Article pubs.acs.org/est
Deepwater Horizon Oil in Gulf of Mexico Waters after 2 Years: Transformation into the Dissolved Organic Matter Pool Thomas S. Bianchi,*,† Christopher Osburn,‡ Michael R. Shields,† Shari Yvon-Lewis,§ Jordan Young,§ Laodong Guo,∥ and Zhengzhen Zhou∥ †
Department of Geological Sciences, University of Florida, Post Office Box 112120, Gainesville, Florida 32611-2120, United States Department of Marine, Earth, and Atmospheric Sciences, North Carolina State University, Raleigh, North Carolina 27695, United States § Department of Oceanography, Texas A&M University, College Station, Texas 77843, United States ∥ School of Freshwater Sciences, University of WisconsinMilwaukee, 600 East Greenfield Avenue, Milwaukee, Wisconsin 53204, United States ‡
S Supporting Information *
ABSTRACT: Recent work has shown the presence of anomalous dissolved organic matter (DOM), with high optical yields, in deep waters 15 months after the Deepwater Horizon (DWH) oil spill in the Gulf of Mexico (GOM). Here, we continue to use the fluorescence excitation−emission matrix (EEM) technique coupled with parallel factor analysis (PARAFAC) modeling, measurements of bulk organic carbon, dissolved inorganic carbon (DIC), oil indices, and other optical properties to examine the chemical evolution and transformation of oil components derived from the DWH in the water column of the GOM. Seawater samples were collected from the GOM during July 2012, 2 years after the oil spill. This study shows that, while dissolved organic carbon (DOC) values have decreased since just after the DWH spill, they remain higher at some stations than typical deep-water values for the GOM. Moreover, we continue to observe fluorescent DOM components in deep waters, similar to those of degraded oil observed in lab and field experiments, which suggest that oil-related fluorescence signatures, as part of the DOM pool, have persisted for 2 years in the deep waters. This supports the notion that some oil-derived chromophoric dissolved organic matter (CDOM) components could still be identified in deep waters after 2 years of degradation, which is further supported by the lower DIC and partial pressure of carbon dioxide (pCO2) associated with greater amounts of these oil-derived components in deep waters, assuming microbial activity on DOM in the current water masses is only the controlling factor of DIC and pCO2 concentrations.
■
INTRODUCTION
entered the particulate organic matter pool when taken up by methanotrophs during the spill. Past work has shown that the complex mixture of constituents found in natural DOM8,9 is very reactive, with many contaminants in natural waters.10−12 Dissolved organic carbon (DOC) represents the largest fraction of DOM in natural waters13 and, along with chromophoric dissolved organic matter (CDOM), represents significant and reactive components of the DOM pool.14 Certain aromatic components of CDOM as well as some oil constituents [e.g., soluble polycyclic aromatic hydrocarbons (PAHs)] absorb light and fluoresce.15 This allows for detection of these compounds in natural waters through a variety of means, including threedimensional excitation−emission matrix (EEM) fluorescence spectra, specific ultraviolet absorbance (SUVA), or spectral
The explosion of the Deepwater Horizon (DWH) MC252 released 4.4 million barrels (6.4 × 108 L ± 20%) of light south Louisiana crude oil, from April to July 2010, into the northern Gulf of Mexico (GOM). This unprecedented event occurred from the seafloor at 1500 m depth and was accompanied by the release of ca. 7 million L of dispersants into the water column.1−5 It has been suggested that dispersants combined with oil derivatives may pose an even greater threat to marine organisms than the parent oil compounds.6 With regard to the fate of DWH oil in the GOM water column, it remains uncertain if all of the oil has been truly removed. If not, it is possible that some of the oil was transformed into derivatives that are now an integral part of the dissolved organic matter (DOM) pool, no longer recognizable as oil or oil derivatives by traditional hydrocarbon analyses. In fact, recent work has shown that 29−43% of the suspended particulate organic matter in DeSoto Canyon, GOM, may be derived from fossil methane following the DWH spill.7 This 14C-depleted carbon © 2014 American Chemical Society
Received: Revised: Accepted: Published: 9288
April 3, 2014 July 26, 2014 August 1, 2014 August 1, 2014 dx.doi.org/10.1021/es501547b | Environ. Sci. Technol. 2014, 48, 9288−9297
Environmental Science & Technology
Article
slope.16−19 Although there have been problems with humic materials interfering with the detection of oil compounds, the advent of modeling EEM spectra by parallel factorial analysis (PARAFAC) has provided a way to deconvolve spectra into an array of scores and loadings that better characterize the materials comprising fluorescent DOM.20,21 More recently, the applications of EEM−PARAFAC have been used in detecting crude and weathered oil in the GOM during and after the DWH spill.22−25 The use of dispersants during the DWH spill appears to have enhanced the aqueous solubility of crude oil, which has resulted in the shift in the EEM spectra as well as degradation of lowmolecular-weight hydrocarbons.22−25 Moreover, ratios of oil components identified by EEM−PARAFAC and attributed to DWH oil are characterized by highly degraded oil, suggesting that, while this oil has changed chemically and optically, it can be tracked in the GOM, even over a year after the DWH explosion. However, this also means that traditional methods of oil detection using fluorescence based on solvent extractions26 may not capture the signals, which have been modeled using native fluorescence of seawater.27 Our objectives here are to (1) determine if oil fluorescence components can still be detected in the water column of the GOM 2 years after the spill and (2) determine if there are any linkages with the metabolism of these oil components, as reflected indirectly by the partial pressure of carbon dioxide (pCO2).
Water samples were collected using Niskin bottles mounted on a CTD rosette system. Samples for DOC, ultraviolet−visible (UV−vis) absorbance, and fluorescence EEMs were gravityfiltered through precombusted GF/F Whatman filters (47 mm diameter and 0.7 μm nominal pore size) into precombusted amber glass vials and stored immediately at −20 °C. Hydrophobic oil components adsorbed onto the filters were therefore not analyzed; rather the signals that we detected were present in dissolved form, nominally 1500 m; 115 ± 106 μM; n = 12) was skewed by a group of three stations with very high DOC concentrations (284 ± 64 μM; n = 3) (see Figures S1 and S2 of the Supporting Information). The DOC concentrations at these three stations were lower than contaminated waters observed by Zhou et al.24 during the oil spill in May/June 2010 (as high as 500 μM) but much higher than the baseline level in deep GOM water (ca. 40−50 μM),33 although specific sampling locations were different. The presence of these high DOC waters may reflect the oil transformation into DOC during degradation25 or a new
were autoscaled, and eigenvectors greater than 1 were used to determine the number of components. The PCA model explained >50% of the variance in the data.
■
RESULTS AND DISCUSSION
The relationship between the DOC concentration and salinity showed a variety of influences (Table 1 and Figure 2A). In the low-salinity surface waters, the DOC concentrations were high because of the Mississippi River and coastal inputs of DOC. A large range of DOC concentrations were found in the 100− 1500 and >1500 m depth intervals. These deep interval values were grouped closely to salinity of 34.96, which was the same salinity reported for the deep seawater anomaly described by Zhou and Guo,22 which consisted of low DOC waters with high optical yields.22 The highest values for pCO2 and SUVA254 were found in these deeper intervals as well (Table 1 and panels B and D of Figure 2), indicating a possible oil influence. Moreover, the high SUVA254 values at depth indicated a higher degree of aromaticity per gram of organic carbon, which could be derived from a variety of sources, including oil, seeps, and DOM in turbidity currents.22 This may suggest the persistent presence of trace oil and oil-derived CDOM components in the 9292
dx.doi.org/10.1021/es501547b | Environ. Sci. Technol. 2014, 48, 9288−9297
Environmental Science & Technology
Article
Figure 5. PCA results are presenting trends driving the variability in samples. Both inorganic carbon (pCO2) and organic carbon (DOC) metrics were used along with PARAFAC components in the analysis to look for common relationships influencing the variability in the data. Over half of the variation in these data was explained by the first two principle components: PC1 (32.6%) and PC2 (20%).
Figure 3. Four components from a PARAFAC model fit to EEMs of GOM samples taken during the July 2012 cruise. Included in the model were EEMs measured in the same region in 2010.22 Components 1 and 4 were matched to NOM from other studies (see the Supporting Information, which also includes the corresponding ex and em spectra for each component).
of our samples (see Figure S1 of the Supporting Information). A comparison of fluorescence component distributions from the PARAFAC model was made between a site in the Zhou et al.24 study and a site sampled in July 2012 that appeared to be near the wellhead (28.25° N, 88.81° W). The fluorescence patterns here also suggested the presence of oil. In general, seawater fluorescence intensities from July 2012 were very low relative to the 2010 and 2011 samples. PARAFAC modeling of the fluorescence EEM for DOM allowed us to uncover four major fluorescent components in the DOM pool and chemical evolution of oil in the water column (Table 2 and Figure 3). These components were matched to other PARAFAC models in the literature, using Tucker’s congruence coefficient, using a value of 0.95 as the threshold for an identical match between spectra (see Table S1 of the Supporting Information). Component 1 exhibited long wave em features typical of terrestrial humic-like DOM and matched components from two PARAFAC models in the OpenFluor database.37 A persistent oil-derived signature, suggesting freshwater-soluble oil from crude oil continues to exist in the deep-water column of the northern GOM, as shown by component 2, similar to a degraded oil component modeled by Zhou et al. (see Figure S4 of the Supporting Information), despite distinct differences between the two sampling periods (see Figure S5 of the Supporting Information).24 Moreover, component 3 was matched to a dominant oil component possibly from released crude oil.22−24 Component 4 resembled amino-acid-like DOM and matched amino-acid-like components from 10 separate PARAFAC models in the OpenFluor database.37 Strong correlations were observed and indeed are expected between PAHs and in situ fluorometers shortly after the DWH spill.38 However, it is important to distinguish between oilderived signals from PAHs and non-oil CDOM. First, many in situ CDOM fluorometers that measure at ex/em wavelength pairing of ca. 370/460 nm, although sensitive at μg L−1 concentrations, completely miss the ex/em pairings of the components that we modeled here. In fact, non-oil CDOM fluorescence interferes in this region. Second, the components that we modeled were attributed to PAH fluorescence yet occurred in the aqueous phase and not as solvent extracts.39
Figure 4. Oil component ratio, C2/C3, of samples taken during the July 2012 cruise compared over three depth ranges.
source of water-soluble oil in the bottom waters of the GOM. After these three high-concentration stations were removed, the average deep-water DOC concentration (59 ± 16 μM; n = 9) was similar to previous post-spill measurements.22 These samples also had high SUVA254 values (3.3 ± 0.9 m2/g of OC; n = 9), even higher than the “deep anomaly” reported by Zhou and Guo22 (
Deepwater Horizon Oil in Gulf of Mexico Waters after 2 Years: Transformation into the Dissolved Organic Matter Pool Thomas S. Bianchi,*,† Christopher Osburn,‡ Michael R. Shields,† Shari Yvon-Lewis,§ Jordan Young,§ Laodong Guo,∥ and Zhengzhen Zhou∥ †
Department of Geological Sciences, University of Florida, Post Office Box 112120, Gainesville, Florida 32611-2120, United States Department of Marine, Earth, and Atmospheric Sciences, North Carolina State University, Raleigh, North Carolina 27695, United States § Department of Oceanography, Texas A&M University, College Station, Texas 77843, United States ∥ School of Freshwater Sciences, University of WisconsinMilwaukee, 600 East Greenfield Avenue, Milwaukee, Wisconsin 53204, United States ‡
S Supporting Information *
ABSTRACT: Recent work has shown the presence of anomalous dissolved organic matter (DOM), with high optical yields, in deep waters 15 months after the Deepwater Horizon (DWH) oil spill in the Gulf of Mexico (GOM). Here, we continue to use the fluorescence excitation−emission matrix (EEM) technique coupled with parallel factor analysis (PARAFAC) modeling, measurements of bulk organic carbon, dissolved inorganic carbon (DIC), oil indices, and other optical properties to examine the chemical evolution and transformation of oil components derived from the DWH in the water column of the GOM. Seawater samples were collected from the GOM during July 2012, 2 years after the oil spill. This study shows that, while dissolved organic carbon (DOC) values have decreased since just after the DWH spill, they remain higher at some stations than typical deep-water values for the GOM. Moreover, we continue to observe fluorescent DOM components in deep waters, similar to those of degraded oil observed in lab and field experiments, which suggest that oil-related fluorescence signatures, as part of the DOM pool, have persisted for 2 years in the deep waters. This supports the notion that some oil-derived chromophoric dissolved organic matter (CDOM) components could still be identified in deep waters after 2 years of degradation, which is further supported by the lower DIC and partial pressure of carbon dioxide (pCO2) associated with greater amounts of these oil-derived components in deep waters, assuming microbial activity on DOM in the current water masses is only the controlling factor of DIC and pCO2 concentrations.
■
INTRODUCTION
entered the particulate organic matter pool when taken up by methanotrophs during the spill. Past work has shown that the complex mixture of constituents found in natural DOM8,9 is very reactive, with many contaminants in natural waters.10−12 Dissolved organic carbon (DOC) represents the largest fraction of DOM in natural waters13 and, along with chromophoric dissolved organic matter (CDOM), represents significant and reactive components of the DOM pool.14 Certain aromatic components of CDOM as well as some oil constituents [e.g., soluble polycyclic aromatic hydrocarbons (PAHs)] absorb light and fluoresce.15 This allows for detection of these compounds in natural waters through a variety of means, including threedimensional excitation−emission matrix (EEM) fluorescence spectra, specific ultraviolet absorbance (SUVA), or spectral
The explosion of the Deepwater Horizon (DWH) MC252 released 4.4 million barrels (6.4 × 108 L ± 20%) of light south Louisiana crude oil, from April to July 2010, into the northern Gulf of Mexico (GOM). This unprecedented event occurred from the seafloor at 1500 m depth and was accompanied by the release of ca. 7 million L of dispersants into the water column.1−5 It has been suggested that dispersants combined with oil derivatives may pose an even greater threat to marine organisms than the parent oil compounds.6 With regard to the fate of DWH oil in the GOM water column, it remains uncertain if all of the oil has been truly removed. If not, it is possible that some of the oil was transformed into derivatives that are now an integral part of the dissolved organic matter (DOM) pool, no longer recognizable as oil or oil derivatives by traditional hydrocarbon analyses. In fact, recent work has shown that 29−43% of the suspended particulate organic matter in DeSoto Canyon, GOM, may be derived from fossil methane following the DWH spill.7 This 14C-depleted carbon © 2014 American Chemical Society
Received: Revised: Accepted: Published: 9288
April 3, 2014 July 26, 2014 August 1, 2014 August 1, 2014 dx.doi.org/10.1021/es501547b | Environ. Sci. Technol. 2014, 48, 9288−9297
Environmental Science & Technology
Article
slope.16−19 Although there have been problems with humic materials interfering with the detection of oil compounds, the advent of modeling EEM spectra by parallel factorial analysis (PARAFAC) has provided a way to deconvolve spectra into an array of scores and loadings that better characterize the materials comprising fluorescent DOM.20,21 More recently, the applications of EEM−PARAFAC have been used in detecting crude and weathered oil in the GOM during and after the DWH spill.22−25 The use of dispersants during the DWH spill appears to have enhanced the aqueous solubility of crude oil, which has resulted in the shift in the EEM spectra as well as degradation of lowmolecular-weight hydrocarbons.22−25 Moreover, ratios of oil components identified by EEM−PARAFAC and attributed to DWH oil are characterized by highly degraded oil, suggesting that, while this oil has changed chemically and optically, it can be tracked in the GOM, even over a year after the DWH explosion. However, this also means that traditional methods of oil detection using fluorescence based on solvent extractions26 may not capture the signals, which have been modeled using native fluorescence of seawater.27 Our objectives here are to (1) determine if oil fluorescence components can still be detected in the water column of the GOM 2 years after the spill and (2) determine if there are any linkages with the metabolism of these oil components, as reflected indirectly by the partial pressure of carbon dioxide (pCO2).
Water samples were collected using Niskin bottles mounted on a CTD rosette system. Samples for DOC, ultraviolet−visible (UV−vis) absorbance, and fluorescence EEMs were gravityfiltered through precombusted GF/F Whatman filters (47 mm diameter and 0.7 μm nominal pore size) into precombusted amber glass vials and stored immediately at −20 °C. Hydrophobic oil components adsorbed onto the filters were therefore not analyzed; rather the signals that we detected were present in dissolved form, nominally 1500 m; 115 ± 106 μM; n = 12) was skewed by a group of three stations with very high DOC concentrations (284 ± 64 μM; n = 3) (see Figures S1 and S2 of the Supporting Information). The DOC concentrations at these three stations were lower than contaminated waters observed by Zhou et al.24 during the oil spill in May/June 2010 (as high as 500 μM) but much higher than the baseline level in deep GOM water (ca. 40−50 μM),33 although specific sampling locations were different. The presence of these high DOC waters may reflect the oil transformation into DOC during degradation25 or a new
were autoscaled, and eigenvectors greater than 1 were used to determine the number of components. The PCA model explained >50% of the variance in the data.
■
RESULTS AND DISCUSSION
The relationship between the DOC concentration and salinity showed a variety of influences (Table 1 and Figure 2A). In the low-salinity surface waters, the DOC concentrations were high because of the Mississippi River and coastal inputs of DOC. A large range of DOC concentrations were found in the 100− 1500 and >1500 m depth intervals. These deep interval values were grouped closely to salinity of 34.96, which was the same salinity reported for the deep seawater anomaly described by Zhou and Guo,22 which consisted of low DOC waters with high optical yields.22 The highest values for pCO2 and SUVA254 were found in these deeper intervals as well (Table 1 and panels B and D of Figure 2), indicating a possible oil influence. Moreover, the high SUVA254 values at depth indicated a higher degree of aromaticity per gram of organic carbon, which could be derived from a variety of sources, including oil, seeps, and DOM in turbidity currents.22 This may suggest the persistent presence of trace oil and oil-derived CDOM components in the 9292
dx.doi.org/10.1021/es501547b | Environ. Sci. Technol. 2014, 48, 9288−9297
Environmental Science & Technology
Article
Figure 5. PCA results are presenting trends driving the variability in samples. Both inorganic carbon (pCO2) and organic carbon (DOC) metrics were used along with PARAFAC components in the analysis to look for common relationships influencing the variability in the data. Over half of the variation in these data was explained by the first two principle components: PC1 (32.6%) and PC2 (20%).
Figure 3. Four components from a PARAFAC model fit to EEMs of GOM samples taken during the July 2012 cruise. Included in the model were EEMs measured in the same region in 2010.22 Components 1 and 4 were matched to NOM from other studies (see the Supporting Information, which also includes the corresponding ex and em spectra for each component).
of our samples (see Figure S1 of the Supporting Information). A comparison of fluorescence component distributions from the PARAFAC model was made between a site in the Zhou et al.24 study and a site sampled in July 2012 that appeared to be near the wellhead (28.25° N, 88.81° W). The fluorescence patterns here also suggested the presence of oil. In general, seawater fluorescence intensities from July 2012 were very low relative to the 2010 and 2011 samples. PARAFAC modeling of the fluorescence EEM for DOM allowed us to uncover four major fluorescent components in the DOM pool and chemical evolution of oil in the water column (Table 2 and Figure 3). These components were matched to other PARAFAC models in the literature, using Tucker’s congruence coefficient, using a value of 0.95 as the threshold for an identical match between spectra (see Table S1 of the Supporting Information). Component 1 exhibited long wave em features typical of terrestrial humic-like DOM and matched components from two PARAFAC models in the OpenFluor database.37 A persistent oil-derived signature, suggesting freshwater-soluble oil from crude oil continues to exist in the deep-water column of the northern GOM, as shown by component 2, similar to a degraded oil component modeled by Zhou et al. (see Figure S4 of the Supporting Information), despite distinct differences between the two sampling periods (see Figure S5 of the Supporting Information).24 Moreover, component 3 was matched to a dominant oil component possibly from released crude oil.22−24 Component 4 resembled amino-acid-like DOM and matched amino-acid-like components from 10 separate PARAFAC models in the OpenFluor database.37 Strong correlations were observed and indeed are expected between PAHs and in situ fluorometers shortly after the DWH spill.38 However, it is important to distinguish between oilderived signals from PAHs and non-oil CDOM. First, many in situ CDOM fluorometers that measure at ex/em wavelength pairing of ca. 370/460 nm, although sensitive at μg L−1 concentrations, completely miss the ex/em pairings of the components that we modeled here. In fact, non-oil CDOM fluorescence interferes in this region. Second, the components that we modeled were attributed to PAH fluorescence yet occurred in the aqueous phase and not as solvent extracts.39
Figure 4. Oil component ratio, C2/C3, of samples taken during the July 2012 cruise compared over three depth ranges.
source of water-soluble oil in the bottom waters of the GOM. After these three high-concentration stations were removed, the average deep-water DOC concentration (59 ± 16 μM; n = 9) was similar to previous post-spill measurements.22 These samples also had high SUVA254 values (3.3 ± 0.9 m2/g of OC; n = 9), even higher than the “deep anomaly” reported by Zhou and Guo22 (
