Article Cite This: Environ. Sci. Technol. XXXX, XXX, XXX−XXX
pubs.acs.org/est
Rapid Degradation of Oil in Mesocosm Simulations of Marine Oil Snow Events Andrew S. Wozniak,*,†,¶ Priscilla M. Prem,†,□ Wassim Obeid,†,△ Derek C. Waggoner,† Antonietta Quigg,‡,⊥ Chen Xu,§ Peter H. Santschi,‡,§ Kathleen A. Schwehr,§ and Patrick G. Hatcher† †
Department of Chemistry and Biochemistry, Old Dominion University, Norfolk, Virginia 23529, United States Department of Marine Biology and §Department of Marine Science, Texas A&M University Galveston Campus, Galveston, Texas 77553, United States ⊥ Department of Oceanography, Texas A&M University, College Station, Texas 77843, United States Environ. Sci. Technol. Downloaded from pubs.acs.org by EAST CAROLINA UNIV on 03/13/19. For personal use only.
‡
S Supporting Information *
ABSTRACT: Following the Deepwater Horizon oil spill in the Gulf of Mexico, natural marine snow interacted with oil and dispersants forming marine oil snow (MOS) that sank from the water column to sediments. Mesocosm simulations demonstrate that Macondo surrogate oil incorporates into MOS and can be isolated, extracted, and analyzed via Fourier transform ion cyclotron resonance mass spectrometry (FTICR-MS). Up to 47% of the FTICR-MS signal from MOS extracts can be attributed to formulas also found in Macondo surrogate oil demonstrating extensive oil incorporation. Additionally, oxygenation patterns for MOS extracts provide evidence for degraded oil compounds. Formulas having similar double bond equivalents but higher oxygen content (MOS CHO: CHO2−9, DBE2−16, MOS CHON: CHO 0−7 N 1, DBE 9−18; Macondo CHO: CHO1−4, DBE 2−15, CHON: CHO0−3N1, DBE9−21) were found in MOS extracts generating isoabundance distributions similar to those of environmentally aged oil. Such shifts in molecular composition are consistent with the transformation of high DBE oil components, unobservable by FTICR-MS until oxygenation in the mesocosms. Low light conditions and the rapid proliferation of hydrocarbon-degraders observed in parallel studies suggest biological activity as the primary cause of oil degradation. MOS may thus represent an important microenvironment for oil degradation especially during its long transit below the euphotic zone to sediments.
■
INTRODUCTION The Deepwater Horizon (DwH) oil spill in the Gulf of Mexico highlighted the need to better understand how natural systems respond to the introduction of petroleum and chemical dispersants to minimize socioeconomic and environmental damage. An important finding in the wake of DwH is the discovery of marine oil snow sedimentation and flocculant accumulation (MOSSFA) as a quantitatively important fate of the oil released into the Gulf of Mexico.1,2 MOSSFA is thought to be initiated by the release of microbial extracellular polymeric substances (EPS) in response to the presence of oil. The sticky EPS aggregates the oil with minerals (e.g., clays) as well as (living and dead) plankton, bacterial, and fungal materials into marine oil snow (MOS) particles that, given enough ballast, sink through the water column to the sediments.2 Estimates suggest that 4−31% of the DwH oil may have become MOS and been deposited in Gulf of Mexico sediments.3,4 The MOSSFA process transfers the oil and any associated negative impacts from the water column to benthic environments. More fully understanding the physical, chemical, and biological factors that promote or retard MOSSFA as well as the impacts that MOSSFA has on biological communities, © XXXX American Chemical Society
the spilled oil itself, and the depositional sediments is an important next step for oil spill science. An area of particular interest is understanding the molecular composition of MOS and how MOSSFA contributes to oil degradation, including molecular-level compositional changes, during water column transit. Field studies in response to DwH and other oil spills have examined the weathering (via evaporative, biodegradative, and photooxidative processes) of oil or oil components in surface waters,5−7 saltmarsh sediments,5,8,9 sands,6,10,11 and oceanic sediments12 and on rock scrapings.6,7 Many of these studies measured suites of biomarker compounds (e.g., n-alkanes, PAHs, hopanes, steranes, isoprenoids) to assess biomarker losses and changes in diagnostic biomarker ratios with increasing environmental exposure.5,8,11 These studies assess weathering extent, but the loss of the biomarker compounds does not equate to the loss of oil mass because many of these Received: Revised: Accepted: Published: A
November 20, 2018 February 18, 2019 March 2, 2019 March 3, 2019 DOI: 10.1021/acs.est.8b06532 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
the added plankton concentrate. (2) WAF treatments contained Macondo surrogate oil mixed with seawater at a ratio of 25 mL:130 L for 18−24 h. (3) CEWAF treatments contained a Corexit−Macondo surrogate oil mixture (1:20 ratio, v:v) mixed with seawater for 18−24 h. (4) DCEWAF treatments were created by mixing CEWAF with the source seawater at a ∼ 1:9 ratio. Replicate tanks contained 77 L (M1) or 80 L (M2) of treatment water plus 2 L of plankton concentrate for final volumes of 79 and 82 L for M1 and M2, respectively. M1 experiments were performed without experimental replicates while M2 experiments were performed in triplicate tanks. Fluorescence analysis (Horiba Aqualog Fluorometer) determined the estimated oil equivalents for each treatment after calibration to the Macondo oil surrogate:18 M1 and M2 control = 0 mg L−1, M1 WAF = 3.4 mg L−1, M1 CEWAF = 36 mg L−1, M1 DCEWAF = 3.6 mg L−1, M2 WAF = 0.26 mg L−1, M2 CEWAF = 41.5 mg L−1, and M2 DCEWAF = 2.74 mg L−1. The M2 DCEWAF treatment estimated oil equivalents were diluted ∼15-fold despite attempts to do a 10-fold dilution. The higher dilution relates to the difficulties controlling WAF oil concentrations in large mesocosm experiments as discussed by Wade et al.18 Experiments ran for 96 h at 19 °C with 12 h light:dark cycles using Gro-lux fluorescent lamps (∼50−100 μmol photons m−2 s−1; ∼864−1728 μmol cm−2 over the duration of the 96 h experiment). Photosynthetically active radiation (PAR) stimulated algal growth. The borosilicate glass tanks do not transmit UV light efficiently; the UVA and UVB light intensity from the lamps was measured at 0.12 and 0 W m−2, respectively, using a radiometer (Solarmeter). The rate of UVA insolation received in our experiments is just 1−2% of the intensity of UVA in daytime natural sunlight. Aggregates formed in all tanks within 24 h and sank to the bottom of each tank as MOS (or simply marine snow in control treatments). For M1, MOS aggregates were collected from the bottom of the tanks at the end of the experiment (96 h). In M2, aggregates that sank during the first 48 h were collected and used for analyses presented elsewhere;16 marine snow and MOS particles that sank during hours 48−96 were collected and combined and are discussed here. Triplicate M2 aggregates were combined to obtain sufficient material for nondestructive analyses in a parallel study,17 and one composite sample for each treatment was available for analysis in this study. Marine snow and MOS particles were subsequently freezedried prior to analysis. The freeze-dried samples were extracted using DCM because it is a solvent frequently used for the analysis of oil composition19 and was shown to effectively isolate the petroleum portion of MOS from the majority of the biological material.17 An aliquot of Macondo surrogate oil was extracted using DCM. The M1 control DCM extract was lost prior to FTICR-MS analysis during method development. As a result, all treatments are compared to the M2 control sample. FTICR-MS Analysis. Marine snow and MOS DCM extracts were diluted with methanol (1:1) and analyzed by negative ion mode electrospray ionization (ESI) FTICR-MS using an Apollo II ESI source coupled to a Bruker Daltonics 12 T Apex Qe FTICR-MS housed at the Old Dominion University COSMIC facility. Samples were infused into the instrument by a syringe pump at 120 μL h−1. Ions accumulated in the hexapole for 3.0 s before being transferred to the ICR cell. Analysis of 300 transients were coadded. The summed free induction decay was zero-filled and sine-bell-apodized prior to
compounds have simply been transformed to new compounds, many of which are not amenable to traditional GC-MS oil characterization techniques.6,11,13 As a result, weathering patterns described by biomarker analyses are an incomplete description of oil transformation/fate. Many of the newly formed compounds are oxygenated hydrocarbons that may actually be more toxic than biomarker precursor compounds.14 Research in the wake of the DwH spill has applied new, powerful analytical techniques including Fourier transform ion cyclotron resonance mass spectrometry (FTICR-MS) to studies of oil degradation in sands and saltmarsh sediments.9,10 These FTICR-MS studies have observed the progressive oxygenation of oil-derived compounds over the course of environmental exposures on the order of months to years thereby elucidating the role of photodegradation for degrading oils exposed to sunlight. Much less is known about how quickly oil-derived compounds may be transformed to oxygenated compounds within MOS. Because microbial communities are components of marine snow and MOS aggregates, there is reason to believe that MOS particles may be hotspots for oil biodegradation.15 MOSSFA events and their impacts on biology and oil chemical composition are difficult to study in the field due to the expense and logistics involved in setting up a field campaign in response to an oil spill. Furthermore, replicating spills in the field is logistically difficult and subjects the ecosystem to unnecessary harm. Mesocosm simulations of MOSSFA events have proven an effective way to examine the processes governing such events in a controlled environment.2,16−18 A major success of these simulations has been the collection of sufficient material to perform suites of chemical and biological analyses for better understanding oil incorporation into MOS. Xu et al.16 demonstrated that oil is incorporated into MOS particles and that relative to water accommodated fraction of oil (WAF) treatments, chemically dispersed (with Corexit) WAF treatments show much higher oil content in MOS. Hatcher et al.17 additionally showed that the 13C NMR chemical characteristics of MOS and marine snow were molecularly different and that the oil component can be extracted with dichloromethane (DCM). Importantly, the extraction of the oil component from MOS enables its characterization by other techniques including electrospray ionization FTICR-MS. DCM extracts of marine snow and MOS sampled from mesocosm simulations of control, WAF, and two chemically enhanced WAF (CEWAF and dilute CEWAF or DCEWAF) treatments were subjected to FTICRMS. The resulting formula assignments are compared to those for a DCM extract of the Macondo oil surrogate, and the results are discussed with respect to their implications for oil degradation during MOSSFA events.
■
METHODS Mesocosm Experiments. Mesocosm experiments simulating a MOSSFA event were conducted in 100 L borosilicate glass tanks in August (mesocosm 1, M1) and October of 2015 (mesocosm 2, M2). Details of these experiments are presented elsewhere.16−18 Seawater was collected from the Gulf of Mexico (near Galveston, TX) and prefiltered to remove large particulates and then transported to Texas A&M Galveston. A plankton concentrate (670 molecular formulas in common with the M2 control which is expected since each treatment began with the same source water and phytoplankton concentrate and thus
fast Fourier transformation and magnitude calculations by the Bruker Daltonics Data Analysis software. Mass spectra were externally calibrated with a polyethylene glycol standard and then internally calibrated using naturally occurring fatty acids and homologous compound series present within each sample.20 After removal of peaks observed in blank spectra, mass-to-charge data for peaks with signal-to-noise ratios greater than 3 were used to assign molecular formulas to peaks in each sample using an in-house Matlab (The MathWorks Inc., Natick, MA) program employing formula assignment rules established and routinely employed by our group.21−23 Only assigned molecular formulas showing exact masses within an error of 0.5 ppm of observed masses were retained in the molecular formula list. Because petroleum does not contain many P-containing or CHONS formulas and because P-containing and CHONS formulas made up a minor fraction of the formula lists in these samples, only CHO, CHOS (including CHS formulas with zero oxygens in the elemental formula), and CHON (including CHN formulas with zero oxygens in the elemental formula) formulas are considered in this manuscript. All percentages reported in the results and discussion reflect percentages of the assigned CHO, CHOS, CHON formulas, i.e., % CHO = 100% × (CHO/ (CHO + CHON + CHOS)). For further methodological details related to experimental setup, sample collection, sample extraction, and sample analysis, readers are referred to the Supporting Information (Supplementary Text S1 and Supplementary Table S1).
■
RESULTS Elemental Formula Groups. Analyses of Macondo oil and M1 and M2 DCM extracts by ESI FTICR-MS yielded thousands of molecular formula assignments for each sample. CHO formulas were the most prevalent formula group accounting for 57% (M2 control) to 81% (M2 WAF) of assigned spectral signal (Supporting Information Table 1). CHON formulas (including CHN formulas with zero oxygen atoms in the formula) were the second most abundant formula C
DOI: 10.1021/acs.est.8b06532 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
Figure 1. van Krevelen diagrams of molecular formulas assigned to peaks identified in negative mode ESI FTICR MS of dichloromethane extracts of (a) M2 control marine snow particles, (b) M2 WAF marine oil snow particles, (c) M2 DCEWAF marine oil snow particles, and (d) the Macondo oil surrogate (MC 252). Data points are color coded for elemental constituents. Formulas assigned to M2 WAF and M2 DCEWAF samples that were also found in the M2 control and/or Macondo oil surrogate are noted accordingly.
treatments. These unique formulas may represent natural and/ or oil compounds not detected in the oil or control treatments due to analytical variability, natural compounds not found in the control treatment due to differing responses of the biological communities in the various treatments, or new compounds produced from oil transformations during the experiments. The large number of new formulas (340−1367 formulas, >21% of formulas in each treatment) and the high reproducibility of the FTICR MS technique26 suggest that analytical variability can explain only a small portion of the unique formulas. Control, WAF, DCEWAF, and CEWAF treatments for M1 and M2 produce conditions resulting in differing bacterial, phytoplankton, and fungal communities.15,27 Hence, these different microbial and fungal communities might produce MOS with molecular characteristics that also differ providing variability between natural components in the M2 control and various treatments. Previous work has also shown that oil degrades due to photochemical and biological processes resulting in new molecular compositions.6,9,13 Analysis of the characteristics and relative abundance of these common and unique formulas sheds light on their sources and processing. Elemental Formula Group Characteristics. The analytical window for the ESI FTICR-MS analyses performed in this study is such that only formulas representing compounds that are both DCM extractable and ionizable in negative ESI can be detected. As a result, the analyses exclude pure
produced marine snow with similar molecular characteristics. MOS produced from the WAF treatments in these mesocosms showed lower oil content than the DCEWAF and CEWAF treatments,16,17 and this is apparent here in the MOS data. The M1 and M2 WAF MOS had the highest percentage of total formulas in common with the M2 control marine snow (41% and 44%, respectively) and just 250 and 105 formulas in common with the Macondo oil corresponding to 15% and 6% of total assigned formulas, respectively. The CEWAF and DCEWAF treatments had 416−1216 formulas in common with the Macondo oil surrogate; the M1 CEWAF (1216) and DCEWAF (850) treatment MOS had more formulas in common with the oil than their counterparts in M2 (CEWAF: 416, DCEWAF: 826). CEWAF treatments have been demonstrated to contain biomarker profiles more indicative of parent oil components as compared to WAF,25 consistent with the higher number of common formulas in the CEWAF and DCEWAF versus WAF treatments here. The M1 MOS represent MOS that accumulated over the course of 96 h whereas the M2 MOS accumulated over hours 48−96, and the longer duration of the M1 experiment may explain the higher number of formulas common to the Macondo oil surrogate in M1. The WAF, CEWAF, and DCEWAF treatment MOS had 340−1367 formulas not found in either the M2 control treatment or the Macondo oil surrogate with the highest contributions found for the M1 CEWAF and DCEWAF D
DOI: 10.1021/acs.est.8b06532 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology hydrocarbons (molecular formulas with only carbon and hydrogen) that are characteristic of petroleum and highly oxygenated, polar compounds that are insoluble in DCM (e.g., compounds with O/C > 0.5). The assigned formulas in the Macondo oil surrogate and the M2 control as well as the unique formulas for the M2 WAF and DCEWAF MOS are color-coded for elemental formula group and presented in van Krevelen (vK) diagrams (Figure 1). Formulas in the M2 WAF and DCEWAF MOS that are common to those assigned in the Macondo oil surrogate or the M2 control marine snow are differentiated from the unique formulas. The remainder of the MOS vK diagrams can be found in Supporting Information Figure 1. The formulas characterizing the Macondo oil surrogate and the M2 control marine snow are quite different (Figure 1a, d). The majority of the formulas in the Macondo oil surrogate are found at O/C ratios 75% of CHO formulas) results from formulas with double bond equivalents (DBE) between 2 and 8 (Figure 2a). While the M2 WAF MOS shows lower relative spectral signal at CHO3 than the M2 control marine snow and a more even distribution of relative spectral signal between CHO2−9, it does not show the high relative spectral signal for CHO1 and CHO2 formulas or spectral signal at DBE between 9 and 20 that distinguishes the Macondo oil surrogate (Figure 2b, d). Thus, the M2 WAF MOS (Figures 1b, 2b) shows CHO formula distributions more similar to what is observed for the M2 control marine snow than the Macondo oil surrogate, indicating it is primarily composed of natural materials. This fact is further evidenced by the number of common formulas between M2 WAF MOS and M2 control marine snow (Table 1). Evidence from Δ14C and 13C NMR have demonstrated the M2 WAF MOS to contain little oil relative to the DCEWAF and CEWAF samples.16,17 The FTICR-MS data for the marine snow confirm that most M2 WAF unique formulas are similar to the natural marine snow. The similarity is observed both in the vK diagrams and in CHO formula and DBE isoabundance plots. CHO formula isoabundance plots and vK diagrams for the M2 DCEWAF MOS (Figures 1c, 2c) are emblematic of those observed in the M1 WAF, CEWAF, and DCEWAF and M2 CEWAF treatment MOS (Supporting Information Figures 1, 2) and show distributions that differ from the M2 control marine snow (Figures 1a, 2a) and the Macondo oil. The primary difference is due to an abundance of formulas that plot
Figure 2. Isoabundance plots of oxygen number (CHO1−14) and double bond equivalents (DBE0−21) for assigned CHO molecular formulas in (a) M2 control treatment, (b) M2 WAF treatment, (c) M2 DCEWAF treatment, and (d) the Macondo oil surrogate bottom particle DCM extracts. The total spectral signals for the formula groups and double bond equivalents are also displayed. See panel d for a key to bar colors. *Note that the relative magnitude scale in panel d is twice that for a−c.
at O/C < 0.2 (Figure 1b) and have 2−4 O atoms over a wide range of DBE (8−15; Figure 2b). The maximum spectral signal (20%) in the M2 DCEWAF MOS is observed in CHO formulas with 2 O atoms, and the relative spectral signal decreases for formula groups with increasing numbers of O atoms (CHO3 = 17%, CHO4 = 9%, CHO5 = 5%), a pattern similar to what was observed by Chen et al.9 for saltmarsh sediments impacted by Macondo well oil exposed to natural conditions. The Macondo oil surrogate CHO molecular formulas are concentrated primarily in two regions of the vK diagram (Figure 1d): (1) 1.0 < H/C < 2.0, O/C < 0.2 and (2) H/C > 1.5, 0.2 < O/C < 0.5 with the vast majority of the spectral signal (>92%) found in the former region. Similar to the M2 DCEWAF MOS, the highest spectral signal among CHO formulas was found for formulas with less than four O atoms. Notably, however, the distribution of CHO formula spectral signal is dominated by CHO2 and to a lesser extent CHO1 formulas. CHO1 formulas represent 9.1% of the total signal, CHO2 formulas represent 36%, and CHO3 and CHO4 E
DOI: 10.1021/acs.est.8b06532 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology formulas represent just 4.8% and 2.2%, respectively. CHO formulas containing DBE from 0 to 20 were detected in the Macondo oil surrogate extract with 74% of the signal having between 2 and 9 DBE. Petroleum oils are characterized by aliphatic and aromatic hydrocarbons, many of which cannot be detected using this method but also contain so-called NSO compounds containing nitrogen, sulfur, and oxygen.28 Figures 1d and 2d suggest that most of the CHO formulas in this oil have very few oxygens in their molecular formulas. Thus, taking relative spectral signal into account, the M2 DCEWAF (and M1 WAF, DCEWAF, CEWAF and M2 CEWAF) MOS CHO molecular formula distributions are influenced more by the Macondo oil surrogate than the M2 control marine snow. Still, the distribution of spectral signal from the M2 DCEWAF MOS indicates it is also different from the Macondo oil. If the molecular differences between the control and WAF, CEWAF, and DCEWAF treatments were due solely to the addition of unaltered oil components to the natural components found in the M2 control, the WAF, CEWAF, and DCEWAF treatments would show CHO molecular distributions different from those observed in Figure 2c (and Supporting Information Figure 2). In particular, because the Macondo oil has essentially no signal for CHO formulas with more than 4 O atoms, one would expect to observe patterns for CHO5−8 formulas similar to what is observed for the M2 control marine snow and M2 WAF MOS treatments which show similar relative spectral signal and a local relative magnitude maximum at CHO7 (Figure 2a). Instead, the MOS treatments show decreased relative magnitudes for CHO1 and CHO2 formulas and decreasing relative spectral signal from CHO2 to at least CHO12. Thus, some other process(es) must produce the oxygenated compounds (CHO3−6) that are uniquely more abundant in the M2 DCEWAF MOS. CHON Formulas. Analysis of CHON formulas assigned to peaks in FTICR-MS mass spectra for the marine snow, MOS, and Macondo oil surrogate samples again demonstrates important differences among the samples. The M2 control marine snow showed formulas plotting in three different areas of the vK diagram (Figure 1a): (1) H/C > 1.6, O/C < 0.25; (2) 1.4 < H/C < 1.6, 0.15 < O/C < 0.3; (3) 0.75 < H/C < 1.25, 0.15 < O/C < 0.5. The first of those regions primarily corresponds to formulas of the form CHO2−9N1 and having DBE0−9 with much of the spectral signal coming from CHO2−4N1 formulas and DBE CHO4N1 > CHO1N1. The Macondo oil surrogate extract showed formulas primarily of the form CHO0−3N1 and DBE7−21 (Figure 3e) which overlap considerably with those in the DCEWAF MOS. However, there is very little signal from CHON formulas with more than one nitrogen in the Macondo oil surrogate spectrum (data not shown). Thus, it is apparent that the CHO0−4N1, DBE>7 formulas in the M1 WAF and M1 and M2 CEWAF and DCEWAF MOS are likely oilderived. However, the relative spectral signal for the Macondo oil surrogate CHON formulas shows a different isoabundance pattern: CHN1 ≫ CHO1N1 > CHO2N1 > CHO3N1 > CHO4N1. The highest abundance for the DCEWAF MOS is again shifted to formulas with a higher number of oxygens compared to the Macondo oil surrogate. We suggest that oilderived CHON compounds are degraded during the mesocosm simulation resulting in the observed distribution for the M1 WAF and M1 and M2 CEWAF and DCEWAF MOS. The higher abundance of CHO3N1, CHO2N1, and CHO4N1 formulas relative to CHN1 and CHO1N1 may indicate that the abundant CHN1 and CHO1N1 formulas in the Macondo oil surrogate are rapidly oxygenated via hydroxylation and/or carboxylation reactions. Alternatively, the microbially mediated addition of nitro or nitrate groups to soluble but nonionizable olefinic or aromatic hydrocarbons may have imparted sufficient polarity to observe them via ESI FTICR-MS. Further work is needed to evaluate the structures of these CHON compounds and the processes that form them.
■
DISCUSSION Collectively, the data demonstrate that oil components associated with MOS have been degraded as observed by the high DBE, low O formulas with isoabundance patterns differing from those of the Macondo oil surrogate and a control treatment. Whereas the Macondo oil surrogate had particularly high relative spectral signal for CHO1−4, DBE3−15, and CHO0−3N1 formulas, the MOS samples show a shift to higher oxygenation. There are three potential mechanisms that might produce the observed oil degradation: (1) biological degradation of the oil in the water column and within the MOS particles themselves, (2) photochemical degradation of the oil in the water column and within the MOS particles themselves, and (3) the combined effects of biological and photochemical degradation. The increase in oxygenated compounds in the CEWAF and DCEWAF MOS is particularly dramatic and rapid (
