The weathering of oil after the Deepwater Horizon oil spill: insights from the chemical composition of the oil from the sea surface, salt marshes and sediments

Chromatographic grade solvents for the analyses were purchased from either Fisher Scientific or Sigma-Aldrich (USA). Chemicals for standard calibration including n-alkane standards from C₈ to C₄₀, pristine (Pr), phytane (Ph), PAH-3 (collective name of 16 EPA priority PAHs), BTEX and alkylbenzenes with the alkyl groups ranging from C₁ to C₃ were purchased from Sigma-Aldrich, except for the alkylated PAH standard (SRM 1491a) which was obtained from the National Institute of Standards and Technology (NIST). Silica gel (100–200 mesh) for purification and fractionation was purchased from Fisher Scientific. The surrogate standards hexadecane-d₃₄,fluorene-d₁₀,benzo(e)pyrene-d₁₂ and ethylbenzene-d₁₀ were purchased from Sigma-Aldrich.

2.2.1. Oil mousse.

In May 2010, sea surface oil mousses, as oil emulsified with water, were collected at stations OSS and CT on board R/V Pelican (figure 1) using an acid-cleaned bucket. The oil mousse was scooped into acid-cleaned polyethylene bottles, sealed in Ziploc bags and frozen immediately at −20 °C until analysis. Stations OSS and CT are located 130 and 85 km away from the accident site, respectively. At station CT, brownish oil mousse was observed in a patchy distribution, while at OSS, the mousse was more extensive with a darker brown color and stronger odor (figure S1 available at stacks.iop.org/ERL/7/035302/mmedia).

Zoom In Zoom Out Reset image size

The brownish oil mousse washing onto salt marshes with tropical systems (Hurricane Alex, TS Bonnie and Tropical Depression #5) was collected at Marsh Point (MP) in Davis Bayou (30.375°N, 88.790°W), Mississippi, on 21 July 2010, about two weeks after the crude oil first came onshore at this location (figure S1 available at stacks.iop.org/ERL/7/035302/mmedia). The marsh plants (Spartina alterniflora) with oil mousse were cut in situ, preserved in glass jars at 4 °C in a cooler and transported to the laboratory immediately. Oil attached heavily to Spartina alterniflora stalks was scraped carefully into a glass vial using a Teflon knife and frozen at −20 °C until analysis. The MC252 reference oil was requested from BP (through Dr Guo). All of the samples were sealed tightly and stored in a freezer (−20 °C).

The mass of total solvent-extractable materials (TSEM) was obtained from an aliquot of the oil mousse. Briefly, about 0.5 g of the oil mousse was weighed and extracted with dichloromethane (DCM), and the extract was evaporated gently with a nitrogen stream. The residue was defined as TSEM (Wang et al 2004). The petroleum hydrocarbons reported in the oil mousse samples were normalized to the TSEM values (as mg g⁻¹ TSEM). The TSEM values were 652, 683 and 750 mg g⁻¹ for OSS, CT and MP mousse, respectively. The water contents of the OSS and CT mousse were quantified gravimetrically. Briefly, about 1 g of oil mousse was dissolved in 20 ml DCM, filtered through a column of anhydrous sodium sulfate, and the DCM was evaporated in the oven at 50 °C overnight. The water contents of the oil mousse at stations OSS and CT were 39% and 41%, respectively. The water contents may have been overestimated considering that volatile hydrocarbons can be lost in the oven.

2.2.2. Oil in sediment.

2.3.1. Sample preparation for alkane and PAH analysis.

2.3.2. Sample preparation for BTEX and C₃-benzene analysis.

2.3.3. Sample preparation for trace metal analysis.

Analysis for n-alkanes (C₈–C₄₀), Pr and Ph was performed on a GC-FID (Shimadzu GC 2014), with a JW scientific DB5 column (30 m × 0.25 mm,0.25 μm film thickness). The injection volume was 1 μl with a split ratio of 20. The temperature of the column was ramped from 40 to 280 °C at a rate of 8 °C min⁻¹, and held at 280 °C for 40 min. Quantification was based on the internal standard hexadecane-d₃₄ and external standards of n-alkanes. Analysis of triplicate samples from the OSS mousse agreed within ±9%.

Sixteen PAHs, 18 alkylated PAH compounds including six alkylated homologues, and BTEX and C₃-benzenes were analyzed with GC-MS (Shimadzu QP2010 plus), with a Restek DB5 column (20 m × 0.18 mm,0.18 m film thickness). The injection volume was 1 μl. The temperature of the column was ramped from 40 to 280 °C at a rate of 8 °C min⁻¹ and held at 280 °C for 40 min. The samples were analyzed by GC-MS with a split ratio of 20 in SIM. Table S1 (available at stacks.iop.org/ERL/7/035302/mmedia) lists the selected characteristic ions used for analysis of target PAHs, alkylated PAH homologues, BTEX and C₃-benzenes.

Trace metal analysis was carried on an ICP-MS (Element 2 Finnigan MAT, Thermo) equipped with a PTFE spray chamber as the sample introduction system. Instrument tuning and mass calibration were performed prior to the analysis. The trace metals Mg, Al, V, Cr, Mn, Fe, Ni, Co, Cu, Zn, As and Pb were measured with a standard calibration method with medium resolution.

Relative response factors for each compound were calculated relative to internal standards. Samples were quantified with deuterated hexadecane-d₃₄, fluorene-d₁₀, benzo(e)pyrene-d₁₂ and ethylbenzene-d₁₀. An instrument blank (GC-FID and GC-MS) and standard solutions, which included authentic n-alkanes with Pr and Ph, or target PAHs, or BTEX and C₃-benzenes, were analyzed before each sample batch to monitor accuracy and precision. The recoveries of hexadecane-d₃₄, fluorene-d₁₀ and benzo(e)pyrene-d₁₂, for alkane and PAH analysis, were 77.1 ± 11.0%, 91.3 ± 8.9% and 93.2 ± 12.1% respectively.

The GC-FID chromatograms of three oil mousse samples (F1 fraction) showed similar distribution patterns for resolved peaks and unresolved complex mixture (UCM) (figure S2 available at stacks.iop.org/ERL/7/035302/mmedia), suggesting that they originated from the same MC252 oil source. The three oil mousse samples had resolved n-alkanes in the range of C₁₄–C₃₈, while the MC252 oil had a typical wide range (from C₉ to C₃₈) of crude oil distribution (figure S2 available at stacks.iop.org/ERL/7/035302/mmedia). Note that n-alkanes less than C₉ were not resolved in our GC-FID protocol, even though these LMW volatile compounds represented a major fraction of the crude oil (Ryerson et al 2011).

The concentrations of individual n-alkanes in OSS mousse ranged from 0.0005 to 2.8 mg g⁻¹ TSEM, with a total 29.3 mg g⁻¹ TSEM; in CT mousse 0.0006–0.58 mg g⁻¹ TSEM, with a total 6.9 mg g⁻¹ TSEM; and in MP mousse 0.001–1.3 mg g⁻¹ TSEM, with a total 14.3 mg g⁻¹ TSEM. The n-alkanes C₁₆–C₃₂ represented about 90% of the total alkanes by mass for the three oil mousse samples. In contrast, the total concentration of individual n-alkanes in MC252 oil was 68.09 mg ml⁻¹, and the n-alkanes C₁₆–C₃₂ constituted 45% of the total alkanes, without considering n-alkanes less than C₉. The loss of low molecular weight n-alkanes in the oil mousse was apparent relative to the crude oil. Concentrations of n-alkanes from C₉ to C₁₅ represented only 0.3%–1.6% of the oil mousse, as compared to 53% in MC252 oil (figure 2). Pristane (Pr) and phytane (Ph) are common isoprenoids in crude oils. The Pr/Ph ratios were 0.94–1.09 for the oil mousse, similar to that of MC252 oil (0.94) (table 1). The carbon preference index (CPI) represents the ratio of relative abundance of odd-numbered n-alkanes to that of even-numbered n-alkanes (Ehrhardt and Petrick 1993). As expected, the CPI is 1 for all of the oil mousse samples, which corresponds to the normal criteria for petroleum hydrocarbons (Requejo and Boehm 1985, Kennicutt et al 1991).

Zoom In Zoom Out Reset image size

The individual n-alkane levels in SC sediment were low, ranging from 2.35 × 10⁻⁵ to 2.94 × 10⁻⁴ mg g⁻¹ dry weight, with a total of 3.63 × 10⁻³ mg g⁻¹. In contrast, concentrations of n-alkanes in SG sediment were much higher, ranging from 0.003 to 0.22 mg g⁻¹ per component, yielding a total of 2.56 mg g⁻¹ (figure 2). This pattern was expected, as station SC (6 km) is further away from the wellhead than station SG (2 km). Consistently, a clear UCM hump was observed in the GC chromatogram of the SG sediment, but not in the SC sediment (figure S3 available at stacks.iop.org/ERL/7/035302/mmedia). However, both of the chromatograms had resolved n-alkanes from C₁₁ to C₃₇. The ratios of n-C₁₇/pristine and n-C₁₈/phytane were 1.31 and 2.44 for SC sediment, and 1.38 and 2.52 for SG sediment, respectively. The CPIs were also different, with 1.0 and 1.2 for the SG and SC sediments, respectively (table 1).

The total concentration of 16 PAHs was 0.23 mg g⁻¹ TSEM in the OSS mouse, 8 times higher than that in the CT mousse (0.03 mg g⁻¹) (figure 3). The MP mousse had the lowest concentration, only 0.01 mg g⁻¹ TSEM. All these concentrations in oil mousse were much lower than the 1.15 mg g⁻¹ (assuming a density of 0.84 g ml⁻¹ for the crude) in the MC252 crude. Naphthalene was the dominant PAH in the crude oil, accounting for 64% of the total PAHs, followed by phenanthrene and fluorene, while chrysene and other PAHs occurred as minor components. In contrast, the naphthalene content in oil mousse decreased greatly, ranging from 3% to 9%. Chrysene represented only 2% of the total PAHs in the crude oil, while it became the dominant PAH in oil mousse, accounting for 38% for the OSS, and 53%–56% for the CT and MP. Phenanthrene is a major PAH in the crude oil (17%), and its percentage increased in the OSS mousse (39%), driven mainly by the loss of naphthalene. The percentage of phenanthrene decreased greatly from OSS (39%), to CT (16%), to MP mousse (4%). Other PAHs were minor and had similar distributions in the oil mousse.

Zoom In Zoom Out Reset image size

The concentrations of alkylated PAH homologous series in the crude oil and oil mousse were higher than their corresponding PAHs (figure 4), in consistency with other studies (Wang et al 1994, 2004). For example, the concentrations of 1- and 2-methylnaphthalene in the crude oil were 2–3 times higher than that of naphthalene, as were the phenanthrene and chrysene series. The total concentration of alkylated PAHs in the oil mousse ranged from 2.6 to 9.4 mg g⁻¹ TSEM, with OSS the highest. This concentration level is similar to that of the crude oil (7.2 mg g⁻¹), suggesting that these components are relatively stable. The distribution of alkylated PAHs in the mousse resembled their parent PAH patterns. Alkylated naphthalenes in the oil mousse became minor components relative to other alkylated PAHs, as did naphthalenes relative to the PAHs. Alkylated phenanthrenes were dominant components in the OSS mousse, but their concentrations decreased gradually from CT to MP mousse. Three-methylchrysene became the dominant component of the alkylated PAHs in the three mousses.

Zoom In Zoom Out Reset image size

Concentrations of total PAHs were 3 times higher in SG sediment (1 μg g⁻¹) than in SC sediment (0.3 μg g⁻¹), consistent with their respective distances to the wellhead. The compositions of the PAHs in the two sediments were similar, becoming enriched with PAHs containing a higher number of rings, such as phenanthrene and chrysene. Consistent with the PAHs, alkylated PAH homologues were much higher in SG sediment (1.7 mg g⁻¹) than in SC sediment (0.03 mg g⁻¹). The concentrations of alkylated PAHs were much higher than their corresponding PAHs in both sediments. For example, concentrations of alkylated phenanthrenes in SG sediment ranged from 0.1 to 0.2 mg g⁻¹, whereas that of phenanthrene was about 0.1 μg g⁻¹. In contrast, concentrations of alkylated phenanthrenes (0.1–0.2 mg ml⁻¹) resembled phenanthrene (0.16 mg ml⁻¹) in the crude oil.

BTEX and C₃-benzenes are important parameters in determining the weathering degrees of oil due to their high volatility and solubility (Wang and Fingas 1995). As expected, the concentrations of BTEX and C₃-benzenes in OSS and CT oil mousses were 105 and 75 μg g⁻¹ TSEM for C₃-benzenes, 449 and 359 μg g⁻¹ TSEM for BTEX, respectively, about one order of magnitude lower than those of the crude oil (figure 5 and table 2). The BTEX and C₃-benzenes were not measured in the MP mousse. For the SG and SC sediments, the concentrations of BTEX and C₃-benzenes were 0.6, 1.8 and 0.2, 0.6 μg g⁻¹ wet weight, respectively (figure 5).

Zoom In Zoom Out Reset image size

The concentration of Mg was 6.4 μg g⁻¹ in the crude oil, but was elevated to 2170 and 2899 μg g⁻¹ TSEM in OSS and CT mousse, respectively (figure 6 and table 3). Concentrations of terrestrial metals such as Al, Mn and Fe in CT and MP oil mousses were significantly higher than those in the crude oil and OSS mousse. The levels of Al were 14 μg g⁻¹ in the crude and 72 μg g⁻¹ TSEM in the OSS mousse, but increased to 432 and 738 μg g⁻¹ TSEM in the CT and MP mousse, respectively. Similarly, Fe concentrations were 8 μg g⁻¹ in the crude oil, 50 μg g⁻¹ TSEM in the OSS mousse, and increased to 603 and 955 μg g⁻¹ TSEM in the CT and MP mousse; the Mn concentration followed the same trend, from 0 (non-detectable) in crude and 4 μg g⁻¹ TSEM in the OSS mousse to 52 and 39 μg g⁻¹ TSEM in the CT and MP mousse, respectively. The concentrations of the other transition metals V, Ni, Co, Cu, As and Pb in the oil mousse followed similar distribution patterns. From the crude, OSS, CT to MP mousse, V concentrations were 0.2 μg g⁻¹, 1.0, 1.5 and 1.7 μg g⁻¹ TSEM; Ni concentrations were 1.5 μg g⁻¹, 4.2, 7.7 and 7.3 μg g⁻¹ TSEM; Co concentrations were 0.2 μg g⁻¹, 0.2, 0.4 and 0.7 μg g⁻¹ TSEM; Cu concentrations were 0.5 μg g⁻¹, 1.7, 4.0 and 3.3 μg g⁻¹ TSEM; Pb concentrations were 0.3 μg g⁻¹, non detectable, 0.6 and 1.5 μg g⁻¹ TSEM; and Zn concentrations were 18 μg g⁻¹, 18.7, 51.8 and 13.3 μg g⁻¹ TSEM, respectively.

Zoom In Zoom Out Reset image size

Chemical analysis of the mousse revealed that the surface oil was subjected to moderate evaporation after rising to the sea surface. Total n-alkane concentrations in the oil mousse (7–30 mg g⁻¹ TSEM) were lower than those found typically in crude oils including MC252 (80 mg g⁻¹), but consistent with oil collected in environments after oil spills in aquatic environments (Wang et al 2004). Low-molecular-weight (LMW) n-alkanes ( < C₁₄) were lost in the oil mousse relative to the crude oil (figure 2), indicating that evaporation was the main weathering processes on the surface oil. These LMW hydrocarbons can be evaporated within 2–3 h after rising to the sea surface (Stiver and Mackay 1984, Fingas 1999, Ryerson et al 2011), while the time scale of evaporation for higher carbon numbered hydrocarbons (C₁₄–C₁₆) is estimated at 10–100 h (de Gouw et al 2011). Similarly, two-ring aromatics, mainly naphthalene and alkylated naphthalene, were lost preferentially ( > 99%) in the oil mousse relative to the crude oil (figures 3 and 4). Acenaphthylene and fluorene were also significantly lost in the oil mousse. Even though the loss of LWM aromatics may indicate evaporation, dissolution during the rise of oil to the sea surface may have also contributed to the loss, considering that the oil rose through a long water column (1500 m) and these LMW aromatics are more soluble than alkanes. For example, the solubility of naphthalene in water is approximately 30 mg l⁻¹. However, the proportional naphthalene content relative to other volatiles detected in the atmosphere above the DWH site (Ryerson et al 2011, 2012) suggested that evaporation on the sea surface might be the main weathering process to these LWM aromatics.

Our chemical analyses suggested that among the three oil mousses, the MP mousse was the most weathered, followed by the CT and OSS mousses. The percentages of LWM alkanes ( < C₁₉) in total alkanes were 25% in OSS, 13% in CT and 7% in MP, indicating an increased weathering degree driven mostly by evaporation (Wang et al 1995). The 16 PAHs measured were dominated by chrysene, accounting for 48%–53% of the total, indicating that the four-ring chrysene is exceptionally resistant to weathering (Wang et al 1994). Percentages of the three-ring phenanthrene, however, showed a clear decrease from OSS (39%), CT (16%), to MP (4%). The ratios of phenanthrene to chrysene were 1.0, 0.3 and 0.1 in OSS, CT and MP mousse, respectively, consistent with the order of weathering degrees derived from n-alkanes. A decreasing ratio of phenanthrene to chrysene indicates weathering processes of oil degradation in aquatic environments (Pastor et al 2001), including biological degradation (Zhou et al 2012). The alkylated phenanthrenes and chrysenes also supported this pattern, as all of the alkylated phenanthrenes decreased from OSS to CT and MP relative to 3-methylchrysene, which was the most resistant of all of the measured hydrocarbons.

Phenanthrene and its alkylated homologues had different time scales of weathering compared to LWM aromatics including naphthalene, acenaphthylene and fluorene. Based on the predicted n-alkane patterns of surface oil with time (Ryerson et al 2012), the OSS mousse was subjected to evaporation for more than 2 d after it rose to the sea surface. If this time scale is right, LMW aromatics were almost completely evaporated within three days as expected, but phenanthrene and its alkylated homologues continued to weather for days to weeks when the oil mousse reached salt marshes at Marsh Point, Mississippi in July 2010. Among the three groups of compounds, PAHs were the most weathered as suggested by the increased ratios of alkanes/PAHs and alkylated PAHs/PAHs from MC252 crude, OSS, CT to MP (table 1). Alkylated PAHs seem to be the most stable, as ratios of alkanes/alkylated PAHs decreased from 11.3 in the crude oil to 5.1 in OSS, 2.7 in CT and 3.4 in MP. The rapid decrease of PAHs relative to alkylated PAHs could be due to dissolution and/or evaporation, because PAHs tend to have higher solubilities and lower boiling points than their corresponding alkylated homologues (Havenga and Rohwer 2002). Photooxidation may not have played a major role, because alkylation tends to increase photooxidation rates (Prince et al 2003).

Many microorganisms can degrade petroleum components. For instance, n-alkanes are readily biodegraded aerobically in marine environments, and in particular n-alkanes with medium chain length (10–22 carbons) can be degraded preferentially (van Beilen et al 1994, Stout and Wang 2007). However, clear signs of biological degradation were not apparent from the oil mousse. For example, the distribution patterns of n-alkanes in the oil mousse relative to crude oil reflected typical evaporation from the progressive loss of LMW n-alkanes (Stout and Wang 2007). The ratios of n-C ₁₇/Pr and n-C ₁₈/Ph can indicate biodegradation (Pritchard and Costa 1991, de Jonge et al 1997). These ratios were generally higher in the oil mousse than in the crude oil (table 1), suggesting that biodegradation was not important in oil weathering. However, it is not clear why these ratios were higher in the oil mousse than in the reference crude oil. As mentioned above, the decrease in phenanthrene/chrysene ratio from OSS, CT to MP mousse could indicate some biodegradation (Pastor et al 2001, Zhou et al 2012). However, alkylated phenanthrenes in the oil mousse decreased at similar rates, suggesting that the weathering was not controlled by biodegradation, as the biodegradation rate decreases with the number of alkyl groups on the PAH nucleus (Prince 2002). Alternatively, alkylated phenanthrenes might be as labile as phenanthrene in the Gulf waters. Further laboratory research is needed to clarify this observation.

The weathering pattern of OSS < CT < MP is consistent with the surface water trajectory model (Liu et al 2011), showing that the surface oil first moved westward (OSS), then northeastward (CT), and finally reached salt marshes. This conclusion is supported further by a satellite image of the Mississippi River Delta, which shows the general path of the oil slicks on the sea surface during the time period when our oil mousse was collected (figure S4 available at stacks.iop.org/ERL/7/035302/mmedia). The oil color becomes lighter with emulsion in seawater (Jordan and Payne 1980), so the color difference between the two mousses suggests that the oil collected at station OSS (dark brown) was less weathered than that at station CT (light brown) (figure S1 available at stacks.iop.org/ERL/7/035302/mmedia). Our underlying assumption is that these oil samples had the same MC252 oil origin. This assumption is reasonable, considering that these samples were collected during the DWH oil spill, and that the sampling stations were contaminated by this oil spill (figure S4 available at stacks.iop.org/ERL/7/035302/mmedia). For example, the arrival and evolvement of the oil at MP after the DWH oil spill were well documented (Biber et al 2012, Wu et al 2012). Typical biomarkers were not applied to verify that these mousses originated from the MC252 well (Carmichael et al 2012), but their similar but gradually evolving hydrocarbon compositions (figures 2–4), such as the dominance of chrysene and 3-methylchrysene, suggest the same origin for these mousses.

The trace metals V, Cr, Fe, Ni, Cu, Zn and Pb are typical constituents of crude oil (Gohlke et al 2011, Osuji and Onojake 2004), but their levels in oil may change with weathering as oil can absorb metal ions from seawater and the oil mass, relative to metal content, may decrease with oil degradation. Our results showed that concentrations of V, Ni, Co, Cu, As and Pb increased from the crude oil to the MP mousse (table 3 and figure 6), and this pattern is consistent with the intensifying weathering of oil mousse from off shore to coastal salt marshes. Further analysis demonstrated that the concentrations of trace metals correlated inversely with the levels of short-chain alkanes (n < 19) and PAHs from the crude oil to the oil mousse along the projected transport route (figure 7). These correlations may indicate that trace metals were accumulated into the oil mousse at different rates during the loss of LMW components, as suggested by their different slopes of the regression curves. Metal ions might preferentially bind to high-molecular-weight oil components that may possess more chelating functional groups such as carboxyls and/or hydroxyls (Wells 2002). Trace metals in seawater are generally complexed with organic ligands of dissolved or particulate organic matter (Hirose 2006, 2007, Varspir and Butler 2009), which makes incorporation of metals to oil mousse easier by adsorption and/or hydrophobic interactions. Another possible explanation for the metal accumulation is that the trace metals simply became more concentrated as LMW oil components were degraded. However, if this was the case, the same slopes on the regression lines (figure 7) for the different metals would be expected.

Zoom In Zoom Out Reset image size

Of the trace metals measured, the concentrations of Al, Fe and Mn in the oil mousse increased significantly from the crude oil, OSS, CT to MP mousse (figure 6). We propose that clay minerals, presumably abundant in the sampling area due to the impact of the Mississippi River, were aggregated into the oil mousse during the transport from the accident site to the salt marshes. Even though clay minerals were not digested during our analytical procedure for trace metals, the leaching of Al, Fe and Mn from clay minerals during the digestion could occur. Concentrations of Mg were also much higher in OSS and CT mousse than that in the crude oil. This trend can be attributed to the water contained in oil mousse due to the oil emulsion, since Mg is one major element in seawater with a concentration of 52.8 mM. The water contents in the mousse were 0.4 g g⁻¹ for the OSS and CT mousses, and 0.25 g g⁻¹ for the MP mousse. The Mg in the mousse water itself can account for the total Mg in the MP mousse, but only about 30% of the Mg contents in the OSS and CT mousses, suggesting that the oil mousse did accumulate Mg, as discussed above.

Previous reports showed that among 120 sites within a 2.7 km range of the wellhead, 29% of the sediment cores exhibited evidence of oil contamination (Operational Science Advisory Team 2010). The results from two sediments (within 6 km from the wellhead) indicate that both sites were contaminated by the DWH oil spill, based on the enhanced hydrocarbon levels, presence of BTEX and C₃ benzenes, UCM in the GC chromatograms, and visual observation of the cores (figure S1 available at stacks.iop.org/ERL/7/035302/mmedia). However, the oil level decreased significantly from station SG (2 km) to SC (6 km). For example, the concentration of total n-alkanes was more than two orders of magnitude higher in SG than in SC.

The sedimentation of oil is caused mainly by physical interactions between suspended particulate matter and oil (Muschenheim and Lee 2002, Shen and Jaffe 2000). About 20%–30% of the total spill oil can reach sediments in coastal regions with high concentrations of suspended particulate matter (Payne et al 2003). Oil may persist much longer in sediment than at the sea surface (Harayama et al 1999). Two factors may contribute to the slow oil degradation in sediment. Hydrocarbons can be absorbed strongly into minerals and/or sedimentary organic matter. Also, oil on the sediment surface can inhibit the penetration of dissolved oxygen into the sediment, and labile hydrocarbon can stimulate the consumption of dissolved oxygen by bacteria. These two mechanisms can quickly result in oxygen depletion and retard further aerobic degradation of the deposited oil.

Our results showed a clear loss of LMW compounds including C₉–C₁₃ n-alkanes, naphthalene and alkylated naphthalenes in oil deposited in sediments, relative to the crude oil. Most of this weathering can be attributed to dissolution and biodegradation. The LMW compounds tend to have higher solubilities in water. For example, the solubility of naphthalene in water is 30 mg l⁻¹ and that of 1-methylnaphthalene is 26 mg l⁻¹, whereas the solubilities of phenanthrene and 1-methylphenanthrene are 1.6 and 0.26 mg l⁻¹, respectively (Shiu et al 1988). As a result, PAHs with more than three rings were enriched in sediments (figure 3), showing different distribution patterns from those of oil mousse, which was dominated by chrysene. Relative to oil mousse, the n-alkanes C₁₂–C₁₅ and alkylated naphthalenes were enriched in the sediments, further suggesting that evaporation was the main mechanism in losing C₁₂–C₁₅ in the oil mousse on the sea surface.

Biodegradation was expected to modify the chemical composition of the oil that had been deposited in sediment for one year, even though the degradation might be slow due to the low temperature (4 °C) and limitation of available dissolved oxygen (Ward et al 1980). Indeed, relatively high abundance of type I methanotrophic bacteria in SG and SC sediments and overlying waters (unpublished data) indicated the existence of methane oxidation (Hanson and Hanson 1996). In addition, some evidence of biodegradation was apparent. The n-alkanes C₁₆–C₂₇ in the SC sediment and C₁₆–C₂₃ in the SG sediment were lost preferentially (figure 2). This result is consistent with observations from other oil spill studies, where medium chain n-alkanes (∼C₁₀–C₂₂) were biodegraded most rapidly (Stout and Wang 2007). This distribution pattern is more evident in the SC sediment than in the SG sediment, suggesting that the oil in SC was more degraded. Likewise the slightly lower ratios of n-C₁₇/Pr and n-C ₁₈/Ph in SC (1.3 and 2.4) than in SG (1.4 and 2.5) lead to similar conclusions, since such ratio patterns can indicate biodegradation (Christensen and Larsen 1993, Wang et al 1994, 1999). Note that the odd-numbered n-alkanes C ₂₉,C ₃₁ and C₃₃ were enhanced in the SC sediment with a CPI of 1.2, indicating a biogenic input from terrestrial sources (Goni et al 1998). The ratio of n-alkanes/alkylated PAHs was used to compare the relative loss of n-alkanes by biodegradation in sediments because the alkylated PAHs are more resistant to microbial degradation than n-alkanes and PAHs (Fedorak and Westlake 1981, Wang and Fingas 1995). The ratio n-alkanes/alkylated PAHs was more than one order of magnitude smaller at SC (0.1) than that at SG (1.5), suggesting that biodegradation was more intensive at SC. The significantly lower amount of oil contamination may have led to rapid degradation due to substrate availability. The distribution of alkylated PAHs in SG was similar to that of crude oil, dominated by phenanthrene homologues, without considering naphthalene homologues. However, the phenanthrene homologues were more degraded at SC than SG, and, as a result, fluoranthene and pyrene homologues became the dominant ones. This pattern cannot be explained by dissolution, because it would cause lower naphthalene homologues that have much higher solubilities. Phenanthrene is often more biologically labile than chrysene (Wang and Fingas 2003, Prince and Walters 2007, Zhou et al 2012). One possible explanation is that the phenanthrene homologues were labile under the low temperature and low DO concentration in the deep sea sediment. A rapid decrease of alkylated phenanthrenes was also observed in oil mousse from OSS to MP (figure 4).

Even though clear signs of biodegradation were observed for the oil deposited in the sediment adjacent to the wellhead, oil degradation in sediments was quite slow, considering that high concentrations of labile n-alkanes were observed after one year at stations SG and SC. The presence of BTEX and C3-benzenes also suggests that the oil was subjected to a light to moderate degree of weathering ( < 25%) (Wang and Fingas 1995).