Early diagenetic cycling, incineration, and burial of sedimentary organic carbon in the central Gulf of Papua (Papua New Guinea)

Authors


Abstract

[1] The clinoform complex of the Gulf of Papua represents a major deltaic system in Oceania. Two seasons largely control seafloor dynamics and sedimentary C cycling: the relatively quiescent NW monsoon, and the SE trades, characterized by remobilization and reoxidation of topset deposits. Surface sediments (∼20 cm) are reactive with ΣCO2 production fluxes ∼35–42 mmol m−2 d−1 at mangrove channel and topset sites during the monsoon, and ∼10–20 mmol m−2 d−1 on the foreset-bottomset (>40 m). Fluxes decrease by a factor of ∼0.3 on the topset during the transition period and trades. The 13,14C isotopic compositions of pore water ΣCO2 reveal diagenetic fractionation, with dominant utilization of young (Δ14C = 1.4–31.1‰), terrestrial C substrates inshore (channels, topset δ13C = −29 to −25‰) and a progressive increase of young marine C sources seaward (outer topset, foreset; bottomset δ13C = −22.2 to −19.5). Remineralization patterns of terrestrial and marine Corg demonstrate cross-shelf exchange. Multiple tracers show that a suboxic, mobile mud layer, ∼10–60 cm thick (usually ∼10–30 cm), characterizes the central gulf topset and Umuda Valley off the Fly River and unconformably overlies methanic deposits releasing old ΣCO214C = −159 to −229‰). Residual terrestrial Corg delivered to the bioturbated foreset continues to be remineralized slowly, generating ΣCO2 having net Δ14C = −270 within sediments deposited 100–200 years ago. The reactivity of Corg below ∼0.5 m in the foreset is ∼10–20 times lower than expected based on accumulation rates, reflecting loss of >50% of sedimentary Corg on the topset, which functions as a suboxic incinerator.

1. Introduction

[2] Deltaic systems are widely recognized as the major storage sites for inorganic sedimentary debris and organic carbon along continental boundaries [Berner, 1982; Hedges and Keil, 1995; Burdige, 2007]. They are also regions of elevated primary production, intense biogeochemical cycling, and dynamic refluxing of material between multiple depositional and diagenetic facies. Understanding the processing, addition, alteration, and burial of riverine, wetland, and marine derived material within these complex sedimentary ecosystems remains a central goal of coastal ocean biogeochemistry [Benner, 2004; McKee et al., 2004; Blair et al., 2004; Goñi et al., 2005]. Tropical deltas are of particular interest because the tropics supply ∼60 ± 10% of the global river water and sediment delivery to coastal lowlands and the ocean, and a comparable percentage of the riverine particulate and dissolved organic carbon flux [Milliman and Meade, 1983; Alongi, 1998; Meybeck, 1993; Ludwig et al., 1996; Schlünz and Schneider, 2000; Jennerjahn and Ittekkot, 2002]. Because of numerous mountainous rivers with high drainage basin yield, tectonically active Oceania accounts for roughly half of the tropical zone fluxes, with the island of New Guinea alone supplying as much as 1.5× the amount of sediment and organic C (Corg) derived from the much larger Amazon basin [Milliman, 1995; Milliman et al., 1999; Bird et al., 1995; Lyons et al., 2002].

[3] One of the major deltaic systems in Oceania is located in the Gulf of Papua, a semicircular embayment of the continental shelf on the south coast of Papua New Guinea (Figure 1). The coalescence of the Fly, Bamu, Turama, Kikori, and Purari rivers along the northern boundary of the gulf results in a broad swath of mangrove forests and the progradation of deltaic deposits into energetic coastal waters of the Coral Sea [Harris et al., 1993, 1996; Robertson et al., 1998; Walsh and Nittrouer, 2004; Walsh et al., 2004]. The Gulf of Papua system thus combines high-yield mountainous river source regions, which are typical of tectonically active margins known to supply a high proportion of refractory Corg to the ocean, with a broad shelf and energetic deltaic depocenter, which are often associated with passive margins and the efficient diagenetic remineralization of sedimentary Corg [Blair et al., 2004; Komada et al., 2004; Leithold et al., 2006; Aller and Blair, 2006]. This union of conditions has unusual potential to provide new insights into the factors controlling the cycling and burial of both old refractory and recent labile sedimentary Corg in the coastal ocean.

Figure 1.

Location are plotted of stations sampled seasonally in the Gulf of Papua during the monsoon through trades seasons 2003–2004 and during previous studies in 1997–2000 at the same or nearby sites. The primary study transect, GH, lies between transects G and H which were established in 2000, along with transects C → J, and sampled from the R/V Franklin. Stations indicated by HM (squares) were sampled from the R/V Harry Messel during 1999, sites HM50 and HM13 of which are comparable to GS48 and GH14 of the present study [Aller et al., 2004; Aller and Blair, 2004]. The bottom right inset shows inshore-offshore bathymetric profiles along G and H, extrapolated to the shoreline, illustrating the prograding clinoform character of the system [Walsh et al., 2004]. Primary wind directions during the NW monsoon and SE trades periods are indicated. The top right inset shows the location of the Gulf of Papua relative to the island of New Guinea and Australia.

[4] In the present study, we substantially extend previous quantitative investigations of the relationship between depositional environment, early diagenetic cycling, and preservation of Corg in the central Gulf of Papua clinoform delta. Our primary purposes are to constrain further the overall patterns of remineralization and net burial of terrestrial and marine Corg in the subaqueous deposits, to examine seasonal variability, to refine conceptual models of the primary factors controlling these patterns, and to construct facies-specific sedimentary Corg cycling budgets. In addition to direct measurements of remineralization rates and Corg fluxes, we utilize extensive new data on 13C/12C and 14C/12C isotopic distributions of pore water ΣCO2 and solid Corg, and diagenetic modeling to infer source, reactivity, and fate of sedimentary Corg in the different major facies of the deltaic complex. It is demonstrated that a wide spectrum of terrestrial soil, mangrove (vascular plant), and marine planktonic Corg debris introduced to the central gulf is remineralized on the shallow clinoform topset and within active channels, apparently as a consequence of sediment refluxing coupled with oxic/suboxic diagenesis. Although both terrestrial and marine-derived Corg substrates are decomposed throughout the clinoform, there is a progressive increase of marine substrate utilization offshore, accounting for nearly 100% of remineralization in the bottomset (75 m). Diagenetic fractionation of Corg substrates during remineralization is the general rule with preferential loss of young relative to old Corg; however, slow remineralization of aged Corg (>2000 years) is evident throughout the central gulf, and a wide range of labile and refractory Corg (>modern to >4800 years) can be decomposed within migrating mangrove channel deposits. The residual sedimentary Corg escaping seaward from the dynamic, shallow water incineration zone is relatively refractory. As a result, bulk Corg supports lower rates of decomposition in the delta foreset than might otherwise be expected based on the rapid accumulation rates that occur there and on previously established correlations between Corg reactivity and sediment accumulation in nondeltaic environments.

2. Study Area

[5] The rivers entering the Gulf of Papua are estimated to deliver ∼384 Mt a−1 of sediment and 470–690 km3 fluid per years into lowland and coastal areas (Mt = 106 t) [Milliman et al., 1999; Pickup and Chewings, 1983; Salomons and Eagle, 1990; Wolanski et al., 1995]. The Fly is the largest of these rivers and supplies ∼85–115 Mt a−1 and 190–220 km3 fluid a−1. These sediment/water delivery ratios (Mt km−3) are about 4–10× higher than other major tropical rivers such as the Amazon and Congo, reflecting the high-relief drainage basin [Milliman and Meade, 1983; Milliman and Syvitski, 1992]. Although river runoff is relatively constant seasonally, major interannual decreases occur during El Niño drought conditions [Pickup and Chewings, 1983; Wolanski et al., 1984, 1995; Moi, 2001; Walsh et al., 2004]. Such climatic events may significantly affect sediment storage and delivery patterns [Ogston et al., 2008]. The quantity of sediment actually making it into the estuarine zones and seaward is uncertain and may be substantially lower than the estimated drainage basin yields and gauged transport, likely averaging ∼156 Mt a−1 over decadal timescales [Brunskill et al., 2007a, 2007b]. This latter estimate of the realized sediment supply to the gulf is consistent with sediment accumulation budgets in the mangrove forests, delta plain, and shelf, which can account for ∼138 Mt a−1 [Walsh et al., 2004; Walsh and Nittrouer, 2004; Brunskill et al., 2003, 2007a, 2007b].

[6] The shelf extends up to ∼150 km in the central gulf, narrowing eastward toward Kerema and westward into the Torres Strait, both of which are regions characterized by coral reefs. Most sediment accumulates on the inner shelf of the central gulf in <50 m of water as part of an overall prograding clinoform deposit [Harris et al., 1996; Walsh et al., 2004]; however, off the Fly, a portion moves seaward within incised relict river channels such as the subaqueous Umuda Valley [Martin et al., 2008]. Approximately 65% of the sediment flux accumulates in 30% of the inner shelf area between ∼30 and 50 m depth within the clinoform foreset. Net accumulation rates on the foreset are ∼1–4 cm a−1 (1–3.8 g cm−2 a−1). Roughly 25% of the flux accumulates in the broad, low-gradient topset between 0 and 20 m, representing nearly 70% of the inner shelf area [Brunskill et al., 2003; Walsh et al., 2004]. The upper 10–100 cm of silty mud deposits in the topset are highly reworked and mobile (usually 10–30 cm), making estimates of accumulation in this zone uncertain. Mangrove forest accretion apparently accounts for ∼5–10% of the sediment flux [Walsh and Nittrouer, 2004]. The seabed becomes progressively more carbonate rich with sands and local hard grounds seaward of the bottomset zone at depths >75–100 m [Brunskill et al., 1995; Harris et al., 1996]. Only a small fraction of river sediment input, <5%, is believed to be exported off shelf into the Pandora Trough [Brunskill et al., 2003; Walsh and Nittrouer, 2003; Muhammad et al., 2008].

[7] The river mouths and proximal delta regions are characterized by migrating sandy estuarine channels, mud banks, and highly productive mangrove forests [Harris et al., 1993, 1996; Alongi et al., 1992; Robertson et al., 1998; Walsh and Nittrouer, 2004]. Before entering these coastal areas, riverbed and suspended matter particulate organic carbon (POC) typically ranges from ∼900 to 1200 μmol Corg g−1, solid C/N ∼ 13 to 15 (mol mol−1), and dissolved organic carbon (DOC) ∼250 μM. [Bird et al., 1994, 1995; Salomons and Eagle, 1990; Robertson et al., 1998; Goñi et al., 2006]. Net addition of mangrove forest detritus and planktonic production within the estuarine and delta plain zones locally augments suspended particulate Corg to levels exceeding 8000 μmol C g−1, elevates C/N to ∼20–45, and depletes DOC by ∼30% [Robertson et al., 1998; Goñi et al., 2006]. In contrast, sedimentary Corg accumulating in seaward deposits on the topset and foreset ranges between ∼200 and 1200 μmol Corg g−1 with typical C/N ∼ 7–15 [Brunskill et al., 1995, 1996; Bird et al., 1995; Goñi et al., 2006]. The lowest Corg and highest C/N occur off the southern channel of the Fly, associated with relatively coarse deposits. Low Corg concentrations are also found in carbonate deposits and hard grounds offshore. Bulk Corg isotope distributions in surface sediment show regular inshore–offshore changes in the central gulf from predominantly terrestrially derived Corg (δ13C = −26.5 ± 0.5‰) over much of the inner topset to dominantly marine planktonic sources in the deeper foreset and bottomset (δ13C = −20.5 ± 0.5‰) [Bird et al., 1995; Aller and Blair, 2004; Goñi et al., 2006]. The average 14C age of bulk sedimentary Corg in the upper 0–3 m on the topset and foreset ranges between ∼5750 years in the western gulf and as young as 300 years in surface sediment in the east, however, most ages are >2000 years, reflecting the input of either recycled sedimentary rock Corg (kerogen) or aged soil Corg [Aller and Blair, 2004; Goñi et al., 2006].

[8] Balances between tidal currents, estuarine flow, seasonal wind wave forcing, a large-scale clockwise gyre, and local geomorphologic features determine sediment accumulation patterns and transport dynamics in the gulf. Tidal range decreases from macrotidal in the west, reaching up to ∼5 m in the mouth of the Fly, to mesotidal off the Purari [Wolanski and Eagle, 1991; Thom and Wright, 1983]. This regular variation in range is reflected in the progressive dominance of funnel-shaped river mouth morphologies in the west [Dalrymple et al., 2003]. Tide- and wave-generated fluid muds and resuspended sediment move in and out of the coastal channels, extensively refluxing and exchanging sediment with inner shelf deposits, and complicating estimates of net sediment flux [Wolanski et al., 1995; Harris et al., 2004]. Wind wave forcing and associated disturbance of the seabed change seasonally. During the NW monsoon period (January to March) winds average 2–5 m s−1 with typical wave heights of ∼0.3 m, whereas during the SE trades period (May to October) winds average 5–8 m s−1, with maximum sustained wind typically 10–12 m s−1 (gusts >20), and average wave heights ∼1.3 m (with individual sets >3 m on the topset, based on our direct observations) [McAlpine et al., 1983; Thom and Wright, 1983; http://www.ncdc.noaa.gov]. Resuspension on the topset and the generation of fluid muds during the SE trades, and subsequent pulsed export to the foreset are believed to be a primary means of clinoform progradation [Walsh et al., 2004].

[9] The shelf water column is warm (∼28 ± 2°C), well oxygenated, and bottom sediments diagenetically reactive, with diffusive uptake of O2 by bottom deposits averaging ∼23 ± 15 mmol m−2 d−1 [Mitchell, 1982; Alongi et al., 1992; Alongi, 1995; Aller et al., 2004]. Benthic biological and diagenetic properties are closely tied to sediment dynamics and depositional facies. Despite high remineralization and net sediment accumulation rates, topset and upper foreset deposits are typically suboxic, dominated by Fe, Mn reduction, and nonsulfidic over 0.1–1 m depth intervals [Alongi, 1995; Aller et al., 2004]. Because of physical reworking, large areas of the topset region (<20 m depth) are characterized by relatively depauperate macrobenthic communities and are dominated by microbial biomass [Alongi and Robertson, 1995; Aller and Aller, 2004; Aller et al., 2008]. Macrobenthic activity increases substantially at depths >30 m and at locally protected sites inshore, as expressed by both increasing biomass and the progressive occurrence of biogenic sedimentary structures.

3. Sampling

[10] Five sampling campaigns spanning the NW monsoon and SE trades periods were carried out from February 2003 to May 2004. The R/V Cape Ferguson (Australian Institute of Marine Science) was used for both shallow water (<15 m) and offshore sampling in February 2003 (monsoon) and November 2003 (end of trades). The R/V Melville (Scripps Institution of Oceanography) was used during August–September 2003 (trades), January 2004 (monsoon), and May 2004 (monsoon → trades transition) at sites deeper than ∼15 m. A primary inshore–offshore transect, GH, was established in the central gulf and augmented by additional sites on the foreset (GS48), topset, mangrove and estuarine channels, and the incised Umuda Valley off the northern entrance of the Fly (Table 1 and Figure 1) (see Aller et al. [2008] for a more complete location listing). The GH transect and station GS48 were close to or overlapped with sites sampled during previous studies in 1997–2000 on the R/V Franklin and R/V Harry Messel [Walsh et al., 2004; Aller et al., 2004; Aller and Blair, 2004], and were designed to seasonally sample the shallow topset and deeper foreset facies of the clinoform. The number accompanying station designations indicates approximate water depth (e.g., GH50 = 50 m depth).

Table 1. Core Station Locationsa
StationCruiseDateLatitude °SLongitude °EDepth, mSite
GH1CF0301 monsoon19 Feb 20037.483144.5061–3Wame River, Aird-Purari delta
GH8CF0302 trades9 Nov 20038.001144.2298inner topset
GH14CF0301 monsoon12 Feb 20038.079144.33614–15topset
GH14CF030211 Nov 2003    
GH14MV0104 monsoon18 Jan 2004    
GH14MV0404 transition12 May 2004    
GH25CF030114 Feb 20038.147144.477 topset
GH25MV040418 Sept 2003  25 
GH35CF030120 Feb 20038.181144.50535outer topset-foreset
GH35MV040410 May 2004    
GH50CF030116 Feb 20038.213144.54150foreset
GH50MV0803 trades21,25 Sep 2003    
GH75CF030122 Feb 20038.255144.58575bottomset
GS48CF030124 Feb 20038.037144.79448foreset
GS48MV010414 Jan 2004    
Pai'a10CF03027 Nov 20037.567144.53510Pai'a Inlet, Aird-Purari delta
Bamu2CF030213 Nov 20037.998143.6052Bamu River channel
Bamu10CF030214 Nov 20037.997143.6318Bamu River channel
Purutu5CF030216 Nov 20038.381143.5733Purutu Island channel
H5CF030210 Nov 20038.013144.0695inner topset
H14CF030218 Nov 20038.180144.30014topset
HI5CF030217 Nov 20038.302143.9216topset
I5CF030215 Nov 20038.346143.8806inner topset
T8-18MV080323 Sep 20038.6153143.975518Umuda Valley
T8-18MV010421 Jan 2004    
T8-18MV04048 May 2004    

3.1. Overlying Water

[11] Conductivity-temperature-depth (CTD)/dissolved O2 casts were made at each station using either a Seabird SBE25 (R/V Cape Ferguson) or Seabird 911plus/SBE43 O2 sensor (R/V Melville). Surface, middepth, and near-bottom water samples were obtained on the upcast using 10 L Niskin bottles, individually or in a rosette. In some cases, a Niskin was mounted at the base of a sediment multicorer to obtain bottom water within <30 cm of the seabed (R/V Melville). Winkler titrations on selected unfiltered samples were used to check the calibration of O2 sensors. Water was filtered through 0.2 μm pore size Whatman Puradisc25 AS polyethersulfone inline filters for nutrient and δ13C-ΣCO2 (dissolved inorganic carbon) analyses. Samples for Δ14C-ΣCO2 analyses were unfiltered. Samples were transferred under a stream of N2 into 2 mL or 10 mL glass ampoules (previously roasted 6 h at 450°C), flame sealed, and frozen (−20°C) for later isotopic analyses of ΣCO2.

3.2. Seabed

[12] The seabed was sampled using a combination of gravity, kasten, multicorer, and piston corers. A wide-diameter gravity corer (15 cm ID, cellulose acetate butyrate (CAB) tubing) (R/V Cape Ferguson) or 8X multicorer (R/V Melville) were used to obtain the upper 0.5 m of the bottom for high-resolution sampling (1 to 5 cm depth intervals) and incubations of undisturbed material near the sediment-water interface. The gravity corer, configured with 2-m-long CAB barrels, or kasten corer (stainless steel barrels, 15 × 15 cm × 3 m and 12 × 12 × 3 m) were used to sample the upper 0–3 m at 10 cm intervals [Kuehl et al., 1985; Brunskill et al., 2002]. The piston corer (6.5 cm ID) sampled lengths up to 8 m, and cores were selectively subsectioned over 10 cm intervals.

3.2.1. Pore Water

[13] Sediment cores were subsampled without occluded air using 60 mL cutoff plastic syringes as minipiston corers. Detailed handling procedures for each core type followed Aller et al. [2004]. Subcores were extruded under N2 into acid-washed (1N HCl) and distilled-water-rinsed 50, 125, or 250 mL centrifuge tubes or bottles, capped, and centrifuged at 5000 rpm (∼4100 G) for 10–15 min using a gimbaled Sorval SS3 centrifuge and GSA rotor. Pore water was removed under N2 into rubber-free plastic syringes through a short section of Tygon tubing and filtered through 0.2 μm pore size Puradisc25 AS inline filters directly into a second rubber-free plastic syringe without exposure to air. The filtered pore water was divided and stored in plastic vials as acidified (to ∼ 0.1 N HCl), unacidified (refrigerated), or frozen subsamples for a range of analyses, including ΣCO2 analyses onboard ship. Filtered pore water for δ13C-ΣCO2 determination was transferred under N2 into 2 mL glass ampoules, sealed, and stored frozen as described for overlying water. For selected depth intervals, a portion of centrifuged but unfiltered pore water was transferred into 10 mL glass ampoules, sealed and stored frozen for combined δ13C, Δ14C-ΣCO2 determinations. Great care was taken to avoid possible contamination with modern C sources at every handling step.

[14] When sediment gas was evident, piston core samples for CH4 analysis were subsampled with 60 mL cutoff plastic syringes, and the wet sediment sections extruded under N2 directly into 4 oz glass jars. Twenty milliliters of N2 degassed distilled water was added under a flow of N2, the jars sealed with metal lids, and the samples frozen for later CH4 concentration and δ13C–CH4 analyses.

3.2.2. Sediment Incubations

[15] Net remineralization rates of ΣCO2 in surface sediment (0–20 cm) were determined at 28°C using time series incubations of both whole cores and individual sediment sections. Two to three glass tubes (4 cm ID, 25 cm length) were used to vertically subcore the upper 0–20 cm of sediment at each site for measurement of ΣCO2 production [Aller et al., 1996]. The tubes were capped and placed in gas impermeable metalized plastic bags containing food-grade O2 scrubbers, heat sealed, and stored at ambient air temperature (∼28°C). These subcores were sectioned in 2 cm intervals and sampled for pore water and solids (as described subsequently) within 5–14 d, and then again after 20–50 d, depending on the build up rate of decomposition products. A diffusion–reaction model was utilized to calculate the optimal ΣCO2 production rate function versus depth from the time-dependent pore water profiles in the incubated cores, assuming a free solution diffusion coefficient for HCO3 of 1.02 cm2 d−1 [Aller et al., 1996; Boudreau, 1997]. In some cases, serial anoxic incubations of individual 10 cm intervals were sampled without additional depth resolution of pore water profiles, and the depth- integrated production rates of ΣCO2 within the 10 cm intervals determined from the slope of a ΣCO2 concentration versus time plot (2–16 weeks).

3.2.3. Sediment Solids

[16] Subsamples of sediment were stored in sealed 25 mL vials for water content and porosity determinations. Following the removal of pore water, residual sediment was retained in respective centrifuge bottles and frozen for later analyses.

[17] Cores for radiochemical analyses were sectioned at 2 cm intervals through the upper 20 cm and 4 cm intervals thereafter. Sediment samples were stored in double plastic bags in the dark before further processing for γ counting.

[18] Sedimentary structures at each site were documented using X radiography of core sections obtained by vertical insertion of acrylic tray subcorers (2.5 × 12 cm in cross section) into gravity and kasten cores.

4. Methods

4.1. Pore Water

[19] The initial ΣCO2 was measured on board ship and in all incubation samples using flow injection analysis/conductivity detection with a typical precision of ∼2% (standard deviation relative precision) [Hall and Aller, 1992]. Cl was determined with 1% precision on unacidified samples using a Radiometer CMT 10 titrator. Total dissolved S (SO42−) was analyzed in acidified pore water using Inductively Coupled Plasma Atomic Emission Spectroscopy (ICP-AES) (Varion Liberty 200), precision 3%, and a subset checked for equivalence to SO42− using ion chromatography (HCO3/CO32− eluent, Dionex AS4A column). Dissolved Fe and Mn were determined colorimetrically using ferrozine and formaldoxime, with precisions of ∼2–3% [Stookey, 1970; Goto et al., 1962] and using ICP-AES.

[20] Water samples for ΣCO2 isotopic analyses were treated as described by Aller and Blair [2006]. For some samples, 0.1–0.4 mL of pore water were injected into preflushed, sealed 3 mL Wheaton vials containing the H3PO4/CuSO4 mixture. The CO2 was stripped with helium, dried via passage through MgClO4 and Nafion™ tubing (Perma-Pure), and delivered to a Conflo III open split interface connected to a Thermo Delta V IRMS for δ13C measurements. Samples for 14C-ΣCO2 measurements (10 mL ampoules) were treated as described by Aller and Blair [2004]. Splits of these samples were also used for δ13C measurements. Graphite conversions and 14C analyses were made by the National Ocean Sciences Accelerator Mass Spectrometry (AMS) Facility at the Woods Hole Oceanographic Institution. The 14C contents are reported as the fraction modern relative to the National Bureau of Standards (NBS) Oxalic Acid I standard or as Δ14C [Olsson, 1970; Stuiver and Polach, 1977]. Modern is defined as 95% of the radiocarbon concentration (in A.D. 1950) of the NBS standard normalized to a δ13C of −19‰ [Olsson, 1970]. Corrections for natural fractionations were made by normalizing the δ13C values of the samples to −25 ‰. The relative precisions for the NBS-22 hydrocarbon standard were 12% for fraction modern and 2% for 14C age.

[21] Methane samples were homogenized and headspace samples removed using a can-piercing sampler. Methane was determined using a Shimadzu Mini2 GC equipped with a Valco 6-port loop injector, a 1/8″o.d. × 3′ long molecular sieve 5A column (100–120 mesh) maintained at room temperature and a flame ionization detector operated at 250°C. Gas samples were also processed through a combustion line (CuO at 780–790°C) [Chanton and Martens, 1988], and the resulting CO2 purified for isotopic measurements as described previously.

4.2. Sediment

[22] Weighed sediment samples for Corg isotopic analyses (particulate organic carbon) were treated with 4N HCl for 4 d at room temperature to remove carbonates, dried under vacuum, and reweighed. Subsamples were placed in tin boats and analyzed for Corg and N concentrations (mg g−1 dry wt sample) with a Carlo Erba 1108 CHNS analyzer. Precision was 2%. The CO2 produced via the oxidation of the Corg was trapped cryogenically for both 13C/12C and 14C/12C analyses as described previously for water samples. The most recent POC δ13C analyses were made using a CE 1112 EA interfaced to a Thermo Delta V IRMS. Corg, total C, and total N were measured on additional samples using a Perkin Elmer 2400 CHNS/O Series II Analyzer and a Shimadzu TOC-5000, the latter after sample acidification, precision typically 3–5%. Carbonate carbon was determined by difference between the total carbon before and after acidification.

[23] In some cases, organic carbon concentrations were normalized to the specific surface area of the sample to minimize variations due strictly to grain size and differential transport rather than net reaction [Mayer, 1994a, 1994b; Hedges and Keil, 1995]. Subsamples of the sediment were rinsed with deionized water to remove salts, dried and then roasted in air for 12–14 h at 350°C. After degassing at 150°C for 30 min., the surface area was determined by the multipoint method on a Beckman Coulter SA 3600 analyzer. Precision of the measurements was 2%.

[24] Highly reactive Fe minerals and Fe oxidation states were estimated by leaching freshly thawed, wet sediment for 15 min in 6N HCl at 22°C. Total highly reactive Fe and Fe(II) were measured immediately after leaching using ferrozine with and without hydroxylamine reductant [Aller and Blair, 1996; Viollier et al., 2000]. A separate sample was used to determine dry/wet ratios for conversion of concentrations to dry weight basis. Al was determined using sediment or suspended matter digestion with concentrated HNO3 and HClO4 for 3 h at 120°C and then refluxed 3 h at 180°C to eliminate HNO3. Standards were prepared in a comparable solution matrix.

[25] Measurements of 210Pb, 226Ra, and 137Cs were made at AIMS on 50–150 g of dried and ground sediment packed into gas-tight Perspex containers. Gamma counting was done using one well and four planar germanium detectors in lead-shielded castles. Estimation of 210Pb used the 46.5-keV gamma emission. After radon daughter ingrowth (3–4 weeks), 226Ra was determined from the gamma emission of 214Pb at 295 and 351 keV, and 214Bi at 609 keV. The 137Cs was estimated from the 661.6-keV gamma emission of 137mBa. Energy spectra were calibrated with Amershamand CANMET low-activity standards in cleaned silica sand of geometry and mass comparable to the sediment samples. International Atomic Energy Agency (IAEA) marine sediment reference material IAEA-315 was used to check calibrations. Total propagated counting errors were ∼5–10%, except for very low activity 137Cs samples where errors were ∼30%. Only 137Cs activities >1.5 times the total propagated error were considered detectable (>0).

[26] Sediment sections were X-radiographed in acrylic trays using a Kramex portable X-ray unit (20 mA 60 kV−1) and Fujifilm IX industrial X-ray film.

5. Results

5.1. Water Column

[27] CTD casts showed minor seasonal ranges of bottom water temperatures and salinities from 26.2 to 27.7°C and 35 to 35.2, respectively, at GH75, the most seaward bottomset site. Topset bottom temperatures varied more substantially from 26.8 to 29.5°C, with salinities ranging from 16 over the inner (5 m) to 35 on the outer topset (25 m). As exemplified by GH14, topset bottom waters tended to be warmer during the NW monsoon (29.5°C) relative to trades (27.2°C), and, at some sites, more saline [Aller et al., 2008]. Bottom waters were well oxygenated throughout the gulf during all seasons, with topset and foreset concentrations 160–180 μM (∼75 to >80% saturation), in general agreement with previous observations (Figure 2) [Mitchell, 1982; Aller et al., 2004]. Oxygen in surface waters ranged from 180 to 200 μM. The lowest O2 concentrations and coldest temperatures were found in the bottom water in channels of the Wame River in the Aird (Purari) delta (100–120 μM; T = 23°C). The lowest O2 layers within the water column on the foreset (40–50 m) correlated with decreased light transmission (suspended matter). Detailed examples of water column property profiles at or near the coring sites are given by McKinnon et al. [2007], Ogston et al. [2008], and Aller et al. [2008].

Figure 2.

Overlying water column was generally well oxygenated at all times of year throughout the clinoform system (>75% saturation), as illustrated by bottom water O2 concentrations as a function of (left) salinity or (right) bathymetric depth. The dashed (monsoon) and dotted (trades) lines connect the range of O2 saturation concentrations at the temperatures and salinities of the sampling sites (generally lower temperatures during trades). Bottom water in the Wame River mangrove-lined, distributary channels (Aird-Purari delta region) showed the greatest depletion (<50% saturation).

5.2. Sediment Properties and Diagenetic Environment

5.2.1. Sedimentary Structures

[28] Previous studies demonstrated a general dominance of physically formed sedimentary structures and lack of well-developed macrobenthic communities within tidal channels and across much of the topset region in the central gulf between depths of ∼5 and 20 m [Alongi et al., 1992; Alongi and Robertson, 1995; Walsh et al., 2004; Aller and Aller, 2004; Dalrymple et al., 2003; Goñi et al., 2008]. Interbedded sands and muds characterize the inshore (<10 m), with sand deposits and shoals common off the river mouths and within tidal channels. Sandy layers become progressively subordinate to silt and clay offshore (5–20 m), but increase again within the outer topset (25–35 m) [Walsh et al., 2004]. Watery, tidally mobile muds often overlie the sandy channel bottoms, or are present as a transient drape over sandy intertidal flats. The available coring gear did not permit sampling of relatively sandy bottoms, and thus seabed sampling is biased to muddy deposits. Truncated biogenic structures and tubiculous benthic species are occasionally present in the mud deposits of the topset; evidence of episodic but temporary colonization of an otherwise physically reworked seabed by macrobenthos [Alongi et al., 1992; Aller and Aller, 2004]. Biogenic structures, and the macrobenthos that form them, become progressively more common at depths >20 m on the outer topset, and dominate sedimentary fabric in the foreset and bottomset [Walsh et al., 2004; Aller et al., 2008]. These overall patterns were evident at the range of inshore–offshore sites sampled in the present study. Representative examples from areas or sampling times that were not included in earlier reports are shown in Figure 3. One feature not emphasized in previous reports is the common occurrence of methane gas-bubble structures below ∼40–100 cm depth in the tidal channels and topset deposits (Figure 3). These gas-rich zones were usually found in relatively consolidated, firm muds unconformably underlying a more watery and visually oxidized surface layer (Fe oxide rich) between 10 and 40 cm thick.

Figure 3.

Representative X-radiographs of sedimentary structures at inshore distributary channels (Purutu5, Fly River delta plain, 0–40 cm; Pai'a10, Pai'a inlet, Purari delta plain, 40–80 cm), inner topset (GH8, central gulf, 0–40 cm), outer topset (GH25, central gulf, 30–60 cm) and foreset (GH50, central gulf, 60–90 cm) sites illustrate the typical regular pattern of domination by physically formed structures in channel and topset deposits inshore (interbedded muds, sands), and the increasing dominance of macrobenthic biogenic structures offshore. (X-ray negative images, lightly colored regions contain sediment with relatively high bulk densities [see also Alongi et al., 1992; Walsh et al., 2004; Dalrymple et al., 2003; Aller and Aller, 2004].) The dark flecks visible below ∼50 cm in the Pai'a10 X-radiograph are formed by methane bubbles and are commonly observed below the physically reworked layer in the channel bed and topset deposits.

5.2.2. Radiochemical Distributions

[29] The distributions of 210Pb and 137Cs distributions varied substantially between the sampling regions, indicative of the variety of sedimentation regimes within the subaqueous area. As illustrated by station GH50, the foreset and bottomset facies were characterized by exponentially decreasing excess 120Pb (210Pbxs) and smoothly varying 137Cs activities (Figure 4), implying a relatively steady supply and accumulation of debris. Sedimentary structures and detailed examination of the activity distributions indicate that on the central gulf foreset, deposition likely takes place episodically in layers 5–10 cm thick, but sufficiently regularly so as to produce an overall time-averaged steady accumulation (GH50, 1.7 cm a−1; Figure 4), in the range expected based on general spatial patterns [Brunskill et al., 2003, 2007b; Walsh et al., 2004]. Net accumulation decreases substantially on the bottomset (GH75, ≤0.56 cm a−1; data not shown), and activity distributions are strongly influenced by bioturbation, making estimates of accumulation maxima if only steady advective transport is assumed to affect the 210Pbxs distribution. In contrast to the foreset and bottomset, sites on the inner topset region (<20 m) typically have a seasonally variable surficial layer of sediment ∼10–40 cm in thickness with relatively uniform or sometimes irregular 210Pbxs and 137Cs. The layer unconformably overlies firm sediment lacking analytically significant activities (e.g., GH14, Figure 4); GH8, H5, HI5 (data not shown) (see also examples from Brunskill et al. [2003] and Walsh et al. [2004]). This surficial layer is physically reworked sufficiently often that no steady 210Pbxs or 137Cs activity gradients are formed relative to analytical error, compromising estimates of net accumulation, if it occurs.

Figure 4.

Example vertical profiles of excess 210Pb (210Pbxs) and 137Cs activities illustrate the widely ranging, but regularly varying, depositional conditions found across the delta facies inshore approaching offshore. Purutu5 (mangrove distributary channel, Purutu Island, Fly delta) has low activities of 210Pbxs characteristic of river suspended matter and shows evidence of episodic deposition. GH1 (Wame River channel, Aird-Purari delta) has high activities of 210Pbxs characteristic of open shelf deposits and shows extremely rapid accretion of a channel bar (advection model fit ∼12 cm a−1). GH14 (midtopset, central gulf) has a physically reworked, seasonally variable surface layer, ∼35 cm thick at the time of 2003 trades sampling (solid symbols) and ∼10 cm thick during the 2003 monsoon (solid symbols). The mobile layer unconformably overlies firmer, methanic deposits that lack 210Pbxs and detectable 137Cs (indicated by horizontal line). GH50 (foreset, central gulf) shows approximately exponential decrease of 210Pbxs with depth, implying steady accumulation at an average rate of ∼1.7 cm a−1, although as suggested by adjacent sampling intervals of near constant activity and X-radiographs, sediment appears to accumulate in episodic pulses of 5–10 cm [see also Walsh et al., 2004; Aller and Aller, 2004].

[30] The inshore tidal channels and channel bars within the Fly and Aird-Purari delta plains show evidence of episodic, sometimes extremely rapid, deposition (Purutu5, GH1, Figure 4) [see also Walsh and Nittrouer, 2004]. Interpreted as steady accumulation, the 210Pbxs distribution at GH1 indicates accretion rates of ∼12 cm a−1; however, this migrating channel bar deposit is clearly transitory. Of significant importance is the fact that 210Pbxs activities at GH1 (Aird-Purari delta) are high and comparable to the offshore marine topset and foreset regions (30–40 Bq kg−1); whereas at Purutu5 (Fly delta plain), 210Pbxs activities are 4–5 times lower (≤10 Bq kg−1) and equivalent to activities in riverine suspended sediment [Brunskill et al., 2007a; Aalto et al., 2008].

5.2.3. Sediment C, C/N, and Al

[31] Average sediment Corg and C/N over the upper 1–2.5 m at the sampling sites varied between 0.6–2 mmol Corg g−1 and 9.3–16.2 mol mol−1, respectively, with most seabed sites 0.8–1.4 mmol g−1 and 9–13 mol mol−1 (Table 2). Carbonate C (Cinorg) ranged between 0.07 and 0.47 mmol g−1 (0.7–4.7% CaCO3 dry weight) with lowest values at the inner topset sites GH8 and H5, and the highest at the bottomset site GH75 (Table 2). Suspended matter collected at zero salinity in the Fly River and Wame River (Aird-Purari delta) ranged from 0.87 to 0.94 mmol Corg g−1 and 12.3 to 13.8 C/N; and suspended matter at the mouth of the Turama was 1.19 mmol g−1 with C/N = 13.2. These ranges agree well with other studies [Bird et al., 1995; Brunskill et al., 1995; Aller and Blair, 2004; Goñi et al., 2006, 2008]. The highest Corg (2 mmol g−1) and C/N (16.2) values were found at the tidal channel bar site GH1 in the Aird-Purari delta plain. Vertical profiles of sediment Corg and C/N demonstrated that, except at the inner topset and channel sites where the presence of distinct sand layers locally lower Corg, these values showed very little variation with depth at most stations (Figure 5) [see also Alongi et al., 1992; Aller and Blair, 2004; Goñi et al., 2008]. A regular decrease, however, of Corg from 0.9 to 0.7 mmol g−1 was observed with depth at foreset site GH50, suggestive of progressive diagenetic loss (Figure 5; implies remineralization ∼55 μmol C m−2 d−1). The lowest Corg was found at GH25, where relatively sand-rich deposits characteristic of the scoured outer topset occur [Walsh et al., 2004], and where Al content, which largely reflects fine-grained clay mineral content, was lowest (2 mmol g−1). Although Corg content is poorly correlated with Al within the relatively restricted station set examined here (Table 2), a general relationship obtained from 69 surface sediment samples taken throughout the gulf during February and November 2003 is Corg = 0.171*[Al]1.89 (mmol g−1) (data not shown, (r2 = 0.81; P < 0.001) includes all stations in Table 2).

Figure 5.

Vertical changes in Corg concentration and C/N ratios are minimal in the upper ∼2 m of sites examined in distributary channels and on the topset, with variations largely related to grain size differences between interbedded mud and sand layers [see also Goñi et al., 2008]. Small but regular decreases in Corg with depth on the foreset imply slow, progressive diagenetic loss (e.g., GH50 ↔ 55 μmol m−2 d−1 over upper 2 m, k = 0.019 a−1).

Table 2. Seabed and Suspended Matter Propertiesa
StationInterval, cmCorg, mmol g−1C/N, mol mol−1Cinorg, mmol g−1Al, mmol g−1Corgδ13C ‰ (PDB)Cl, mMSA, m2 g−1Corg/m2, mg m−2
  • a

    The δ13C and SA from 0 to 10 cm, all other properties, including pore water Cl, averaged over entire core depth interval unless indicated.

  • b

    From same or nearby station of Goñi et al. [2008].

  • c

    Mean of the five intervals listed in Table 3.

Fly Riversurface0.7712.20.212.30−26.870 salinity16.480.56
Wame Riversurface0.8810.70.082.84−25.370 salinity22.370.47
Turama River (mouth)bottom1.1913.60.052.41−27.00<2 salinity28.080.51
GH10–1602.01 ± 0.2216.2 ± 1.70.294 ± 0.1082.84 ± 0.15−27.58159 ± 1930.110.80
Purutu50–1470.813 ± 0.1212.7 ± 1.70.136 ± 0.0282.45 ± 0.22−26.77290 ± 3419.790.49
Pai'a100–1661.45 ± 0.4413.69 ± 1.780.241 ± 0.1172.72 ± 0.52−27.68184 ± 2436.360.48
H50–1401.122 ± 0.12911.5 ± 2.270.069 ± 0.0352.90 ± 0.40−24.6b289 ± 1322.5b0.60
HI-50–1110.819 ± 0.27812.3 ± 2.50.121 ± 0.0542.59 ± 0.37−26.5b351 ± 21  
GH80–1681.02 ± 0.2813.23 ± 2.10.079 ± 0.0592.85 ± 0.32−26.78 ± 0.15c360 ± 2224.3 ± 9.1c0.49 ± 0.05
GH14 (monsoon)0–1111.11 ± 0.3513.6 ± 2.30.111 ± 0.0612.80 ± 0.36−27.04487 ± 1129.730.45
GH14 (trades)0–1101.23 ± 0.2814.1 ± 2.60.078 ± 0.0413.06 ± 0.33−26.79467 ± 1230.390.49
GH250–1320.602 ± 0.12312.1 ± 2.020.429 ± 0.0502.06 ± 0.15−26.37542 ± 1524.990.29
GH35 (M)0–1680.883 ± 0.04911.44 ± 1.10.127 ± 0.0312.83 ± 0.12−25.94548 ± 632.090.33
GH500–1720.820 ± 0.0399.27 ± 0.600.150 ± 0.0263.16 ± 0.22−25.60 ± 0.13c555 ± 423.1 ± 1.3c0.42 ± 0.05
GS48 (HM50)0–2501.387 ± 0.31212.8 ± 1.700.189 ± 0.0342.93 ± 0.15−27.08c547.5 ± 531.5c0.53
GH750–1360.868 ± 0.0999.44 ± 0.530.469 ± 0.123.13 ± 0.12−23.70556 ± 431.60.33

[32] Bulk sediment surface areas (SA) varied between 10.4 m2 g−1 (GH8, 90–100 cm) and 33.4 m2 g−1 (GH8, 30–40 cm, surface mobile layer), the extremes occurring within different depth intervals of interbedded muds and sands at the same inner topset site GH8. The highest SAs were otherwise found inshore at GH1 (mangrove channel bar) and on the foreset-bottomset (GS48; GH75). Much of the inner and midtopset is characterized by bulk sediment SAs between 20 and 25 m2 g−1 [Goñi et al., 2008; Aller and Blair, 2004] (Table 2). SA (m2 g−1) correlates directly with Al content (mmol g−1), giving a geometric mean regression over the restricted ranges 10–34 m2 g−1 and Al = 2–3 mmol g−1, SA = 26.7[Al]–50.9; (r2 = 0.44; P = 0.014) or, assuming a nonlinear fit passing through the origin SA = 1.8[Al]2.47. The correlations between Corg, SA, and Al demonstrate the importance of the fine-grained fraction dominated by clay minerals in determining the distribution of Corg. The average Corg/SA ratio expected in surface sediments based on the Corg and SA correlations with mineral Al over the Al range 2–3.25 mmol g−1 is 0.64 mg Corg m−2 (i.e., ratio of Corg(Al) and SA(Al) averaged over Al concentration). Significant deviations from this predicted mean loading ratio occur, with elevated measured ratios at the mangrove channel site GH1 (0.8 mg Corg m−2) and lower measured ratios at the foreset and bottomset sites GH50 and GH75 (0.4 to 0.3 mg Corg m−2).

5.2.4. Corg Isotopic Distributions

[33] River surface suspended matter samples from zero-salinity end-members in the Fly River and Wame River (Aird-Purari delta) gave Corgδ13C values of −26.87 and −25.37 ‰, respectively (Table 2). The Fly River value, which was derived from a ∼1000 L sample, agrees well with those reported by Keil et al. [1997] for suspended matter (−26.77 ± 0.25 ‰) and for the river bed (−26.85 ± 0.1‰) by Bird et al. [1994, 1995]. Goñi et al. [2006] measured lighter values at two sites in the Fly apex region −29.8 and −29.3‰, possibly reflecting local input of mangrove detritus. The value measured here for the Wame River is comparable to those found in surface soil humic layers in the regional upland drainage basin (Purari), however, reported soil values were from samples that purposely discriminated against mineral material common (as indicated by Al) in river suspended samples [Bird et al., 1994]. The Turama River mouth sample, −27.00‰, was obtained at a salinity greater than zero.

[34] Seabed bulk Corgδ13C samples were analyzed from the 0–10 cm depth interval at a subset of the stations. Values varied from a low of −27.55‰ at GH1 (mangrove/Nypa palm fringed tidal channel), to a high of −23.70‰ at GH75 (bottomset) (Table 2). Inner and midtopset sites (5–20 m) range between −27.2‰ (GH14) and −26.8‰ (GH8), consistent with typical values of −26.5 ± 0.1‰ measured over much of the topset in the more extensive spatial studies of Bird et al. [1995] and Goñi et al. [2008] and the topset sites of Aller and Blair [2004]. The progressive increase in bulk Corgδ13C across the foreset (−25.6‰; GH50) and bottomset (−23.5‰; GH75), either without significant change or with a decrease in total Corg also agrees with the general depth-dependent patterns resolved previously [Bird et al., 1995; Brunskill et al., 1996].

[35] Down-core variation in Corgδ13C over depths of 2–6 m is minimal at representative inner topset site GH8 and foreset site GH50, in agreement with measurements at most, but not all, inner topset and foreset locations reported to date [Aller and Blair, 2004; Goñi et al., 2008]. In contrast, Corg Δ14C decreases substantially at both GH50, where there is an initial progressive change with depth and then an apparent stabilization of Δ14C around ∼−500‰ (5500–6350 years) between 4 and 6 m. At inner topset site GH8, Corg Δ14C has a stepwise distribution consistent with a two-layer diagenetic regime: a reworked surface layer 0–40 cm thick having mean Corg age ∼2500 years within which 210Pbxs is homogeneous (data not shown), discontinuously overlying an older, methanic zone characterized by Corg >4000 years old (Table 3).

Table 3. Inner Topset and Foreset 14,13C Vertical Profiles
Interval, cmCorg, mmol g−1Corgδ13C (PDB), ‰Corg Δ 14C, ‰Fraction ModernSA, m2 g−1Corg/m2, mg m−2
GH8
0–101.47−26.79−371.30.6287 ± 0.004931.150.57
30–401.28−26.90−476.60.5234 ± 0.003833.510.46
70–800.95−26.95−503.60.4964 ± 0.003924.030.47
90–1000.38−26.61−546.30.4537 ± 0.003210.350.44
150–1600.93−26.64−498.60.5014 ± 0.002922.600.49
 
GH50
50–600.99−25.69−269.70.7303 ± 0.00524.970.48
140–1500.85−25.51−273.70.7263 ± 0.003722.550.45
275–2850.70−25.55−399.20.6008 ± 0.003721.670.39
365–3750.67−25.63−557.20.4428 ± 0.003122.460.36
592–6020.84−26.26−436.40.5636 ± 0.004824.060.42

5.2.5. Pore Water SO42−

[36] Dissolved SO42− is initially available (see pore water Cl concentrations, Table 2) and depleted with depth at all sites, consistent with net diagenetic reduction (Figure 6). The depths of zero SO42− at topset sites usually occur within ∼0.5–1 m of the surface and are clearly associated with depositional and diagenetic discontinuities. These discontinuities are indicated by a sharp color change from light to dark, a jump in sediment firmness, visually evident erosional contacts, and obvious methanic compositions (gas bubbles) juxtaposed with the suboxic layer. The lack of substantial curvature in the SO42− profiles (i.e., linearity) within the nonbioturbated, upper few decimeters of sediment on the topset implies that much of the reduction takes place near the lower boundary of the SO42− gradient in association with anaerobic CH4 oxidation. In contrast to topset sites, the foreset is characterized by extended zones of SO42− reduction with nonzero terminal concentrations of SO42− over the depth scales sampled by piston cores (6–8 m) (Figure 6).

Figure 6.

Pore water SO42− profiles at topset sites, e.g., (left) GH8 and (middle) GH14, show approximately constant concentrations or linear decreases with depth, implying that net reduction of SO42− is focused toward the base of the physically reworked surface layer, and that anaerobic oxidation of CH4 occurs near the basal depositional unconformity. (right) As illustrated by GH50, net reduction of SO42− that takes place below the bioturbated zone in foreset deposits and SO42− decreases exponentially to a constant, nonzero value over the sampled interval (the fitted curve represents the one-dimensional diagenetic model fit discussed in text with k/ω = 0.0084 cm−1, k = 0.014 a−1). The two sets of symbols (open, solid) represent samples from different cores at the same sites (e.g., kasten and piston cores at GH14 and GH50; high-resolution and long gravity cores at GH8).

5.2.6. Sediment Fe Distributions and Seasonal Dynamics

[37] Pore water Fe and solid phase reactive Fe oxidation states reflect suboxic diagenetic redox conditions and sedimentary dynamics in the study area, selected examples of which are given here for the topset site GH14 and foreset site GH50 (Figure 7). The topset site GH14 shows evidence of substantial seasonal variation in diagenetic conditions in the upper 15–30 cm, with relatively high subsurface dissolved Fe (100–250 μM) maxima and a high proportion of reduced solid phase Fe (Fe(II)/FeR ∼ 0.6–0.8) at all depths during the monsoon periods, and relatively oxidized solid phase Fe (Fe(II)/FeR ∼ 0.2) and low dissolved Fe (<20 μM) in surface sediment during the late trades. The upper ∼30 cm of sediment was also visually oxidized and very watery in the trades cores, consistent with recent physical reworking, exposure to oxygenated overlying water, and reoxidation. In contrast, although the foreset site GH50 shows differences in Fe distributions between sampling times and slightly more oxidized conditions during the trades compared to the monsoon period, the seasonal variations are relatively subdued, with slightly lowered pore water Fe and solid phase oxidized Fe concentrations. The sediment was bioturbated at the GH50 foreset site, and no visual evidence of physical disturbance was obvious during seasonal sampling. Neither site has detectable dissolved sulfide. The monsoon distributions are comparable to those measured previously in the gulf [Aller et al., 2004; Alongi, 1995].

Figure 7.

Pore water Fe and solid phase reactive Fe oxidation states at middle topset site GH14 vary seasonally, consistent with suboxic diagenetic conditions in the physically reworked surface zone, seasonal oxidation of reworked sediment during the trades period, and the ingrowth of reduced Fe2+ during the more quiescent monsoon period. Suboxic, nonsulfidic redox conditions also characterize the bioturbated zone at foreset site GH50, and any seasonal changes, if they occur, are relatively subdued compared to the topset.

5.2.7. Remineralization Rates

[38] For the present purposes, the ΣCO2 net production rate distributions obtained from whole core incubations were integrated over the upper 0–20 cm and converted to equivalent areal ΣCO2 production fluxes using the measured sediment porosities (0.7–0.86). There were well defined spatial and seasonal differences in net ΣCO2 production fluxes (Figure 8). The highest values, ranging from ∼30 to 42 mmol m−2 d−1, were found during the monsoon periods at tidal channel and topset sites <35 m, and the lowest, ranging from 3 to 20 mmol m−2 d−1, were measured at topset sites <35 m during the late trades and transition periods. With some exceptions, the production rates generally decreased seaward into the foreset and bottomset sites, and showed no evidence of seasonal variation at foreset site GH50. The magnitudes of the ΣCO2 production fluxes agreed well with previous measurements of ΣCO2 production rates and with estimates of benthic O2 fluxes, most of which were made during monsoon periods [Aller et al., 2004; Alongi, 1991, 1995; Alongi et al., 1993].

Figure 8.

Benthic ΣCO2 production rates in the upper 0–20 cm ranged between ∼3 and 43 mmol m−2 d−1 and varied substantially seasonally, particularly on the topset and in the Umuda Valley (solid symbols). The highest remineralization rates were measured during the monsoon periods, as were the highest concentrations of sedimentary Chl a [Aller et al., 2008]. Benthic O2 fluxes estimated in previous studies (before 2003; open or gray symbols) in the Fly delta and central gulf clinoform are also plotted and used in averages for Corg budgets [Alongi, 1995; Alongi et al., 1992, 1993; Aller et al., 2004].

5.2.8. Pore Water ΣCO2 and CH4 Distributions and Isotopic Compositions

[39] Pore water concentrations of ΣCO2 increase substantially immediately below the sediment–water interface at all sites except GH75 (Figures 9–12), consistent with generally high remineralization rates and minimal bioturbation. The highest concentrations, reaching >40 mM in the upper 1 m, are found in methanic zones below the episodically reworked layer on the topset. The lowest concentrations, 2–5 mM, are found at GH75 in the highly bioturbated bottomset deposits. A net decrease of ΣCO2 concentration within the underlying methanic zone is evident only at Purutu5 in the Fly delta plain, where total concentrations are relatively low (∼10 mM). As shown previously, pore waters throughout the gulf are at saturation or are supersaturated with respect to a range of common carbonate minerals below a thin surface zone of undersaturation which is usually much less than 10 cm [Aller et al., 2004].

Figure 9.

(left) Pore water ΣCO2 concentration (solid circles) and ΣCO2δ13C (open squares) profiles are shown from Wame River channel bar site GH1. ΣCO2 Δ14C measured within discrete depth intervals are indicated as numbers down core. (right) Pore water ΣCO2 concentration (solid circles) and ΣCO2δ13C (open squares) profiles at Fly delta channel site Purutu5. ΣCO2 Δ14C indicates remineralization of Corg aged 1560–4800 years in a deposit containing 210Pbxs a few tens of meters from modern mangroves. In both cases, significant ΣCO2 concentrations and 13C enrichments are attained in the upper few centimeters. The δ13C increases substantially in the methanic zone.

Figure 10.

Pore water ΣCO2 concentration (solid circle) and ΣCO2δ13C (open squares) profiles at (left) GH8, (middle) H5, and (right) HI-5, represent an approximate along-isobath transect on the inner topset. ΣCO2 Δ14C measured within discrete depth intervals are indicated as numbers down core. CH4 was not evident in the sampled interval at HI-5. (Site H5 is the same as FF3 of Goñi et al. [2008].)

Figure 11.

Pore water ΣCO2 concentration (solid circles) and ΣCO2δ13C (open squares) profiles from the middle topset site GH14: (a) kasten core and (b) piston core. The ΣCO2 Δ14C measured within discrete depth intervals are indicated as numbers down core. Δ14C values in italics are from a gravity core sampled previously at the same site (HM13 [Aller and Blair, 2004]). (c) CH4 concentrations (solid diamonds) and CH4δ13C (open diamonds) in the same piston core as Figure 11b.

Figure 12.

Pore are shown from water ΣCO2 concentration (solid circle) and ΣCO2δ13C (open squares) profiles (a) kasten core from Umuda Valley site T8-18 (18 m; mobile layer inferred from 210Pbxs [Martin et al., 2008] and solute profiles), (b) piston core from foreset site GH50, and (c) gravity core at bottomset site GH75. The ΣCO2 Δ14C measured within discrete depth intervals are indicated as numbers down core.

[40] The δ13C of pore water ΣCO2 initially decreases with depth at all sites, reflecting metabolic inputs from remineralized Corg. ΣCO2δ13C reaches minimum basal values of −12‰ on the bottomset, −21 on the foreset, −25 to −28‰ on the topset, and −26‰ in the mangrove channels. ΣCO2 becomes isotopically heavier in the deeper zones (>40–100 cm) at the mangrove channel and topset sites due to methanogenesis. The isotopic composition of CH4 at topset site GH14 (δ13C = −81 to −90) is consistent with a biogenic source. CH4 was not evident within the upper 7 m at foreset site GH50, as expected based on the presence of relatively high, constant concentrations of pore water SO42− at depth (Figure 7). At GS48 (HM50), CH4 was present below 4.5 m at 0.10–0.83 mM with isotopic compositions −83 to −90 (not plotted).

[41] The Δ14C of pore water ΣCO2 in the upper few meters varied widely from +62 to −449.5‰. At most sites, ΣCO2 Δ14C in the upper ∼30–40 cm was >0, indicating the dominant decomposition of modern Corg substrates formed since atmospheric testing of nuclear weapons (∼1960). A significant exception is mangrove channel site Purutu5, where ΣCO2 Δ14C is −176.7‰ at ∼30 cm depth and reaches −449.5‰ at 1 m, indicating remineralization of aged Corg in migrating channel deposits. In contrast, at the mangrove channel bar site GH1, modern bomb signature 14C is released throughout the rapidly accumulating sediment pile (∼2 m). Modern ΣCO2 Δ14C also characterizes the mobile upper 1.5 m in the Umuda Valley, with substantial Δ14C decreases in deeper intervals. On the topset, there is a significant decrease of Δ14C in the methanic zone, reaching values <−150‰ in the upper 1–2 m at sites along the GH transect. A similar decrease occurs in the nonmethanic sediment of the foreset site GH50, and multiple other sites along the foreset (Figure 12) [Aller and Blair, 2004]. In the case of the foreset, however, depleted Δ14C ΣCO2 is measured in steadily accreting deposits within intervals containing 210Pbxs, whereas on the topset, the oldest Δ14C is measured beneath an erosional and diagenetic unconformity within zones lacking 210Pbxs (Figures 3 and 11).

5.3. Net Reaction and Mixing Models

[42] Mixing models are used here to derive information on net changes and reactions associated with compositional variations in both solids and pore water during transport and diagenesis in the delta. Assuming multicomponent mixing between samples having concentrations, Cj, of individual components with fixed property, Pj, then the respective mass and property compositional balances for the mixture are given by

equation image
equation image

where total concentration and its integrated property composition are CΣ and PΣ, respectively, (e.g., bulk Corg and δ13C). In these cases, the components are assumed normalized to a nonreactive constituent such as Al, or to total weight if nonconservative components are a relatively minor contribution. If progressive variation in concentration and property composition occurs from an initial mixture of reactants Cj(i):

equation image

In equation (3), ΔCΣ corresponds to the mass change in CΣ from an initial condition (CΣ(i) = equation image Cj(i)), and PΔ defines its net value. For the case where there are net changes in the quantity CΣ, (ΔCΣ ≠ 0), combining equations (1)–(3) gives

equation image
equation image
equation image

Equation (4) demonstrates that the slopes of linear portions of a plot of PΣCΣ versus CΣ are direct estimates of the net change in bulk property P during progressive addition or removal processes [Sayles and Curry, 1988; Martin et al., 2000; Blair et al., 2003; Aller and Blair, 2006; Leithold et al., 2006]. For example, in a simple two-component isotopic composition model with terrestrial and marine end-members having fixed isotopic composition, the slope of a plot for δ13C*Corg versus Corg reflects the net isotopic value of the Corg added or removed as terrestrial and marine contributions to bulk Corg vary. Alternatively, dividing both sides of equation (3) by CΣ results in a linear relation between PΣ and (1/CΣ) with intercept PΔ as (1/CΣ) → 0 (equation (5)) As shown by equations (4) and (5), it is not necessary to know the number of contributing components in order to derive a value for the net property change PΔ, whereas equations (2) and (6) provide a means of interpreting specific contributions.

6. Discussion

6.1. Sources and Inputs of Sedimentary Corg

[43] The sedimentary Corg initially entering or produced within the Gulf of Papua is derived from multiple sources having widely varying metabolic reactivities. Typical river suspended matter and riverbed Corg concentrations within the lower reaches range between ∼0.6 and 1.3 mmol g−1 (0.75–1.5 wt %; average 1.1%) with bulk δ13C of −26.8‰ (Fly) to −25.37‰ (Wame-Purari), the latter isotopic value is based on a single sample (Table 2) [Bird et al., 1995; Keil et al., 1997; Brunskill et al., 2007b]. The total, individual river weighted flux of particulate Corg to the gulf is estimated to be 152 × 109 mol a−1, with an additional DOC flux of 62 × 109 mol a−1 [Brunskill et al., 2007a]. Spread evenly over the clinoform area <50 m depth (21.7 × 109 m2), this particulate terrestrial Corg flux corresponds to an average input of ∼19 mmol C m−2 d−1. If this flux were partitioned into the inner to mid topset (0–20 m; 15.5 × 109 m2) and the outer topset-foreset (20–50 m; 6.2 × 109 m2) on a basis simply proportional to the percentage of net accumulation of sediment in each region (34 and 66%), then respective inputs of 9.1 and 44.3 mmol m−2 d−1 would be predicted. As shown subsequently, however, extensive modification of the river flux disproportionate to net accumulation patterns occurs during progressive transit and diagenetic processing on the energetic topset.

[44] The riverine Corg is a mixture of recycled rock kerogen, soil humus, vascular plant debris, and freshwater plankton [Blair et al., 2003; Goñi et al., 2008]. Given the mountainous, high-yield terrain (Fly and Purari, 1–3 Kt km−2 a−1), kerogen is expected to contribute a portion of the Corg, possibly ∼0.2–0.5 wt %, in the suspended sediment of this system [Komada et al., 2004; Blair et al., 2003; Leithold et al., 2006]. Aged soil humus derived from C3 vascular plants and modern C3 vascular plant debris dominates the remaining river particulate fractions [Bird et al., 1994; Goñi et al., 2006]. Suspended matter in five small rivers from eastern and northern Papua New Guinea averaged 3.6 ± 5.7% Corg (median 1.9%), Δ14C = −162 ± 87; and δ13C = −26.5 ± 3.4, consistent with old soil Corg sources in at least a subset of rivers draining interior highlands [Raymond, 1999]. The bulk and mineral fraction Corg in surface deposits (0–0.5 m) seaward of the Fly River have Δ14C between −515 and −644‰, perhaps reflecting the delivery of substantially aged, kerogen-rich debris, however, redistribution of older material within the delta or aging and reexposure in situ may also account for these values [Aller and Blair, 2004; Goñi et al., 2008].

[45] The initial suspended matter load is substantially augmented by inputs from mangrove and Nypa palm forests, and marine plankton production within the delta plain, resulting in concentrations for distributary channel surface suspended matter >10% Corg, and creating a dramatic spatial halo of Corg-enriched, isotopically light (δ13C = −29 to −30‰) suspended particulates around the Fly delta plain distributary channel region [Robertson and Alongi, 1995; Robertson et al., 1998; Goñi et al., 2006]. Virtually all of the vascular plant Corg introduced to the particulate load within this region is remineralized in the proximal delta water column and in surfacemost sediment, with little incorporated into the seabed or reaching the central gulf topset. The process of loss and addition of Corg within this inshore zone and its average isotopic composition can be readily discerned from the mixing models outlined previously (equations (1)–(6)). A plot of δ13C*Corg versus Corg for suspended matter in the Fly River distributary region and underlying surface seabed shows water column addition to the initial river suspended matter Corg, modest depletion of Corg in the surface seabed, and reveals that the dominant isotopic composition of the Corg added and lost in this region is −29.2‰ (Figure 13). This net isotopic value is indicative of C3 mangrove and Nypa palm detritus and of a minor role for marine plankton [Cifuentes et al., 1996; Bouillon et al., 2003]. A comparable model plot for pore water ΣCO2 in the upper 40 cm at the Purutu5 mangrove channel site also implies remineralization of substrate with a net isotopic value of −29. 1‰, directly demonstrating the reactivity and remineralization of isotopically light Corg in delta plain channel deposits (Figure 14). The isotopic value of bulk Corg over large areas of the surface seabed in the proximal Fly delta averages ∼−26.3‰ with Corg = 0.4–1.2% [Goñi et al., 2006], comparable to the initial Fly River suspended matter, and consistent with no substantial net addition and burial of mangrove forest detritus in offshore sediments. Thus the large introduction of reactive terrestrial C3 plant debris in the delta plain distributary channels and inshore mangrove forest regions is relatively transient and apparently has little net effect on the quantity of sedimentary Corg exported seaward to the deeper subaqueous delta topset and the central gulf.

Figure 13.

Addition and virtually immediate removal of sedimentary Corg in the Fly River delta plain distributary zone are demonstrated by a combined plot of Corg in initial Fly River suspended matter (0 salinity) (data from Keil et al. [1997] and Table 2 of this study), suspended matter in the surface and bottom water of the distributary channels, and the underlying surface seabed (0–1 cm) in the proximal Fly delta topset (data from Goñi et al. [2006]). The surface seabed generally shows modest depletion relative to suspended matter, which is substantially enriched within the delta plain. The net δ13C value of Corg added and removed in the process is −29.2‰ (geometric mean slope), consistent with a dominant source from mangrove and Nypa palm vascular plant detritus derived from forests fringing the distributary channels, and with the δ13C of respiratory ΣCO2 released from channel sediments in the same region (Figures 9 (right) and 14).

Figure 14.

Pore water reaction model plots illustrate the net δ13C values (geometric mean slope) of ΣCO2 added over the modeled intervals. On the basis of assumed end-members of ≤−26.8 and ≥−20‰ for terrestrial and marine sources, respectively, the estimated% contribution of terrestrial Corg substrate ranges from 100 to 0% across the clinoform facies. The inshore distributary channel sites GH1 and Bamu10 show evidence of significant import of marine Corg, as do inner topset sites GH8, H5, and HI 5. The only site with no apparent contribution of light terrestrial Corg substrates is bottomset site GH75 (−19.5‰). An average net addition of +23‰ relative to an initial ΣCO2 pore water value of −26‰ is shown in the upper region of the methanic zone at GH14. Note the nonlinear increase in the model slope (not fitted) as methanogenesis proceeds and light CO2 is progressively removed. (Standard error of slopes is 37% Purutu5 and 2.3 ± 1.9% otherwise.)

[46] Net marine primary production occurs throughout the distributary channel and open shelf system, with measurements averaging ∼4.3 ± 2.5 mmol C m−2 d−1 in the Fly delta distributary region, and 21 ± 6 to 52.2 ± 0.8 mmol C m−2 d−1 over the outer topset to bottomset region of the central gulf [Robertson et al., 1998; McKinnon et al., 2007]. The δ13C composition of this planktonic source has not been measured directly but is inferred to be −19.5 to −20.5‰, based on measurements in the Great Barrier Reef to the south and on the inshore–offshore spatial patterns of bulk sediment Corg in the central Gulf of Papua [Bird et al., 1995]. These net planktonic Corg production rates are either substantially less than or in the same range as the benthic remineralization fluxes measured seasonally across the gulf clinoform (Figure 8) [Aller et al., 2004], implying the potential to consume the entire autochthonous planktonic production in underlying bottom waters and seabed. The lack of any substantial net buildup of isotopically heavy Corg in bottom sediments over the topset is consistent with complete remineralization of marine planktonic sources in this region (Table 2) [Bird et al., 1995; Aller and Blair, 2004; Goñi et al., 2008].

6.2. Sources of Remineralized ΣCO2

[47] Mixing model evaluation of pore water ΣCO2 and isotopic distributions demonstrate that a broad range of Corg substrates from multiple sources are decomposed across the clinoform facies. The net isotopic values of decomposing substrates range from a minimum of −29.1‰ in the Purutu Island mangrove channel to −19.5‰ in the bottomset (GH75) (Figure 14). Anaerobic oxidation of CH4 and carbonate mineral precipitation may contribute to these calculated values in some cases, making them possible minimum estimates [e.g., Sivan et al., 2002]. However, anaerobic oxidation of CH4 is unlikely to be significant except in restricted intervals similar to those at topset site GH14 (Figures 6 and 11), and early diagenetic carbonate mineral precipitation is of minor importance in gulf sediments over the depth zones considered [Aller et al., 2004; T. Fang et al., manuscript in preparation, 2007]. The deeper methanic zones are characterized by a net input of heavy ΣCO2, reflecting removal of light CO2 during methanogenesis (e.g., at GH14, net addition of ∼+23 relative to −26‰ ΣCO2 pool; Figure 14).

[48] The fact that both the inshore Bamu and Wame River channel deposits, surrounded by mangroves and Nypa palm forests, release relatively heavy ΣCO2, −24 to −25.4‰, implies the availability of reactive marine planktonic substrates within the mangrove distributary channel facies. The importation of marine planktonic debris from the topset into the mangrove distributary channels is consistent with cross-shelf particle exchange during estuarine flow and with the unusually high 210Pbxs activities at GH1 (Figure 4). River suspended matter normally has 210Pbxs < 10–20 Bq kg−1, indicating import of 210Pbxs tagged particles from the shelf [Brunskill et al., 2007b; Aalto et al., 2008]. With exceptions such as GH14, many inner-outer topset sites are remineralizing Corg with a net isotopic composition between −22.9 and −24.1‰, increasing progressively to −22.0 to −23.0‰ on the foreset. The net isotopic value of remineralized ΣCO2 in the Umuda Valley is −25.1‰, implying a primarily terrestrial Corg metabolic source in these highly mobile sediments.

[49] Approximate estimates of the relative contributions of terrestrial and marine sources to diagenetic remineralization can be made assuming simple two-component mixing between terrestrial and marine end-members having average isotopic values of ≤−26.8 ± 0.5‰ and ≥−20 ± 0.5‰, respectively. The contributions of terrestrial organic matter to remineralization range from 100% inshore to 0% offshore, with most topset and foreset sites ∼40%; decreasing to ∼35% if a terrestrial end-member of −28‰ is assumed (Figure 14). These estimates demonstrate that marine Corg dominates early diagenetic remineralization at many sites but that except at the deepest bottomset sites, terrestrial Corg components are a significant, sometimes major, proportion of the decomposing substrate throughout the subaqueous clinoform (topset–Umuda Valley average = 61 ± 26% terrestrial). Assuming seasonally averaged benthic ΣCO2 production rates of ∼24 ± 10 mmol m−2 d−1 in the upper 20 cm of the topset (Figure 8; time weighting values 2/3 monsoon; 1/3 trades and transition), these percentages indicate a terrestrially sourced benthic Corg remineralization rate ∼15 ± 6 mmol m−2 d−1. The increased abundance of Chl a in surface sediment during the monsoons relative to the trades suggests that much of the seasonal excursions in remineralization rates are related to inputs of labile marine Corg [Aller et al., 2008].

6.3. Ages and Reactivity of Decomposing Sedimentary Corg

[50] The wide range of terrestrial and marine Corg substrates decomposed during early diagenesis in the upper few meters across the clinoform facies also have a broad spectrum of ages, as shown by ΣCO2 with Δ14C ranging from >0 (modern) in the surface 40 cm at most sites, to −229‰ (2030 years) at depth in the topset and foreset. The inshore mangrove channels show the widest variation in ΣCO2 ages with >modern in the Wame channel bar and ∼1600 to 4800 years in Purutu channel sediments (Figure 9), the latter presumably reflecting channel migration and undercutting of old deposits. The relationships between Δ14C and ΣCO2 are determined largely by local depositional environment, with rapidly deposited sediment often characterized by Δ14C > 0 across all ranges of ΣCO2, and steadily accumulating deposits on the foreset characterized by regular inverse relationships between Δ14C and ΣCO2 (Figure 15). Sediment Corg also directly shows evidence for net loss of old components having a mean Δ14C of −378‰ (3926 years), with much of the removal on the topset (Figure 15). These mean values presumably reflect mixtures of metabolic substrates with widely varying ages.

Figure 15.

(a) Pore water ΣCO2 has a wide range of Δ14C in the upper 1–5 m of the clinoform, from more than modern to <−230‰. The relation of ΣCO2 concentration and ΣCO2 Δ14C depends on the local depositional conditions, with reworked or rapidly deposited sediments such as GH1 (open diamonds) showing input of greater than modern 14C and little variation of Δ14C as ΣCO2 increases, while steadily accreting deposits such as GH50 (solid triangles) show substantial input of ancient ΣCO2 (e.g., intercept value > −250‰; model equation (5)). (b) As shown by the slopes of plots of Δ14C * Corg/SA versus Corg/SA for either the topset sites alone or topset-foreset sites combined, in addition to the obvious rapid respiratory loss of young Corg substrates into pore and overlying water. There is a slow net loss of ancient Corg (average age ∼4000 years) as sediment moves across the clinoform facies and the overall Corg loading decreases (see also Figure 19; data from Aller and Blair [2004] and Goñi et al. [2008]). (Standard errors of slopes are 13% topset and 16% topset-foreset.)

[51] Relatively old Corg is clearly also decomposing in the rapidly accumulating foreset facies, as demonstrated by a plot of Δ14C ΣCO2 versus ΣCO2 from 0.5–6 m depth at GH50, which indicates net release of ΣCO2 exceeding a 2500 year conventional 14C age (∼2100 years, Δ14C = −230‰, in the upper 0.5–4 m). As shown by the 210Pbxs distribution and sediment accumulation rate at this site (1.7 cm a−1), the remineralization of ancient Corg is occurring in sediment deposited within the last 100–200 years (Figure 16). Similarly old ΣCO2 is released in the upper 2–3 m throughout the rapidly deposited foreset region [Aller and Blair, 2004]. This fact shows that old Corg is remineralized in the SO42− reduction zone during steady accretion (Figure 6) and is consistent with the observed solid-phase losses (Figure 15). It further implies that remineralization of aged pools must occur at a slow rate continuously in other clinoform facies and during the process of sediment transport through oxygenated waters to the foreset [e.g., Keil et al., 2004; Ogston et al., 2008]. The remineralization of aged Corg (ΣCO2 ∼ 1600 years) in migrating mangrove channel deposits containing excess 210Pb (Purutu5, Figures 4 and 9), also supports the concept of continuous loss of refractory components during particle transport from source to sink.

Figure 16.

(a) Net Δ14C value of ΣCO2 added to pore water in the SO42− reduction zone below ∼0.5 m at foreset site GH50 is −274‰, implying virtually complete absence of young labile components in material delivered to the foreset (or less likely, mixing of much older ΣCO2 to a labile pool). (b) Net Δ14C of ΣCO2 in the surface mobile zone at GH14 is −39‰, essentially modern, whereas in the underlying relict methanic deposits the net value is −759‰. This latter extremely depleted value, which is smaller than any measured bulk Corg in the upper 2 m at topset sites, implies the remineralization of kerogen, carbonate precipitation, substantially aged sediment below ∼2 m, or the transport (diffusion upward) of 14C depleted ΣCO2 into this zone. (c) Surface ∼1.5 m of suboxic sediment in the Umuda Valley remineralizing modern Corg substrate (Δ14C = +20.3‰). Extrapolation of a single Δ14C value in the underlying deposit to a likely initial value in overlying water (open circle), implies a distinctly different, older source (−99‰) and a likely depositional unconformity.

[52] The reactivity of the decomposing Corg pool on the foreset can be estimated from the SO42− pore water distributions below the bioturbated zone by assuming steady accumulation, no compaction, and a single reactive Corg pool over the interval of interest [Berner, 1980]. The attenuation of the SO42− profiles with depth is determined in this case by the ratio k/ω, with k = reaction rate constant and ω = sediment accumulation rate. Excellent fits of this classic diagenetic model in zones below the bioturbated zone are obtained in piston cores at both GH50 and GS48 (SO42− data not shown), resulting in k/ω = 0.0084 and 0.0052 cm−1, respectively (Figure 6). Given the measured accumulation rates, k equals 0.014 a−1 and 0.019 a−1 at GH50 and GS48, which are 10 to 15 times lower than predicted based on nondeltaic correlations between k and ω [Toth and Lerman, 1977; Tromp et al., 1995]. These low reactivities demonstrate that the sedimentary Corg delivered to the foreset from the topset is relatively depleted in reactive components, as also shown previously by initial SO42− concentration gradient modeling at additional foreset sites [Aller et al., 2004]. The initial decomposition rates measured in the foreset region (ΣCO2 ∼ 20 mmol m−2 d−1) must therefore be sustained by a relatively small quantity of reactive, marine organic matter that is largely decomposed within the bioturbated zone (Figures 6, 8, and 12).

[53] Topset sites and the Umuda Valley are characterized by a reworked, suboxic surface layer of variable extent, generally ∼10–30 cm thick over much of the topset and up to ∼150 cm in the Umuda Valley (Figure 12; Fe2+ profiles (data not shown)) (210Pbxs [Martin et al., 2008]). As noted previously, the ΣCO2 being released in this mobile suboxic zone is generally modern (Δ14C > 0), although older material is also clearly degraded but is not a dominant source. For example, at least during some periods, the surface zone at GH14 is characterized by net input of ΣCO2 with Δ14C of −39 (∼330 years) (Figure 16). At H5 (equivalent to station FF3 of Goñi et al. [2008]), a ΣCO2 Δ14C = −3.6‰ at 20–30 cm implies net input of ΣCO2 with Δ14C ∼ −12 ‰, if an initial overlying water value of + 67‰ is assumed for the topset region [Aller and Blair, 2004]. Unlike the foreset, however, sites with a surface reworked zone show a large, discontinuous change in the age of ΣCO2 introduced below the reworked layer, consistent with a depositional and diagenetic unconformity between the suboxic layer and the underlying, often methanic, deposit. The net addition of ΣCO2 with extremely depleted Δ14C (−759‰) below 40 cm at GH14 implies that either ancient kerogen-like Corg is being remineralized locally or, more likely, that 14C-depleted ΣCO2 is diffusing upward from deeper deposits. Removal of CO2 by carbonate precipitation without isotopic fractionation would minimize the net Δ14C of added ΣCO2.

6.4. Diagenetic Fractionation

[54] Although a broad spectrum of sedimentary Corg is remineralized in the water column and seabed throughout the clinoform, at many sites there is preferential remineralization of marine relative to terrestrial Corg, and young relative to old Corg, consistent with observations in numerous soil and water column studies [Trumbore and Zheng, 1996; McCallister et al., 2004; Blair et al., 2003; Raymond and Bauer, 2001]. Comparison of δ13C in diagenetically released ΣCO2 to the associated bulk Corg demonstrates that, metabolic ΣCO2 released in the upper 0.5–1 m is often preferentially enriched in 13C by 1–4‰ (Figure 17a). This enrichment is consistent with the presence of a relatively small quantity of marine planktonic Corg sufficient to dominate early diagenetic remineralization and the initial production rates of ΣCO2, but which is not abundant enough to strongly influence bulk Corg isotopic compositions at most sites, nor to be preserved [Aller and Blair, 2004]. As shown by the earlier diagenetic model estimates of Corg reactivity in the foreset and by the bulk isotopic values of sedimentary Corg across the topset, this relatively heavy (isotopically), reactive pool is lost within either the physically or biologically reworked surface zones.

Figure 17.

(a) Comparison of the net δ13C value of released ΣCO2 (Figure 14) with δ13C of Corg at the same site demonstrates a typical diagenetic fractionation of 1–4‰, consistent with previous findings of Aller and Blair [2004]. Exceptions occur when abundant labile vascular plant detritus is present, such as at distributary channel site Purutu5 and, presumably, middle topset site GH14. (b) Curvilinear nature of a plot of 1/ΣCO2 in pore water versus its δ13C value on the foreset (see locations in Figure 1) demonstrates preferential release of marine relative to terrestrial-sourced ΣCO2, as ΣCO2 increases. For 1/ΣCO2 > 0.25, the intercept of the tangent is −21.4‰, whereas for 1/ΣCO2 <0.25 the intercept is −25.4‰ (see equation (5)) (data at sites other than GH50 from T. Fang et al. (manuscript in preparation, 2007)).

[55] A compilation of ΣCO2 concentrations and isotopic analyses obtained over the upper 3–6 m on the outer topset and foreset in this and earlier studies demonstrates that as a general rule diagenetic ΣCO2 becomes progressively lighter with burial depth. The initial compositional changes are characterized by an average net input of ΣCO2 with δ13C = −21.4 grading to an average of −25.4‰ (Figure 17b). These patterns imply early diagenetic loss of relatively reactive marine substrates and continued decomposition of residual refractory Corg, the latter dominated by old terrestrial sources. In the case of 14C, similar depth-dependent relative discrimination patterns occur with respect to substrate age: Typical enrichments of ΣCO2 Δ14C compared to the solid phase are ∼300‰ or greater (Figure 18) [Aller and Blair, 2004]. It is clear, however, that despite time-dependent preferential reaction, virtually every component of sedimentary Corg is eventually subject to decomposition.

Figure 18.

Relationships between both Δ14C and δ13C in the solid phase Corg and respiratory ΣCO2 in pore water demonstrate that diagenetic fractionation is common. In general, young substrates are utilized preferentially to older substrates and marine (heavy) relative to terrestrial (light). (Data were summarized from this study, Aller and Blair [2004], and Goñi et al. [2006].) The marine particulate organic matter and dissolved inorganic carbon fields are based on the work by Druffel et al. [2001, 2005].

6.5. Bulk 14Corg and Diagenetic Aging

[56] Diagenetic fractionation can result in the apparent aging of the bulk Corg independently of radioactive decay. Assuming steady initial activities of 14C, vertical profiles of Corg Δ14C on the foreset show a far more rapid change in 14C age with depth than predicted on the basis of 210Pbxs distributions at the same sites [Aller and Blair, 2004]. In the case of GH50, where a well resolved 14C profile is available for example, the 14C predicted accumulation rate is ∼0.12 cm a−1 compared to a 210Pbxs-determined rate of 1.7 cm a−1 (Figures 19 and 4). Time-dependent forms of the mass balance equations (1)–(2) can be used to estimate 14C activity profiles consistent with differential diagenetic remineralization of Corg pools of varying age over finite depth intervals. Assuming that over a restricted depth interval a single reactive Corg pool (dCorg(j)/dt = −kjCorg(j)) and a nonreactive pool (k = 0) are present, each with a Δ14Cj value changing with time only as a function of radioactive decay, then vertical profiles of total Corg(t) and total Δ14C(t) can be calculated as a function of both reactive pool remineralization and radioactive decay in all pools. Model profiles for the case when a reactive pool is lost with a time averaged Δ14C = −230‰; are shown in Figure 19, assuming an accumulation rate of 1.7 cm a−1, a decomposing Corg pool (0.32 mmol g−1) with first-order reaction rate constant k = 0.014 a−1 (obtained from SO42− profile model (Figure 6, right); also comparable to the rate constant estimated directly from the total Corg profile, k = 0.019 a−1 (Figure 5)), and a relatively unreactive Corg pool (0.67 mmol g−1) over the same interval with a terminal Δ14C = −546‰ at 4 m. In addition to being consistent with the phenomenon of diagenetic aging, these model profiles also illustrate the agreement between processes implied by both pore water (SO42−, ΣCO2 Δ14C) and solid phase data when steady accumulation occurs.

Figure 19.

(a) Vertical profile of Corg Δ14C (open triangles) and corresponding conventional 14C age (solid circles) distributions at GH50 are consistent with apparent aging during diagenesis. The initial activity gradient (0–4 m) implies an accumulation rate of 0.12 cm a−1, ∼15 times lower than the 210Pbxs profile over the same interval (Figure 4). The model curves represent the predicted profiles in the case where a reactive pool (Corg = 0.32 mmol g−1; k = 0.014 a−1) with a time-averaged Δ14C of −230‰ is lost from the solid during remineralization assuming an accumulation rate of 1.7 cm a−1. Both the reactive and nonreactive pools age by Δ14C ∼ −18‰ due to radioactive decay during the burial period, whereas the total Corg pool changes by ∼−115‰ due to diagenetic loss of the reactive portion. (b) Two-zone diagenetic regime typical of the topset expressed in the Corg Δ14C distribution at GH8 (inner topset; 210Pbxs homogeneous 0–40 cm). The stepwise change in ages (∼1500 years) at the zonal boundary is consistent with a depositional unconformity. The age of the material in the surface zone is comparable to surface sediment on the foreset, reflecting the role of topset as source.

[57] Although diagenetic aging must occur continuously during remineralization throughout the deltaic system, depositional conditions do not allow its expression in progressive vertical changes of Corg Δ14C except in cases of regular accumulation such as occur on the foreset. In contrast to the foreset, Corg Δ14C profiles on the topset reflect the two-layer diagenetic regime typical of that facies: a mobile surface, batch reactor layer unconformably overlying diagenetically distinct, stable deposits (Figure 19b). The conventional 14C age of bulk Corg in the suboxic surface zone at inner topset site GH8, ∼2500 years, averages ∼1500 years younger than in the more consolidated underlying methanic deposits (>4000 years). The stepwise transition in ages reflects the depositional discontinuity. The 14C age of bulk Corg in the topset mobile layer is comparable to that commonly found in the upper 0.5 m on the foreset, consistent with the initial processing of sediment on the topset and its progressive movement seaward.

6.6. Corg Incineration and Burial

[58] The spatial patterns of Corg loading and progressive changes in the bulk isotopic composition of Corg across the topset-bottomset facies (shift from ≤ −26.8 to ≥−23.5‰) demonstrate an overall net loss of terrestrial Corg within these central gulf regions relative to initial riverine suspended matter (Figure 20) [Bird et al., 1995; Aller and Blair, 2004; Goñi et al., 2008]. The net stable isotopic value of the lost sedimentary Corg averages −27.2‰, consistent with removal of terrestrially sourced Corg (Figure 20). As shown earlier, the large flux to the seabed of modern vascular plant debris in the delta plain distributary channels is transient and essentially eliminated within its area of input (Figure 13) [Robertson et al., 1998;Goñi et al., 2006]. The lack of major shifts in bulk sediment isotopic composition over large areas of the inner and middle topset of the central gulf (−27.2 to −26.4‰) (Table 3) [Bird et al., 1995; Goñi et al., 2008] also confirms that fresh marine plankton introduced to that region, which supports 25–60% of the early diagenetic ΣCO2 production, is largely remineralized, leaving little or no record in the seabed organic phase. Of course, the balances between governing processes, sources, and specific reactions of sedimentary material may vary in other subregions of the system such as southwest of the Fly or within the shelf valleys [Goñi et al., 2008; Martin et al., 2008].

Figure 20.

With the exception of the distributary channel delta plain regions where transient introduction of reactive vascular plant Corg occurs (Figure 13), deposits in the Gulf of Papua generally become progressively depleted in Corg relative to initial river suspended matter. The net isotopic value of the Corg removed is −27.2‰, consistent with a dominant net loss of terrestrial Corg without substantial replacement by marine Corg in the region examined. The various Δ14C and δ13C relationships between ΣCO2 and Corg demonstrate that Corg of virtually all ages and sources is remineralized at a wide range of rates, although relatively young Corg substrates dominate ΣCO2 production at any given time (Figures 15–18).

[59] The measured benthic remineralization rates, which are minima because only the upper 20-cm layer is considered and because incubation methodology minimizes rates, are nevertheless sufficiently high relative to the inputs of material to account for loss of a substantial proportion of the terrestrial sedimentary Corg delivered to the gulf. As shown earlier, remineralization rates on the inner midtopset (0–20 m) have an approximate weighted seasonal average of ∼24 ± 10 mmol m−2 d−1 (assuming 2/3 monsoon, 1/3 trades and transition seasonal weighting), of which ∼15 ± 6 mmol m−2 d−1 is derived from terrestrial sources (Figures 8 and 12; assuming simple average% terrestrial at stations <20 m). In comparison, if the river particulate Corg flux were initially focused entirely onto the inner–midtopset region (15.5 × 109 m2), then the input flux to that area is 26.9 mmol m−2 d−1. Thus, although it is not the primary clinoform depocenter, the inner–midtopset alone remineralizes ∼56% of the entire riverine particulate Corg flux on a steady basis as sediment transits and is refluxed within this region.

[60] Because of physical reworking, net accumulation rates of sediment on the inner midtopset are uncertain, whereas accumulation rates measured within the outer topset (20–30 m) and foreset (30–50 m) are far better constrained (e.g., Figure 4) [Brunskill et al., 2003; Walsh et al., 2004]. The outer topset and foreset store 24.4 × 109 and 19.1 × 109 mol Corg a−1, respectively, or 30.7 and 13.2 mmol Corg m−2 d−1 on an areal basis; assuming mean accumulation rates of 18.5 and 9.6 kg m−2 a−1 and areas of 2.18 and 3.97 × 109 m2 [Brunskill et al., 2007a, 2007b]. On the basis of δ13C distributions, virtually all of the Corg buried on the inner midtopset is of terrestrial origin, ∼88% is terrestrial on the outer topset and ∼69% on the foreset (assuming mean values of −26 and −24.7 ± 1.9 ‰ on the outer topset and foreset, respectively, and terrestrial-marine end-members −26.8 and −20 ‰ [Bird et al., 1995; Aller and Blair, 2004; Goñi et al., 2008]). By adding the measured burial fluxes from the outer topset and foreset together with the annual average ΣCO2 remineralization rate fluxes in the three facies, approximate Corg budgets can be derived for each region and a net burial of 3.5 mmol Corg m−2 d−1 in the inner topset necessary to balance the initial river supply to the system can be obtained by difference (Figure 21). This latter estimated burial flux compares reasonably well with an estimate of 7.3 mmol Corg m−2 d−1 on the inner midtopset made from apparent sediment accumulation rates [Brunskill et al., 2007a].

Figure 21.

Approximate diagenetic Corg budgets on the topset and foreset were estimated assuming progressive movement of sediment from the inner topset to the foreset regions and successive areal focusing of residual material (terrestrial component in italics). The burial rate of Corg on the topset was determined by the difference between the initial river supply to the gulf, the burial rates in the outer topset and foreset, and the estimated remineralization rates in each facies. The marine inputs to the seabed were estimated from the seasonally averaged remineralization rates (Figure 8) and the average percent terrestrial ΣCO2 released at the sampling sites in each region (Figure 14). The marine input to the seabed varies substantially seasonally, and the values used are conservative. These budgets, although approximate, demonstrate the central role of the topset in remineralizing Corg, which oxidizes >50% of the river input while storing ∼13–27%. The budget units are mmol Corg m−2 d−1. The areas of the inner midtopset, outer topset, and foreset regions are 15.5, 2.18, and 3.97 × 109 m2, respectively [Brunskill et al., 2003].

[61] These diagenetic Corg cycle budgets have large uncertainties and are subject to substantial seasonal excursions, particularly in the inputs to the seabed and remineralization of labile marine Corg and macrodetritus (Figure 8) [Aller et al., 2008]. Nevertheless, they provide a sense of how the system operates in terms of processing sedimentary Corg as it cascades progressively through the clinoform facies to the foreset and bottomset depocenters. The inner and midtopset store ∼13–27% of the initial river supply of terrestrial Corg, remineralize ∼56%, and accumulate ∼34% of the total sediment. In contrast, the outer topset and foreset (20–50 m) store ∼23% of the river Corg flux, remineralize ∼8.6% of the terrestrial Corg, and accumulate ∼66% of the total detrital debris. These relative percentages, the ΣCO2 production rates, and the isotopic release patterns on the topset demonstrate that the topset region plays a far more critical role in the processing and remineralization of terrestrial Corg than in storage (Figures 12, 15, 19, and 20). Net sediment accumulation is thus largely decoupled from remineralization patterns in the topset facies.

[62] The episodic refluxing of sedimentary debris and reexposure to oxygenated overlying water within and across the topset region promote suboxic diagenetic conditions in the reworked surface layer, exchange metabolites, and entrain reactive terrestrial and marine substrates (Figures 7 and 12) [Ogston et al., 2008; Aller et al., 2008]. These conditions optimize Corg decomposition as a result of oxygen exposure, the episodic resupply of Fe, Mn oxides as secondary oxidants, the removal of inhibitory metabolites, and the priming of refractory organic decomposition by the presence of labile substrates [Hedges et al., 1999; Keil et al., 2004; Aller, 1998; Aller and Blair, 2006]. The refluxing of particles vertically into the topset photic zone and horizontally across shelf onto periodically exposed tidal flats, must also enhance photochemical degradation processes [Mayer et al., 2006]. Consistent with this conceptual model, the sedimentary debris that eventually escapes the topset sedimentary incinerator is depleted in reactivity and in a substantial portion of its original terrestrial Corg load (Figures 6 and 20). The latter decreases from a range of ∼0.5–0.7 mg Corg m−2 initially supplied by the rivers to ∼0.19–0.3 mg Corg m−2 on the foreset and bottomset (calculated from the total Corg load of 0.36–0.42 mg m−2, Table 2, assuming −26.8 and −20‰ terrestrial and marine end-members).

[63] Therefore, despite receiving a high proportion of relatively old refractory material from a mountainous drainage area [Komada et al., 2004; Leithold et al., 2006], the diagenetic regime in the Gulf of Papua is effective at eliminating a large fraction of this input. Although not as efficient, as massive, or as energetic as the Amazon-Guianas region, which reduces initial terrestrial Corg loading to <0.12 mg C m−2 over transport scales of ∼600 km [Aller and Blair, 2006], the Gulf of Papua deltaic topset system operates as a functionally similar sedimentary incinerator. Residence time of sediment within this facies must be a primary determinant of remineralization and preservation of sedimentary Corg.

7. Conclusions

[64] Terrestrial and marine Corg having a wide spectrum of ages are remineralized throughout the deltaic complex, reflecting extensive cross-shelf particle exchange. As expected from boundary inputs, terrestrial Corg dominates as a primary substrate at many sites inshore, and marine substrates dominate offshore.

[65] In general, diagenetic fractionation is the rule with average respiratory ΣCO2 relatively young and heavy (e.g., labile marine) compared to bulk sedimentary Corg, although all components are remineralized and terrestrial Corg dominates when labile sources are locally available.

[66] The thin (∼10–40 cm), physically reworked layer on the topset is a critical component of the delta diagenetic system, acting as a suboxic incinerator and eliminating >50% of the terrestrial Corg in the initial river supply while storing only 13–27% as long-term accumulation. Sediment exported to the outer topset and foreset is depleted in reactive components.

[67] The net loss of sedimentary Corg on the delta, while spatially variable, is dominated by the oxidation of terrestrial organic matter, including aged components >4000 years old, and its remineralization continues throughout the burial depth examined (7 m).

[68] Sediment refluxing and associated diagenetic conditions within the shallow delta topset zone are capable of efficiently incinerating the aged, refractory Corg from the high-yield drainage basins characteristic of regions such as Oceania.

Acknowledgments

[69] Irena Zagorskis, John Pfitzner, Caterina Panzeca, Megan Dantzler, Vanessa Madrid, Lynn Abramson, Vasso (Bessy) Alexandratos, Christina Heilbrun, Steve Boyle, Miguel Goñi, Chuck Nittrouer, Josephine Aller, Catherine Thompson, Laurel Childress, and numerous additional participants and technical staff provided invaluable aid in the field and laboratory. The expertise and efforts of the Masters and crews of the R/V Harry Messel, R/V Franklin, and R/V Cape Ferguson from AIMS and R/V Melville, Scripps Institute of Oceanography were, of course, critical to the success of the project. We thank the reviewers for critical comments. NOSAMS analyses reflect facility support from NSF Cooperative Agreement OCE-9807266. Research support came from NSF Oceanography Program, grant OCE0219919.

Ancillary

Advertisement