thermogenic organic matter dissolved in the abyssal ocean

10
Thermogenic organic matter dissolved in the abyssal ocean Thorsten Dittmar a, , Boris P. Koch b a Florida State University, Department of Oceanography, Tallahassee, FL 32306-4320, USA b Alfred Wegener Institute for Polar and Marine Research, Am Handelshafen 12, 27570 Bremerhaven, Germany Received 30 November 2005; received in revised form 30 March 2006; accepted 19 April 2006 Available online 30 May 2006 Abstract Formation and decay of thermogenic organic matter are important processes in the geological carbon cycle, but little is known about the fate of combustion-derived and petrogenic compounds in the ocean. We explored the molecular structure of marine dissolved organic matter (DOM) for thermogenic signatures in different water masses of the Southern Ocean. Ultrahigh-resolution mass spectrometry via the Fourier transform-ion cyclotron resonance technique (FT-ICR-MS) revealed the presence of polyaromatic hydrocarbons (PAHs) dissolved in the abyssal ocean. More than 200 different PAHs were discerned, most of them consisting of seven condensed rings with varying numbers of carboxyl, hydroxyl, and aliphatic functional groups. These unambiguously thermogenic compounds were homogenously distributed in the deep sea, but depleted at the sea surface. Based on the structural information alone, petrogenic and pyrogenic compounds cannot be distinguished. Surface depletion of the PAHs and first estimates for their turnover rate (> 1.2 · 10 12 mol C per year) point toward a primarily petrogenic source, possibly deep-sea hydrothermal vents, which is thus far speculative because the fluxes of combustion-derived and petrogenic matter to the ocean are not well constrained. We estimate that >2.4% of DOM are thermogenic compounds, and their global inventory in the oceans is >1.4·10 15 mol C, significantly impacting global biogeochemical cycles. © 2006 Elsevier B.V. All rights reserved. Keywords: Aromatic hydrocarbons; Dissolved organic matter; Petrogenic organic matter; Black carbon; Carbon cycle; Deep water; Antarctica; Weddell Sea 1. Introduction The oxygen content of the earth's atmosphere and the global cycle of organic carbon are invariably linked (Lasaga and Ohmoto, 2002). Free oxygen in the atmosphere is generated via burial of organic carbon in marine sediments. Sedimentary organic carbon then cycles through the lithosphere. Over geological time scales tectonic and volcanic processes carry petrogenic carbon back to the continents' surface where it is oxi- dized into CO 2 , consuming atmospheric oxygen. It is this delicate balance between photosynthesis, organic carbon burial, tectonic forces, volcanism and sedimentary rock weathering that ultimately controls atmospheric compo- sition (Lasaga and Ohmoto, 2002). The time petrogenic organic carbon is protected from oxidation is a crucial parameter. In periods of enhanced tectonic activity, turnover time accelerates and increased amounts of petro- genic organic carbon are oxidized which causes atmo- spheric oxygen to drop. Decreased oxygen levels, in turn, enhance organic matter preservation in sediments and reduce oxidative rock weathering, providing a negative feedback which is vital for the habitability of earth. The spatial separation between organic matter burial in the oceans and its oxidation at the continents' Marine Chemistry 102 (2006) 208 217 www.elsevier.com/locate/marchem Corresponding author. Tel.: +1 850 645 1887. E-mail address: [email protected] (T. Dittmar). 0304-4203/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.marchem.2006.04.003

Upload: thorsten-dittmar

Post on 26-Jun-2016

220 views

Category:

Documents


4 download

TRANSCRIPT

Page 1: Thermogenic organic matter dissolved in the abyssal ocean

(2006) 208–217www.elsevier.com/locate/marchem

Marine Chemistry 102

Thermogenic organic matter dissolved in the abyssal ocean

Thorsten Dittmar a,⁎, Boris P. Koch b

a Florida State University, Department of Oceanography, Tallahassee, FL 32306-4320, USAb Alfred Wegener Institute for Polar and Marine Research, Am Handelshafen 12, 27570 Bremerhaven, Germany

Received 30 November 2005; received in revised form 30 March 2006; accepted 19 April 2006Available online 30 May 2006

Abstract

Formation and decay of thermogenic organic matter are important processes in the geological carbon cycle, but little is knownabout the fate of combustion-derived and petrogenic compounds in the ocean. We explored the molecular structure of marinedissolved organic matter (DOM) for thermogenic signatures in different water masses of the Southern Ocean. Ultrahigh-resolutionmass spectrometry via the Fourier transform-ion cyclotron resonance technique (FT-ICR-MS) revealed the presence ofpolyaromatic hydrocarbons (PAHs) dissolved in the abyssal ocean. More than 200 different PAHs were discerned, most of themconsisting of seven condensed rings with varying numbers of carboxyl, hydroxyl, and aliphatic functional groups. Theseunambiguously thermogenic compounds were homogenously distributed in the deep sea, but depleted at the sea surface. Based onthe structural information alone, petrogenic and pyrogenic compounds cannot be distinguished. Surface depletion of the PAHs andfirst estimates for their turnover rate (>1.2 ·1012 mol C per year) point toward a primarily petrogenic source, possibly deep-seahydrothermal vents, which is thus far speculative because the fluxes of combustion-derived and petrogenic matter to the ocean arenot well constrained. We estimate that >2.4% of DOM are thermogenic compounds, and their global inventory in the oceans is>1.4 ·1015 mol C, significantly impacting global biogeochemical cycles.© 2006 Elsevier B.V. All rights reserved.

Keywords: Aromatic hydrocarbons; Dissolved organic matter; Petrogenic organic matter; Black carbon; Carbon cycle; Deep water; Antarctica;Weddell Sea

1. Introduction

The oxygen content of the earth's atmosphere and theglobal cycle of organic carbon are invariably linked(Lasaga and Ohmoto, 2002). Free oxygen in theatmosphere is generated via burial of organic carbon inmarine sediments. Sedimentary organic carbon thencycles through the lithosphere. Over geological timescales tectonic and volcanic processes carry petrogeniccarbon back to the continents' surface where it is oxi-dized into CO2, consuming atmospheric oxygen. It is this

⁎ Corresponding author. Tel.: +1 850 645 1887.E-mail address: [email protected] (T. Dittmar).

0304-4203/$ - see front matter © 2006 Elsevier B.V. All rights reserved.doi:10.1016/j.marchem.2006.04.003

delicate balance between photosynthesis, organic carbonburial, tectonic forces, volcanism and sedimentary rockweathering that ultimately controls atmospheric compo-sition (Lasaga and Ohmoto, 2002). The time petrogenicorganic carbon is protected from oxidation is a crucialparameter. In periods of enhanced tectonic activity,turnover time accelerates and increased amounts of petro-genic organic carbon are oxidized which causes atmo-spheric oxygen to drop. Decreased oxygen levels, in turn,enhance organic matter preservation in sediments andreduce oxidative rock weathering, providing a negativefeedback which is vital for the habitability of earth.

The spatial separation between organic matter burialin the oceans and its oxidation at the continents'

Page 2: Thermogenic organic matter dissolved in the abyssal ocean

209T. Dittmar, B.P. Koch / Marine Chemistry 102 (2006) 208–217

surface links atmospheric oxygen content to the rockcycle. Volcanic activity occurs equally on the con-tinents and the ocean floor. Some of the most activevolcanoes and hydrothermal fields exist along theextensive mid-oceanic ridges and even on the older,tectonized portions of the oceanic crust (Kelley et al.,2002). A global hydrothermal fluid reservoir resideswithin the uppermost warm oceanic crust (Johnson andPruis, 2003). The global effect is that a huge amountof water, exceeding the river runoff by an order ofmagnitude, passes through hydrothermal plumes,equivalent to an ocean water recycling time of 4–8 · 103 years (Elderfield and Schultz, 1996; andreferences therein). Some of the hydrothermal waterscan reach temperatures of 350 °C and more, leading toextensive chemical reactions between seawater con-stituents and rocks (Elderfield and Schultz, 1996).Abundant deposits of hydrothermal petroleum werefound in Mid-Atlantic Ridge sediments (Simoneit etal., 2004). The mobilization of soluble organiccompounds into hydrothermal fluids is likely (Simo-neit et al., 2004), but has gained very little scientificattention (Cowen et al., 2004).

Residues from biomass combustion (“black carbon”)can be highly resistant to further degradation (Seiler andCrutzen, 1980). The abundance of large-scale wildfiresthat can be triggered by increased volcanism could thusprovide an additional negative feedback in the oxygencycle. However, it is unclear to what degree black carbonpersists on geological time scales. It has recently beenshown that black carbon in soils can be remobilized intoactive cycles and transferred into riverine dissolvedorganic matter (Kim et al., 2004; Mannino and Harvey,2004).

We hypothesize that a significant fraction ofpetrogenic or combustion-derived organic matter isremobilized out of marine or terrestrial deposits andfinally enters the marine dissolved organic matter pool(DOM). If our hypothesis is true, the chemical signatureof thermogenesis should be universally imprinted intothe molecular structure of DOM. In particular polyaro-matic hydrocarbons (PAHs) with five or more con-densed ring systems would unambiguously indicatethermal alteration (e.g. Simoneit, 2002; Kramer et al.,2004). The organic matter that is dissolved in theoceanic water column amounts to ∼700 Gt of carbon,similar to the quantity of atmospheric CO2 or living(terrestrial and marine) biota (Siegenthaler and Sar-miento, 1993; Hedges et al., 1997). Up to 90% of thelipophilic DOM fraction has a fossil radiocarbon ageand may arise from inputs of pre-aged petrogenicprecursors (Loh et al., 2004).

In order to identify a thermogenic molecular signaturein marine DOM we used ultrahigh-resolution massspectrometry. Major advances in molecular fingerprint-ing and chemical characterization of DOM have recentlybeen achieved by ultrahigh-resolution mass spectrome-try via the Fourier transform-ion cyclotron resonancetechnique (FT-ICR-MS; Kujawinski et al., 2002; Sten-son et al., 2002, 2003). The primary strength of the FT-ICR-MS technique is the ability to identify individualmolecules in complex natural mixtures without priorseparation, e.g. through chromatography. It has beenshown by Koch et al. (2005) that FT-ICR-MS is suitablefor obtaining detailed compositional information onmarine DOM. On the basis of ultrahigh-resolved massspectrometry data, the authors identified >1500 molec-ular formulae in marine DOM. Because of theextraordinary mass accuracy of high-field FT-ICR-MS,molecules with the same nominal mass can bedistinguished. For instance, N2O and CO2 have thesame nominal mass (44 Da), but because of a distinctivemass defect, i.e. different number in protons andneutrons, their actual masses differ slightly (44.001versus 43.990 Da). This feature can be exploited in orderto assign mass differences to specific differences inmolecular formulae, and discrete formulae can bedetermined for each mass detected via FT-ICR-MS.The data presented by Koch et al. (2005) providemolecular insights into the polydisperse and complexmixtures of marine DOM.

The Southern Ocean (Antarctica) is uniquely suitedfor the purpose of this study because the deep watermasses are particularly important for global oceancirculation (Orsi et al., 1999; Mackensen, 2001; andreferences therein): the Warm Deep Water (WDW)branches from the deep Antarctic circumpolar currentand is largely composed of deep water from the NorthernAtlantic and even older water masses derived from thePacific that have traveled all the way through the oceanicconveyor belt, thereby integrating global information.Weddell Sea Deep Water (WSDW) underlies the WDW;it has partly been in contact with the atmosphere and isthe water mass that ultimately ventilates the abyssalplains of the three oceans. Antarctic surface water(AASW), on the other hand, is comprised primarily ofold (pre-industrial) water masses that have reached thesurface from the deep through regional upwelling.Anthropogenic impact is minimal in the Weddell Seaand due to its unique location, aerosol deposition isnegligible (e.g. Jickells et al., 2005) and no riversdischarge into it.

The objective of the present study was to elaborate onour hypothesis, the release of thermogenic organic

Page 3: Thermogenic organic matter dissolved in the abyssal ocean

210 T. Dittmar, B.P. Koch / Marine Chemistry 102 (2006) 208–217

matter into the marine DOM pool, by exploring themolecular structure of DOM. For this purpose wesearched the extensive mass spectrometry data setobtained for the Southern Ocean via FT-ICR-MS(Koch et al., 2005) on specific thermogenic molecularstructures. This data set is the first of its kind publishedfor open ocean DOM.

2. Material and methods

2.1. Sampling and sample preparation

Antarctic seawater was sampled from a rosettesampler connected to a CTD probe during theexpedition ANT XIX/2 of the research icebreakerPolarstern in the central and western Weddell Sea.Five samples of the major water masses were taken attwo different sites (Fig. 1): 69.6°S, 3.7°W (AASW at30 m and 100 m, and WSDW at 3500 m water depths)and 69.0°S, 28.5°W (WDW at 200 m, and WSDW at4600 m water depths). Ice cover during this early australsummer was moderate. Immediately after sampling, allwater samples were filtered with a sequence of 3.0 and0.2 μmWhatman Nuclepore membrane filter cartridges.For mass spectrometry, DOM was extracted onboardfrom the saline seawater matrix via solid phaseextraction (C18 cartridges, 60 mL, Varian Mega BondElut). Before extraction the cartridges were rinsed withacidified (pH=2) Milli-Q water and methanol. Then thefiltrate (20 L seawater) was acidified with hydrochloric

Fig. 1. Location of the sampling sites in the

acid to pH=2 and pumped through the cartridges at aflow rate of <50 mL min−1. For complete removal ofsalt, the cartridges were rinsed with 100 mL acidifiedMilli-Q water prior to elution. Immediately thereafter,DOM was eluted with 30 mL methanol and stored at−18 °C in the dark. For the determination of extractionefficiencies, DOC concentrations were measured in theoriginal samples and in redissolved freeze-dried extracts(in Milli-Q water) by high temperature catalyticoxidation (HTCO, Shimadzu, Model 5000). The DOCextraction efficiency of this procedure was 26±4%.DOC detection limit was 5 μM. No DOC was detectablein the blank samples (solid phase extracts of artificialseawater).

2.2. Mass spectrometry and data analysis

For electrospray ionization (ESI), which is the softestavailable ionization technique for mass spectrometry, analiquot of the DOM methanol extract was mixed withMilli-Q water (50:50 v/v). Formic acid was added(0.2% final concentration) to enhance ionizationefficiency. The solution was introduced into theelectrospray ion source by infusion with a flow rate of2 mL min−1. Prior to injection small amounts of NaCl(final concentration < 1 mM) were added to supportsodium adduct formation during ionization. All analy-ses were performed on an APEX-Q Fourier TransformIon Cyclotron Resonance Mass Spectrometer equippedwith a 7 T superconducting magnet (Bruker Daltonics)

Atlantic sector of the Southern Ocean.

Page 4: Thermogenic organic matter dissolved in the abyssal ocean

Fig. 2. Ultrahigh-resolution mass spectra of marine DOM samples fromAntarctica: (a) complete spectrum (300–600 m/z) of Antarctic SurfaceWater (30 m water depth), the grey shaded area marks (b) an enlargedsection of the full spectrum (455–540 m/z), the asterisk marks (c) a verynarrowmass range of 499.09–499.17m/z for three different water depths.Due to the ultrahigh resolution, molecular formulae can be assigned toeach mass. For details on the proposed structural formulae see Fig. 4.

211T. Dittmar, B.P. Koch / Marine Chemistry 102 (2006) 208–217

according to the procedure described in Koch et al.(2005). The spectra were internally calibrated with apolyethylene glycol standard and measured in thepositive ionization mode. To increase peak resolutionthe ions were mass selected with a quadrupole filterbetween ion source and the FT-ICR-MS analyzer usinga setting for ion selection of about 100 mass units. Thus,several mass spectra were acquired for each sample andsubsequently merged to one continuous peak list. Ionaccumulation time was set to 3 s for each scan.200 scans were added for each mass spectrum.

All detected ions were singly charged, as deter-mined by the m/z (mass-to-charge ratio) spacingbetween 12Cn and 13C12Cn−1 peaks of the individualmolecules. The overall peak intensities were similar inall samples. Once the exact masses of the moleculeshad been determined, their molecular formulae werecalculated by arbitrarily combining any possiblecombination of atoms (Koch et al., 2005). All possiblechemical formulae were computed for each detectedmass in ±0.001 Da mass windows. Typical peakdistances reached accuracies better than 0.0003 Da(for comparison: the mass of one electron is0.0005 Da). In this work we focused on the mostabundant odd ions (signal to noise ratio > 30) whichwere all nitrogen-free. The following elements (andnumber of atoms of each element) were considered inthe calculation: 12C (0–100), 1H (0–200), 14N (0–10),16O (0–50) and 23Na (0–1). After applying the rulesand assumptions described in Koch et al. (2005) alldetected masses could be assigned to one unambig-uous chemical formula. The molecular formulae weresorted into molecular families using the KendrickMass Defect (KMD; Kendrick, 1963) and the methoddescribed by Stenson et al. (2003). The KMDidentifies (pseudo-) homologous series to which anion belongs. The members of a homologous seriesdiffer from each other by the number of CH2-groupsand thus multiple addition of 14.01565 Da. The KMDwill be the same for all members of a homologousseries. Using this approach, a computer program canidentify and sort the thousands of individual formulaethat can be calculated from a single spectrum into amore manageable data array.

Another convenient way of visualizing the enor-mous amount of information are van Krevelendiagrams. In a van Krevelen diagram, atomic hydro-gen-to-carbon ratios (H/C) of each identified moleculeare plotted versus the respective oxygen-to-carbonratios (O/C). These plots are useful tools for identifyingthe various types of organic compounds (e.g. hydro-carbons, carbohydrates, aromatics, carboxylic acids) in

a DOM mixture (Kim et al., 2003). However, it must beconsidered that different molecular formulae can haveidentical O/C and H/C ratios and occupy the samelocation in the plot. Molecular formulae also provideinformation about the sum of double bonds and rings ineach molecule, especially in relation to the oxygencontent. The sum of rings and double bonds in eachmolecule or “double bond equivalents” (DBE) can becalculated from the number of atoms and the valence ofeach element (e.g. Koch and Dittmar, 2006). In asimilar fashion, an aromaticity index (AI) can becalculated from molecular formulae of naturallyoccurring compounds containing C, H, O, N, S and P(Koch and Dittmar, 2006). AI provides an unequivocalminimum criterion for the existence of either aromatic(AI>0.5) or condensed aromatic structures (AI≥0.67)in NOM.

Page 5: Thermogenic organic matter dissolved in the abyssal ocean

212 T. Dittmar, B.P. Koch / Marine Chemistry 102 (2006) 208–217

3. Results and discussion

3.1. Evidence for functionalized polycyclic aromatichydrocarbons (PAHs)

A total of 1458 different molecular formulae wereresolved and identified in the five samples. An exampleillustrating the resolving power of the mass spectrom-etry is given in Fig. 2. The relatively narrow range ofO/C and H/C ratios of all identified molecules (Fig. 3a;and Koch et al., 2005) reflects the absence of mostcommon lipids, proteins and carbohydrates. Comparedto these biomolecules, H/C ratios are low, indicatingdehydrogenation or condensation of precursors or se-lective preservation of more unsaturated compounds.

Most strikingly in the context of a possible thermo-genic origin is the presence of molecules with very lowH/C ratios. A group of molecules in a very distinctregion in the van Krevelen diagram (Fig. 3) ischaracteristically separated from the other compoundsby their low H/C and O/C ratios of <0.9 and <0.25,respectively. A total of 224 different molecular formulaewere identified in this region. Owing to the low H/Cratios, these molecules have a high number of doublebonds with up to 35 double bond equivalents (DBE)which restricts the number of possible structuralisomers. Whether this high unsaturation results inaromatic or even condensed polyaromatic structures,which might be indicative of thermogenic sources,depends on the molecular size and the number ofcarbonyl-oxygen in the molecule: a large number ofdouble bonds can be distributed over a large molecule,without necessarily forming aromatic structures, and

Fig. 3. (a) van Krevelen plot of deep-sea DOM, presenting molecules identifi(b) Enlarged region of condensed polyaromates. White circles are members

each carbonyl-oxygen contributes with one DBE to thetotal number of DBEs. The size of a molecule can beaccounted for by normalizing DBE with the number ofcarbon atoms, i.e. computing DBE/C ratios. Forinstance, benzene has a DBE/C ratio of 0.67 (4/6),thus every molecule with a higher DBE/C ratio containscondensed aromatic ring structures. However, a DBE/Cthreshold value that would serve as an unambiguouscriterion for identifying condensed polyaromatic struc-tures is not possible, because of the potential contribu-tion of oxygen to DBE.

For a most conservative estimate we considered alloxygen as carbonyl, minimizing the amount of C–Cdouble bonds and aromaticity. Under this (unrealistic)assumption, all molecules marked with hs1 to hs6 inFig. 3b would consist of five condensed aromatic ringswith a different number of carbonyl, methyl or vinylfunctional groups. On the other extreme, if all oxygenwere present as hydroxyl and therefore not contributingDBEs, eight condensed rings, again with differentnumbers of hydroxyl, methyl and vinyl functionalgroups, would be needed to explain the mass formulaeof the marked molecules. It is therefore certain that thesemolecules are functionalized polyaromatic compoundswith five to eight condensed aromatic rings. They rangein molecular weight from 428 Da (molecule hs1) to530 Da (molecule hs6), and the aromaticity index (AI;Koch and Dittmar, 2006) for these compounds isAI>0.67. Therefore these substances necessarily con-tain condensed polyaromatic structures.

The most likely scenario concerning oxygen-contain-ing functional groups is probably in between the twoextremes with a similar proportion of carbonyl and

ed in all deep-sea samples, i.e. both at 3500 m and 4600 m water depth.of (CH2)n homologous series with n≥4.

Page 6: Thermogenic organic matter dissolved in the abyssal ocean

213T. Dittmar, B.P. Koch / Marine Chemistry 102 (2006) 208–217

hydroxyl groups, or predominance of carboxyl groups.Microbial degradation and partial ring cleavage reducecarboxyl functional groups in PAHs (e.g. Sepic et al.,1997) which is the likely reason for the presence ofoxygen in the PAHs identified in our samples. Also thesolubility in seawater points towards acidic functional-ity. If we assign all oxygen to carboxyl groups and, inthe case of an odd number of oxygen in a molecularformula, include one hydroxyl group, all moleculeswould consist of a polyaromatic core with the massformula C27H16. This core comprises six or sevencondensed rings and a different number of carboxyl,hydroxyl and aliphatic functional groups (Fig. 4).

The molecules hs1 to hs6 are the smallest membersof the (CH2)n homologous series that extend, with up toseven members, over the entire H/C range of thedistinct area in the van Krevelen plot depicted in Fig.3b. Even though these homologous series may consistof different isomers within one series, their existenceshows that they are chemically related to each other,

Fig. 4. Proposed molecular structures and alternative isomers for each (CH2

maximum number of oxygen is present in carboxyl functional groups.

indicating a common origin. Since none of thehomologous series of the polyaromatic region in Fig.3 extends into the region where the bulk of thecompounds were found, it is likely that the functiona-lized PAHs have a different biogeochemical source andhistory than the main body of organic moleculesidentified in our samples.

3.2. The source of PAHs in the ocean

There exist no data on high-molecular weight,functionalized PAHs in marine systems. The availableliterature on PAHs in the marine environment is almostentirely restricted to anthropogenic contaminants, inparticular smaller unsubstituted PAHs, which have verylow solubility in seawater. Their background concen-tration in remote marine environments is <0.5 nM C(Nemirovskaya and Novigatskii, 2003). Owing to thelarge number of carboxyl, carbonyl and/or hydroxylfunctional groups, the PAHs in our samples have

)n homologous series of condensed polyaromates. It is assumed that a

Page 7: Thermogenic organic matter dissolved in the abyssal ocean

Fig. 5. Depth profiles of PAHs. Number of identified PAHs andcumulative PAHs signal intensity in themass spectra plotted versus waterdepth (exponential scale). PAHs that occurred in both of the two deepestsamples (3500 m and 4600 m water depth) are termed deep-sea PAHs.

214 T. Dittmar, B.P. Koch / Marine Chemistry 102 (2006) 208–217

comparatively high solubility in seawater and thereforeprobably do not accumulate in sedimentary deposits,which could explain their ubiquitous presence through-out the water column. It is, however, likely that thesePAHs were mobilized from a more apolar, insolublesource by (microbial) oxidation.

The molecular structure of the PAHs (Fig. 4), andthe virtual absence of unsubstituted, small PAHs inour samples gives us some indication of their ultimatesource. Biotic processes that lead to the formation ofhigh molecular weight PAHs are not known. Incom-plete combustion and pyrolysis at high temperature,including biomass combustion, fossil-fuel burning orvolcanism, are significant sources of PAHs to theenvironment. While combustion-derived PAHs fromfossil-fuel utilization are mostly characterized by lowmolecular weights (Simoneit, 2002), black carbonfrom biomass combustion can contain a widespectrum of functionalized, high-molecular weightPAHs (Kramer et al., 2004). These compounds wereabundant in a volcanic soil, but were carboxylated to amuch higher degree than the PAHs in our samples.They had approximately twice the O/C ratios than oursamples, with O/C values exceeding 0.5 (Kramer etal., 2004). Functionalized, large and probably com-bustion-derived PAHs have also been identified inhigh-molecular weight DOM (>1 kDa) from the RioNegro in Brazil (Kim et al., 2004). Combustion-derived compounds can be transported to the oceansvia rivers (as suspended solids or DOM) and aerosols.Once deposited on the oceans' floor, particles may beoxidized and remobilized into the water column asDOM.

High molecular weight PAHs are also generated inearth's crust during catagenesis through high tempera-ture alteration and reworking (aromatization) of bio-genic organic matter or de novo synthesis from freeradical recombination (Simoneit, 2002). Free radicals,such as acetylene and butadiene, recombine successive-ly. Once formed, the simple low-molecular weightunsubstituted PAHs undergo further build-up to largerand thermodynamically more stable PAH series. Theabiogenic PAHs occur primarily as the parent com-pounds without substituents, contrary to thermallyaltered biogenic material. However, random substitutionwith methyl groups may still occur in the course ofcatagenesis (Simoneit, 2002), which makes the presenceof methyl substitution ambiguous as an indicator forbiogenic precursors of the PAHs.

While it is likely that the PAHs are derived fromthermally altered biogenic material and probably notfrom fossil-fuel combustion and that they underwent

considerable microbial modification, petrogenic andpyrogenic compounds cannot be distinguished based onthe structural information alone. The vertical distribu-tion of PAHs in the water column may provideadditional information on their ultimate source (Fig. 5).

The presence of PAHs in the different water massesof the abyssal ocean contradicts a local point source ofthese compounds and indicates a rather homogenousdistribution over the world oceans, considering the timescale over which complete vertical mixing of theocean's water column takes place. Even though thehigh degree of oxygen-containing functional groupsindicates microbial reworking, the PAHs seem to berefractory in the oceans' interior. At the sea surface,however, the number of identified PAHs as well as theircumulative signal intensity in the mass spectrumdecreased sharply. Both parameters correlated withwater density (Fig. 6). In order to explain the surfacedepletion, a removal process must be inferred becausethe contribution of freshwater (ice melt, σst≈0) is fartoo small to have any detectable effect on PAHsconcentration. A plausible removal process is photo-degradation by sunlight (e.g. Fasnacht and Blough,2002). The conjugated π-electron system makes PAHs

Page 8: Thermogenic organic matter dissolved in the abyssal ocean

Fig. 6. Number of identified PAHs and cumulative PAHs signalintensity plotted versus potential water density (σst).

215T. Dittmar, B.P. Koch / Marine Chemistry 102 (2006) 208–217

particularly receptive to photochemical reactions, espe-cially under absence of protective particle surfaces.Photochemical reactions may lead to ring cleavage ordefunctionalization which in turn may decrease theionization efficiency in ESI. Adsorption onto suspendedorganic matter in the upper water column where sinkingparticles originate, could provide another explanationfor the observed vertical distribution of the PAHs. Themechanisms behind the observed surface depletionremain speculative and require further experimentaltesting, but it is unlikely that the observed surfaceremoval is a local phenomenon restricted to the WeddellSea. If the PAHs are indeed labile at the sea surface ingeneral, then PAHs probably do not survive large-scaletransport over continental shelves. This points toward awidespread source at the floor of the deep ocean basins.This source could be either petrogenic (hydrothermalfluids) or pyrogenic (remobilization of combustion-derived deposits).

3.3. Potential implications for global biogeochemicalcycles

The cycling of thermogenic DOM through theoceanic water column may have major implicationsfor global biogeochemical cycles, depending on amountand turnover rate. FT-ICR-MS is considered semi-quantitative, mainly because the ionization efficiency inthe electrospray (ESI) differs between molecules. Forinstance, no sodium adducts were observed for PAHs,which had comparatively low O/C ratios, whereas mostmolecules in the higher O/C range (O/C>0.3) weresodiated during electrospray (see also Koch et al., 2005).

As a general rule, ionic and polar compounds ionizemore efficiently in positive ionization mode than apolarmolecules. PAHs were therefore less efficiently ionizedthan the average DOM molecule observed in FT-ICR-MS. The signal intensity of PAHs thus underestimatestheir actual contribution to total DOM. For the samereason, solid phase extraction from seawater throughhydrophobic adsorbers is probably more efficient forPAHs than for the bulk of DOM. In order to provide alower limit for the concentration of PAHs in seawater wemake the following (conservative) assumptions: (i)PAHs were completely recovered from seawater throughsolid phase extraction, whereas extraction efficiencywas only 26% for the bulk of DOM (see Material andmethods). (ii) Ionization efficiency during ESI was thesame for PAHs and bulk DOM.

Based on these most conservative assumptions>2.4±0.5% of dissolved organic carbon in the darkocean (>100 m) were high-molecular weight functio-nalized PAHs. On a hydrogen basis, this translates into>1.3±0.2% of organic hydrogen, matching very wellresults from 1H-NMR analyses that identified approx-imately double that amount as aromatic compounds(1H-NMR shift range of 10.0–6.0 ppm; Dittmar et al.,unpublished).

The deep water masses of the Atlantic sector of theSouthern Ocean integrate to a certain degree globalinformation. It is therefore reasonable to extrapolateour findings onto a larger scale in order to obtain afirst estimate on the global inventory of thermogenicPAHs dissolved in the oceans. The oceans contain∼60 ·1015 mol C of dissolved organic carbon. If weassume that >2.4±0.5% are thermogenic PAHs, theglobal inventory of this compound class would be>1.4±0.3 ·1015 mol C. Their turnover is probablylinked to sea surface processes (Fig. 5) and thus to theturnover of the oceans' water column. It takes∼ 850 yearsto ventilate the major ocean basins (Broecker et al., 1999).If we assume steady state of ocean circulation over the lastfew hundred years, and that 70% of PAHs are removed atthe sea surface (Fig. 5), the turnover of thermogeniccarbon in the ocean is >1.2·1012 mol C per year. Thisestimated turnover rate for marine PAHs is considerablyhigher than current estimates of the annual input of blackcarbon to the global ocean.

The global emission of black carbon particles frombiomass burning is ∼0.44 ·1012 mol C per year (IPCC,2001) of which approximately one third (∼0.15 ·1012 mol C) is deposited on the ocean surface (Suman,1999). Rivers may transport 1.0 ·1012 mol C of blackcarbon per year to the ocean (Suman, 1999), of which>90% is deposited in coastal ocean sediments (Hedges et

Page 9: Thermogenic organic matter dissolved in the abyssal ocean

216 T. Dittmar, B.P. Koch / Marine Chemistry 102 (2006) 208–217

al., 1997). The total input of black carbon to the openocean is therefore probably <0.25 ·1012 mol C per yearor even lower because of common methodologicalartifacts, which inadvertently overestimate the amount ofblack carbon (Dickens et al., 2004; Simpson andHatcher, 2004).

In order to explain our turnover estimates withpetrologic processes alone, the water that passes throughhydrothermal plumes (1.8–3.4 · 1017 kg per year;Elderfield and Schultz, 1996) must contain 3.5–6.7 μM DOC (as PAHs) above marine background.Simoneit and Sparrow (2002) found comparatively highDOC concentrations of 308 μM, which is 267 μM abovethe marine background, in near-bottom waters at theseafloor spreading center of Juan de Fuca Ridge(Northern Pacific). The release of polar organiccompounds from thermal alteration of organic matterwas a likely reason for these elevated DOC concentra-tions (Simoneit and Sparrow, 2002), and could explainthe presence of thermogenic PAHs in the deep ocean.

3.4. Conclusions

The unsurpassed mass-resolving power of FT-ICR-MS enabled us to identify thermogenic organiccompounds in the abyssal ocean on a molecularlevel. These compounds were depleted at the seasurface. Based on the unique location of our samplingstations we provided a first estimate for the globalinventory of dissolved PAHs in the ocean (>1.4±0.3 ·1015 mol C). The release of thermogenic com-pounds from deep-sea hydrothermal systems canexplain our estimates for the inventory and turnoverof thermogenic DOM, but this explanation is thus farspeculative and requires further evidence. The discov-ery of a major thermogenic DOM component in thedeep sea and its depletion at sea surface shows thatmarine processes may significantly mobilize refractorycompounds back into active cycles, affecting long-termbiogeochemical cycles in a way not previouslyconsidered. FT-ICR-MS is one of the most expensiveand time-consuming analytical techniques available,and only a small number of samples can be processed.We increased the significance of a limited number ofsamples by choosing a most representative locationwhere anthropogenic influence is small and wherewater masses integrate global information, and werestricted any quantitative estimates to the mostconservative case. Future research should include theestablishment of routine analytical techniques forcarboxylated large PAHs, in order to cover the majorocean basins on a high spatial resolution.

Acknowledgements

We thank Ralph Engbrodt and Gerhard Kattner forhelpful discussions, Matthias Witt (Bruker Daltonik,Bremen) for FT-ICR-MS measurements, and JeffChanton, the Associate Editor Dennis Hansell andthree anonymous reviewers for constructive commentson the manuscript. This work was financially sup-ported by the Petroleum Research Fund (ACSPRF#41515-G2), the National Oceanic and Atmo-spheric Administration (NOAA GC05-099), DeutscheForschungsgemeinschaft (DFG KO 2164/3-1), and theGerman Academic Exchange Service (DAAD PPPUSA 315/ab).

References

Broecker, W.A., Sutherland, S., Peng, T.-H., 1999. A possible 20th-century slowdown of Southern Ocean deep water formation.Science 286, 1132–1135.

Cowen, J.P., Thomson, R.E., Kadko, D.C., Wakeham, S.G., Bertram,M.A., Lam, P., Popp, B.N., 2004. Biogeochemical processes in mid-ocean ridge hydrothermal plumes. American Geophysical Union,Fall Meeting 2004, San Francisco CA, USA, 13–17 December.

Dickens, A.F., Gelinas, Y., Masiello, C.A., Wakeham, S.G., Hedges,J.I., 2004. Reburial of fossil organic carbon in marine sediments.Nature 427, 336–339.

Elderfield, H., Schultz, A., 1996. Mid-ocean ridge hydrothermal fluxesand the chemical composition of the ocean. Annu. Rev. EarthPlanet. Sci. 24, 191–224.

Fasnacht, M.P., Blough, N.V., 2002. Aqueous photodegradation ofpolycyclic aromatic hydrocarbons. Environ. Sci. Technol. 36,4364–4369.

Hedges, J.I., Keil, R.G., Benner, R., 1997. What happens to terrestrialorganic matter in the ocean? Org. Geochem. 27, 195–212.

IPCC, 2001. Climate Change 2001, The Scientific Basis. Contributionof Working Group I to the Third Assessment Report of theIntergovernmental Panel on Climate Change. Cambridge Univer-sity Press, Cambridge, UK.

Jickells, T.D., An, Z.S., Andersen, K.K., Baker, A.R., Bergametti, G.,Brooks, N., Cao, J.J., Boyd, P.W., Duce, R.A., Hunter, K.A.,Kawahata, H., Kubilay, N., laRoche, J., Liss, P.S., Mahowald, N.,Prospero, J.M., Ridgwell, A.J., Tegen, I., Torres, R., 2005. Globaliron connections between desert dust, ocean biogeochemistry, andclimate. Science 308, 67–71.

Johnson, H.P., Pruis, M.J., 2003. Fluxes of fluid and heat from theoceanic crustal reservoir. Earth Planet. Sci. Lett. 216, 565–574.

Kelley, D.S., Baross, J.A., Delaney, J.R., 2002. Volcanoes, fluids, andlife at mid-ocean ridge spreading centers. Annu. Rev. Earth Planet.Sci. 30, 385–491.

Kendrick, E., 1963. A mass scale based on CH2=14,0000 for highresolution mass spectrometry of organic compounds. Anal. Chem.35, 2146–2154.

Kim, S., Kramer, R.W., Hatcher, P.G., 2003. Graphical method foranalysis of ultrahigh-resolution broadband mass spectra of naturalorganic matter, the Van Krevelen Diagram. Anal. Chem. 75,5336–5344.

Kim, S., Kaplan, L.A., Benner, R., Hatcher, P.G., 2004. Hydrogen-deficient molecules in natural riverine water samples—evidence

Page 10: Thermogenic organic matter dissolved in the abyssal ocean

217T. Dittmar, B.P. Koch / Marine Chemistry 102 (2006) 208–217

for the existence of black carbon in DOM. Mar. Chem. 92,225–234.

Koch, B.P., Dittmar, T., 2006. From mass to structure: an aromaticityindex for high-resolution mass data of natural organic matter.Rapid Commun. Mass Spectrom. 20, 926–932.

Koch, B.P., Witt, M., Engbrodt, R., Dittmar, T., Kattner, G., 2005.Molecular formulae of marine and terrigenous dissolved organicmatter detected by electrospray ionisation Fourier transform ioncyclotron resonance mass spectrometry. Geochim. Cosmochim.Acta 69, 3299–3308.

Kramer, R.W., Kujawinski, E.B., Hatcher, P.G., 2004. Identification ofblack carbon derived structures in a volcanic ash soil humic acid byFourier transform ion cyclotron resonance mass spectrometry.Environ. Sci. Technol. 38, 3387–3395.

Kujawinski, E.B., Freitas, M.A., Zang, X., Hatcher, P.G., Green-Church, K.B., Jones, R.B., 2002. The application of electrosprayionization mass spectrometry (ESI MS) to the structural character-ization of natural organic matter. Org. Geochem. 33, 171–180.

Lasaga, A.C., Ohmoto, H., 2002. The oxygen geochemical cycle:dynamics and stability. Geochim. Cosmochim. Acta 66, 361–381.

Loh, A.N., Bauer, J.E., Druffel, E.R.M., 2004. Variable ageing andstorage of dissolved organic components in the open ocean. Nature430, 877–881.

Mackensen, A., 2001. Oxygen and carbon stable isotope tracers ofWeddell Sea water masses: new data and some paleoceanographicimplications. Deep-Sea Res., Part I 48, 1401–1422.

Mannino, A., Harvey, H.R., 2004. Black carbon in estuarine and coastalocean dissolved organic matter. Limnol. Oceanogr. 49, 735–740.

Nemirovskaya, I.A., Novigatskii, A.N., 2003. Hydrocarbons in thesnow and ice cover and waters of the Arctic Ocean. Geochem. Int.41, 585–594.

Orsi, A.H., Johnson, G.C., Bullister, J.L., 1999. Circulation, mixing,and production of Antarctic Bottom Water. Prog. Oceanogr. 43,55–109.

Seiler, W., Crutzen, P.J., 1980. Estimates of gross and net fluxes ofcarbon between the biosphere and the atmosphere from biomassburning. Clim. Change 2, 207–247.

Sepic, E., Bricelj, M., Leskovsek, H., 1997. Biodegradation studies ofpolyaromatic hydrocarbons in aqueous media. J. Appl. Microbiol.83, 561–568.

Siegenthaler, U., Sarmiento, J.L., 1993. Atmospheric carbon dioxideand the ocean. Nature 365, 119–125.

Simoneit, B.R.T., 2002. Molecular indicators (biomarkers) of past life.Anat. Rec. 268, 186–195.

Simoneit, B.R.T., Sparrow, M.A., 2002. Dissolved organic carbon ininterstitial waters from sediments of Middle Valley and EscanabaTrough, Northeast Pacific, ODP Legs 139 and 169. Appl.Geochem. 17, 1495–1502.

Simoneit, B.R.T., Lein, A.Y., Peresypkin, V.I., Osipov, G.A., 2004.Composition and origin of hydrothermal petroleum and associatedlipids in the sulfide deposits of the Rainbow Field (Mid-AtlanticRidge at 36°N). Geochim. Cosmochim. Acta 68, 2275–2294.

Simpson, M.J., Hatcher, P.G., 2004. Overestimates of black carbon insoils and sediments. Naturwissenschaften 91, 436–440.

Stenson, A.C., Landing, W.M., Marshall, A.G., Cooper, W.T., 2002.Ionization and fragmentation of humic substances in electrosprayionization Fourier transform-ion cyclotron resonance mass spec-trometry. Anal. Chem. 74, 4397–4409.

Stenson, A.C., Marshall, A.G., Cooper, W.T., 2003. Exact massesand chemical formulae of individual Suwannee River fulvicacids from Ultrahigh resolution electrospray ionization Fouriertransform ion cyclotron resonance mass spectra. Anal. Chem. 75,1275–1284.

Suman, D., 1999. Marine sediments: a black carbon reservoir andrecord of combustion. 9th Goldschmidt Conference, CambridgeMA, USA, 22–27 August.