ARE OXYGENATED HYDROCARBONS IMPORTANT OIL DEGRADATION PRODUCTS AFTER OIL SPILLS?

Typically, environmental scientists focus on the fate, transport and effect of petroleum hydrocarbons (mainly PAHs) to assess environmental impact after an oil spill. Different weathering processes such as biodegradation, photoxidation and physical weathering (evaporation, dissolution, partitioning to other phases) are believed to alleviate the environmental impacts of oil spills (Marine Board et al., 2003). However, processes such as biodegradation and photochemistry have been shown to convert hydrocarbons into transformation products, which may be environmentally important (Aeppli et al., 2012). These transformation products are generally overlooked during the risk assessment of an oil spill. For example, regulatory limits are available for PAHs but not for oxygenated PAHs (OxPAHs). To understand the fate and effect of transformation products of hydrocarbons, the following questions need to be addressed: (i) What are the dominating transformation products of hydrocarbons that arise due to oil weathering? (ii) What are environmental levels of transformation products relative to their parent hydrocarbons? (iii) Are they intrinsically bioaccumulative and toxic?

In samples collected in the aftermath of the DHOS, a surprisingly large amount of OxHCs have been found (Aeppli et al., 2012; Hall et al., 2013; Ruddy et al., 2014). More than 50% of these samples’ mass were non GC-amenable and had a high oxygen content, and several hydroxyl and carbonyl functional groups were identified (Aeppli et al., 2012). The appearance of OxHCs in the weathered oil samples were found to be correlated with the disappearance of saturated naphthenic-type hydrocarbons and alkyl benzenes (Hall et al., 2013). In addition, Tidwell et al. (2016) reported several PAHs along with OxPAHs using passive samplers during and after shoreline oiling from the DHOS, and hypothesized that different biotic and abiotic processes transformed PAHs to OxPAHs (Tidwell et al., 2016). Furthermore, the transformation of hydrocarbons into OxHCs increased as a function of weathering time (Chen et al., 2016). These observations call for a deeper understanding into the transformation pathways, molecular composition, environmental levels and the assessment of bioaccumulation and toxicity potential of OxHCs.

The OxHCs are likely to have the attributes of persistence, bioaccumulation and toxicity (PBT). The metrics used for PBT assessment are the environmental half-lives (t_1/2), bioconcentration factors (BCF), defined as equilibrium partitioning between water and the lipid pool (i.e. membrane plus storage lipid) of an organism, and the median lethal concentrations (LC₅₀) required to kill 50% of organisms (e.g., the fathead minnow). According to regulatory agencies such as the European Chemicals Agency (ECHA) and the US EPA, chemicals are classified as PBT when they have t_1/2 > 180 days, BCF > 2,000 and LC₅₀ < 0.1 mg/L. The ECHA further refers to chemicals with BCF ≥ 2,000 and ≥5,000 respectively as “bioaccumulative” (B) and “very bioaccumulative” (vB) (EU REACH, 2007) (ECHA, 2007).

Because of their long-term prevalence in these samples OxHCs may be considered as a persistent substance. OxHCs are also likely bioaccumulative and toxic, based on their moderate to high hydrophobicity (log Kow > 3, where Kow is octanol-water partition coefficient). However, there are currently not many PBT data available on OxHC. it is therefore important to screen the representative chemical families of OxHCs under the light of these regulatory guidelines.

The Abraham Solvation Model (ASM) is a widely-used approach to reliably estimate environmentally relevant properties (Abraham et al., 1994). The ASM comprises of five or six molecular descriptors that account for polarity, polarizability, hydrogen-bond donating/accepting capacity, and the solvation energy of cavity formation (Abraham et al., 1994). Experimental values for these descriptors are freely available for approximately 8,000 chemicals in the UFZ LSER database (Endo et al., 2015). The relevant environmental partitioning properties required to understand bioaccumulation and toxicity potential can be estimated using reported ASM equations, which along with several other properties have been compiled in literature (Nabi & Arey, 2017; Nabi et al., 2014).

In this paper, we use the ASM to calculate the bioaccumulation and toxicity potential of major OxHC families that are expected to be present in the weathered oils. We also present the concentrations of linear-chain carboxylic acids, and alcohols together with n-alkanes and PAHs measured in weathered oil samples from the 2010 DHOS. Based on these data we calculated the overall toxicity of the investigated oil samples.

To begin with, we investigated the bioaccumulation and toxicity potential of representative hydrocarbons and their oxygenated analogues that have been reported or are likely to be present in the weathered oil. These included three main groups: (i) aliphatic compounds: n-alkanes, n-alkanal, n-alkanone, n-alkanol, n-alkane-diol, n-alkanoic acids, n-alkanoic ions, n-alkandioc acids, n-alkandioc ions, n-alkane-dions, n-alkyl-ester, n-alkyl-di-ester, n-alkyl-ether; (ii) polycyclic aromatic compounds: PAHs and OxPAHs; and (iii) naphthenic acids, naphthenic acid ions, and naphthenic hydrocarbons. We collected the experimental Abraham solute descriptors of chemicals from UFZ LSER Database (Endo et al., 2015). Due to lack of experimental data, we estimated Abraham solute descriptors of naphthenic compounds using the UFZ online tool (Endo et al., 2015). Solute descriptors for acid ions were calculated using approach described elsewhere (Bittermann et al., 2016). We calculated aqueous solubility (S_wL), BCF, and LC₅₀ using solute descriptors as input in the model Abraham solvation equations for these systems.

The toxicity potential of a chemical were defined as toxicity of a compound at its solubility limit. In other words, it is the maximum toxicity a chemical can exert and is described as maximum toxic unit, TU_max, (Di Toro et al., 2007):

Toxicity of complex mixture comprising n components with mole fraction x was calculated as:

To evaluate the contribution of OxHCs in real-world oil spills we analyzed three samples: First, a surrogate of Macondo Well Oil (SMWO), which was provided by BP and has a very similar chemical composition as the spilled oil during the DHOS. Second, a slick oil (“Juniper Oil”) that was collected on the 90^th day of oil spill. Third, oil extracted from oil/sand aggregates (“sand patties”) collected on August 2016 on Alabama beaches (SP2016). Chemical fingerprinting shows that Juniper as well as SP2016 samples are from the DHOS. These three samples represent up to six years of weathering. We quantified BTEX (benzene, toluene, ethylbenzene and xylenes), PAHs, n-alkanes, nalkanols and n-alkanoic acids for these samples using gas chromatography/mass spectrometry methods (Aeppli et al., 2012). The average molar masses of SMWO, Juniper and SP2016 and weathered oil samples--which are required to calculate the mole fraction—were estimated from simulated distillation gas chromatography data of the SMWO and n-alkane concentrations from Juniper oil and SP2016.

The addition of oxygen in the normal-chain hydrocarbon decreased the bioaccumulation potential of the oxygenated compound with respect to its corresponding n-alkane (Fig 1a). This decrease in BCF was dependent on the type and number of oxygenated functional groups and decreased in the order as shown in Fig 1a. Interestingly, BCF values decrease by more than one order of magnitude for chemical families containing one oxygen atom. For chemical families containing two oxygen atoms, this decrease in BCF was even more pronounced (2.74–3.48 log units). The highest decrease was observed in ionized n-alkanoic and n-alkandoic acids (≥ 4.37 log unit) compared to corresponding n-alkanes. The decrease in BCF with the addition of oxygen on the carbon-chain can be attributable to the increase in the polarity due to the formation of polarized carbon to oxygen bond. This results in increase aqueous solubility and thus reduced affinity to the lipid, which causes the reduction in the BCF.

Increasing the carbon number within a homologous series increases the bioaccumulation potential at a different rate for each chemical family (Fig 1b). Such an increase is expected since elongating the carbon-chain length increases the size of the molecule, which increases the solvation energy of cavity formation in water phase relative to that in the lipid phase of the organism. N-alkanes with 3–8 carbons and ≥9 are categorized as “bioaccumulative” and “very bioaccumulative”, respectively, based on their estimated BCF values (Fig 1b and Table 1). For the oxygenated analogs of nalkanes, the decrease in BCF attributed to the addition of oxygen can be counterbalanced by the increase in the carbon chain length of molecules to keep them in the category of B or vB (Fig 1b and Table 1).

Oxygenation affected the BCF of PAHs differently than aliphatic chemicals (Fig 1c). Overall, the bioaccumulation potential of PAHs still decreased with increase in to oxygenation. However, the decrease in BCF is less pronounced than that observed for the aliphatic families.

For example, oxygenation of anthracene with four oxygen atoms resulted in a decrease of 2.7 log unit, compared to a 3.5 log unit decrease in BCF for an aliphatic compound with the same degree of oxygenation and carbon number (Fig 1b–c). This can be attributed to the strong dispersion interactions offered by the multiple rings of PAH molecule compared with the aliphatic chemicals. For example, hexadecane-air partition coefficient, an Abraham solute descriptor for dispersion, is 10.7 for anthracene and 8.7 for octadecane (Endo et al., 2015). Moreover, there was no systematic trend observed in the decrease of BCF for aromatic compounds. The position and type of the oxygenated functional group on the PAH ring has a strong effect on the electron donating capacity, attributing to the decrease in BCF (Fig 1d).

Many oxygenated PAHs still have BCF ≥ 3,000 and thus classified as bioaccumulative according to the REACH regulation (ECHA, 2007). The high molecular weight oxPAHs can be categorized as vB because their BCF ≥ 5,000. Fig 1e shows that the BCF value of anthracene and its different oxygenated analog is dependent on the position and type of oxygen group on anthracene backbone.

For naphthenic structural analogs, BCF values decreased upon oxygenation in a similar way as for aliphatics and aromatics (Fig 1e). Interestingly, almost all naphthenic chemicals considered in this study can be classified as “bioaccumulative” because their BCF ≥ 2,000. High molecular naphthenic acids with C_n ≥ 14 have BCF ≥ 5,000 and can be classified as “very bioaccumulative”. For those hydrocarbons with the same number of carbon and oxygen atoms, BCF values of naphthenic hydrocarbon and acids were less than that of linear-chain n-alkane and n-alkanoic acid (Fig 1b and 1e). This can be attributed to the smaller size of naphthenic analogue than that for the linear chain analog with the same carbon and oxygen atoms. For example, the McGowan volume—a Abraham solute descriptor for the size of molecule—for Bicyclo[3.3.1]nonane-1-carboxylic acid (C₁₀H₁₆O₂) is 1.4 and for n-decanoic acid (C₁₀H₂₀O₂) is 1.6. Hence, because of their compact geometries naphthenic hydrocarbons and acids have a lower cost of cavity formation in water than their aliphatic counterparts, which results in smaller BCF value for the naphthenics. Fig 1f shows the variation in BCF for naphthenics containing C₉ with different structural arrangements and position of carboxylic group. It indicates the knowledge of the exact molecular structure is critical in assessing the bioaccumulation potential of naphthenics.

In contrast to BCFs, the TU_max decreases with increasing the carbon number (Fig 2a). Interestingly, the degree of oxygenation has no systematic effect, and most oxygenated compounds have a higher TU_max than the corresponding n-alkanes. This can be attributed to the fact that while the increase in carbon chain length decreases solubility and LC₅₀ values, the degree of oxygenation increases solubility and LC₅₀. Amongst the investigated aliphatic compound classes, n-alkyl-di-esters have the highest toxic potential (TU_max= 10⁴−10⁵) and n-alkanes have the lowest toxic potential (TU_max=10⁻²−10¹). N-alkanol, n-alkanal, n-alkanone and n-alkyl ether and n-alkyl ester have similar toxic potential (TU_max= 10⁻¹−10²). The toxic potential of n-alkandiols and n-alkandioic acids with C_n > 5 was surprisingly similar to that of the corresponding n-alkane. The toxic potential of n-alkanoic acid decreased from >3,500 for C_n =1 to 0.69 for C_n =24.

For carboxylic acids calculating TU_max values is complex, since disentangling contribution of ionic species from neutral fraction towards overall toxicity is challenging. While ionization is predicated to enhance the toxicity potential by several factors, the ionized species are not expected to follow non-specific mode of toxic action; a critical assumption in the calculation of toxic potential (Verhaar et al., 1996). For oxygenated hydrocarbons, we can expect a non-specific mode of toxic action for the chemicals that were classified above as bioaccumulative and very bioaccumulative. This is further corroborated by findings that most of reactive hydrophobic chemicals with log K_OW > 4 were classified as baseline toxicants (Maeder et al., 2004). This implies that overall toxicity of mixture can be calculated by summing toxicities of individual chemicals, an approach that we followed in the application section. In summary, our computational results show that oxygenated aliphatic hydrocarbons are more toxic than their parent hydrocarbons.

Similar to aliphatic compounds, the toxic potential of oxygenated PAHs is higher than their non-oxygenated counterparts (Fig 2b). This is as expected because oxygenation increases the subcooled liquid solubility of OxPAHs. This effect makes them more bioavailable, even though their LC₅₀ values are higher than those of parent PAHs. However, toxicity potential of OxPAHs depends on the number and position of oxygenated substituents on the PAH backbone (Fig 2c). For example, the toxic potential of anthracene (C₁₄H₁₆) is 6.6 TU and 2,6-dihydroxyanthraquinone (C₁₄H₈O₄) is 43.4 TU (Fig 2c). However, for 1,8-dihydroxyanthraquinone (C₁₄H₈O₄) toxic potential is seven times less than that of 2,6-dihydroxyanthraquinone (Fig 2c). As stated above, the substituent position on the PAH ring has a strong effect on the overall hydrogen bond donating/accepting capability influencing the S_wl and LC₅₀ values according.

Similar to the other investigated compound classes, oxygenation of naphthenic hydrocarbons increased toxic potential of molecules by approximately two orders of magnitude (Fig 2d). In general, toxic potential decreased with increasing number of carbon atoms in the naphthenic molecule. Even though naphthenic hydrocarbons are intrinsically more toxic than their oxygenated counterparts (lower LC₅₀ values), they become less bioavailable due to their reduced aqueous subcooled liquid solubility. However, the increment in TU_max value due to oxygen addition is variable and depends on the structure of naphthenic backbone (Fig 2e).

The whole oil sample analysis shows that concentrations of hydrocarbons decrease and those of OxHCs increase as a function of weathering (Fig 3a). As expected, BTEX were not detected in Juniper Oil (oil slick sample) and SP2016 (sandy patty sample), because highly volatile BTEX compounds generally evaporate within hours of an oil spill. No n-alkanes were detected in the sand patty sample, which can be attributed to their complete degradation and/or transformation to oxygenated species. Benzothiophenes remained relatively persistent in the weathered oil and slightly degraded in the SP2016 sample.

We used Raoult’s law to predict dissolved-phase seawater concentrations for compounds detected in the whole oil analysis (Fig 3b). Note that the relative abundance of different chemical classes in water phase is different than that in the whole oil phase. This is because of differences in the partitioning and solubilities of the polar and nonpolar chemical classes.

We then calculated the toxicity of these dissolved phase mixtures using equation 2 in the material and method section. As expected, the toxicity is dominated by BTEX and PAHs for the crude oil, with only minor contributions from OxHCs (Fig 3a–b). Furthermore, weathering decreases the overall toxicity of the oil (Fig 3c): The toxicity of the Juniper oil (TU_Juniper = 0.024) and SP2016 sample (TU = _SP2016 = 0.013) decreased by a factor of 10 and 20 compared to the crude SMWO sample (TU_{BP Surrogate} = 0.25). Interestingly, the toxicity of SP2016, which has been collected in 2016, is still half that of Juniper, collected during the spill in 2010 (Fig 3d). This implies that toxicity decreases rapidly during the first month of an oil spill, but decreases only slowly the following years. Furthermore, the relative contribution of each chemical class towards toxicities changes as a function of weathering (Fig 3d). While BTEX and PAH contribute 93% to the overall toxicity for crude oil, this value decreased to 60% for Juniper and 21% for SP2016. Carboxylic acids increased total toxicity from 0.16% in crude oil and 24% in Juniper oil to >70% for the SP2016 sample (Fig 3d).

While the approach we adopted in this study for the assessment of bioaccumulation and toxicity is very useful to screen chemicals, it is based on certain assumptions and therefore not free from limitations. For instance, the bioaccumulation and toxicity model is purely based on the partitioning and diffusion properties. It does not take into account the active uptake of chemicals and metabolism. It also assumes that equilibrium is reached between the organism and water phase, which may not be the case for very hydrophobic chemicals (log Kow > 6.5) in short experiment duration (e.g. 96 hours for acute toxicity tests).

Furthermore, the toxicity assessment approach assumes that the chemicals follow the baseline narcosis mode of toxic action and disregards the specific toxic modes of actions. Finally, while applying the Raoult’s law, we assume that the components in the crude and weathered oil samples form ideal mixture, which is often a valid assumption for non-polar chemicals but may not be the case for polar chemicals.

The results indicate that oxygenated transformation products of petroleum hydrocarbons in weathered oil are environmentally important and further investigation would help further our insight into this research area. Despite oxygenation generally decreasing the overall bioaccumulation, many moderate to high molecular weight OxHCs (C_n >5) still fall under the category of bioaccumulative and very bioaccumulative, respectively. Oxygenation tends to increase the aqueous solubilities of OxHCs making them more bioavailable, thereby increasing their toxic potential compared to their hydrocarbon counterparts. Although weathering considerably decreases the overall toxicity of the oil samples, even highly weathered samples are predicted to still exhibit toxicity that are in the same order of magnitude as slightly weathered slick oil, and this toxicity originates mainly from OxHCs. Future study can focus on investigating and overcoming the limitations of this study, specifically the effect of reactive toxicities and biotransformation.