients, oil, and dispersant) that mimic those of the environment of interest. In particular, experimenters must bear in mind that dispersants aid the dispersion of oil slicks, but have the potential to leave the dispersed droplets due to their water solubility. Furthermore, components of Corexit are degraded at similar rates to that of dispersed oil (McFarlin et al., 2018). In summary, dispersants have no significant impact on the biodegradation of physically dispersed oil (Tables 1 and 2), but that dispersed oil, however generated, is degraded much more rapidly than when it is in a slick. Field samples of spilled oil have shown that floating oil slicks experience minimal biodegradation at sea (Aeppli et al., 2014).

Biodegradation experiments that aim to inform the fate of oil in the environment should carefully consider the experimental design as some environmental variables can profoundly impact the results and conclusions of laboratory experiments. For example, it is important to include a treatment mimicking a surface oil slick if the objective of the experiment is to determine the effect of dispersant application on a surface spill. Furthermore, it is important for researchers to acknowledge the limitations of experiments conducted with ‘water-accommodated fractions’. Initial oil concentrations in biodegradation experiments should be as environmentally realistic as possible to ensure that natural nutrient concentrations can sustain biodegradation and prevent nutrient limitation. In conclusion, experimental design should account for natural dilution and be conducted at the temperature of interest and with indigenous microorganisms from the environment of interest. Experimental results should not be used to support environmental predictions if the experiment does not mimic the environment of interest.