DISCUSSION

Evidence for Anaerobic Methane Oxidation

Together with a strongly concave-up methane profile (Martens and Berner, 1974), several additional lines of evidence indicate that sulfate and methane co-consumption occurs at the sulfate-methane interface: (1) CO₂ and alkalinity profiles indicate localized CO₂ production there; (2) ¹³CCO₂ values are extremely depleted in ¹³C, and are lowest within the logical depth-zone for AMO; and (3) sulfate gradients are linear throughout the sulfate-reduction zone.

The magnitude and position of isotopic values of CO₂ provide the most compelling evidence for AMO (e.g., Reeburgh, 1980). The ¹³CCO₂ values more negative than -30 PDB most likely indicate derivation from methane because the isotopic carbon composition of marine phytoplankton is generally heavier than -30 (Deines, 1980). In addition, ¹³CCO₂ values reach their lightest (-37.7) at the SMI, where CO₂ production is indicated by deviation from linearity in the CO₂ and alkalinity profiles (Fig. 2). This local increase in CO₂ is consistent with production of ¹²C-rich CO₂ through sulfate and methane co-consumption, and the diffusion of light CO₂ away from the SMI. Near invariant alkalinity and CO₂ concentrations some meters below the SMI are consistent with a net sink for interstitial CO₂ in the uppermost methanogenic zone due to carbonate mineral authigenesis (Rodriguez et al., 1997, and Chap. 30, this volume).

Assuming steady state conditions, linear sulfate gradients within the sulfate-reduction zone are circumstantial evidence for focused consumption of sulfate at the sulfate-methane interface by AMO. Linear profiles imply sulfate diffusion through the sulfate-reduction zone and consumption at the interface where sulfate and methane co-occur (Borowski et al., 1996).

Microbial consumption of sulfate to form interstitial dissolved sulfide (HS^-_dissolved = S^2- + HS^- + H₂S) preferentially uses light (³²S) sulfur by both sulfate reduction of SOM (Eq. 1) and AMO (Eq. 2). These fractionations create a dissolved sulfide pool depleted in heavy sulfur (³⁴S; typically negative ³⁴S values), and create progressive enrichments of ³⁴S in the residual sulfate pool (³⁴S values more positive than +20) (e.g., Chambers and Trudinger, 1979; Goldhaber and Kaplan, 1974). Although the sulfur isotopic composition of interstitial sulfate indicates microbial fractionation at all sites (11-8, 994, 995; Fig. 2), either or both sulfate-depleting mechanisms could qualitatively account for the observed ¹⁴CSO₄ profiles.

Intensity of AMO

The intensity of anaerobic methane oxidation operating at the sulfate-methane interface can be gauged by appraising how much methane carbon resides in the CO₂ pool, or by estimating the proportion of sulfate that AMO consumes relative to total sulfate flux into the sediments. Thus, we assess the importance of AMO at Site 995 using two independent methods: (1) a simple mixing model that considers the sources of CO₂ carbon (Eq. 5); and (2) a diagenetic model that utilizes the interstitial methane concentration profile to estimate methane consumption at the SMI (Eq. 3).

Mixing Model

The CO₂ mixing model suggests that 24% of the carbon within the CO₂ pool has come from methane, but there are biases that affect this calculation. The largest potential bias is the diffusion of ¹²C-rich CO₂ away from the sulfate-methane interface as suggested by the ¹³CCO₂ profiles (Fig. 2). Such diffusive processes would act to lower the quantity of light carbon within the CO₂ pool and to lower the estimate of methane carbon.

Alternatively, carbonate mineral formation and the assumption of no carbon fractionation during AMO result in an overestimation of methane carbon within the CO₂ pool. Fractionation during AMO allows entry of more ¹²C into the CO₂ pool, but from a smaller proportion of methane. Thus, the model calculation overestimates the quantity of methane carbon in the CO₂ pool, however, an isotopic fractionation factor for AMO of 1.002-1.014 (Whiticar and Faber, 1986; Alperin et al., 1988) changes the calculation by only a few percent.

Carbonate mineral formation occurs within the zone of AMO (Rodriguez et al., Chap. 30, this volume), affecting CO₂ concentration. Large losses of CO₂ only slightly affect the isotopic mass balance by changing the proportion of CO₂ derived from seawater (X_sw), and therefore altering the mix of CO₂ carbon derived from organic matter (X_om) and methane (X_amo). Thus, the major effect of an open system on the model calculation lies not in the change of CO₂ concentration, but in its effect on the isotopic composition of the CO₂ pool. Accordingly, because the carbon fractionation involved in carbonate mineral formation is small (e.g., Emrich et al., 1970; Anderson and Arthur, 1983), the isotopic balance calculation is unaffected by this process.

The above factors interact to slightly vary the results of the mixing model, but the largest bias--diffusion of ¹²C-rich CO₂—acts to underestimate the contribution of methane carbon to the CO₂ pool. The mixing model is relatively insensitive to CO₂ concentration and the AMO fractionation factor, cumulatively changing the estimate of the fraction of methane carbon to between 19% and 24%. A comprehensive interpretation is that at least one-fifth to one-fourth of the carbon within the CO₂ pool is derived from methane through AMO.

Diagenetic Model

The diagenetic model based on methane concentrations does consider diffusive processes and is unaffected by fractionations inherent in microbially mediated reactions because it contains no isotopic inputs. Thus, it is potentially a better tool in assessing the intensity of AMO. Based on methane concentration values, the model predicts that 35% of the total sulfate flux is consumed by AMO, but this may be an underestimate.

The linear sulfate profile of PC 11-8 (Fig. 2C) was reproduced by a finite difference diagenetic model that consumed 95% of the calculated sulfate flux through AMO (Borowski et al., 1996; Borowski, 1998). Increasing contributions of sulfate depleted by oxidation of SOM (Eq. 1), cause increasing curvature in the model results and increasing deviation from the observed sulfate concentration values. A similar predominant proportion of sulfate depletion due to AMO is necessary to approximate the sulfate data at Site 995 (see fit to sulfate data in Fig. 2A). The conundrum is that the model results based on methane data predict a much smaller proportion of sulfate consumed by AMO than do model results based on sulfate concentrations, although each model is internally consistent.

To adjust model results based on methane data to predict higher amounts of sulfate depletion due to AMO, in situ methane concentrations would have to be significantly higher than measured values. Methane loss may occur through outgassing, diffusion, and consumption by aerobic or anaerobic methane oxidation. However, the coincidence of modeled peak methane oxidation rates with the most ¹³C-depleted ¹³CCO₂ values suggest that the measured methane concentrations faithfully reproduce at least relative changes in methane concentration.

To adjust sulfate data to reflect lower amounts of sulfate depletion due to AMO, in situ sulfate concentrations must be lower than measured concentrations, or sulfate must be renewed in situ. There are three known mechanisms that could add additional sulfate: (1) oxidation of interstitial dissolved sulfide during sediment and pore-water processing (e.g., Almgren and Hagstrom, 1974); (2) oxidation of sulfide minerals during processing; or (3) in situ anaerobic oxidation of dissolved sulfide by manganese and/or iron oxide minerals (Sorensen and Jorgensen, 1987; Aller and Rude, 1988; Fossing and Jorgensen, 1990; Elsgaard and Jorgensen, 1992; Schulz et al., 1994).

However, sulfide concentration and sulfur isotopic data show no obvious indication that these processes occur in Blake Ridge sediments:

a. Typical interstitial dissolved sulfide concentrations in Blake Ridge pore waters are usually <10 µM and almost always less than 120 µM (except at methane seep sites, Paull et al., 1995; Borowski, 1998). Thus, only a very small amount of oxidized sulfide could be added to the sulfate pool as an artifact of sampling.

b. Sulfide minerals in the sulfate reduction zone of Sites 994, 995, and 997 have a mean ³⁴S value of -38.8 ± 5.3 (N = 30, Borowski, 1998). Pore-water ¹⁴CSO₄ values consistently show ³⁴S enrichment at all sites (Fig. 2), rather than showing any addition of light sulfur from light sulfide derived from oxidized sulfide minerals. Thus, the data do not indicate that the oxidation of very ³⁴S-depleted sulfide minerals during core handling has occurred.

c. To significantly alter sulfate concentration gradients, sizeable amounts of dissolved sulfide, derived from sulfate consumed by sulfate reduction processes, must be oxidized in situ. Because of the argument in (b), if this process occurs it is masked by sulfur disproportionation reactions (Canfield and Thamdrup, 1994). Although there is ample evidence for anoxic sulfide oxidation elsewhere (see references above), we cannot identify its effects in Blake Ridge sediments. Moreover, we infer that anoxic sulfide oxidation is most likely to occur preferentially within the uppermost portion of anoxic sediments (where iron and manganese oxides are readily available; e.g., Aller and Rude, 1988), not deep within the sulfate-reduction zone.

To summarize, we have confidence in our data, but have not resolved the inconsistencies between the different model results. Either there is some undiscovered defect in the models or model assumptions, or there are unrecognized processes that we cannot adequately account for, nor quantify. Future modeling attempts may resolve the issue, especially by testing how modified diagenetic models predict the magnitude and distribution of ¹³CCO₂ and ¹³CSH₄ values. Even with the inconsistencies between model results using methane versus sulfate data, a conservative interpretation is that 35% of the total sulfate flux is consumed by AMO at Site 995.

Areal Variation

There is evidence for a still larger role for AMO, relative to the results at Site 995, at other sites of the CR-BR area. Fully 40% (17 of 42) of the nondiapir sites over the Carolina Rise and Blake Ridge have sulfate gradients steeper than those observed at Site 995 (Borowski, 1998). If sulfate consumption through AMO induces linearity in the sulfate profiles, we infer that these sites may have a larger proportion of sulfate depleted via AMO versus sulfate reduction of SOM. For example, Site PC 11-8 (Fig. 2C) shows the largest amount of sulfate depletion recorded at nondiapir sites with a gradient of 2.88 mM m^-1 (vs. 1.30 mM m^-1 at Site 995), a SMI depth of 10.3 mbsf (vs. 21 msbf at Site 995), and a predicted sulfate flux of 18 × 10^-4 mmol cm^-2 yr^-1, or 2.2 times that of Site 995 (8.2 × 10^-4 mmol cm^-2 yr^-1).

Implications

The indication that AMO is such a significant sink for interstitial sulfate in these deep-sea sediments is surprising. Large fractions of sulfate consumed by AMO have been documented elsewhere (Table 4), but only at seep or advective sites (not tabulated; e.g., Suess and Whiticar, 1989; Masuzawa, et al., 1992; Cragg et al., 1995; Paull et al., 1995) in organic-rich coastal marine environments (e.g., Saanich Inlet, see Table 4 for references), or in marine basins that have anoxic water columns such as the Black Sea (Reeburgh et al., 1991) and Cariaco Trench (Reeburgh, 1976). Unlike these anoxic basins, the overlying water of the CR-BR area is oxygenated, implying a different impetus for intensified AMO.

Borowski et al. (1996) suggested that the magnitude of upward methane flux controls the intensity of AMO occurring at the sulfate-methane interface, which in turn influences sulfate concentrations (and perhaps produces linear sulfate gradients). Pore-water data presented here, especially ¹³CCO₂ data, firmly establish AMO as an important diagenetic process at Site 995. Although independent diagenetic modeling results are inconsistent, or do not fully explain the data, a model using methane concentrations (Site 995) suggests that 35% of the total sulfate flux is consumed by AMO. Such large proportions of AMO must affect sulfate concentration gradients. If such an enlarged role for AMO is indeed fueled by upward methane flux, we infer that similar interstitial chemistries should also occur at other continental rise, methane-rich, gas hydrate-associated settings. Thus, steep (and linear) sulfate gradients may be a geochemical indicator of methane-rich sediments, and may also be an indicator for the presence of gas hydrates in continental rise sediments, given that conditions are suitable for gas hydrate formation.