Continental Shelf Research 24 (2004) 2373–2394 Organic biomarkers for tracing carbon cycling in the Gulf of Papua (Papua New Guinea) K.A. Burns a, , P. Greenwood b , R. Benner c , D. Brinkman a , G. Brunskill a , S. Codi a , I. Zagorskis a a Australian Institute of Marine Science, Box 3, Townsville, Qld. 4810, Australia b Geosciences Australia, Box 378 Canberra, ACT 2601, Australia c School of Biological Sciences, University South Carolina, Columbia, SC 29208, USA Available online 19 October 2004 Abstract Sediment traps were deployed in the Gulf of Papua in June–July 1997, to determine fluxes of organic matter and inorganic elements from the photic zone to deeper waters at the base of the continental slope and in the northern Coral Sea. Three stations, ranging from 900 to 1500 m depth, had ‘‘shallow’’ traps at 300 m below the water surface and ‘‘deep’’ traps set 100 m above the bottom. Infiltrex II water samplers collected particulate and dissolved organic matter from the Fly, Purari and Kikori rivers, and near-surface water from the shelf of the Gulf of Papua. Samples were analysed for molecular organic biomarkers to estimate the sources of organic carbon and its cycling processes. Dry weight fluxes from the shallow traps ranged from 115 to 181 mg m 2 day 1 and particulate organic carbon (POC) fluxes ranged from 1.2 to 1.9 mM OC m 2 d 1 with molar organic carbon to particulate nitrogen ratios (C/N) ranging from 6.0 to 6.5. Fluxes in deep traps were likely influenced by both early diagenesis and entrapment of resuspended shelf sediments. Dry weight fluxes in deep traps ranged from 106 to 574 mg m 2 day 1 and POC fluxes ranged from 0.6 to 1.5 mM OC m 2 d 1 , with C/N ratios ranging from 8.5 to 10.8. 13 C/ 12 C ratios were 20.2% to 21.7% in all trap samples, indicating that most of the settling POC was ‘‘marine-derived’’. Shallow traps had d 15 N values of 6.3% to 7.2% while the values in deep traps were 4.9–5.0%, indicating the N-rich near-surface OC was less degraded than that in the deep traps. The biogenic lipids consisted of hydrocarbon, sterol and fatty acid biomarkers indicative of marine zooplankton, phytoplankton and bacteria. Sterol markers for diatoms and dinoflagellates were abundant in the water samples. Highly branched isoprenoid alkenes, usually attributable to diatoms, were also detected in both water and shallow traps. Traces of C 26 –C 34 n-alcohols indicative of land–plant biomarkers, were found in river water samples and in the shallow sediment traps. A large unresolved complex mixture (UCM) of hydrocarbons, and a uniform distribution of n-alkanes, indicative of petroleum hydrocarbons, were also detected in the traps. Hopane and sterane biomarkers detected in the trap oil were characteristic of a marine carbonate source, and the aromatic hydrocarbon composition distinguished at least two different oil signatures. ARTICLE IN PRESS www.elsevier.com/locate/csr 0278-4343/$ - see front matter Crown Copyright r 2004 Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.csr.2004.07.014 Corresponding author. Tel.: +61-747534376; fax: +61-747725852. E-mail address: [email protected] (K.A. Burns).
Transcript
ARTICLE IN PRESS
0278-4343/$ - se
doi:10.1016/j.cs
�CorrespondiE-mail addre
Continental Shelf Research 24 (2004) 2373–2394
www.elsevier.com/locate/csr
Organic biomarkers for tracing carbon cycling in the Gulf ofPapua (Papua New Guinea)
K.A. Burnsa,�, P. Greenwoodb, R. Bennerc, D. Brinkmana, G. Brunskilla,S. Codia, I. Zagorskisa
aAustralian Institute of Marine Science, Box 3, Townsville, Qld. 4810, AustraliabGeosciences Australia, Box 378 Canberra, ACT 2601, Australia
cSchool of Biological Sciences, University South Carolina, Columbia, SC 29208, USA
Available online 19 October 2004
Abstract
Sediment traps were deployed in the Gulf of Papua in June–July 1997, to determine fluxes of organic matter and
inorganic elements from the photic zone to deeper waters at the base of the continental slope and in the northern Coral
Sea. Three stations, ranging from 900 to 1500m depth, had ‘‘shallow’’ traps at 300m below the water surface and
‘‘deep’’ traps set �100m above the bottom. Infiltrex II water samplers collected particulate and dissolved organic
matter from the Fly, Purari and Kikori rivers, and near-surface water from the shelf of the Gulf of Papua. Samples were
analysed for molecular organic biomarkers to estimate the sources of organic carbon and its cycling processes.
Dry weight fluxes from the shallow traps ranged from 115 to 181mgm�2 day�1 and particulate organic carbon (POC)
fluxes ranged from 1.2 to 1.9mM OCm�2 d�1 with molar organic carbon to particulate nitrogen ratios (C/N) ranging
from 6.0 to 6.5. Fluxes in deep traps were likely influenced by both early diagenesis and entrapment of resuspended shelf
sediments. Dry weight fluxes in deep traps ranged from 106 to 574mgm�2 day�1 and POC fluxes ranged from 0.6 to
1.5mM OCm�2 d�1, with C/N ratios ranging from 8.5 to 10.8. 13C/12C ratios were �20.2% to �21.7% in all trap
samples, indicating that most of the settling POC was ‘‘marine-derived’’. Shallow traps had d15N values of 6.3% to
7.2% while the values in deep traps were 4.9–5.0%, indicating the N-rich near-surface OC was less degraded than that
in the deep traps. The biogenic lipids consisted of hydrocarbon, sterol and fatty acid biomarkers indicative of marine
zooplankton, phytoplankton and bacteria. Sterol markers for diatoms and dinoflagellates were abundant in the water
samples. Highly branched isoprenoid alkenes, usually attributable to diatoms, were also detected in both water and
shallow traps. Traces of C26–C34 n-alcohols indicative of land–plant biomarkers, were found in river water samples and
in the shallow sediment traps. A large unresolved complex mixture (UCM) of hydrocarbons, and a uniform distribution
of n-alkanes, indicative of petroleum hydrocarbons, were also detected in the traps. Hopane and sterane biomarkers
detected in the trap oil were characteristic of a marine carbonate source, and the aromatic hydrocarbon composition
distinguished at least two different oil signatures.
e front matter Crown Copyright r 2004 Published by Elsevier Ltd. All rights reserved.
r.2004.07.014
ng author. Tel.: +61-747534376; fax: +61-747725852.
K.A. Burns et al. / Continental Shelf Research 24 (2004) 2373–23942374
We concluded that mass and POC fluxes were similar to those reported for other continental shelves and marginal
oceans in tropical and subtropical regions. There was a dramatic decrease in POC as particles sank, due to zooplankton
repackaging and photochemical and bacterial decomposition. Carbon isotopic and biomarker patterns showed most of
the POC in the sediment traps was marine-sourced with only traces of terrestrial input. There was a significant flux of
petroleum, which may signal the existence of natural petroleum seeps in this region.
Crown Copyright r 2004 Published by Elsevier Ltd. All rights reserved.
Keywords: Sediment traps; Vertical flux; Petroleum hydrocarbons; Lipid biomarkers; Petroleum seeps; Gulf of Papua; New Guinea
1. Introduction
This work is part of Project TROPICS (TropicalRiver Ocean Processes in Coastal Settings), a jointcoastal oceanographic study by Australia, Indo-nesia, Papua New Guinea, and the USA (http://www.aims.gov.au/tropics). Interest in New Guineais intense because much of the wet tropicalmountainous land-mass of New Guinea is beinguplifted at rates of 2–4 cmyr�1 (Loffler, 1977;Milliman and Sytvitsky, 1992) and with3–10myr�1 rainfall (McAlpine et al., 1983), thisresults in high erosion rates. Suspended sedimentdischarge from the island of New Guinea is greaterthan that of the Amazon River, despite its muchsmaller catchment area (Nittrouer et al., 1995;Milliman, 1995). The Fly, Bamu, Turama, Kikori,and Purari Rivers contribute approximately470 km3 yr�1 of muddy water to the estuaries andinner shelf of the Gulf of Papua (GOP). Riversediment supply and shelf sediment distributionhas been described by Davies (2001); Salomonsand Eagle (1990), Harris (1988), Harris and Baker(1991), Harris et al. (1993, 1996), Wolanski andAlongi (1995), Bird et al. (1995), and Brunskill etal. (1995). The latter studies show that most of theriverine sediment in the GOP is retained along theinner shelf at water depthso60m. They also showthe outer shelf is largely a relict carbonateplatform, and that the slope and trough sedimentscontain organic carbon of largely marine origin.Wolanski et al. (1984) described the circulation onthe inner shelf of the GOP, and Ayukai andWolanski (1997) demonstrated the importance ofbiologically mediated removal of fine sedimentsfrom the Fly River plume.The work described here comprises organic
geochemical studies intended to examine the
magnitude of carbon fluxes through the watercolumn of the GOP at the base of the continentalslope (�1000m) and into the northern Coral Sea,the sources contributing to these organic carbonfluxes, and a description of the important redis-tribution and degradation processes affecting theorganic matter.
2. Methods
2.1. Sediment traps
Six replicate sediment trap arrays were attachedto wire moorings just below the euphotic zone at300m below the water surface, and at 100m abovethe sea bottom, in water depths ranging from 960to 1460m (Fig. 1). Each trap array consisted of sixstainless steel cylinders fitted in a weightedaluminium frame. The frames were fitted with atrailing vane to orient the array in an uprightposition and facing any current. Each cylinder was25 cm diameter, 2m high. The bottom end waswelded to a 0.5m tall funnel arrangement, whichwas threaded to permit the attachment of a 4LNalgeneTM polyethylene collection bottle. Thecollection bottles were filled with salinity 70%brine solution made from pre-combusted NaCland ultra pure Milli-QTM water. A few mg ofHgCl2 was added as a preservative to prevent in-trap bacterial degradation, because the traps wereto be deployed for 2 months time. The cylinderswere meticulously pre-cleaned by detergent scrubsand then rinses with water and organic solvents.The traps were sealed with polyethylene bags andthen capped with specially made fibreglass clothcaps to ensure the integrity of the seals duringtransport. Covers were removed only as the traps
Fig. 1. Map of the Gulf of Papua (south coast of PNG) showing locations of sediment traps at the base of the continental slope and in
the northern Coral Sea. Also shown are the water sampling sites in the Purari, Kikori and Fly Rivers and the GOP.
K.A. Burns et al. / Continental Shelf Research 24 (2004) 2373–2394 2375
were lowered over the stern of the R/V Franklin.The mooring wires were clean plastic-coatedstainless steel and were wound around drumsseparate from the main ship winch. For deploy-ment, the dump hook on the main winch wire wasattached to the trap framework well above theopenings to the traps. As soon as the traps sankbelow the surface, the dump hook was releasedand the traps sank quickly. During deployment,the ship was positioned stern into the wind toprevent any contamination from smoke stacks.Traps were deployed in June–July 1997 at loca-tions shown as Fly Slope, Kerema Canyon andSouth East Fields in Fig. 1.When retrieved, caps were replaced and the
water in the upper portion of the traps was drainedfrom the cylinders through vents positioned abovethe collection bottles. The collection bottles werecarefully removed and capped. The subsamples forsalinity determinations were scanned with aGuildline salinometer capable of a maximumsalinity reading of �41%. All samples read off-scale, confirming that brine still remained in thecollection bottles although some mixing with waterin the cylinders could be expected. The water in thecollection bottles was filtered through pre-weighed
and pre-combusted GFA filters. Salt wasremoved with Milli-Q water rinses. The filterswere wrapped in solvent-cleaned aluminium foil,packed in labelled polyethylene bags and frozenuntil analysis.
2.2. Water samples
Water samples were collected by Infiltrex IITM
semi-high volume samplers (Green et al., 1986)which have a stainless steel holder for two 147mmdiameter, GFF filters in-line before an absorptioncolumn filled with approximately 200ml of Am-berlite XAD-2 resin (Rohm and Haas Corp.TM).The XAD-2 columns were glass and the resinswere pre-cleaned for several days in a modifiedSoxhlet apparatus until blank analyses wereacceptable (Ehrhardt, 1987; Ehrhardt and Burns,1999a). Columns were then capped and sealed inpolyethylene bags and stored in the refrigeratoruntil used in the samplers. Samplers were cleanedand loaded with filters and absorption column justbefore use. Filters were pre-combusted at 450 1Cbefore use. Water was collected at 0% salinitystations in the Purari and Kikori rivers and fromthe salinity gradient of the Fly river (Fig. 1). Clean
ARTICLE IN PRESS
K.A. Burns et al. / Continental Shelf Research 24 (2004) 2373–23942376
4L brown glass bottles were mounted in a stainlesssteel framework and tossed away from theresearch vessel. These sank to a depth of 1–2mand filled with water (UNESCO, 1984). On theship, 24–36 L of water were pumped through aTeflon inlet tube on the top of the sampler,through the filters and absorption column andthen into the body of the pump which contains aflow meter, the controlling electronics and bat-teries. After sampling, the filters were removed andpacked in glass jars with foil-lined lids and frozen.The XAD-2 columns were resealed and storedunder refrigeration. Organic residues absorbed tothe resin have been shown to be stable for longperiods of time (Green and La Pape, 1987).
2.3. C/N and stable isotope analysis
Subsamples of the dried trap particles werescrapped from the filters and analysed for C/N and13C/12C and 15N/14N ratios after vapour phaseacidification and in-line combustion using a CarloErba 1108 CHN (Benner et al., 1997) interfaced toa Finnigan Delta Plus ion ratio mass spectrometer.Stable carbon and nitrogen isotope compositionsare reported as
dNE ¼ ½ðRsample=RstandardÞ � 1�1000;
where NE is the heavy isotope of an element (13Cor 15N) and R is the ratio of 13C:12C or 15N:14N inthe sample to that of the standard (PDB carbonateand atmospheric nitrogen). Typical reproducibilityof analyses was 70.04% for d13C and 70.1%for d15N.
2.4. Sample preparation and extraction
‘‘Dissolved’’ lipids were eluted from XAD-2columns with the modified version of a Soxhletextractor using a 50% mixture of CH2Cl2 andCH3OH. Each XAD-2 column had two extrac-tions of 8 h duration plus overnight soaks with thesolvent. The extracts were reduced in volume on arotary evaporator until the CH2Cl2 was removed.The remaining methanol extract was poured into aglass separatory funnel and the lipids werepartitioned 3 times into hexane. The combined
hexane extract was dried with Na2SO4 andreduced to approximately 1ml.The GFF filters from the Infiltrex II samplers
were thawed and cut into very small pieces usingstainless steel scissors and tweezers. Approxi-mately 20 g of pre-combusted Na2SO4 was addedto the filter pieces in a glass beaker and thoroughlymixed with the salt using a stainless steel spatula.The filter/salt mixture was then loaded into aconventional Soxhlet extractor and extracted for 2days with CH2Cl2. Samples soaked in the solventovernight between the 8 h extractions.The filters from the sediment trap samples were
gently dried by opening their foil packets andplacing them in a clean polyethylene box filledwith moisture-indicating silica gel. The box wassealed and returned to the freezer for several days,until samples were dry. The filters were inspectedfor macro-fauna. Small copepod type crustaceanswere found on some filters. These were removedwith tweezers and weighed separately. The crusta-ceans were not included in the organic analyses.The filters were then weighed to determine massdry weight of sediment recovered. Filters werecarefully re-wrapped in their foil. To facilitategamma counting for radio-isotopic tracer analysis(reported separately) all six filters from one arraywere packed into a perspex chamber and countedfor periods averaging 10 days. Samples were keptin the freezer when not in the counter, which wasat room temperature.After radio-isotopic analysis, the trap filters
were cut into small pieces and placed into 50mlTeflon screw cap centrifuge tubes. Fifteen mlCH2Cl2 was added. Samples were extracted byplacing the tightly capped tubes in a beaker ofwater into which a probe sonicator was lowered.After 15min of sonication, extracts were filteredthrough a Pasteur pipette filled with Na2SO4 toensure extracts were dry. The sonic extraction wasrepeated twice.All extractions had a hexane solution of ortho-
terphenyl (OTP) added to the initial extractionsolvent as a surrogate internal surrogate standardto track extract losses during the procedures.Recoveries were generally greater than 80% forall samples, thus the analyses were not correctedfor recovery of the OTP standards.
ARTICLE IN PRESS
K.A. Burns et al. / Continental Shelf Research 24 (2004) 2373–2394 2377
Final extracts from all samples were reduced involume using a rotary evaporator and transferredto graduated glass centrifuge tubes with Teflonstoppers. Final volume reduction was made with agentle stream of high purity N2 gas. The amount ofextractable organic matter (EOM) was determinedby gently evaporating 10 ml aliquots from a knownvolume of the extracts onto the pan of anelectronic microbalance. All crude extracts wereanalysed by ultraviolet fluorescence (UVF) todetermine if any aromatic hydrocarbons werepresent (UNEP, 1992). Hexane dilutions of a lightcrude oil from the NWS oil fields (Harriet A) wasused to construct a calibration graph for the UVFanalysis.
2.5. Lipid class separation
Sediment trap and water sample extracts wereevaporated to just dry, using a gentle stream ofultra pure N2 gas, and then saponified in 5%KOH/CH3OH solution under nitrogen with gentleheating at 60 1C for 30min. After cooling, theneutral lipids were extracted with hexane/chloro-form (4:1V/V). After acidification, the CH3OHwas again extracted to obtain the fatty acids (F5),which were then methylated to produce thecorresponding fatty acid methyl esters (FAMES).The neutral lipids were separated on a smallcolumn of 5% deactivated silica gel into hydro-carbon (F1+F2), ester (F3) and alcohol (F4)fractions (Ehrhardt and Burns, 1999b).
2.6. Instrumental analysis
Initial analysis at the AIMS laboratories of thehydrocarbon composition of the non-polar frac-tions was performed by full scan GC-MS (m/z
50–500) using a 30m� 0.32mm� 0.11 mm J&WDB5-MS fused silica column in a Hewlett Packard6890/5972 GC-MS system. Mass spectral para-meters included ionisation energy of 70 eV andsource temperature of 280 1C. For ester andalcohol analysis by GC-MS, the scan range wasincreased to m/z 650.The extracts were also analysed by GC-FID
using a Fisons 8000 GC with an automated Grobon-column injector and fitted with a similar
column, for quantification of all peaks plus theunresolved complex mixture (UCM). In GC-MSanalyses, the GC temperature program was 701 to300 1C at 41min�1 with a 15min hold time. In GC-FID analysis the program was 50–1001 at61min�1, then to 3001 at 41min�1 with a 15minhold time. Helium was used as carrier gas for GC-MS and N2 for GC-FID analysis.Peak identifications were based on elution order
and MS spectra interpretation, including wherepossible comparison with published spectra. Sterolidentifications were based on relative retentiontimes and mass spectra according to the identifica-tion scheme suggested by Jones et al. (1989).Quantification of resolved hydrocarbons and
UCM was achieved by applying response factorsderived from C14 to C34 n-alkanes on the GC-FID.The alcohols were converted into TMS-deriva-tives. Quantification of sterols was based on aresponse factor derived from the base peak of aTMS-cholesterol standard on the GC-MS. Thebase peak areas of other sterols were quantifiedagainst the base peak area of the TMS-cholesterolstandard. The acid fraction was methylated andanalysed as FAMES by GC-FID. Quantificationof FAMES and n-alcohols was based on responsefactors derived from pure FAMES standardswithin the C11:0–C34:0 elution range as required.The F1+F2 fractions were further purified by
column chromatography on 5% AgNO3 impreg-nated silica gel (with hexane eluant) to removeunsaturated hydrocarbons comprised of clusters ofbranched alkenes. The saturated fractions werethen analysed for trace diagnostic biomarkers atthe AGSO laboratories by selected ion monitoring(SIM) GC-MS using an Agilent 6890/5973 instru-ment and by metastable reaction monitoring(MRM) GC-MS with a Carlo Erba CE8000 GCinterfaced to a Micromass Autospec (UltimaQ)MS. With the former, an HP-5 fused silicacapillary column (50m� 0.2mm� 0.11 mm) wasused with He carrier and the GC oven wastemperature programmed to increase at 2 1Cmin�1
from an initial 150 1C to final 300 1C (held for12min). Mass spectral parameters included anionisation energy of 70 eV and source temperatureof 250 1C. The MRM experiments were conductedwith a DB-5 fused silica capillary column
ARTICLE IN PRESS
enfluxes,C/N
ratios,
C/N
(M/M
)
d13C
d15N
6.3
�20.4
6.3
10.0
�20.2
4.9
6.5
�20.6
6.4
8.5
�20.6
4.9
6.0
�21.7
7.2
10.8
�21.2
5.0
K.A. Burns et al. / Continental Shelf Research 24 (2004) 2373–23942378
(50m� 0.25mm� 0.25 mm), H2 carrier gas and aGC oven temperature program of an initial 60 1C(held for 2min) increased at 4 1Cmin�1 to a final300 1C (held for 25min). These data were com-pared to other known Papua New Guinean, SouthEast Asian and Northern Australian oils.
Table1
Deploymentdays,location,depth,totalmgdry
weightrecovered,coefficientofvariation,dry
wtflux,inorganicandorganiccarbonandnitrog
stablecarbonandnitrogen
isotopedata
forsedimenttrapmaterial
DaysLocation
Depth
(m)
Total
(mg)
Dry
wt.
(mgflux)
CVaPIC
(mgg�1)
PIC
(mgflux)
PIC
(mM
flux)
POC
(mgg�1)
POC
(mgflux)
POC
(mM
flux)
PN
(mgg�1)
PN
(mgflux)
PN
(mM
flux)
50
Fly
Slope
300
1627
115
0.25
19
2.2
0.18
144
16.6
1.4
26.9
3.09
0.22
50
Fly
Slope
960
3278
232
0.17
41
9.4
0.78
39
9.0
0.7
4.5
1.04
0.07
44
Kerem
a300
1127
181
0.35
13
2.4
0.20
80
14.5
1.2
14.3
2.59
0.18
44
Kerem
a870
7142
574
0.13
20
11.3
0.94
31
17.9
1.5
4.3
2.47
0.18
29
SEFields
300
970
118
0.17
b191
22.5
1.9
37.1
4.38
0.31
29
SEFields1360
868
106
0.10
b64
6.8
0.6
6.9
0.73
0.05
mM
ismmol,M
isMolar.
Fluxes
are
inmgm
�2d�1ormmolm
�2d�1.Other
unitsasshown.
aCVorcoefficientofvariationisthestandard
deviationresultsfrom
sixreplicatestrapsdivided
bythemean.
bParticulateinorganiccarbonwasnotdetermined
intheSEField
traps.
3. Results
3.1. Sediment trap particles
3.1.1. Dry weight and organic carbon fluxes, C/N
and stable isotopes
Dry weight fluxes from the shallow traps rangedfrom 115 to 181mgm�2 day�1 and POC fluxesranged from 1.2 to 1.9mM POCm�2 d�1 (Table1). POC values were �65% reduced from the300m to the deep traps in all cases. The molarorganic carbon to nitrogen ratios, C/N, rangedfrom 6.0 to 6.5 in the 300m traps, which is justbelow the Redfield ratio of 6.6–7.0. Fluxes in deeptraps were likely influenced by re-suspended shelfsediments and had C/N ratios ranging from 8.5 to10.8. The d13C ratios were �20.2% to �21.7% inall trap samples, consistent with marine-derivedorganic matter, similar to the sediments under-neath, as reported by Bird et al. (1995). Shallowtraps had higher d15N values of 6.3–7.2%, whilethose in deep traps were �5.0%.
3.1.2. Summary lipid biomarkers
Emphasis was placed on the lipid classesbecause these are extremely useful in sourceidentification. Table 2 shows the concentrationsand fluxes for total lipid classes in the sedimenttrap samples. The extractable organic matter(EOM) was reduced 84–91% between shallowand deep traps.The trap samples contained both resolved
hydrocarbons and an UCM of hydrocarbons,indicative of petroleum. The UCM accounted formost of the hydrocarbon content. Traces ofbiogenic markers were visible in the resolvedcomponents, as will be subsequently detailed. Trapsamples contained both fatty acid methyl estersand wax esters. Total sterol concentrations werevariable in the traps without a clear distribution
K.A. Burns et al. / Continental Shelf Research 24 (2004) 2373–2394 2379
pattern, although concentrations were highestin shallow traps. The total concentration offatty acids (FA) was highest in shallow trapsand these increased offshore. Concentrations ofFA decreased by 92–96% from shallow to deeptraps. In general, there was a one to two orderof magnitude decrease in concentrations ofall the planktonic lipid classes from shallow todeep traps.
3.1.3. F1+F2 hydrocarbons
Hydrocarbon compositions of the sediment trapsamples are summarized in Table 3. Among theresolved peaks, the n-alkenes, including C17:1,C19:1, C21:5 and C21:6, were readily recognizableas general phytoplankton biomarkers (e.g. Blumeret al., 1970, 1971; Sinninghe Damste et al., 1999).The highly branched isoprenoid alkenes, such asbrC30:5, brC25:5 and related isomers are generallyattributed to diatoms (e.g. Volkman et al., 1994;Belt et al., 2000; Schouten et al., 2000). Squalene,phytadiene and pristane were zooplankton mar-kers (e.g. Avigan and Blumer, 1968; Blumer et al.,1969; Wakeham and Canuel, 1986, 1988). Therewas an even distribution of n-alkanes, plusphytane and steranes associated with the oilcontent of the samples (e.g. Peters and Moldowan,1993).The extracts were also scanned by UVF, which
is sensitive to the aromatics and associatedoxidation products. This is a semi-quantitativemeasure of the oil in the samples as calibratedagainst a light crude oil. The technique tends toover estimate the amount of oil in some samples,especially when the oil is degraded and containsaromatic oxidation products (Burns, 1993). Ex-periments have shown that oil in seawater rapidlydegrades by both photochemical and biologicaldegradation processes (e.g. Ehrhardt et al., 1992).For example, the aromatic hydrocarbon fluoreneis quantitatively converted to its ketone, fluore-none in sunlight. Fluorenone was found in all ofthe sediment trap extracts and the ratio of theketone to the parent hydrocarbon providedevidence that the oil was photo-degraded.Gas chromatograms of oil in the Kerema
Canyon traps are shown in Fig. 2. The pattern ofthe shallow trap shows the n-alkanes are clearly
ARTICLE IN PRESS
Table 3
F1+F 2 hydrocarbons in Gulf of Papua sediment trap particulates
Degraded oil Fluorenone/fluorened ratio 0.5 0.6 1.3 4.2 2.2 1.0
aUCM is unresolved complex mixture indicative of petroleum.bHBI alkenes are highly branched alkenes made by diatoms.cUVF oil estimate is based on calibration against a light crude from the Northwest Shelf of Australia.dFluorenone to fluorene ratio is indicative of photo-degradation of oil.
K.A. Burns et al. / Continental Shelf Research 24 (2004) 2373–23942380
visible above the unresolved hydrocarbons. Thedeep trap has less n-alkanes compared to the UCMand represents a more degraded pattern. Bothtraces show a significant content of methylphe-nanthrenes.After further purification, the saturated hydro-
carbons were analysed by MRM–GC–MS todetermine the composition of hopane and steranepetroleum biomarkers. These high molecularweight polycyclic hydrocarbons are often highlyspecific to the source of the oil, which combinedwith a high resistance to biodegradation makesthem particularly useful diagnostic markers (Petersand Moldowan, 1993). Fig. 3 shows the ion tracesfor the hopane (m/z 191) and sterane (m/z 217)biomarkers in the SE Fields 300m trap sample.Peak identifications are given in Table 4a. Severalbiomarker ratios, useful for source identification,are listed in Table 4b. Each of the six sediment trapsamples showed very similar biomarker patterns,indicative of a carbonate-based oil. These includehigh norhopane to hopane (e.g., C29H/C30H41), ahigh proportion of C27 steranes compared to C29
steranes, low diasteranes/steranes and a relativelylow proportion of diahopanes compared to theirunarranged (regular) analogues (Peters and Mol-dowan, 1993). On the basis of these biomarkerdistributions, no differentiation between the sixsediment trap samples was possible.
Further compositional information on the sedi-ment trap hydrocarbons was ascertained from theseries of alkylated aromatic hydrocarbons. Fig. 4shows the distribution of the aromatic hydrocar-bons normalized against the C30-hopane based onpeak areas from reconstructed ion current (RIC)peak areas determined by SIM GC–MS. KeremaCanyon samples contained a much higher percen-tage of the aromatic hydrocarbons and they werethe only samples that contained any of the fourringed chrysene series.Also detected, were cadalene (C), retene (R) and
perylene (P) which are 2, 3 and 5 ringed aromatichydrocarbons of land-derived sources (Peters andMoldowan, 1993). Benzopyrene is from forestfires. These markers were present in the traps nearthe shelf edge but low or absent in traps set in thenorthern Coral Sea.
3.1.4. F3 esters
The FAMES and wax esters obtained in F3 arederived from zooplankton, and possibly fromsome marine bacterial lipids (Boon and de Leeuw,1979; Wakeham, 1985; Rontani et al., 1999). Table5 shows that the FAMES and wax esters werepresent in all trap samples with no discernabledifference in the patterns of abundance.The alkenones are made by the planktonic
Fig. 2. Total Ion Current (TIC) reconstructed chromatograms of the hydrocarbon fractions from the Kerema Canyon sediment trap
samples showing the relatively degraded petroleum pattern of unresolved complex mixture, with some methylphenanthrene, alkane and
biogenic hydrocarbon peaks visible.
K.A. Burns et al. / Continental Shelf Research 24 (2004) 2373–2394 2381
1980; Marlowe et al., 1984). Identifications werebased on published spectra (de Leeuw et al., 1979;McCaffrey, 1990). We found only the C37:2 ethyl-and methyl- and the C38:2 methyl-alkenones. Thesewere highest near the shelf and found mostly in theshallow traps. Concentrations in deep traps were46–70% less.
3.1.5. F4 alcohols
The detailed composition of the alcohol fractionis given in Table 6. The sterols consisted mostly ofmarine phyto- and zooplankton biomarkers andwere highest in shallow traps and decreased in
deep traps. The steroid ketones are formed bypassage through crustacean guts (Wakeham andCanuel, 1986, 1988). We found only two steroidketones, which were two of the most abundantsteroid ketones reported in sediment trap materialfrom the Atlantic (Gagosian et al., 1982). Theseketones were a significant portion of the trapsterols and were not found in the water samples.Source information for sterols was reviewed byVolkman (1986). Tetrahymanol was identifiedfrom spectra in Simoneit (1977) and ten Havenet al. (1989) and is considered to be a land deriveddiagenesis product (Peters and Moldowan, 1993).
ARTICLE IN PRESS
Ts Tm
C28(29, 30)
C29H
C30H
C29Ts
C29βα 30-nor
C31H(S+R)
C32H(S+R)
C33H(S+R)
C27 20S
C27 20R
C27 20S
Retention Time
C30βα
C35H(S+R)
C34H(S+R)
C27 20R
C27 20R+S
C28 20R+S
C29 20R+S
C29 20S
C29 20R
S E Fields300 m
αββαββ
αββ
βα
βα
αααααα
αααααα
(a)
(b)
Fig. 3. Reconstructed Ion Chromatograms (RIC) for the m/z 191 hopane and m/z 217 sterane petroleum biomarkers often used for
source identification. Labelled peaks are listed in Table 4A.
K.A. Burns et al. / Continental Shelf Research 24 (2004) 2373–23942382
Phytol is derived from the side chain of chlor-ophyll and the C26–C34 even carbon n-alcohols arefrom higher terrestrial plants (Eglinton andHamilton, 1967). The phytol and n-alcohols werehighest in the shallow traps and decreased to nearnon-detectable in the deep traps. This data issummarized in Table 7 to illustrate the sourceinformation for the alcohol lipid class.
3.1.6. F5 fatty acids
The fatty acids consisted of marine phyto- andzooplankton biomarkers (Table 8). Within theshallow traps, most individual fatty acids in-creased from the traps near the shelf to the trapfurther offshore. However, concentrations in deeptraps were dramatically reduced. Based on thehigh content of polyunsaturated fatty acids(%PUFA and C18:3o6+C22:6o3), the relative con-centrations of the C18:0 oo C18:1 and C16:0�C16:1,and the very low ratios of phytoplankton markersto zooplankton markers (C20:5o3/C16:0 and C16:1o7/
C16:0), the organic matter was more similar tozooplankton than phytoplankton lipids (Kattnerand Krause, 1989; Cripps et al., 1999). The C18:1o7
isomer has been linked to bacterial origin (Matsu-da and Koyama, 1977) and the ratio of C18:1o7/C18:1o9 has been used to estimate the contributionof bacterial lipids in sediments (e.g. Shi et al.,2001). The iso- and anteiso- C15:0 and C17:0 are alsobacterial markers, as are other branched and cyclicfatty acids (e.g. Perry et al., 1979). All of thesebacterial biomarkers were low in the trap samplesindicating minimal bacterial contribution. Thismay result from the HgCl2 that was added as apreservative. Thus alteration of the fatty acidsfrom the photic zone to the shallow and deep trapsmust have occurred during the downward passage.The high ratio of F3 FAMES to the F5 fatty acidswould indicate that the organic carbon in the deeptrap had undergone much more alteration byzooplankton digestion as well as by bacterialdecomposition than that in shallow traps.
ARTICLE IN PRESS
Table 4
(a) Major hopane and sterane biomarkers detected by GC–MS analysis of the sediment trap samples (peak assignments as in Fig. 3)
Sum C26:C34 even n-alcohols (mg g�1 drywt.) 40.9 0.0 16.3 0.6 19.6 0.3
K.A. Burns et al. / Continental Shelf Research 24 (2004) 2373–2394 2385
3.2. Water samples
3.2.1. F1+F2 hydrocarbons in water samples
To determine if the oil in the sediment traps wascoming from river sources, Infiltrex II watersamplers were used to collect particulate anddissolved organic matter from the rivers, andnear-surface water from the shelf in the GOP. The‘‘dissolved’’ and ‘‘particulate’’ hydrocarbons wereall resolved peaks that could be identified (Table9). Some of these were known phytoplanktonmarkers such as C21:5 and C21:6 plus the C25 andC30 highly branched (HBI) alkenes attributed to
diatoms. Zooplankton markers included pristaneand squalene. However, there was also a homo-logous distribution of n-alkanes with a carbonpreference index (CPI) near to 1. This wouldindicate some trace oil content. Thus, GC–MS wasused to search for the individual aromatic hydro-carbons and quantify the alkylnaphthalene andalkylphenanthrene series of aromatics. Heavierringed aromatics were not detected in the watersamples. The extracts were scanned by UVF andthere was a statistically significant correlation(po0.05) between the UVF estimate and thequantity of aromatics determined by GC–MS.
ARTICLE IN PRESS
Table 7
Summary content of F4 alcohol content of sediment trap particles as % of total sterols or as mg g�1
Likely Source Biomarker Fly Slope Kerema Canyon SE Fields
Land plants C26–C34 even n-alcohols mg g�1 41 16 1 20 Tr
Tr means o1mg g�1, but visible in the GC-MS ion traces.aReviewed by Volkman (1986).bVenkatesan (1989), Peters and Moldowan (1993).cWakeham and Canuel (1986).
Table 8
Summary fatty acid concentrations (mg g�1), plus some diagnostic ratios and relative abundances in Gulf of Papua sediment trap
samples
Fly Slope Kerema Canyon SE Fields
300m 960m 300m 870m 300m 1300m
S Saturated FA 118 6 284 8 477 62
S MUFA 193 10 541 12 1038 89
S PUFA 119 3 275 6 525 26
S Branched FA 63 1 144 2 174 7
18:3o6+22:6o3 40 1 90 2 209 9
16:1o7 41 3 103 3 229 22
16:0 68 4 159 5 293 36
i-+ai�15:0+i�+ai�17:0 4 0.3 11 0.4 22 3
% PUFA 24 16 22 20 24 14
% Branched 13 5 12 7 8 4
% 18:3o6+22:6o3 8 6 7 7 9 5
% 16:1o7 8 13 8 12 14 12
% 16:0 14 18 13 17 13 20
% 18:0 of 18:0+18:1 11 8 10 11 10 11
% i-+ai-15:0+i-+ai-17:0 1 1 1 1 1 1
20:5o3/16:0 ratio 0.5 0.2 0.4 0.4 0.6 0.2
16:1o7/16:0 ratio 0.6 0.7 0.6 0.7 1.0 0.6
18:1o7c/18:1o9c ratio 0.1 0.1 0.1 0.1 0.1 0.1
MUFA are monounsaturated fatty acids.
PUFA are polyunsaturated fatty acids.
Branched are iso- and anteiso- and some unidentified fatty acids.
18:1o7c co-eluted with 18:1o9t but we assumed the trans isomer was negligible.
K.A. Burns et al. / Continental Shelf Research 24 (2004) 2373–23942386
Fluorenone was found in all the water samples.The ketone to aromatic ratio increased offshore.Thus, the rivers may contain traces of petroleum,
but this is unlikely to be a source to the traps. Thewater samples did not contain the same oilcomposition as the traps. Water samples lacked
ARTICLE IN PRESS
Table 9
Summary of F1+F2 hydrocarbons in GOP water samples of dissloved (D=XAD-2) and particulate (P=GFF filters) fractions
Location salinity Kikori R 0% Fly R 0% Fly R 6% Fly R 12% Fly R 18% GOP 27%
D P D P D P D P D P D P
Total HC (ngL�1) 274 57 383 91 761 38 28 8 604 120 361 148
aCPI is carbon preference index calculated as (C27+C29+C31+C33)/(C26+C28+C30+C32).bSum of alkylnaphthalenes and alkylphenanthrenes (only found in dissolved phase).cUVF estimate based on calibration with light crude oil from Harriet A platform on Northwest Shelf of Australia.dFluorenone/fluorene ratio is indicative of photo-oxidation of oil.
K.A. Burns et al. / Continental Shelf Research 24 (2004) 2373–2394 2387
the series of m/z 191 hopane and 217 steranebiomarkers, which are typical of a crude oilcontent. Thus the trace oil residues from the riverswould likely be a light fuel and would be photo-and bacterially degraded during water flow off-shore. The traps were set more than 200 kmoffshore from the rivers.
3.2.2. F4 alcohols in water samples
The sterols in the water samples containeda much higher percentage of the phyto-plankton biomarkers from both diatoms anddinoflagellates than trap samples (Table 10). Therewere no steroid ketones in water samples, as hadbeen seen in trap samples. Phytol and then-alcohols were highest in the rivers and decreasedoffshore.
3.2.3. F5 fatty acids in water samples
Summary fatty acid data for the water samplesare given in Table 11. Values of total FA werehighest in the Kikori and Purari rivers. Within theFly estuary, concentration of total FA generallyincreased from the 0% salinity water out to theGOP at 27%. The Kikori was the only sample tocontain the high molecular weight (C26–C36) evencarbon number chains indicative of terrestrialplant matter. The C16:0 and C18:0 were in verylow concentration and were non-detectable in
some samples, compared with the predominantmonounsaturated counterparts. This means thefatty acids were freshly derived from phytoplank-ton (e.g. Cripps et al., 1999). The C16:1o7 is adiatom marker, and is in much higher relativeproportion than in the trap samples. The branchedbacterial markers and the C18:1o7/C18:1o9 ratioswere low in water samples.In the low salinity river samples, a significant
fraction of the higher molecular weight fatty acidswere found on the glass fiber filters. However, athigher salinities, a much lower fraction wasretained on the filters. Berge et al. (1995) showedthat living phytoplankton do not contain signifi-cant quantities of free fatty acids. Rather, the acidsare contained within intact membrane lipids. Thusto detect such a high content of polyunsaturatedand other fatty acids in the ‘‘dissolved phase’’ theplankton cells must have lysed and their lipidspassed through the GFFs.Further contrast with the trap samples is the
very small content of F3 FAMES in watersamples. This is despite the fact that the solventmixture used for extraction of the XAD-2 resinwas a 50% mix of methanol and methylenechloride. There was concern that this mixturewould methylate the fatty acids during extractionof the organics off the resin. But this does notappear to have happened.
ARTICLE IN PRESS
Table 10
F4 alcohols found in GOP river and estuary water samples (XAD-2 dissolved phase)
K.A. Burns et al. / Continental Shelf Research 24 (2004) 2373–23942388
4. Discussion
4.1. Mass and POC fluxes
Total mass fluxes determined in the GOPsediment traps were similar to those reported forother tropical or semi-tropical coastal continentaland ocean areas. Noriki and Tsnunogai (1986)summarized total shallow particulate mass fluxesin the deep water column of the Pacific off Hawaiias 6–82mgm�2 d�1; 300–420mgm�2 d�1 off Cali-fornia, and 790–1200mgm�2 d�1 in the Antarctic.Our mass fluxes of 115–181mgm�2 d�1 in the300m traps were slightly less than the coastalCalifornia range but slightly higher than theoceanic Pacific fluxes. Heussner et al. (1999)
reported mean seasonal total mass flux of�500mgm�2 d�1 in traps set at 380m off theBay of Biscay. These authors showed inputs todeeper trap fluxes by sediment from the uppershelf. The increases in mass flux in the deep trapsat the shelf edge of the GOP (115–232 and181–574mgm�2 d�1) indicates input from re-sus-pended shelf sediments. There appeared to be littleinfluence from shelf sediments in traps set in thenorthern Coral Sea (118–106mgm�2 d�1).Lee et al. (1998) summarized the POC flux from
the US JGOFS studies in the Arabian Sea. Fromtheir Fig. 2, a trap set at 200–300m depth wouldshow an average organic carbon flux of1–5mMm�2 d�1. The average POC flux in the380m depth traps in the Bay of Biscay was
ARTICLE IN PRESS
Table 11
Summary fatty acids (ngL�1) detected on GFF filters and XAD-2 columns from the Infiltrex II samplers in water samples from the
Kikori, Purari and Fly rivers and the Gulf of Papua, plus some diagnostic ratios and percentages
Branched are iso- and anteiso- and some unidentified fatty acids.
ND is not determined because the GFF sample was lost during analysis.
N means no 16:0 was detected so ratio is invalid.
K.A. Burns et al. / Continental Shelf Research 24 (2004) 2373–2394 2389
�2.3mMm�2 d�1 (Heussner et al., 1999). Our300m POC fluxes ranged from 1.2 to1.9mMm�2 d�1. Thus it appears that even thoughour study was a single ‘‘snap shot’’ in the GOP, weobtained POC fluxes within the ranges of othercomparable long-term studies. Comparison usingthe same traps is possible from similar work on theNorthwest Shelf of Australia, in which POC fluxesfrom 200 to 300m traps ranged from 2.0 to3.5mMm�2 d�1 (Burns et al., 2003).
4.2. Alterations during descent
Altabet (1988) determined that the d15N value ofsurface ocean phytoplankton was about –1% insamples from the Atlantic. Several authors havedescribed an average increase of �3.4% in theisotopic value with passage through the food chain
(Minagawa and Wada, 1984; Checkley and Miller,1989). This study did not measure d15N insuspended phytoplankton. However, using an aver-age of 1% for phytoplankton, and the values for the300m traps, we can estimate that the POC in theGOP traps had been reprocessed by zooplankton�2 times during the descent from the photic zone.Nakatsuka et al. (1997) suggested that decreasingd15N values in deep sinking POC was likely due tothe preferential removal of nitrogen rich organicmatter during bacterial remineralization. Aminoacids are examples of labile nitrogen-rich POCcomponents (Lee and Cronin, 1984). Altabet (1988)estimated this change in d15N to be between �1.5%and �2.1%. The change from shallow to deep trapsin this data set ranged from –1.4% to �2.2%.The organic biomarkers in the ester, sterol and
fatty acid classes confirmed the POC was altered
ARTICLE IN PRESS
K.A. Burns et al. / Continental Shelf Research 24 (2004) 2373–23942390
by zooplankton reprocessing and bacterial decom-position during descent. This was seen in the orderof magnitude decreases in concentration of EOM,FA and sterols between shallow and deep traps.The F3 ester fraction/F5 fatty acid fraction rangedfrom 0.3 to 1.3 in shallow traps and 6.7–16.2 indeep traps consistent with increased zooplanktonlipid contribution in the deep traps. Phytol and theC26–C34 n-alcohols were found mostly in surfacetraps with only traces visible in deep traps.
4.3. Biogenic lipid fluxes
The GOP fluxes can be compared with indivi-dual biomarker results summarized for the Ara-bian Sea study (Prahl et al., 2000). This long-termstudy was conducted in tropical waters on thewestern edge of the Indian Ocean. The GOP fluxesranged from 0.04 to 0.31 mgm�2 d�1 for HBIalkenes, from 1.9 to 5.5 gm�2 d�1 for C37–C38
alkenones, and 0–3.9 mgm�2 d�1 for dinosterol.Fluxes of these lipids reported by Prahl et al.(2000) vary widely over the seasons but generallybracket these values. Thus despite the vastseasonal changes detected in the Arabian Gulfstudy, the GOP data showed a similar mixture ofthe three classes of biomarkers for diatoms,haptophytes and dinoflagellates. Oil interferedwith the estimation of C27–C31 alkanes to be usedas terrestrial markers, which were reported in theArabian Gulf study. We did however detectterrestrial biomarkers such as the C26–C34 n-
alcohols, in trace amounts in water samples andthe shallow traps set near the shelf edge indicatingtrace input of terrestrial organic matter whichcould have reached the traps through river out-flows or from atmospheric transport.
4.4. Petroleum hydrocarbon fluxes
The presence of petroleum hydrocarbons hasbeen previously reported in many sediment trapstudies. Crisp et al. (1979) reported fluxes of UCMhydrocarbons that ranged from 7.2 to173.7mgm�2 yr�1 in four basins off the Californiacoast. Based on a long-term sediment trapstudy in the coastal Mediterranean near Monaco,Burns et al. (1985) reported UCM hydrocarbon
fluxes of between 8 and 223 and averaging90766mgm�2 yr�1. In a separate study, Saliot etal. (1985) estimated the water-column sinking rateof petroleum hydrocarbons at four stations in thewestern basin by analysing particles (463 mm)from below the photic zone and by applying theparticle-settling model of McCave (1975). Theseflux estimates ranged from 3 to 77 and averaged35733mgm�2 yr�1. A further estimate of petro-leum flux to the seabed under 2900m water depthin the western basin of the Mediterranean wasbased on sediment cores and calculated to be152mgm�2 yr�1 (Burns and Villeneuve, 1983;Burns and Saliot, 1986). More recent estimatesof hydrocarbon flux based on sediment trap fluxesin the Southern Ocean were reported by Ternois etal. (1998). These authors report the presence ofUCM hydrocarbons visible only during times ofrelatively low biogenic fluxes and they reportedalkane fluxes from 36 to 3029, which averaged547784 mgm�2 yr�1. Thus, the UCM fluxes calcu-lated in the present work (i.e. 6–36mgm�2 yr�1)are less than those reported from relativelycontaminated areas off the California coast andin the Mediterranean. But the GOP alkane fluxesranged from 161 to 3137 mgm�2 yr�1 or similar tothose seen in the Southern Ocean. Estimates ofUCM flux on the Northwest Shelf of Australiaranged from 38 to 87mgm�2 yr�1 (Burns et al.,2001).
4.5. Oil source
The n-alkane content of the oil in the trapsamples is too waxy to be a lubricating oil and it istoo weathered and microbiologically degradedwith missing alkanes to be a contamination bythe ship’s fuel. Thus we believe the oil content tobe indicative of the local environment. The steraneand hopane biomarkers detected in each of thesediment traps were characteristic of a marinecarbonate sourced oil (Peters and Moldowan,1993). The most common PNG oils are from thePapuan fold belt. They are characterized byabundant diahopanes, neohopane and retene,reflecting an origin from Jurassic sediments withmixed marine and terrestrial organic matter, notunlike the oils of the Australian NW shelf (Murray
ARTICLE IN PRESS
K.A. Burns et al. / Continental Shelf Research 24 (2004) 2373–2394 2391
et al., 1993). Petroleum seeps discovered on theNew Guinea Mobile Belt reflected an additionaloil source, of Canozoic age, characterized byabundant higher plant biomarkers including olea-nane and bicadinane (Murray et al., 1993).However, carbonate sourced oils are not alien tothe region. The oils of Seram (Price et al., 1987)and several coastal bitumens (Summons et al.,1993; e.g., Barclay Point, Bathurst Is.) from theNorthern Territory, Australia are examples of oilscontaining biomarker features consistent withcarbonate sourced oils, including high C29-norho-pane/C30-hopane ratio and relatively high C29-steranes/C27-steranes. The aromatic content of theGOP samples suggested at least two types of oil inthe traps, with the Kerema Canyon traps having amuch higher aromatic hydrocarbon content thanthe oil in the other traps. This could indicatemultiple oil sources with different aromatic hydro-carbon compositions over the GOP area.The occurrence of the petroleum signal over
several hundred square kilometers with no obvioussource, suggests natural seeps as a likely source ofthe oil in the traps, as was also discussed in Burnset al. (2001). There is evidence of widespreadnatural seepage of hydrocarbons from submarinepetroleum formations on the Northwest Shelf ofAustralia and Timor Sea (O’Brien et al., 1998;Cowley and O’Brien, 2000). Thus, even in remoteand seemingly pristine areas, oil seeps influence thecycling of lipids in many coastal oceans. Our datasuggests that this oil does not just float in tar balls,but settles in zooplankton faeces along with thebiogenic components and undergoes the sameprocesses of degradation.
5. Conclusions
Gulf of Papua mass and POC fluxes were typicalof other continental slopes or marginal oceans atsimilar latitudes. Stable carbon isotope analysisshowed that most of the POC was marine-derived.Changes in C/N and d15N values suggests theparticulate N was recycled through two trophictransfers before reaching the 300m traps and thatthere was further microbial degradation thatselectively removed N during transit to deep traps.
There was also entrapment of re-suspended sedi-ments in traps set on the continental slope. Lipidbiomarker analysis showed the rapid degradationof several classes, including the n-alkanes, betweenshallow and deep traps. Lipid analysis also showedthat a significant fraction of the lipid classes of thetrap particulates were of zooplankton origin, incontrast to river water and inner shelf seawatersamples which were dominated by phytoplanktonlipids.Lipid analysis confirmed that the major source
of the organic carbon in the GOP was marineproduction, with an input of fossil petroleum, andtrace land–plant input potentially from river flow,plus benzopyrene which was likely due to aeoliantransport and deposition from fires.There was a significant flux of petroleum in the
GOP, in the range of approximately half of thatmeasured on the Northwest Shelf of Australia.The oil appeared to be of marine carbonate source,was weathered and degraded, and was comprisedof at least two types based on the content ofaromatic hydrocarbons. The rivers are an unlikelysource for oil on the shelf edge and beyond. Withno other obvious sources, natural oil seeps areimplicated.Since much of the hydrocarbon and fatty acid
lipid classes quantified in water samples passedthrough the GFF filters and were retained on theXAD-2 resin, this indicates that much informationon the lipids of natural plankton assemblages ismissed, if only the particulate fractions areanalysed.
Acknowledgements
We thank the Masters and Crews of the R/VFranklin and R/V Harry Messel, during the 1997studies on the TROPICS cruises to PNG. AIMSWorkshop built the sediment traps according todesign by Gregg Brunskill and Kathryn Burns.John Soles, Cary McLean (AIMS) and IanHelmond (CSIRO) did mooring design, deploy-ments and recoveries. Special thanks to GeorgeCresswell (CSIRO) who altered his cruise plans toretrieve the traps after the weather provedimpossible on our earlier cruise. John Pfitzner of
ARTICLE IN PRESS
K.A. Burns et al. / Continental Shelf Research 24 (2004) 2373–23942392
AIMS helped on the ships and in the laboratory.Ron Szymczak of ANSTO collected the watersamples. Peter Nichols and John Volkman ofCSIRO, Hobart, kindly provided calibrationstandards and advice for the analysis of marinefatty acids and sterols. Stuart Wakeham graciouslyprovided RT and GC–MS spectral data from theVERTEX II studies, so that we could confirm theidentities of the sterol ketones in our samples. Wethank three anonymous reviewers for suggestionson revision. This is AIMS Contribution No. 1074.PFG publishes with the permission of the Execu-tive Director, GA.
References
Altabet, M.A., 1988. Variations in nitrogen isotopic composi-
tion between sinking and suspended particles: implications
for nitrogen cycling and particle transformation in the open
ocean. Deep-Sea Research 35 (4), 535–554.
Avigan, J., Blumer, M., 1968. On the origin of pristine in
marine organisms. Journal of Lipid Research 9, 350–352.
Ayukai, T., Wolanski, E., 1997. Importance of biologically
mediated removal of fine sediments from the Fly River
plume, PNG. Estuarine, Coastal and Shelf Science 44,
629–639.
Belt, S.T., Allard, W.G., Rintatalo, J., Johns, L., van Duin,
A.C.T., Rowland, S.J., 2000. Clay and acid catalysed
isomerisation and cyclisation reactions of highly branched
isoprenoid (HBI) alkenes: implications for sedimentary
reactions and distributions. Geochimica et Cosmochimica
Acta 64 (19), 3337–3345.
Benner, R., Biddanda, Black, B., McCarthy, M., 1997.
Abundance, size distribution, and stable carbon and
nitrogen isotopic compositions of marine organic matter
isolated by tangential-flow ultrafiltration. Marine Chemistry
